Merge branch 'linux-2.6'

301 files changed, 15889 insertions, 7045 deletions

diff --git a/drivers/acpi/sleep/main.c b/drivers/acpi/sleep/main.c
index f8c63410bcbf..52b23471dd69 100644
--- a/drivers/acpi/sleep/main.c
+++ b/drivers/acpi/sleep/main.c
@@ -29,7 +29,6 @@ static u32 acpi_suspend_states[] = {
 	[PM_SUSPEND_ON] = ACPI_STATE_S0,
 	[PM_SUSPEND_STANDBY] = ACPI_STATE_S1,
 	[PM_SUSPEND_MEM] = ACPI_STATE_S3,
-	[PM_SUSPEND_DISK] = ACPI_STATE_S4,
 	[PM_SUSPEND_MAX] = ACPI_STATE_S5
 };
 
@@ -94,14 +93,6 @@ static int acpi_pm_enter(suspend_state_t pm_state)
 		do_suspend_lowlevel();
 		break;
 
-	case PM_SUSPEND_DISK:
-		if (acpi_pm_ops.pm_disk_mode == PM_DISK_PLATFORM)
-			status = acpi_enter_sleep_state(acpi_state);
-		break;
-	case PM_SUSPEND_MAX:
-		acpi_power_off();
-		break;
-
 	default:
 		return -EINVAL;
 	}
@@ -157,12 +148,13 @@ int acpi_suspend(u32 acpi_state)
 	suspend_state_t states[] = {
 		[1] = PM_SUSPEND_STANDBY,
 		[3] = PM_SUSPEND_MEM,
-		[4] = PM_SUSPEND_DISK,
 		[5] = PM_SUSPEND_MAX
 	};
 
 	if (acpi_state < 6 && states[acpi_state])
 		return pm_suspend(states[acpi_state]);
+	if (acpi_state == 4)
+		return hibernate();
 	return -EINVAL;
 }
 
@@ -189,6 +181,49 @@ static struct pm_ops acpi_pm_ops = {
 	.finish = acpi_pm_finish,
 };
 
+#ifdef CONFIG_SOFTWARE_SUSPEND
+static int acpi_hibernation_prepare(void)
+{
+	return acpi_sleep_prepare(ACPI_STATE_S4);
+}
+
+static int acpi_hibernation_enter(void)
+{
+	acpi_status status = AE_OK;
+	unsigned long flags = 0;
+
+	ACPI_FLUSH_CPU_CACHE();
+
+	local_irq_save(flags);
+	acpi_enable_wakeup_device(ACPI_STATE_S4);
+	/* This shouldn't return.  If it returns, we have a problem */
+	status = acpi_enter_sleep_state(ACPI_STATE_S4);
+	local_irq_restore(flags);
+
+	return ACPI_SUCCESS(status) ? 0 : -EFAULT;
+}
+
+static void acpi_hibernation_finish(void)
+{
+	acpi_leave_sleep_state(ACPI_STATE_S4);
+	acpi_disable_wakeup_device(ACPI_STATE_S4);
+
+	/* reset firmware waking vector */
+	acpi_set_firmware_waking_vector((acpi_physical_address) 0);
+
+	if (init_8259A_after_S1) {
+		printk("Broken toshiba laptop -> kicking interrupts\n");
+		init_8259A(0);
+	}
+}
+
+static struct hibernation_ops acpi_hibernation_ops = {
+	.prepare = acpi_hibernation_prepare,
+	.enter = acpi_hibernation_enter,
+	.finish = acpi_hibernation_finish,
+};
+#endif /* CONFIG_SOFTWARE_SUSPEND */
+
 /*
  * Toshiba fails to preserve interrupts over S1, reinitialization
  * of 8259 is needed after S1 resume.
@@ -227,14 +262,18 @@ int __init acpi_sleep_init(void)
 			sleep_states[i] = 1;
 			printk(" S%d", i);
 		}
-		if (i == ACPI_STATE_S4) {
-			if (sleep_states[i])
-				acpi_pm_ops.pm_disk_mode = PM_DISK_PLATFORM;
-		}
 	}
 	printk(")\n");
 
 	pm_set_ops(&acpi_pm_ops);
+
+#ifdef CONFIG_SOFTWARE_SUSPEND
+	if (sleep_states[ACPI_STATE_S4])
+		hibernation_set_ops(&acpi_hibernation_ops);
+#else
+	sleep_states[ACPI_STATE_S4] = 0;
+#endif
+
 	return 0;
 }
 
diff --git a/drivers/acpi/sleep/proc.c b/drivers/acpi/sleep/proc.c
index 5a76e5be61d5..76b45f0b8341 100644
--- a/drivers/acpi/sleep/proc.c
+++ b/drivers/acpi/sleep/proc.c
@@ -60,7 +60,7 @@ acpi_system_write_sleep(struct file *file,
 	state = simple_strtoul(str, NULL, 0);
 #ifdef CONFIG_SOFTWARE_SUSPEND
 	if (state == 4) {
-		error = pm_suspend(PM_SUSPEND_DISK);
+		error = hibernate();
 		goto Done;
 	}
 #endif
diff --git a/drivers/ata/Kconfig b/drivers/ata/Kconfig
index 45dbdc14915f..c7219663f2b9 100644
--- a/drivers/ata/Kconfig
+++ b/drivers/ata/Kconfig
@@ -435,7 +435,7 @@ config PATA_OPTIDMA
 	help
 	  This option enables DMA/PIO support for the later OPTi
 	  controllers found on some old motherboards and in some
-	  latops
+	  laptops.
 
 	  If unsure, say N.
 
diff --git a/drivers/ata/libata-acpi.c b/drivers/ata/libata-acpi.c
index 03a0acff6cfa..cb3eab6e379d 100644
--- a/drivers/ata/libata-acpi.c
+++ b/drivers/ata/libata-acpi.c
@@ -489,8 +489,7 @@ static void taskfile_load_raw(struct ata_port *ap,
 
 	/* convert gtf to tf */
 	tf.flags |= ATA_TFLAG_ISADDR | ATA_TFLAG_DEVICE; /* TBD */
-	tf.protocol = atadev->class == ATA_DEV_ATAPI ?
-		ATA_PROT_ATAPI_NODATA : ATA_PROT_NODATA;
+	tf.protocol = ATA_PROT_NODATA;
 	tf.feature = gtf->tfa[0];	/* 0x1f1 */
 	tf.nsect   = gtf->tfa[1];	/* 0x1f2 */
 	tf.lbal    = gtf->tfa[2];	/* 0x1f3 */
diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c
index a7950885d18e..4595d1f8cf60 100644
--- a/drivers/ata/libata-core.c
+++ b/drivers/ata/libata-core.c
@@ -895,6 +895,7 @@ static u64 ata_read_native_max_address(struct ata_device *dev)
 /**
  *	ata_set_native_max_address_ext	-	LBA48 native max set
  *	@dev: Device to query
+ *	@new_sectors: new max sectors value to set for the device
  *
  *	Perform an LBA48 size set max upon the device in question. Return the
  *	actual LBA48 size or zero if the command fails.
@@ -932,6 +933,7 @@ static u64 ata_set_native_max_address_ext(struct ata_device *dev, u64 new_sector
 /**
  *	ata_set_native_max_address	-	LBA28 native max set
  *	@dev: Device to query
+ *	@new_sectors: new max sectors value to set for the device
  *
  *	Perform an LBA28 size set max upon the device in question. Return the
  *	actual LBA28 size or zero if the command fails.
@@ -1316,7 +1318,7 @@ void ata_port_flush_task(struct ata_port *ap)
 	spin_unlock_irqrestore(ap->lock, flags);
 
 	DPRINTK("flush #1\n");
-	flush_workqueue(ata_wq);
+	cancel_work_sync(&ap->port_task.work); /* akpm: seems unneeded */
 
 	/*
 	 * At this point, if a task is running, it's guaranteed to see
@@ -1327,7 +1329,7 @@ void ata_port_flush_task(struct ata_port *ap)
 		if (ata_msg_ctl(ap))
 			ata_port_printk(ap, KERN_DEBUG, "%s: flush #2\n",
 					__FUNCTION__);
-		flush_workqueue(ata_wq);
+		cancel_work_sync(&ap->port_task.work);
 	}
 
 	spin_lock_irqsave(ap->lock, flags);
@@ -6475,9 +6477,9 @@ void ata_port_detach(struct ata_port *ap)
 	/* Flush hotplug task.  The sequence is similar to
 	 * ata_port_flush_task().
 	 */
-	flush_workqueue(ata_aux_wq);
+	cancel_work_sync(&ap->hotplug_task.work); /* akpm: why? */
 	cancel_delayed_work(&ap->hotplug_task);
-	flush_workqueue(ata_aux_wq);
+	cancel_work_sync(&ap->hotplug_task.work);
 
  skip_eh:
 	/* remove the associated SCSI host */
diff --git a/drivers/ata/pata_pcmcia.c b/drivers/ata/pata_pcmcia.c
index 75dc84797ff3..11245e331f77 100644
--- a/drivers/ata/pata_pcmcia.c
+++ b/drivers/ata/pata_pcmcia.c
@@ -357,6 +357,7 @@ static struct pcmcia_device_id pcmcia_devices[] = {
 	PCMCIA_DEVICE_MANF_CARD(0x000a, 0x0000),	/* I-O Data CFA */
 	PCMCIA_DEVICE_MANF_CARD(0x001c, 0x0001),	/* Mitsubishi CFA */
 	PCMCIA_DEVICE_MANF_CARD(0x0032, 0x0704),
+	PCMCIA_DEVICE_MANF_CARD(0x0032, 0x2904),
 	PCMCIA_DEVICE_MANF_CARD(0x0045, 0x0401),	/* SanDisk CFA */
 	PCMCIA_DEVICE_MANF_CARD(0x0098, 0x0000),	/* Toshiba */
 	PCMCIA_DEVICE_MANF_CARD(0x00a4, 0x002d),
diff --git a/drivers/ata/pata_qdi.c b/drivers/ata/pata_qdi.c
index 27685ce63ceb..fb8c9e14b8d4 100644
--- a/drivers/ata/pata_qdi.c
+++ b/drivers/ata/pata_qdi.c
@@ -375,7 +375,7 @@ static __init int qdi_init(void)
 				res = inb(port + 3);
 				if (res & 1) {
 					/* Single channel mode */
-					if (qdi_init_one(port, 6580, ide_port[r & 0x01], ide_irq[r & 0x01], r & 0x04))
+					if (qdi_init_one(port, 6580, ide_port[r & 0x01], ide_irq[r & 0x01], r & 0x04) == 0)
 						ct++;
 				} else {
 					/* Dual channel mode */
diff --git a/drivers/ata/pata_scc.c b/drivers/ata/pata_scc.c
index 5df354d573e8..203f463ac39f 100644
--- a/drivers/ata/pata_scc.c
+++ b/drivers/ata/pata_scc.c
@@ -1142,14 +1142,14 @@ static int scc_init_one (struct pci_dev *pdev, const struct pci_device_id *ent)
 	static int printed_version;
 	unsigned int board_idx = (unsigned int) ent->driver_data;
 	const struct ata_port_info *ppi[] = { &scc_port_info[board_idx], NULL };
-	struct device *dev = &pdev->dev;
+	struct ata_host *host;
 	int rc;
 
 	if (!printed_version++)
 		dev_printk(KERN_DEBUG, &pdev->dev,
 			   "version " DRV_VERSION "\n");
 
-	host = ata_port_alloc_pinfo(&pdev->dev, ppi, 1);
+	host = ata_host_alloc_pinfo(&pdev->dev, ppi, 1);
 	if (!host)
 		return -ENOMEM;
 
diff --git a/drivers/ata/sata_nv.c b/drivers/ata/sata_nv.c
index e2e795e58236..a097595d4dc7 100644
--- a/drivers/ata/sata_nv.c
+++ b/drivers/ata/sata_nv.c
@@ -257,6 +257,8 @@ static void nv_adma_port_stop(struct ata_port *ap);
 static int nv_adma_port_suspend(struct ata_port *ap, pm_message_t mesg);
 static int nv_adma_port_resume(struct ata_port *ap);
 #endif
+static void nv_adma_freeze(struct ata_port *ap);
+static void nv_adma_thaw(struct ata_port *ap);
 static void nv_adma_error_handler(struct ata_port *ap);
 static void nv_adma_host_stop(struct ata_host *host);
 static void nv_adma_post_internal_cmd(struct ata_queued_cmd *qc);
@@ -444,8 +446,8 @@ static const struct ata_port_operations nv_adma_ops = {
 	.bmdma_status		= ata_bmdma_status,
 	.qc_prep		= nv_adma_qc_prep,
 	.qc_issue		= nv_adma_qc_issue,
-	.freeze			= nv_ck804_freeze,
-	.thaw			= nv_ck804_thaw,
+	.freeze			= nv_adma_freeze,
+	.thaw			= nv_adma_thaw,
 	.error_handler		= nv_adma_error_handler,
 	.post_internal_cmd	= nv_adma_post_internal_cmd,
 	.data_xfer		= ata_data_xfer,
@@ -815,8 +817,16 @@ static irqreturn_t nv_adma_interrupt(int irq, void *dev_instance)
 			u16 status;
 			u32 gen_ctl;
 			u32 notifier, notifier_error;
+			
+			/* if ADMA is disabled, use standard ata interrupt handler */
+			if (pp->flags & NV_ADMA_ATAPI_SETUP_COMPLETE) {
+				u8 irq_stat = readb(host->iomap[NV_MMIO_BAR] + NV_INT_STATUS_CK804)
+					>> (NV_INT_PORT_SHIFT * i);
+				handled += nv_host_intr(ap, irq_stat);
+				continue;
+			}
 
-			/* if in ATA register mode, use standard ata interrupt handler */
+			/* if in ATA register mode, check for standard interrupts */
 			if (pp->flags & NV_ADMA_PORT_REGISTER_MODE) {
 				u8 irq_stat = readb(host->iomap[NV_MMIO_BAR] + NV_INT_STATUS_CK804)
 					>> (NV_INT_PORT_SHIFT * i);
@@ -826,7 +836,6 @@ static irqreturn_t nv_adma_interrupt(int irq, void *dev_instance)
 					    command is active, to prevent losing interrupts. */
 					irq_stat |= NV_INT_DEV;
 				handled += nv_host_intr(ap, irq_stat);
-				continue;
 			}
 
 			notifier = readl(mmio + NV_ADMA_NOTIFIER);
@@ -912,22 +921,77 @@ static irqreturn_t nv_adma_interrupt(int irq, void *dev_instance)
 	return IRQ_RETVAL(handled);
 }
 
+static void nv_adma_freeze(struct ata_port *ap)
+{
+	struct nv_adma_port_priv *pp = ap->private_data;
+	void __iomem *mmio = pp->ctl_block;
+	u16 tmp;
+
+	nv_ck804_freeze(ap);
+
+	if (pp->flags & NV_ADMA_ATAPI_SETUP_COMPLETE)
+		return;
+
+	/* clear any outstanding CK804 notifications */
+	writeb( NV_INT_ALL << (ap->port_no * NV_INT_PORT_SHIFT),
+		ap->host->iomap[NV_MMIO_BAR] + NV_INT_STATUS_CK804);
+
+	/* Disable interrupt */
+	tmp = readw(mmio + NV_ADMA_CTL);
+	writew( tmp & ~(NV_ADMA_CTL_AIEN | NV_ADMA_CTL_HOTPLUG_IEN),
+		mmio + NV_ADMA_CTL);
+	readw( mmio + NV_ADMA_CTL );	/* flush posted write */
+}
+
+static void nv_adma_thaw(struct ata_port *ap)
+{
+	struct nv_adma_port_priv *pp = ap->private_data;
+	void __iomem *mmio = pp->ctl_block;
+	u16 tmp;
+
+	nv_ck804_thaw(ap);
+
+	if (pp->flags & NV_ADMA_ATAPI_SETUP_COMPLETE)
+		return;
+
+	/* Enable interrupt */
+	tmp = readw(mmio + NV_ADMA_CTL);
+	writew( tmp | (NV_ADMA_CTL_AIEN | NV_ADMA_CTL_HOTPLUG_IEN),
+		mmio + NV_ADMA_CTL);
+	readw( mmio + NV_ADMA_CTL );	/* flush posted write */
+}
+
 static void nv_adma_irq_clear(struct ata_port *ap)
 {
 	struct nv_adma_port_priv *pp = ap->private_data;
 	void __iomem *mmio = pp->ctl_block;
-	u16 status = readw(mmio + NV_ADMA_STAT);
-	u32 notifier = readl(mmio + NV_ADMA_NOTIFIER);
-	u32 notifier_error = readl(mmio + NV_ADMA_NOTIFIER_ERROR);
-	void __iomem *dma_stat_addr = ap->ioaddr.bmdma_addr + ATA_DMA_STATUS;
+	u32 notifier_clears[2];
 
-	/* clear ADMA status */
-	writew(status, mmio + NV_ADMA_STAT);
-	writel(notifier | notifier_error,
-	       pp->notifier_clear_block);
+	if (pp->flags & NV_ADMA_ATAPI_SETUP_COMPLETE) {
+		ata_bmdma_irq_clear(ap);
+		return;
+	}
+
+	/* clear any outstanding CK804 notifications */
+	writeb( NV_INT_ALL << (ap->port_no * NV_INT_PORT_SHIFT),
+		ap->host->iomap[NV_MMIO_BAR] + NV_INT_STATUS_CK804);
 
-	/** clear legacy status */
-	iowrite8(ioread8(dma_stat_addr), dma_stat_addr);
+	/* clear ADMA status */
+	writew(0xffff, mmio + NV_ADMA_STAT);
+	
+	/* clear notifiers - note both ports need to be written with
+	   something even though we are only clearing on one */
+	if (ap->port_no == 0) {
+		notifier_clears[0] = 0xFFFFFFFF;
+		notifier_clears[1] = 0;
+	} else {
+		notifier_clears[0] = 0;
+		notifier_clears[1] = 0xFFFFFFFF;
+	}
+	pp = ap->host->ports[0]->private_data;
+	writel(notifier_clears[0], pp->notifier_clear_block);
+	pp = ap->host->ports[1]->private_data;
+	writel(notifier_clears[1], pp->notifier_clear_block);
 }
 
 static void nv_adma_post_internal_cmd(struct ata_queued_cmd *qc)
diff --git a/drivers/ata/sata_promise.c b/drivers/ata/sata_promise.c
index f56549b90aa6..3a7d9b5332af 100644
--- a/drivers/ata/sata_promise.c
+++ b/drivers/ata/sata_promise.c
@@ -45,7 +45,7 @@
 #include "sata_promise.h"
 
 #define DRV_NAME	"sata_promise"
-#define DRV_VERSION	"2.05"
+#define DRV_VERSION	"2.07"
 
 
 enum {
@@ -653,6 +653,8 @@ static void pdc_error_intr(struct ata_port *ap, struct ata_queued_cmd *qc,
 	qc->err_mask |= ac_err_mask;
 
 	pdc_reset_port(ap);
+
+	ata_port_abort(ap);
 }
 
 static inline unsigned int pdc_host_intr( struct ata_port *ap,
@@ -924,6 +926,7 @@ static int pdc_ata_init_one (struct pci_dev *pdev, const struct pci_device_id *e
 	struct ata_host *host;
 	void __iomem *base;
 	int n_ports, i, rc;
+	int is_sataii_tx4;
 
 	if (!printed_version++)
 		dev_printk(KERN_DEBUG, &pdev->dev, "version " DRV_VERSION "\n");
@@ -962,10 +965,23 @@ static int pdc_ata_init_one (struct pci_dev *pdev, const struct pci_device_id *e
 	}
 	host->iomap = pcim_iomap_table(pdev);
 
-	for (i = 0; i < host->n_ports; i++)
+	is_sataii_tx4 = 0;
+	if ((pi->flags & (PDC_FLAG_GEN_II|PDC_FLAG_4_PORTS)) == (PDC_FLAG_GEN_II|PDC_FLAG_4_PORTS)) {
+		is_sataii_tx4 = 1;
+		dev_printk(KERN_INFO, &pdev->dev, "applying SATAII TX4 port numbering workaround\n");
+	}
+	for (i = 0; i < host->n_ports; i++) {
+		static const unsigned char sataii_tx4_port_remap[4] = { 3, 1, 0, 2};
+		int ata_nr;
+
+		ata_nr = i;
+		if (is_sataii_tx4)
+			ata_nr = sataii_tx4_port_remap[i];
+
 		pdc_ata_setup_port(host->ports[i],
-				   base + 0x200 + i * 0x80,
-				   base + 0x400 + i * 0x100);
+				   base + 0x200 + ata_nr * 0x80,
+				   base + 0x400 + ata_nr * 0x100);
+	}
 
 	/* initialize adapter */
 	pdc_host_init(host);
diff --git a/drivers/ata/sata_via.c b/drivers/ata/sata_via.c
index 305ab7c68ca5..939c9246fdd1 100644
--- a/drivers/ata/sata_via.c
+++ b/drivers/ata/sata_via.c
@@ -93,6 +93,10 @@ static struct pci_driver svia_pci_driver = {
 	.name			= DRV_NAME,
 	.id_table		= svia_pci_tbl,
 	.probe			= svia_init_one,
+#ifdef CONFIG_PM
+	.suspend		= ata_pci_device_suspend,
+	.resume			= ata_pci_device_resume,
+#endif
 	.remove			= ata_pci_remove_one,
 };
 
@@ -112,6 +116,10 @@ static struct scsi_host_template svia_sht = {
 	.slave_configure	= ata_scsi_slave_config,
 	.slave_destroy		= ata_scsi_slave_destroy,
 	.bios_param		= ata_std_bios_param,
+#ifdef CONFIG_PM
+	.suspend		= ata_scsi_device_suspend,
+	.resume			= ata_scsi_device_resume,
+#endif
 };
 
 static const struct ata_port_operations vt6420_sata_ops = {
diff --git a/drivers/base/devres.c b/drivers/base/devres.c
index e177c9533b6c..e1c0730a3b99 100644
--- a/drivers/base/devres.c
+++ b/drivers/base/devres.c
@@ -101,19 +101,6 @@ static void add_dr(struct device *dev, struct devres_node *node)
 	list_add_tail(&node->entry, &dev->devres_head);
 }
 
-/**
- * devres_alloc - Allocate device resource data
- * @release: Release function devres will be associated with
- * @size: Allocation size
- * @gfp: Allocation flags
- *
- * allocate devres of @size bytes.  The allocated area is zeroed, then
- * associated with @release.  The returned pointer can be passed to
- * other devres_*() functions.
- *
- * RETURNS:
- * Pointer to allocated devres on success, NULL on failure.
- */
 #ifdef CONFIG_DEBUG_DEVRES
 void * __devres_alloc(dr_release_t release, size_t size, gfp_t gfp,
 		      const char *name)
@@ -128,6 +115,19 @@ void * __devres_alloc(dr_release_t release, size_t size, gfp_t gfp,
 }
 EXPORT_SYMBOL_GPL(__devres_alloc);
 #else
+/**
+ * devres_alloc - Allocate device resource data
+ * @release: Release function devres will be associated with
+ * @size: Allocation size
+ * @gfp: Allocation flags
+ *
+ * Allocate devres of @size bytes.  The allocated area is zeroed, then
+ * associated with @release.  The returned pointer can be passed to
+ * other devres_*() functions.
+ *
+ * RETURNS:
+ * Pointer to allocated devres on success, NULL on failure.
+ */
 void * devres_alloc(dr_release_t release, size_t size, gfp_t gfp)
 {
 	struct devres *dr;
@@ -416,7 +416,7 @@ static int release_nodes(struct device *dev, struct list_head *first,
 }
 
 /**
- * devres_release_all - Release all resources
+ * devres_release_all - Release all managed resources
  * @dev: Device to release resources for
  *
  * Release all resources associated with @dev.  This function is
@@ -600,7 +600,7 @@ static int devm_kzalloc_match(struct device *dev, void *res, void *data)
 }
 
 /**
- * devm_kzalloc - Managed kzalloc
+ * devm_kzalloc - Resource-managed kzalloc
  * @dev: Device to allocate memory for
  * @size: Allocation size
  * @gfp: Allocation gfp flags
@@ -628,7 +628,7 @@ void * devm_kzalloc(struct device *dev, size_t size, gfp_t gfp)
 EXPORT_SYMBOL_GPL(devm_kzalloc);
 
 /**
- * devm_kfree - Managed kfree
+ * devm_kfree - Resource-managed kfree
  * @dev: Device this memory belongs to
  * @p: Memory to free
  *
diff --git a/drivers/base/platform.c b/drivers/base/platform.c
index eb84d9d44645..869ff8c00146 100644
--- a/drivers/base/platform.c
+++ b/drivers/base/platform.c
@@ -360,7 +360,7 @@ EXPORT_SYMBOL_GPL(platform_device_unregister);
  *	This function creates a simple platform device that requires minimal
  *	resource and memory management. Canned release function freeing
  *	memory allocated for the device allows drivers using such devices
- *	to be unloaded iwithout waiting for the last reference to the device
+ *	to be unloaded without waiting for the last reference to the device
  *	to be dropped.
  *
  *	This interface is primarily intended for use with legacy drivers
diff --git a/drivers/base/topology.c b/drivers/base/topology.c
index 067a9e8bc377..8d8cdfec6529 100644
--- a/drivers/base/topology.c
+++ b/drivers/base/topology.c
@@ -126,10 +126,13 @@ static int __cpuinit topology_cpu_callback(struct notifier_block *nfb,
 
 	switch (action) {
 	case CPU_UP_PREPARE:
+	case CPU_UP_PREPARE_FROZEN:
 		rc = topology_add_dev(cpu);
 		break;
 	case CPU_UP_CANCELED:
+	case CPU_UP_CANCELED_FROZEN:
 	case CPU_DEAD:
+	case CPU_DEAD_FROZEN:
 		topology_remove_dev(cpu);
 		break;
 	}
diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index af6d7274a7cc..18cdd8c77626 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -243,17 +243,13 @@ static int do_lo_send_aops(struct loop_device *lo, struct bio_vec *bvec,
 		transfer_result = lo_do_transfer(lo, WRITE, page, offset,
 				bvec->bv_page, bv_offs, size, IV);
 		if (unlikely(transfer_result)) {
-			char *kaddr;
-
 			/*
 			 * The transfer failed, but we still write the data to
 			 * keep prepare/commit calls balanced.
 			 */
 			printk(KERN_ERR "loop: transfer error block %llu\n",
 			       (unsigned long long)index);
-			kaddr = kmap_atomic(page, KM_USER0);
-			memset(kaddr + offset, 0, size);
-			kunmap_atomic(kaddr, KM_USER0);
+			zero_user_page(page, offset, size, KM_USER0);
 		}
 		flush_dcache_page(page);
 		ret = aops->commit_write(file, page, offset,
diff --git a/drivers/block/nbd.c b/drivers/block/nbd.c
index 090796bef78f..069ae39a9cd9 100644
--- a/drivers/block/nbd.c
+++ b/drivers/block/nbd.c
@@ -366,20 +366,25 @@ static struct disk_attribute pid_attr = {
 	.show = pid_show,
 };
 
-static void nbd_do_it(struct nbd_device *lo)
+static int nbd_do_it(struct nbd_device *lo)
 {
 	struct request *req;
+	int ret;
 
 	BUG_ON(lo->magic != LO_MAGIC);
 
 	lo->pid = current->pid;
-	sysfs_create_file(&lo->disk->kobj, &pid_attr.attr);
+	ret = sysfs_create_file(&lo->disk->kobj, &pid_attr.attr);
+	if (ret) {
+		printk(KERN_ERR "nbd: sysfs_create_file failed!");
+		return ret;
+	}
 
 	while ((req = nbd_read_stat(lo)) != NULL)
 		nbd_end_request(req);
 
 	sysfs_remove_file(&lo->disk->kobj, &pid_attr.attr);
-	return;
+	return 0;
 }
 
 static void nbd_clear_que(struct nbd_device *lo)
@@ -569,7 +574,9 @@ static int nbd_ioctl(struct inode *inode, struct file *file,
 	case NBD_DO_IT:
 		if (!lo->file)
 			return -EINVAL;
-		nbd_do_it(lo);
+		error = nbd_do_it(lo);
+		if (error)
+			return error;
 		/* on return tidy up in case we have a signal */
 		/* Forcibly shutdown the socket causing all listeners
 		 * to error
diff --git a/drivers/block/rd.c b/drivers/block/rd.c
index 43d4ebcb3b44..a1512da32410 100644
--- a/drivers/block/rd.c
+++ b/drivers/block/rd.c
@@ -151,7 +151,7 @@ static int ramdisk_commit_write(struct file *file, struct page *page,
 }
 
 /*
- * ->writepage to the the blockdev's mapping has to redirty the page so that the
+ * ->writepage to the blockdev's mapping has to redirty the page so that the
  * VM doesn't go and steal it.  We return AOP_WRITEPAGE_ACTIVATE so that the VM
  * won't try to (pointlessly) write the page again for a while.
  *
diff --git a/drivers/char/drm/drm_dma.c b/drivers/char/drm/drm_dma.c
index 892db7096986..32ed19c9ec1c 100644
--- a/drivers/char/drm/drm_dma.c
+++ b/drivers/char/drm/drm_dma.c
@@ -65,7 +65,7 @@ int drm_dma_setup(drm_device_t * dev)
  * \param dev DRM device.
  *
  * Free all pages associated with DMA buffers, the buffers and pages lists, and
- * finally the the drm_device::dma structure itself.
+ * finally the drm_device::dma structure itself.
  */
 void drm_dma_takedown(drm_device_t * dev)
 {
diff --git a/drivers/char/drm/drm_vm.c b/drivers/char/drm/drm_vm.c
index 35540cfb43dd..b5c5b9fa84c3 100644
--- a/drivers/char/drm/drm_vm.c
+++ b/drivers/char/drm/drm_vm.c
@@ -157,7 +157,7 @@ static __inline__ struct page *drm_do_vm_nopage(struct vm_area_struct *vma,
  * \param address access address.
  * \return pointer to the page structure.
  *
- * Get the the mapping, find the real physical page to map, get the page, and
+ * Get the mapping, find the real physical page to map, get the page, and
  * return it.
  */
 static __inline__ struct page *drm_do_vm_shm_nopage(struct vm_area_struct *vma,
diff --git a/drivers/char/drm/r300_reg.h b/drivers/char/drm/r300_reg.h
index a881f96c983e..ecda760ae8c0 100644
--- a/drivers/char/drm/r300_reg.h
+++ b/drivers/char/drm/r300_reg.h
@@ -293,7 +293,7 @@ I am fairly certain that they are correct unless stated otherwise in comments.
 #       define R300_PVS_CNTL_1_PROGRAM_START_SHIFT   0
 #       define R300_PVS_CNTL_1_POS_END_SHIFT         10
 #       define R300_PVS_CNTL_1_PROGRAM_END_SHIFT     20
-/* Addresses are relative the the vertex program parameters area. */
+/* Addresses are relative to the vertex program parameters area. */
 #define R300_VAP_PVS_CNTL_2                 0x22D4
 #       define R300_PVS_CNTL_2_PARAM_OFFSET_SHIFT 0
 #       define R300_PVS_CNTL_2_PARAM_COUNT_SHIFT  16
diff --git a/drivers/char/genrtc.c b/drivers/char/genrtc.c
index 49f914e79216..9e1fc02967ff 100644
--- a/drivers/char/genrtc.c
+++ b/drivers/char/genrtc.c
@@ -12,7 +12,7 @@
  *
  *	This driver allows use of the real time clock (built into
  *	nearly all computers) from user space. It exports the /dev/rtc
- *	interface supporting various ioctl() and also the /proc/dev/rtc
+ *	interface supporting various ioctl() and also the /proc/driver/rtc
  *	pseudo-file for status information.
  *
  *	The ioctls can be used to set the interrupt behaviour where
@@ -377,7 +377,7 @@ static int gen_rtc_release(struct inode *inode, struct file *file)
 #ifdef CONFIG_PROC_FS
 
 /*
- *	Info exported via "/proc/rtc".
+ *	Info exported via "/proc/driver/rtc".
  */
 
 static int gen_rtc_proc_output(char *buf)
diff --git a/drivers/char/hw_random/Kconfig b/drivers/char/hw_random/Kconfig
index 5f3acd8e64b8..7cda04b33534 100644
--- a/drivers/char/hw_random/Kconfig
+++ b/drivers/char/hw_random/Kconfig
@@ -91,3 +91,17 @@ config HW_RANDOM_OMAP
 	  module will be called omap-rng.
 
  	  If unsure, say Y.
+
+config HW_RANDOM_PASEMI
+	tristate "PA Semi HW Random Number Generator support"
+	depends on HW_RANDOM && PPC_PASEMI
+	default HW_RANDOM
+	---help---
+	  This driver provides kernel-side support for the Random Number
+	  Generator hardware found on PA6T-1682M processor.
+
+	  To compile this driver as a module, choose M here: the
+	  module will be called pasemi-rng.
+
+	  If unsure, say Y.
+
diff --git a/drivers/char/hw_random/Makefile b/drivers/char/hw_random/Makefile
index c41fa19454e3..c8b7300e2fb1 100644
--- a/drivers/char/hw_random/Makefile
+++ b/drivers/char/hw_random/Makefile
@@ -10,3 +10,4 @@ obj-$(CONFIG_HW_RANDOM_GEODE) += geode-rng.o
 obj-$(CONFIG_HW_RANDOM_VIA) += via-rng.o
 obj-$(CONFIG_HW_RANDOM_IXP4XX) += ixp4xx-rng.o
 obj-$(CONFIG_HW_RANDOM_OMAP) += omap-rng.o
+obj-$(CONFIG_HW_RANDOM_PASEMI) += pasemi-rng.o
diff --git a/drivers/char/hw_random/pasemi-rng.c b/drivers/char/hw_random/pasemi-rng.c
new file mode 100644
index 000000000000..fa6040b6c8f2
--- /dev/null
+++ b/drivers/char/hw_random/pasemi-rng.c
@@ -0,0 +1,156 @@
+/*
+ * Copyright (C) 2006-2007 PA Semi, Inc
+ *
+ * Maintained by: Olof Johansson <olof@lixom.net>
+ *
+ * Driver for the PWRficient onchip rng
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/platform_device.h>
+#include <linux/hw_random.h>
+#include <asm/of_platform.h>
+#include <asm/io.h>
+
+#define SDCRNG_CTL_REG			0x00
+#define   SDCRNG_CTL_FVLD_M		0x0000f000
+#define   SDCRNG_CTL_FVLD_S		12
+#define   SDCRNG_CTL_KSZ		0x00000800
+#define   SDCRNG_CTL_RSRC_CRG		0x00000010
+#define   SDCRNG_CTL_RSRC_RRG		0x00000000
+#define   SDCRNG_CTL_CE			0x00000004
+#define   SDCRNG_CTL_RE			0x00000002
+#define   SDCRNG_CTL_DR			0x00000001
+#define   SDCRNG_CTL_SELECT_RRG_RNG	(SDCRNG_CTL_RE | SDCRNG_CTL_RSRC_RRG)
+#define   SDCRNG_CTL_SELECT_CRG_RNG	(SDCRNG_CTL_CE | SDCRNG_CTL_RSRC_CRG)
+#define SDCRNG_VAL_REG			0x20
+
+#define MODULE_NAME "pasemi_rng"
+
+static int pasemi_rng_data_present(struct hwrng *rng)
+{
+	void __iomem *rng_regs = (void __iomem *)rng->priv;
+
+	return (in_le32(rng_regs + SDCRNG_CTL_REG)
+		& SDCRNG_CTL_FVLD_M) ? 1 : 0;
+}
+
+static int pasemi_rng_data_read(struct hwrng *rng, u32 *data)
+{
+	void __iomem *rng_regs = (void __iomem *)rng->priv;
+	*data = in_le32(rng_regs + SDCRNG_VAL_REG);
+	return 4;
+}
+
+static int pasemi_rng_init(struct hwrng *rng)
+{
+	void __iomem *rng_regs = (void __iomem *)rng->priv;
+	u32 ctl;
+
+	ctl = SDCRNG_CTL_DR | SDCRNG_CTL_SELECT_RRG_RNG | SDCRNG_CTL_KSZ;
+	out_le32(rng_regs + SDCRNG_CTL_REG, ctl);
+	out_le32(rng_regs + SDCRNG_CTL_REG, ctl & ~SDCRNG_CTL_DR);
+
+	return 0;
+}
+
+static void pasemi_rng_cleanup(struct hwrng *rng)
+{
+	void __iomem *rng_regs = (void __iomem *)rng->priv;
+	u32 ctl;
+
+	ctl = SDCRNG_CTL_RE | SDCRNG_CTL_CE;
+	out_le32(rng_regs + SDCRNG_CTL_REG,
+		 in_le32(rng_regs + SDCRNG_CTL_REG) & ~ctl);
+}
+
+static struct hwrng pasemi_rng = {
+	.name		= MODULE_NAME,
+	.init		= pasemi_rng_init,
+	.cleanup	= pasemi_rng_cleanup,
+	.data_present	= pasemi_rng_data_present,
+	.data_read	= pasemi_rng_data_read,
+};
+
+static int __devinit rng_probe(struct of_device *ofdev,
+			       const struct of_device_id *match)
+{
+	void __iomem *rng_regs;
+	struct device_node *rng_np = ofdev->node;
+	struct resource res;
+	int err = 0;
+
+	err = of_address_to_resource(rng_np, 0, &res);
+	if (err)
+		return -ENODEV;
+
+	rng_regs = ioremap(res.start, 0x100);
+
+	if (!rng_regs)
+		return -ENOMEM;
+
+	pasemi_rng.priv = (unsigned long)rng_regs;
+
+	printk(KERN_INFO "Registering PA Semi RNG\n");
+
+	err = hwrng_register(&pasemi_rng);
+
+	if (err)
+		iounmap(rng_regs);
+
+	return err;
+}
+
+static int __devexit rng_remove(struct of_device *dev)
+{
+	void __iomem *rng_regs = (void __iomem *)pasemi_rng.priv;
+
+	hwrng_unregister(&pasemi_rng);
+	iounmap(rng_regs);
+
+	return 0;
+}
+
+static struct of_device_id rng_match[] = {
+	{
+		.compatible      = "1682m-rng",
+	},
+	{},
+};
+
+static struct of_platform_driver rng_driver = {
+	.name		= "pasemi-rng",
+	.match_table	= rng_match,
+	.probe		= rng_probe,
+	.remove		= rng_remove,
+};
+
+static int __init rng_init(void)
+{
+	return of_register_platform_driver(&rng_driver);
+}
+module_init(rng_init);
+
+static void __exit rng_exit(void)
+{
+	of_unregister_platform_driver(&rng_driver);
+}
+module_exit(rng_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Egor Martovetsky <egor@pasemi.com>");
+MODULE_DESCRIPTION("H/W RNG driver for PA Semi processor");
diff --git a/drivers/char/mmtimer.c b/drivers/char/mmtimer.c
index c09160383a53..6e55cfb9c65a 100644
--- a/drivers/char/mmtimer.c
+++ b/drivers/char/mmtimer.c
@@ -705,15 +705,13 @@ static int __init mmtimer_init(void)
 	maxn++;
 
 	/* Allocate list of node ptrs to mmtimer_t's */
-	timers = kmalloc(sizeof(mmtimer_t *)*maxn, GFP_KERNEL);
+	timers = kzalloc(sizeof(mmtimer_t *)*maxn, GFP_KERNEL);
 	if (timers == NULL) {
 		printk(KERN_ERR "%s: failed to allocate memory for device\n",
 				MMTIMER_NAME);
 		goto out3;
 	}
 
-	memset(timers,0,(sizeof(mmtimer_t *)*maxn));
-
 	/* Allocate mmtimer_t's for each online node */
 	for_each_online_node(node) {
 		timers[node] = kmalloc_node(sizeof(mmtimer_t)*NUM_COMPARATORS, GFP_KERNEL, node);
diff --git a/drivers/char/pcmcia/Kconfig b/drivers/char/pcmcia/Kconfig
index 27c1179ee527..f25facd97bb4 100644
--- a/drivers/char/pcmcia/Kconfig
+++ b/drivers/char/pcmcia/Kconfig
@@ -21,6 +21,7 @@ config SYNCLINK_CS
 config CARDMAN_4000
 	tristate "Omnikey Cardman 4000 support"
 	depends on PCMCIA
+	select BITREVERSE
 	help
 	  Enable support for the Omnikey Cardman 4000 PCMCIA Smartcard
 	  reader.
diff --git a/drivers/char/pcmcia/cm4000_cs.c b/drivers/char/pcmcia/cm4000_cs.c
index e91b43a014b0..fee58e03dbe2 100644
--- a/drivers/char/pcmcia/cm4000_cs.c
+++ b/drivers/char/pcmcia/cm4000_cs.c
@@ -31,6 +31,7 @@
 #include <linux/init.h>
 #include <linux/fs.h>
 #include <linux/delay.h>
+#include <linux/bitrev.h>
 #include <asm/uaccess.h>
 #include <asm/io.h>
 
@@ -194,41 +195,17 @@ static inline unsigned char xinb(unsigned short port)
 }
 #endif
 
-#define	b_0000	15
-#define	b_0001	14
-#define	b_0010	13
-#define	b_0011	12
-#define	b_0100	11
-#define	b_0101	10
-#define	b_0110	9
-#define	b_0111	8
-#define	b_1000	7
-#define	b_1001	6
-#define	b_1010	5
-#define	b_1011	4
-#define	b_1100	3
-#define	b_1101	2
-#define	b_1110	1
-#define	b_1111	0
-
-static unsigned char irtab[16] = {
-	b_0000, b_1000, b_0100, b_1100,
-	b_0010, b_1010, b_0110, b_1110,
-	b_0001, b_1001, b_0101, b_1101,
-	b_0011, b_1011, b_0111, b_1111
-};
+static inline unsigned char invert_revert(unsigned char ch)
+{
+	return bitrev8(~ch);
+}
 
 static void str_invert_revert(unsigned char *b, int len)
 {
 	int i;
 
 	for (i = 0; i < len; i++)
-		b[i] = (irtab[b[i] & 0x0f] << 4) | irtab[b[i] >> 4];
-}
-
-static unsigned char invert_revert(unsigned char ch)
-{
-	return (irtab[ch & 0x0f] << 4) | irtab[ch >> 4];
+		b[i] = invert_revert(b[i]);
 }
 
 #define	ATRLENCK(dev,pos) \
@@ -1114,7 +1091,7 @@ static ssize_t cmm_write(struct file *filp, const char __user *buf,
 	/*
 	 * wait for atr to become valid.
 	 * note: it is important to lock this code. if we dont, the monitor
-	 * could be run between test_bit and the the call the sleep on the
+	 * could be run between test_bit and the call to sleep on the
 	 * atr-queue.  if *then* the monitor detects atr valid, it will wake up
 	 * any process on the atr-queue, *but* since we have been interrupted,
 	 * we do not yet sleep on this queue. this would result in a missed
@@ -1881,8 +1858,11 @@ static int cm4000_probe(struct pcmcia_device *link)
 	init_waitqueue_head(&dev->readq);
 
 	ret = cm4000_config(link, i);
-	if (ret)
+	if (ret) {
+		dev_table[i] = NULL;
+		kfree(dev);
 		return ret;
+	}
 
 	class_device_create(cmm_class, NULL, MKDEV(major, i), NULL,
 			    "cmm%d", i);
@@ -1907,7 +1887,7 @@ static void cm4000_detach(struct pcmcia_device *link)
 	cm4000_release(link);
 
 	dev_table[devno] = NULL;
- 	kfree(dev);
+	kfree(dev);
 
 	class_device_destroy(cmm_class, MKDEV(major, devno));
 
@@ -1956,12 +1936,14 @@ static int __init cmm_init(void)
 	if (major < 0) {
 		printk(KERN_WARNING MODULE_NAME
 			": could not get major number\n");
+		class_destroy(cmm_class);
 		return major;
 	}
 
 	rc = pcmcia_register_driver(&cm4000_driver);
 	if (rc < 0) {
 		unregister_chrdev(major, DEVICE_NAME);
+		class_destroy(cmm_class);
 		return rc;
 	}
 
diff --git a/drivers/char/pcmcia/cm4040_cs.c b/drivers/char/pcmcia/cm4040_cs.c
index f2e4ec4fd407..af88181a17f4 100644
--- a/drivers/char/pcmcia/cm4040_cs.c
+++ b/drivers/char/pcmcia/cm4040_cs.c
@@ -636,8 +636,11 @@ static int reader_probe(struct pcmcia_device *link)
 	setup_timer(&dev->poll_timer, cm4040_do_poll, 0);
 
 	ret = reader_config(link, i);
-	if (ret)
+	if (ret) {
+		dev_table[i] = NULL;
+		kfree(dev);
 		return ret;
+	}
 
 	class_device_create(cmx_class, NULL, MKDEV(major, i), NULL,
 			    "cmx%d", i);
@@ -708,12 +711,14 @@ static int __init cm4040_init(void)
 	if (major < 0) {
 		printk(KERN_WARNING MODULE_NAME
 			": could not get major number\n");
+		class_destroy(cmx_class);
 		return major;
 	}
 
 	rc = pcmcia_register_driver(&reader_driver);
 	if (rc < 0) {
 		unregister_chrdev(major, DEVICE_NAME);
+		class_destroy(cmx_class);
 		return rc;
 	}
 
diff --git a/drivers/char/tpm/Kconfig b/drivers/char/tpm/Kconfig
index fe00c7dfb649..11089be0691b 100644
--- a/drivers/char/tpm/Kconfig
+++ b/drivers/char/tpm/Kconfig
@@ -33,7 +33,7 @@ config TCG_NSC
 	tristate "National Semiconductor TPM Interface"
 	depends on TCG_TPM && PNPACPI
 	---help---
-	  If you have a TPM security chip from National Semicondutor 
+	  If you have a TPM security chip from National Semiconductor 
 	  say Yes and it will be accessible from within Linux.  To 
 	  compile this driver as a module, choose M here; the module 
 	  will be called tpm_nsc.
diff --git a/drivers/char/tty_io.c b/drivers/char/tty_io.c
index 7710a6a77d97..fc662e4ce58a 100644
--- a/drivers/char/tty_io.c
+++ b/drivers/char/tty_io.c
@@ -934,13 +934,6 @@ restart:
 		return -EINVAL;
 
 	/*
-	 *	No more input please, we are switching. The new ldisc
-	 *	will update this value in the ldisc open function
-	 */
-
-	tty->receive_room = 0;
-
-	/*
 	 *	Problem: What do we do if this blocks ?
 	 */
 
@@ -951,6 +944,13 @@ restart:
 		return 0;
 	}
 
+	/*
+	 *	No more input please, we are switching. The new ldisc
+	 *	will update this value in the ldisc open function
+	 */
+
+	tty->receive_room = 0;
+
 	o_ldisc = tty->ldisc;
 	o_tty = tty->link;
 
@@ -1573,11 +1573,11 @@ void no_tty(void)
 
 
 /**
- *	stop_tty	-	propogate flow control
+ *	stop_tty	-	propagate flow control
  *	@tty: tty to stop
  *
  *	Perform flow control to the driver. For PTY/TTY pairs we
- *	must also propogate the TIOCKPKT status. May be called
+ *	must also propagate the TIOCKPKT status. May be called
  *	on an already stopped device and will not re-call the driver
  *	method.
  *
@@ -1607,11 +1607,11 @@ void stop_tty(struct tty_struct *tty)
 EXPORT_SYMBOL(stop_tty);
 
 /**
- *	start_tty	-	propogate flow control
+ *	start_tty	-	propagate flow control
  *	@tty: tty to start
  *
  *	Start a tty that has been stopped if at all possible. Perform
- *	any neccessary wakeups and propogate the TIOCPKT status. If this
+ *	any neccessary wakeups and propagate the TIOCPKT status. If this
  *	is the tty was previous stopped and is being started then the
  *	driver start method is invoked and the line discipline woken.
  *
diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
index 893dbaf386fb..eb37fba9b7ef 100644
--- a/drivers/cpufreq/cpufreq.c
+++ b/drivers/cpufreq/cpufreq.c
@@ -1685,9 +1685,11 @@ static int cpufreq_cpu_callback(struct notifier_block *nfb,
 	if (sys_dev) {
 		switch (action) {
 		case CPU_ONLINE:
+		case CPU_ONLINE_FROZEN:
 			cpufreq_add_dev(sys_dev);
 			break;
 		case CPU_DOWN_PREPARE:
+		case CPU_DOWN_PREPARE_FROZEN:
 			if (unlikely(lock_policy_rwsem_write(cpu)))
 				BUG();
 
@@ -1699,6 +1701,7 @@ static int cpufreq_cpu_callback(struct notifier_block *nfb,
 			__cpufreq_remove_dev(sys_dev);
 			break;
 		case CPU_DOWN_FAILED:
+		case CPU_DOWN_FAILED_FROZEN:
 			cpufreq_add_dev(sys_dev);
 			break;
 		}
diff --git a/drivers/cpufreq/cpufreq_stats.c b/drivers/cpufreq/cpufreq_stats.c
index d1c7cac9316c..d2f0cbd8b8f3 100644
--- a/drivers/cpufreq/cpufreq_stats.c
+++ b/drivers/cpufreq/cpufreq_stats.c
@@ -313,9 +313,11 @@ static int cpufreq_stat_cpu_callback(struct notifier_block *nfb,
 
 	switch (action) {
 	case CPU_ONLINE:
+	case CPU_ONLINE_FROZEN:
 		cpufreq_update_policy(cpu);
 		break;
 	case CPU_DEAD:
+	case CPU_DEAD_FROZEN:
 		cpufreq_stats_free_table(cpu);
 		break;
 	}
diff --git a/drivers/crypto/Kconfig b/drivers/crypto/Kconfig
index f21fe66c9eef..f4c634504d1a 100644
--- a/drivers/crypto/Kconfig
+++ b/drivers/crypto/Kconfig
@@ -51,7 +51,7 @@ config CRYPTO_DEV_GEODE
 	default m
 	help
 	  Say 'Y' here to use the AMD Geode LX processor on-board AES
-	  engine for the CryptoAPI AES alogrithm.
+	  engine for the CryptoAPI AES algorithm.
 
 	  To compile this driver as a module, choose M here: the module
 	  will be called geode-aes.
diff --git a/drivers/hwmon/coretemp.c b/drivers/hwmon/coretemp.c
index 03b1f650d1c4..75e3911810a3 100644
--- a/drivers/hwmon/coretemp.c
+++ b/drivers/hwmon/coretemp.c
@@ -309,9 +309,11 @@ static int coretemp_cpu_callback(struct notifier_block *nfb,
 
 	switch (action) {
 	case CPU_ONLINE:
+	case CPU_ONLINE_FROZEN:
 		coretemp_device_add(cpu);
 		break;
 	case CPU_DEAD:
+	case CPU_DEAD_FROZEN:
 		coretemp_device_remove(cpu);
 		break;
 	}
diff --git a/drivers/i2c/chips/tps65010.c b/drivers/i2c/chips/tps65010.c
index 7ed92dc3d833..3c3f2ebf3fc9 100644
--- a/drivers/i2c/chips/tps65010.c
+++ b/drivers/i2c/chips/tps65010.c
@@ -354,7 +354,7 @@ static void tps65010_interrupt(struct tps65010 *tps)
 			 * also needs to get error handling and probably
 			 * an #ifdef CONFIG_SOFTWARE_SUSPEND
 			 */
-			pm_suspend(PM_SUSPEND_DISK);
+			hibernate();
 #endif
 			poll = 1;
 		}
diff --git a/drivers/ide/Kconfig b/drivers/ide/Kconfig
index 5bdf64b77913..1d06b415ede9 100644
--- a/drivers/ide/Kconfig
+++ b/drivers/ide/Kconfig
@@ -291,6 +291,17 @@ config IDE_TASK_IOCTL
 
 	  If you are unsure, say N here.
 
+config IDE_PROC_FS
+	bool "legacy /proc/ide/ support"
+	depends on IDE && PROC_FS
+	default y
+	help
+	  This option enables support for the various files in
+	  /proc/ide.  In Linux 2.6 this has been superseded by
+	  files in sysfs but many legacy applications rely on this.
+
+	  If unsure say Y.
+
 comment "IDE chipset support/bugfixes"
 
 config IDE_GENERIC
@@ -360,6 +371,9 @@ config IDEPCI_SHARE_IRQ
 	  It is safe to say Y to this question, in most cases.
 	  If unsure, say N.
 
+config IDEPCI_PCIBUS_ORDER
+	def_bool PCI && BLK_DEV_IDE=y && BLK_DEV_IDEPCI
+
 config BLK_DEV_OFFBOARD
 	bool "Boot off-board chipsets first support"
 	depends on PCI && BLK_DEV_IDEPCI
diff --git a/drivers/ide/Makefile b/drivers/ide/Makefile
index d9f029e8ff74..75dc6969e0a7 100644
--- a/drivers/ide/Makefile
+++ b/drivers/ide/Makefile
@@ -20,7 +20,7 @@ ide-core-$(CONFIG_BLK_DEV_CMD640)	+= pci/cmd640.o
 # Core IDE code - must come before legacy
 ide-core-$(CONFIG_BLK_DEV_IDEPCI)	+= setup-pci.o
 ide-core-$(CONFIG_BLK_DEV_IDEDMA)	+= ide-dma.o
-ide-core-$(CONFIG_PROC_FS)		+= ide-proc.o
+ide-core-$(CONFIG_IDE_PROC_FS)		+= ide-proc.o
 ide-core-$(CONFIG_BLK_DEV_IDEPNP)	+= ide-pnp.o
 ide-core-$(CONFIG_BLK_DEV_IDEACPI)	+= ide-acpi.o
 
diff --git a/drivers/ide/arm/bast-ide.c b/drivers/ide/arm/bast-ide.c
index 9d474e5fd8dc..f7449d04114a 100644
--- a/drivers/ide/arm/bast-ide.c
+++ b/drivers/ide/arm/bast-ide.c
@@ -45,7 +45,7 @@ bastide_register(unsigned int base, unsigned int aux, int irq,
 	hw.io_ports[IDE_CONTROL_OFFSET] = aux + (6 * 0x20);
 	hw.irq = irq;
 
-	ide_register_hw(&hw, hwif);
+	ide_register_hw(&hw, 0, hwif);
 
 	return 0;
 }
diff --git a/drivers/ide/arm/icside.c b/drivers/ide/arm/icside.c
index e2953fc1fafb..1fe0457243db 100644
--- a/drivers/ide/arm/icside.c
+++ b/drivers/ide/arm/icside.c
@@ -342,7 +342,7 @@ static int icside_dma_check(ide_drive_t *drive)
 	 * Enable DMA on any drive that has multiword DMA
 	 */
 	if (id->field_valid & 2) {
-		xfer_mode = ide_dma_speed(drive, 0);
+		xfer_mode = ide_max_dma_mode(drive);
 		goto out;
 	}
 
@@ -591,7 +591,8 @@ icside_register_v5(struct icside_state *state, struct expansion_card *ec)
 	state->hwif[0] = hwif;
 
 	probe_hwif_init(hwif);
-	create_proc_ide_interfaces();
+
+	ide_proc_register_port(hwif);
 
 	return 0;
 }
@@ -679,7 +680,9 @@ icside_register_v6(struct icside_state *state, struct expansion_card *ec)
 
 	probe_hwif_init(hwif);
 	probe_hwif_init(mate);
-	create_proc_ide_interfaces();
+
+	ide_proc_register_port(hwif);
+	ide_proc_register_port(mate);
 
 	return 0;
 
diff --git a/drivers/ide/arm/ide_arm.c b/drivers/ide/arm/ide_arm.c
index 23488c4d1fcd..a3d6744e870a 100644
--- a/drivers/ide/arm/ide_arm.c
+++ b/drivers/ide/arm/ide_arm.c
@@ -38,6 +38,6 @@ void __init ide_arm_init(void)
 		memset(&hw, 0, sizeof(hw));
 		ide_std_init_ports(&hw, IDE_ARM_IO, IDE_ARM_IO + 0x206);
 		hw.irq = IDE_ARM_IRQ;
-		ide_register_hw(&hw, NULL);
+		ide_register_hw(&hw, 1, NULL);
 	}
 }
diff --git a/drivers/ide/arm/rapide.c b/drivers/ide/arm/rapide.c
index 9c6c49fdd2b1..890ea3fac3c6 100644
--- a/drivers/ide/arm/rapide.c
+++ b/drivers/ide/arm/rapide.c
@@ -76,7 +76,7 @@ rapide_probe(struct expansion_card *ec, const struct ecard_id *id)
 		hwif->gendev.parent = &ec->dev;
 		hwif->noprobe = 0;
 		probe_hwif_init(hwif);
-		create_proc_ide_interfaces();
+		ide_proc_register_port(hwif);
 		ecard_set_drvdata(ec, hwif);
 		goto out;
 	}
diff --git a/drivers/ide/cris/ide-cris.c b/drivers/ide/cris/ide-cris.c
index 5e8efc89255a..c04cb25a01ff 100644
--- a/drivers/ide/cris/ide-cris.c
+++ b/drivers/ide/cris/ide-cris.c
@@ -796,7 +796,7 @@ init_e100_ide (void)
 		                ide_offsets,
 		                0, 0, cris_ide_ack_intr,
 		                ide_default_irq(0));
-		ide_register_hw(&hw, &hwif);
+		ide_register_hw(&hw, 1, &hwif);
 		hwif->mmio = 1;
 		hwif->chipset = ide_etrax100;
 		hwif->tuneproc = &tune_cris_ide;
@@ -1004,7 +1004,7 @@ static int cris_ide_build_dmatable (ide_drive_t *drive)
 
 static int cris_config_drive_for_dma (ide_drive_t *drive)
 {
-	u8 speed = ide_dma_speed(drive, 1);
+	u8 speed = ide_max_dma_mode(drive);
 
 	if (!speed)
 		return 0;
diff --git a/drivers/ide/h8300/ide-h8300.c b/drivers/ide/h8300/ide-h8300.c
index 88750a300337..6d26ad7360d5 100644
--- a/drivers/ide/h8300/ide-h8300.c
+++ b/drivers/ide/h8300/ide-h8300.c
@@ -101,7 +101,7 @@ void __init h8300_ide_init(void)
 	hw_setup(&hw);
 
 	/* register if */
-	idx = ide_register_hw(&hw, &hwif);
+	idx = ide_register_hw(&hw, 1, &hwif);
 	if (idx == -1) {
 		printk(KERN_ERR "ide-h8300: IDE I/F register failed\n");
 		return;
diff --git a/drivers/ide/ide-cd.c b/drivers/ide/ide-cd.c
index 638becda81c6..252ab8295edf 100644
--- a/drivers/ide/ide-cd.c
+++ b/drivers/ide/ide-cd.c
@@ -3059,10 +3059,14 @@ int ide_cdrom_probe_capabilities (ide_drive_t *drive)
 	return nslots;
 }
 
+#ifdef CONFIG_IDE_PROC_FS
 static void ide_cdrom_add_settings(ide_drive_t *drive)
 {
-	ide_add_setting(drive,	"dsc_overlap",		SETTING_RW, -1, -1, TYPE_BYTE, 0, 1, 1,	1, &drive->dsc_overlap, NULL);
+	ide_add_setting(drive, "dsc_overlap", SETTING_RW, TYPE_BYTE, 0, 1, 1, 1, &drive->dsc_overlap, NULL);
 }
+#else
+static inline void ide_cdrom_add_settings(ide_drive_t *drive) { ; }
+#endif
 
 /*
  * standard prep_rq_fn that builds 10 byte cmds
@@ -3274,7 +3278,7 @@ int ide_cdrom_setup (ide_drive_t *drive)
 	return 0;
 }
 
-#ifdef CONFIG_PROC_FS
+#ifdef CONFIG_IDE_PROC_FS
 static
 sector_t ide_cdrom_capacity (ide_drive_t *drive)
 {
@@ -3291,7 +3295,7 @@ static void ide_cd_remove(ide_drive_t *drive)
 {
 	struct cdrom_info *info = drive->driver_data;
 
-	ide_unregister_subdriver(drive, info->driver);
+	ide_proc_unregister_driver(drive, info->driver);
 
 	del_gendisk(info->disk);
 
@@ -3321,7 +3325,7 @@ static void ide_cd_release(struct kref *kref)
 
 static int ide_cd_probe(ide_drive_t *);
 
-#ifdef CONFIG_PROC_FS
+#ifdef CONFIG_IDE_PROC_FS
 static int proc_idecd_read_capacity
 	(char *page, char **start, off_t off, int count, int *eof, void *data)
 {
@@ -3336,8 +3340,6 @@ static ide_proc_entry_t idecd_proc[] = {
 	{ "capacity", S_IFREG|S_IRUGO, proc_idecd_read_capacity, NULL },
 	{ NULL, 0, NULL, NULL }
 };
-#else
-# define idecd_proc	NULL
 #endif
 
 static ide_driver_t ide_cdrom_driver = {
@@ -3355,7 +3357,9 @@ static ide_driver_t ide_cdrom_driver = {
 	.end_request		= ide_end_request,
 	.error			= __ide_error,
 	.abort			= __ide_abort,
+#ifdef CONFIG_IDE_PROC_FS
 	.proc			= idecd_proc,
+#endif
 };
 
 static int idecd_open(struct inode * inode, struct file * file)
@@ -3517,7 +3521,7 @@ static int ide_cd_probe(ide_drive_t *drive)
 
 	ide_init_disk(g, drive);
 
-	ide_register_subdriver(drive, &ide_cdrom_driver);
+	ide_proc_register_driver(drive, &ide_cdrom_driver);
 
 	kref_init(&info->kref);
 
@@ -3534,7 +3538,7 @@ static int ide_cd_probe(ide_drive_t *drive)
 	g->flags = GENHD_FL_CD | GENHD_FL_REMOVABLE;
 	if (ide_cdrom_setup(drive)) {
 		struct cdrom_device_info *devinfo = &info->devinfo;
-		ide_unregister_subdriver(drive, &ide_cdrom_driver);
+		ide_proc_unregister_driver(drive, &ide_cdrom_driver);
 		kfree(info->buffer);
 		kfree(info->toc);
 		kfree(info->changer_info);
diff --git a/drivers/ide/ide-disk.c b/drivers/ide/ide-disk.c
index 37aa6ddd9702..7fff773f2df7 100644
--- a/drivers/ide/ide-disk.c
+++ b/drivers/ide/ide-disk.c
@@ -559,8 +559,7 @@ static sector_t idedisk_capacity (ide_drive_t *drive)
 	return drive->capacity64 - drive->sect0;
 }
 
-#ifdef CONFIG_PROC_FS
-
+#ifdef CONFIG_IDE_PROC_FS
 static int smart_enable(ide_drive_t *drive)
 {
 	ide_task_t args;
@@ -678,12 +677,7 @@ static ide_proc_entry_t idedisk_proc[] = {
 	{ "smart_thresholds",	S_IFREG|S_IRUSR,	proc_idedisk_read_smart_thresholds,	NULL },
 	{ NULL, 0, NULL, NULL }
 };
-
-#else
-
-#define	idedisk_proc	NULL
-
-#endif	/* CONFIG_PROC_FS */
+#endif	/* CONFIG_IDE_PROC_FS */
 
 static void idedisk_prepare_flush(request_queue_t *q, struct request *rq)
 {
@@ -737,6 +731,9 @@ static int set_multcount(ide_drive_t *drive, int arg)
 {
 	struct request rq;
 
+	if (arg < 0 || arg > drive->id->max_multsect)
+		return -EINVAL;
+
 	if (drive->special.b.set_multmode)
 		return -EBUSY;
 	ide_init_drive_cmd (&rq);
@@ -749,6 +746,9 @@ static int set_multcount(ide_drive_t *drive, int arg)
 
 static int set_nowerr(ide_drive_t *drive, int arg)
 {
+	if (arg < 0 || arg > 1)
+		return -EINVAL;
+
 	if (ide_spin_wait_hwgroup(drive))
 		return -EBUSY;
 	drive->nowerr = arg;
@@ -800,6 +800,9 @@ static int write_cache(ide_drive_t *drive, int arg)
 	ide_task_t args;
 	int err = 1;
 
+	if (arg < 0 || arg > 1)
+		return -EINVAL;
+
 	if (ide_id_has_flush_cache(drive->id)) {
 		memset(&args, 0, sizeof(ide_task_t));
 		args.tfRegister[IDE_FEATURE_OFFSET]	= (arg) ?
@@ -835,6 +838,9 @@ static int set_acoustic (ide_drive_t *drive, int arg)
 {
 	ide_task_t args;
 
+	if (arg < 0 || arg > 254)
+		return -EINVAL;
+
 	memset(&args, 0, sizeof(ide_task_t));
 	args.tfRegister[IDE_FEATURE_OFFSET]	= (arg) ? SETFEATURES_EN_AAM :
 							  SETFEATURES_DIS_AAM;
@@ -855,6 +861,9 @@ static int set_acoustic (ide_drive_t *drive, int arg)
  */
 static int set_lba_addressing(ide_drive_t *drive, int arg)
 {
+	if (arg < 0 || arg > 2)
+		return -EINVAL;
+
 	drive->addressing =  0;
 
 	if (HWIF(drive)->no_lba48)
@@ -866,23 +875,27 @@ static int set_lba_addressing(ide_drive_t *drive, int arg)
 	return 0;
 }
 
+#ifdef CONFIG_IDE_PROC_FS
 static void idedisk_add_settings(ide_drive_t *drive)
 {
 	struct hd_driveid *id = drive->id;
 
-	ide_add_setting(drive,	"bios_cyl",		SETTING_RW,					-1,			-1,			TYPE_INT,	0,	65535,				1,	1,	&drive->bios_cyl,		NULL);
-	ide_add_setting(drive,	"bios_head",		SETTING_RW,					-1,			-1,			TYPE_BYTE,	0,	255,				1,	1,	&drive->bios_head,		NULL);
-	ide_add_setting(drive,	"bios_sect",		SETTING_RW,					-1,			-1,			TYPE_BYTE,	0,	63,				1,	1,	&drive->bios_sect,		NULL);
-	ide_add_setting(drive,	"address",		SETTING_RW,					HDIO_GET_ADDRESS,	HDIO_SET_ADDRESS,	TYPE_INTA,	0,	2,				1,	1,	&drive->addressing,	set_lba_addressing);
-	ide_add_setting(drive,	"bswap",		SETTING_READ,					-1,			-1,			TYPE_BYTE,	0,	1,				1,	1,	&drive->bswap,			NULL);
-	ide_add_setting(drive,	"multcount",		id ? SETTING_RW : SETTING_READ,			HDIO_GET_MULTCOUNT,	HDIO_SET_MULTCOUNT,	TYPE_BYTE,	0,	id ? id->max_multsect : 0,	1,	1,	&drive->mult_count,		set_multcount);
-	ide_add_setting(drive,	"nowerr",		SETTING_RW,					HDIO_GET_NOWERR,	HDIO_SET_NOWERR,	TYPE_BYTE,	0,	1,				1,	1,	&drive->nowerr,			set_nowerr);
-	ide_add_setting(drive,	"lun",			SETTING_RW,					-1,			-1,			TYPE_INT,	0,	7,				1,	1,	&drive->lun,			NULL);
-	ide_add_setting(drive,	"wcache",		SETTING_RW,					HDIO_GET_WCACHE,	HDIO_SET_WCACHE,	TYPE_BYTE,	0,	1,				1,	1,	&drive->wcache,			write_cache);
-	ide_add_setting(drive,	"acoustic",		SETTING_RW,					HDIO_GET_ACOUSTIC,	HDIO_SET_ACOUSTIC,	TYPE_BYTE,	0,	254,				1,	1,	&drive->acoustic,		set_acoustic);
- 	ide_add_setting(drive,	"failures",		SETTING_RW,					-1,			-1,			TYPE_INT,	0,	65535,				1,	1,	&drive->failures,		NULL);
- 	ide_add_setting(drive,	"max_failures",		SETTING_RW,					-1,			-1,			TYPE_INT,	0,	65535,				1,	1,	&drive->max_failures,		NULL);
+	ide_add_setting(drive,	"bios_cyl",	SETTING_RW,	TYPE_INT,	0,	65535,			1,	1,	&drive->bios_cyl,	NULL);
+	ide_add_setting(drive,	"bios_head",	SETTING_RW,	TYPE_BYTE,	0,	255,			1,	1,	&drive->bios_head,	NULL);
+	ide_add_setting(drive,	"bios_sect",	SETTING_RW,	TYPE_BYTE,	0,	63,			1,	1,	&drive->bios_sect,	NULL);
+	ide_add_setting(drive,	"address",	SETTING_RW,	TYPE_BYTE,	0,	2,			1,	1,	&drive->addressing,	set_lba_addressing);
+	ide_add_setting(drive,	"bswap",	SETTING_READ,	TYPE_BYTE,	0,	1,			1,	1,	&drive->bswap,		NULL);
+	ide_add_setting(drive,	"multcount",	SETTING_RW,	TYPE_BYTE,	0,	id->max_multsect,	1,	1,	&drive->mult_count,	set_multcount);
+	ide_add_setting(drive,	"nowerr",	SETTING_RW,	TYPE_BYTE,	0,	1,			1,	1,	&drive->nowerr,		set_nowerr);
+	ide_add_setting(drive,	"lun",		SETTING_RW,	TYPE_INT,	0,	7,			1,	1,	&drive->lun,		NULL);
+	ide_add_setting(drive,	"wcache",	SETTING_RW,	TYPE_BYTE,	0,	1,			1,	1,	&drive->wcache,		write_cache);
+	ide_add_setting(drive,	"acoustic",	SETTING_RW,	TYPE_BYTE,	0,	254,			1,	1,	&drive->acoustic,	set_acoustic);
+ 	ide_add_setting(drive,	"failures",	SETTING_RW,	TYPE_INT,	0,	65535,			1,	1,	&drive->failures,	NULL);
+ 	ide_add_setting(drive,	"max_failures",	SETTING_RW,	TYPE_INT,	0,	65535,			1,	1,	&drive->max_failures,	NULL);
 }
+#else
+static inline void idedisk_add_settings(ide_drive_t *drive) { ; }
+#endif
 
 static void idedisk_setup (ide_drive_t *drive)
 {
@@ -1001,7 +1014,7 @@ static void ide_disk_remove(ide_drive_t *drive)
 	struct ide_disk_obj *idkp = drive->driver_data;
 	struct gendisk *g = idkp->disk;
 
-	ide_unregister_subdriver(drive, idkp->driver);
+	ide_proc_unregister_driver(drive, idkp->driver);
 
 	del_gendisk(g);
 
@@ -1066,7 +1079,9 @@ static ide_driver_t idedisk_driver = {
 	.end_request		= ide_end_request,
 	.error			= __ide_error,
 	.abort			= __ide_abort,
+#ifdef CONFIG_IDE_PROC_FS
 	.proc			= idedisk_proc,
+#endif
 };
 
 static int idedisk_open(struct inode *inode, struct file *filp)
@@ -1140,9 +1155,49 @@ static int idedisk_getgeo(struct block_device *bdev, struct hd_geometry *geo)
 static int idedisk_ioctl(struct inode *inode, struct file *file,
 			unsigned int cmd, unsigned long arg)
 {
+	unsigned long flags;
 	struct block_device *bdev = inode->i_bdev;
 	struct ide_disk_obj *idkp = ide_disk_g(bdev->bd_disk);
-	return generic_ide_ioctl(idkp->drive, file, bdev, cmd, arg);
+	ide_drive_t *drive = idkp->drive;
+	int err, (*setfunc)(ide_drive_t *, int);
+	u8 *val;
+
+	switch (cmd) {
+	case HDIO_GET_ADDRESS:	 val = &drive->addressing;	goto read_val;
+	case HDIO_GET_MULTCOUNT: val = &drive->mult_count;	goto read_val;
+	case HDIO_GET_NOWERR:	 val = &drive->nowerr;		goto read_val;
+	case HDIO_GET_WCACHE:	 val = &drive->wcache;		goto read_val;
+	case HDIO_GET_ACOUSTIC:	 val = &drive->acoustic;	goto read_val;
+	case HDIO_SET_ADDRESS:	 setfunc = set_lba_addressing;	goto set_val;
+	case HDIO_SET_MULTCOUNT: setfunc = set_multcount;	goto set_val;
+	case HDIO_SET_NOWERR:	 setfunc = set_nowerr;		goto set_val;
+	case HDIO_SET_WCACHE:	 setfunc = write_cache;		goto set_val;
+	case HDIO_SET_ACOUSTIC:	 setfunc = set_acoustic;	goto set_val;
+	}
+
+	return generic_ide_ioctl(drive, file, bdev, cmd, arg);
+
+read_val:
+	down(&ide_setting_sem);
+	spin_lock_irqsave(&ide_lock, flags);
+	err = *val;
+	spin_unlock_irqrestore(&ide_lock, flags);
+	up(&ide_setting_sem);
+	return err >= 0 ? put_user(err, (long __user *)arg) : err;
+
+set_val:
+	if (bdev != bdev->bd_contains)
+		err = -EINVAL;
+	else {
+		if (!capable(CAP_SYS_ADMIN))
+			err = -EACCES;
+		else {
+			down(&ide_setting_sem);
+			err = setfunc(drive, arg);
+			up(&ide_setting_sem);
+		}
+	}
+	return err;
 }
 
 static int idedisk_media_changed(struct gendisk *disk)
@@ -1202,7 +1257,7 @@ static int ide_disk_probe(ide_drive_t *drive)
 
 	ide_init_disk(g, drive);
 
-	ide_register_subdriver(drive, &idedisk_driver);
+	ide_proc_register_driver(drive, &idedisk_driver);
 
 	kref_init(&idkp->kref);
 
diff --git a/drivers/ide/ide-dma.c b/drivers/ide/ide-dma.c
index fd213088b06b..5fe85191d49c 100644
--- a/drivers/ide/ide-dma.c
+++ b/drivers/ide/ide-dma.c
@@ -705,6 +705,100 @@ int ide_use_dma(ide_drive_t *drive)
 
 EXPORT_SYMBOL_GPL(ide_use_dma);
 
+static const u8 xfer_mode_bases[] = {
+	XFER_UDMA_0,
+	XFER_MW_DMA_0,
+	XFER_SW_DMA_0,
+};
+
+static unsigned int ide_get_mode_mask(ide_drive_t *drive, u8 base)
+{
+	struct hd_driveid *id = drive->id;
+	ide_hwif_t *hwif = drive->hwif;
+	unsigned int mask = 0;
+
+	switch(base) {
+	case XFER_UDMA_0:
+		if ((id->field_valid & 4) == 0)
+			break;
+
+		mask = id->dma_ultra & hwif->ultra_mask;
+
+		if (hwif->udma_filter)
+			mask &= hwif->udma_filter(drive);
+
+		if ((mask & 0x78) && (eighty_ninty_three(drive) == 0))
+			mask &= 0x07;
+		break;
+	case XFER_MW_DMA_0:
+		mask = id->dma_mword & hwif->mwdma_mask;
+		break;
+	case XFER_SW_DMA_0:
+		mask = id->dma_1word & hwif->swdma_mask;
+		break;
+	default:
+		BUG();
+		break;
+	}
+
+	return mask;
+}
+
+/**
+ *	ide_max_dma_mode	-	compute DMA speed
+ *	@drive: IDE device
+ *
+ *	Checks the drive capabilities and returns the speed to use
+ *	for the DMA transfer.  Returns 0 if the drive is incapable
+ *	of DMA transfers.
+ */
+
+u8 ide_max_dma_mode(ide_drive_t *drive)
+{
+	ide_hwif_t *hwif = drive->hwif;
+	unsigned int mask;
+	int x, i;
+	u8 mode = 0;
+
+	if (drive->media != ide_disk && hwif->atapi_dma == 0)
+		return 0;
+
+	for (i = 0; i < ARRAY_SIZE(xfer_mode_bases); i++) {
+		mask = ide_get_mode_mask(drive, xfer_mode_bases[i]);
+		x = fls(mask) - 1;
+		if (x >= 0) {
+			mode = xfer_mode_bases[i] + x;
+			break;
+		}
+	}
+
+	printk(KERN_DEBUG "%s: selected mode 0x%x\n", drive->name, mode);
+
+	return mode;
+}
+
+EXPORT_SYMBOL_GPL(ide_max_dma_mode);
+
+int ide_tune_dma(ide_drive_t *drive)
+{
+	u8 speed;
+
+	/* TODO: use only ide_max_dma_mode() */
+	if (!ide_use_dma(drive))
+		return 0;
+
+	speed = ide_max_dma_mode(drive);
+
+	if (!speed)
+		return 0;
+
+	drive->hwif->speedproc(drive, speed);
+
+	return ide_dma_enable(drive);
+}
+
+EXPORT_SYMBOL_GPL(ide_tune_dma);
+
 void ide_dma_verbose(ide_drive_t *drive)
 {
 	struct hd_driveid *id	= drive->id;
diff --git a/drivers/ide/ide-floppy.c b/drivers/ide/ide-floppy.c
index 57cd21c5b2c1..f429be88c4f9 100644
--- a/drivers/ide/ide-floppy.c
+++ b/drivers/ide/ide-floppy.c
@@ -1811,18 +1811,22 @@ static int idefloppy_identify_device (ide_drive_t *drive,struct hd_driveid *id)
 	return 0;
 }
 
+#ifdef CONFIG_IDE_PROC_FS
 static void idefloppy_add_settings(ide_drive_t *drive)
 {
 	idefloppy_floppy_t *floppy = drive->driver_data;
 
 /*
- *			drive	setting name	read/write	ioctl	ioctl		data type	min	max	mul_factor	div_factor	data pointer		set function
+ *			drive	setting name	read/write	data type	min	max	mul_factor	div_factor	data pointer		set function
  */
-	ide_add_setting(drive,	"bios_cyl",		SETTING_RW,					-1,			-1,			TYPE_INT,	0,	1023,				1,	1,	&drive->bios_cyl,		NULL);
-	ide_add_setting(drive,	"bios_head",		SETTING_RW,					-1,			-1,			TYPE_BYTE,	0,	255,				1,	1,	&drive->bios_head,		NULL);
-	ide_add_setting(drive,	"bios_sect",		SETTING_RW,					-1,			-1,			TYPE_BYTE,	0,	63,				1,	1,	&drive->bios_sect,		NULL);
-	ide_add_setting(drive,	"ticks",		SETTING_RW,					-1,			-1,			TYPE_BYTE,	0,	255,				1,	1,	&floppy->ticks,		NULL);
+	ide_add_setting(drive,	"bios_cyl",	SETTING_RW,	TYPE_INT,	0,	1023,		1,		1,	&drive->bios_cyl,	NULL);
+	ide_add_setting(drive,	"bios_head",	SETTING_RW,	TYPE_BYTE,	0,	255,		1,		1,	&drive->bios_head,	NULL);
+	ide_add_setting(drive,	"bios_sect",	SETTING_RW,	TYPE_BYTE,	0,	63,		1,		1,	&drive->bios_sect,	NULL);
+	ide_add_setting(drive,	"ticks",	SETTING_RW,	TYPE_BYTE,	0,	255,		1,		1,	&floppy->ticks,		NULL);
 }
+#else
+static inline void idefloppy_add_settings(ide_drive_t *drive) { ; }
+#endif
 
 /*
  *	Driver initialization.
@@ -1873,7 +1877,7 @@ static void ide_floppy_remove(ide_drive_t *drive)
 	idefloppy_floppy_t *floppy = drive->driver_data;
 	struct gendisk *g = floppy->disk;
 
-	ide_unregister_subdriver(drive, floppy->driver);
+	ide_proc_unregister_driver(drive, floppy->driver);
 
 	del_gendisk(g);
 
@@ -1892,8 +1896,7 @@ static void ide_floppy_release(struct kref *kref)
 	kfree(floppy);
 }
 
-#ifdef CONFIG_PROC_FS
-
+#ifdef CONFIG_IDE_PROC_FS
 static int proc_idefloppy_read_capacity
 	(char *page, char **start, off_t off, int count, int *eof, void *data)
 {
@@ -1909,12 +1912,7 @@ static ide_proc_entry_t idefloppy_proc[] = {
 	{ "geometry",	S_IFREG|S_IRUGO,	proc_ide_read_geometry,	NULL },
 	{ NULL, 0, NULL, NULL }
 };
-
-#else
-
-#define	idefloppy_proc	NULL
-
-#endif	/* CONFIG_PROC_FS */
+#endif	/* CONFIG_IDE_PROC_FS */
 
 static int ide_floppy_probe(ide_drive_t *);
 
@@ -1933,7 +1931,9 @@ static ide_driver_t idefloppy_driver = {
 	.end_request		= idefloppy_do_end_request,
 	.error			= __ide_error,
 	.abort			= __ide_abort,
+#ifdef CONFIG_IDE_PROC_FS
 	.proc			= idefloppy_proc,
+#endif
 };
 
 static int idefloppy_open(struct inode *inode, struct file *filp)
@@ -2159,7 +2159,7 @@ static int ide_floppy_probe(ide_drive_t *drive)
 
 	ide_init_disk(g, drive);
 
-	ide_register_subdriver(drive, &idefloppy_driver);
+	ide_proc_register_driver(drive, &idefloppy_driver);
 
 	kref_init(&floppy->kref);
 
diff --git a/drivers/ide/ide-generic.c b/drivers/ide/ide-generic.c
index 99fd56151131..0f72b98d727f 100644
--- a/drivers/ide/ide-generic.c
+++ b/drivers/ide/ide-generic.c
@@ -22,8 +22,6 @@ static int __init ide_generic_init(void)
 	if (ide_hwifs[0].io_ports[IDE_DATA_OFFSET])
 		ide_release_lock();	/* for atari only */
 
-	create_proc_ide_interfaces();
-
 	return 0;
 }
 
diff --git a/drivers/ide/ide-io.c b/drivers/ide/ide-io.c
index 8670112f1d39..8e568143d90d 100644
--- a/drivers/ide/ide-io.c
+++ b/drivers/ide/ide-io.c
@@ -172,15 +172,6 @@ static ide_startstop_t ide_start_power_step(ide_drive_t *drive, struct request *
 
 	memset(args, 0, sizeof(*args));
 
-	if (drive->media != ide_disk) {
-		/*
-		 * skip idedisk_pm_restore_pio and idedisk_pm_idle for ATAPI
-		 * devices
-		 */
-		if (pm->pm_step == idedisk_pm_restore_pio)
-			pm->pm_step = ide_pm_restore_dma;
-	}
-
 	switch (pm->pm_step) {
 	case ide_pm_flush_cache:	/* Suspend step 1 (flush cache) */
 		if (drive->media != ide_disk)
@@ -207,7 +198,13 @@ static ide_startstop_t ide_start_power_step(ide_drive_t *drive, struct request *
 	case idedisk_pm_restore_pio:	/* Resume step 1 (restore PIO) */
 		if (drive->hwif->tuneproc != NULL)
 			drive->hwif->tuneproc(drive, 255);
-		ide_complete_power_step(drive, rq, 0, 0);
+		/*
+		 * skip idedisk_pm_idle for ATAPI devices
+		 */
+		if (drive->media != ide_disk)
+			pm->pm_step = ide_pm_restore_dma;
+		else
+			ide_complete_power_step(drive, rq, 0, 0);
 		return ide_stopped;
 
 	case idedisk_pm_idle:		/* Resume step 2 (idle) */
diff --git a/drivers/ide/ide-iops.c b/drivers/ide/ide-iops.c
index 3caa176b3155..f0be5f665a0e 100644
--- a/drivers/ide/ide-iops.c
+++ b/drivers/ide/ide-iops.c
@@ -571,51 +571,54 @@ EXPORT_SYMBOL(ide_wait_stat);
  */
 u8 eighty_ninty_three (ide_drive_t *drive)
 {
-	if(HWIF(drive)->udma_four == 0)
-		return 0;
+	ide_hwif_t *hwif = drive->hwif;
+	struct hd_driveid *id = drive->id;
+
+	if (hwif->udma_four == 0)
+		goto no_80w;
 
 	/* Check for SATA but only if we are ATA5 or higher */
-	if (drive->id->hw_config == 0 && (drive->id->major_rev_num & 0x7FE0))
+	if (id->hw_config == 0 && (id->major_rev_num & 0x7FE0))
 		return 1;
-	if (!(drive->id->hw_config & 0x6000))
-		return 0;
-#ifndef CONFIG_IDEDMA_IVB
-	if(!(drive->id->hw_config & 0x4000))
-		return 0;
-#endif /* CONFIG_IDEDMA_IVB */
+
 	/*
 	 * FIXME:
 	 * - change master/slave IDENTIFY order
 	 * - force bit13 (80c cable present) check
 	 *   (unless the slave device is pre-ATA3)
 	 */
-	return 1;
-}
+#ifndef CONFIG_IDEDMA_IVB
+	if (id->hw_config & 0x4000)
+#else
+	if (id->hw_config & 0x6000)
+#endif
+		return 1;
+
+no_80w:
+	if (drive->udma33_warned == 1)
+		return 0;
 
-EXPORT_SYMBOL(eighty_ninty_three);
+	printk(KERN_WARNING "%s: %s side 80-wire cable detection failed, "
+			    "limiting max speed to UDMA33\n",
+			    drive->name, hwif->udma_four ? "drive" : "host");
+
+	drive->udma33_warned = 1;
+
+	return 0;
+}
 
 int ide_ata66_check (ide_drive_t *drive, ide_task_t *args)
 {
 	if ((args->tfRegister[IDE_COMMAND_OFFSET] == WIN_SETFEATURES) &&
 	    (args->tfRegister[IDE_SECTOR_OFFSET] > XFER_UDMA_2) &&
 	    (args->tfRegister[IDE_FEATURE_OFFSET] == SETFEATURES_XFER)) {
-#ifndef CONFIG_IDEDMA_IVB
-		if ((drive->id->hw_config & 0x6000) == 0) {
-#else /* !CONFIG_IDEDMA_IVB */
-		if (((drive->id->hw_config & 0x2000) == 0) ||
-		    ((drive->id->hw_config & 0x4000) == 0)) {
-#endif /* CONFIG_IDEDMA_IVB */
-			printk("%s: Speed warnings UDMA 3/4/5 is not "
-				"functional.\n", drive->name);
-			return 1;
-		}
-		if (!HWIF(drive)->udma_four) {
-			printk("%s: Speed warnings UDMA 3/4/5 is not "
-				"functional.\n",
-				HWIF(drive)->name);
+		if (eighty_ninty_three(drive) == 0) {
+			printk(KERN_WARNING "%s: UDMA speeds >UDMA33 cannot "
+					    "be set\n", drive->name);
 			return 1;
 		}
 	}
+
 	return 0;
 }
 
diff --git a/drivers/ide/ide-lib.c b/drivers/ide/ide-lib.c
index 68719314df3f..3be3c69383f2 100644
--- a/drivers/ide/ide-lib.c
+++ b/drivers/ide/ide-lib.c
@@ -69,123 +69,41 @@ char *ide_xfer_verbose (u8 xfer_rate)
 EXPORT_SYMBOL(ide_xfer_verbose);
 
 /**
- *	ide_dma_speed	-	compute DMA speed
- *	@drive: drive
- *	@mode:	modes available
- *
- *	Checks the drive capabilities and returns the speed to use
- *	for the DMA transfer.  Returns 0 if the drive is incapable
- *	of DMA transfers.
- */
- 
-u8 ide_dma_speed(ide_drive_t *drive, u8 mode)
-{
-	struct hd_driveid *id   = drive->id;
-	ide_hwif_t *hwif	= HWIF(drive);
-	u8 ultra_mask, mwdma_mask, swdma_mask;
-	u8 speed = 0;
-
-	if (drive->media != ide_disk && hwif->atapi_dma == 0)
-		return 0;
-
-	/* Capable of UltraDMA modes? */
-	ultra_mask = id->dma_ultra & hwif->ultra_mask;
-
-	if (!(id->field_valid & 4))
-		mode = 0;	/* fallback to MW/SW DMA if no UltraDMA */
-
-	switch (mode) {
-	case 4:
-		if (ultra_mask & 0x40) {
-			speed = XFER_UDMA_6;
-			break;
-		}
-	case 3:
-		if (ultra_mask & 0x20) {
-			speed = XFER_UDMA_5;
-			break;
-		}
-	case 2:
-		if (ultra_mask & 0x10) {
-			speed = XFER_UDMA_4;
-			break;
-		}
-		if (ultra_mask & 0x08) {
-			speed = XFER_UDMA_3;
-			break;
-		}
-	case 1:
-		if (ultra_mask & 0x04) {
-			speed = XFER_UDMA_2;
-			break;
-		}
-		if (ultra_mask & 0x02) {
-			speed = XFER_UDMA_1;
-			break;
-		}
-		if (ultra_mask & 0x01) {
-			speed = XFER_UDMA_0;
-			break;
-		}
-	case 0:
-		mwdma_mask = id->dma_mword & hwif->mwdma_mask;
-
-		if (mwdma_mask & 0x04) {
-			speed = XFER_MW_DMA_2;
-			break;
-		}
-		if (mwdma_mask & 0x02) {
-			speed = XFER_MW_DMA_1;
-			break;
-		}
-		if (mwdma_mask & 0x01) {
-			speed = XFER_MW_DMA_0;
-			break;
-		}
-
-		swdma_mask = id->dma_1word & hwif->swdma_mask;
-
-		if (swdma_mask & 0x04) {
-			speed = XFER_SW_DMA_2;
-			break;
-		}
-		if (swdma_mask & 0x02) {
-			speed = XFER_SW_DMA_1;
-			break;
-		}
-		if (swdma_mask & 0x01) {
-			speed = XFER_SW_DMA_0;
-			break;
-		}
-	}
-
-	return speed;
-}
-EXPORT_SYMBOL(ide_dma_speed);
-
-
-/**
- *	ide_rate_filter		-	return best speed for mode
- *	@mode: modes available
+ *	ide_rate_filter		-	filter transfer mode
+ *	@drive: IDE device
  *	@speed: desired speed
  *
- *	Given the available DMA/UDMA mode this function returns
+ *	Given the available transfer modes this function returns
  *	the best available speed at or below the speed requested.
+ *
+ *	FIXME: filter also PIO/SWDMA/MWDMA modes
  */
 
-u8 ide_rate_filter (u8 mode, u8 speed) 
+u8 ide_rate_filter(ide_drive_t *drive, u8 speed)
 {
 #ifdef CONFIG_BLK_DEV_IDEDMA
-	static u8 speed_max[] = {
-		XFER_MW_DMA_2, XFER_UDMA_2, XFER_UDMA_4,
-		XFER_UDMA_5, XFER_UDMA_6
-	};
+	ide_hwif_t *hwif = drive->hwif;
+	u8 mask = hwif->ultra_mask, mode = XFER_MW_DMA_2;
+
+	if (hwif->udma_filter)
+		mask = hwif->udma_filter(drive);
+
+	/*
+	 * TODO: speed > XFER_UDMA_2 extra check is needed to avoid false
+	 * cable warning from eighty_ninty_three(), moving ide_rate_filter()
+	 * calls from ->speedproc to core code will make this hack go away
+	 */
+	if (speed > XFER_UDMA_2) {
+		if ((mask & 0x78) && (eighty_ninty_three(drive) == 0))
+			mask &= 0x07;
+	}
+
+	if (mask)
+		mode = fls(mask) - 1 + XFER_UDMA_0;
 
 //	printk("%s: mode 0x%02x, speed 0x%02x\n", __FUNCTION__, mode, speed);
 
-	/* So that we remember to update this if new modes appear */
-	BUG_ON(mode > 4);
-	return min(speed, speed_max[mode]);
+	return min(speed, mode);
 #else /* !CONFIG_BLK_DEV_IDEDMA */
 	return min(speed, (u8)XFER_PIO_4);
 #endif /* CONFIG_BLK_DEV_IDEDMA */
diff --git a/drivers/ide/ide-pnp.c b/drivers/ide/ide-pnp.c
index 98410ca044cf..2b8009c50e91 100644
--- a/drivers/ide/ide-pnp.c
+++ b/drivers/ide/ide-pnp.c
@@ -42,7 +42,7 @@ static int idepnp_probe(struct pnp_dev * dev, const struct pnp_device_id *dev_id
 	hw.irq = pnp_irq(dev, 0);
 	hw.dma = NO_DMA;
 
-	index = ide_register_hw(&hw, &hwif);
+	index = ide_register_hw(&hw, 1, &hwif);
 
 	if (index != -1) {
 	    	printk(KERN_INFO "ide%d: generic PnP IDE interface\n", index);
diff --git a/drivers/ide/ide-probe.c b/drivers/ide/ide-probe.c
index 8f15c23aa70d..3cebed77f55d 100644
--- a/drivers/ide/ide-probe.c
+++ b/drivers/ide/ide-probe.c
@@ -1427,6 +1427,9 @@ int ideprobe_init (void)
 				}
 		}
 	}
+	for (index = 0; index < MAX_HWIFS; ++index)
+		if (probe[index])
+			ide_proc_register_port(&ide_hwifs[index]);
 	return 0;
 }
 
diff --git a/drivers/ide/ide-proc.c b/drivers/ide/ide-proc.c
index a9e0b30fb1f2..d50bd996ff22 100644
--- a/drivers/ide/ide-proc.c
+++ b/drivers/ide/ide-proc.c
@@ -3,6 +3,8 @@
  *
  *  Copyright (C) 1997-1998	Mark Lord
  *  Copyright (C) 2003		Red Hat <alan@redhat.com>
+ *
+ *  Some code was moved here from ide.c, see it for original copyrights.
  */
 
 /*
@@ -37,6 +39,8 @@
 
 #include <asm/io.h>
 
+static struct proc_dir_entry *proc_ide_root;
+
 static int proc_ide_read_imodel
 	(char *page, char **start, off_t off, int count, int *eof, void *data)
 {
@@ -121,6 +125,265 @@ static int proc_ide_read_identify
 	PROC_IDE_READ_RETURN(page,start,off,count,eof,len);
 }
 
+/**
+ *	__ide_add_setting	-	add an ide setting option
+ *	@drive: drive to use
+ *	@name: setting name
+ *	@rw: true if the function is read write
+ *	@data_type: type of data
+ *	@min: range minimum
+ *	@max: range maximum
+ *	@mul_factor: multiplication scale
+ *	@div_factor: divison scale
+ *	@data: private data field
+ *	@set: setting
+ *	@auto_remove: setting auto removal flag
+ *
+ *	Removes the setting named from the device if it is present.
+ *	The function takes the settings_lock to protect against
+ *	parallel changes. This function must not be called from IRQ
+ *	context. Returns 0 on success or -1 on failure.
+ *
+ *	BUGS: This code is seriously over-engineered. There is also
+ *	magic about how the driver specific features are setup. If
+ *	a driver is attached we assume the driver settings are auto
+ *	remove.
+ */
+
+static int __ide_add_setting(ide_drive_t *drive, const char *name, int rw, int data_type, int min, int max, int mul_factor, int div_factor, void *data, ide_procset_t *set, int auto_remove)
+{
+	ide_settings_t **p = (ide_settings_t **) &drive->settings, *setting = NULL;
+
+	down(&ide_setting_sem);
+	while ((*p) && strcmp((*p)->name, name) < 0)
+		p = &((*p)->next);
+	if ((setting = kzalloc(sizeof(*setting), GFP_KERNEL)) == NULL)
+		goto abort;
+	if ((setting->name = kmalloc(strlen(name) + 1, GFP_KERNEL)) == NULL)
+		goto abort;
+	strcpy(setting->name, name);
+	setting->rw = rw;
+	setting->data_type = data_type;
+	setting->min = min;
+	setting->max = max;
+	setting->mul_factor = mul_factor;
+	setting->div_factor = div_factor;
+	setting->data = data;
+	setting->set = set;
+
+	setting->next = *p;
+	if (auto_remove)
+		setting->auto_remove = 1;
+	*p = setting;
+	up(&ide_setting_sem);
+	return 0;
+abort:
+	up(&ide_setting_sem);
+	kfree(setting);
+	return -1;
+}
+
+int ide_add_setting(ide_drive_t *drive, const char *name, int rw, int data_type, int min, int max, int mul_factor, int div_factor, void *data, ide_procset_t *set)
+{
+	return __ide_add_setting(drive, name, rw, data_type, min, max, mul_factor, div_factor, data, set, 1);
+}
+
+EXPORT_SYMBOL(ide_add_setting);
+
+/**
+ *	__ide_remove_setting	-	remove an ide setting option
+ *	@drive: drive to use
+ *	@name: setting name
+ *
+ *	Removes the setting named from the device if it is present.
+ *	The caller must hold the setting semaphore.
+ */
+
+static void __ide_remove_setting (ide_drive_t *drive, char *name)
+{
+	ide_settings_t **p, *setting;
+
+	p = (ide_settings_t **) &drive->settings;
+
+	while ((*p) && strcmp((*p)->name, name))
+		p = &((*p)->next);
+	if ((setting = (*p)) == NULL)
+		return;
+
+	(*p) = setting->next;
+
+	kfree(setting->name);
+	kfree(setting);
+}
+
+/**
+ *	auto_remove_settings	-	remove driver specific settings
+ *	@drive: drive
+ *
+ *	Automatically remove all the driver specific settings for this
+ *	drive. This function may not be called from IRQ context. The
+ *	caller must hold ide_setting_sem.
+ */
+
+static void auto_remove_settings (ide_drive_t *drive)
+{
+	ide_settings_t *setting;
+repeat:
+	setting = drive->settings;
+	while (setting) {
+		if (setting->auto_remove) {
+			__ide_remove_setting(drive, setting->name);
+			goto repeat;
+		}
+		setting = setting->next;
+	}
+}
+
+/**
+ *	ide_find_setting_by_name	-	find a drive specific setting
+ *	@drive: drive to scan
+ *	@name: setting name
+ *
+ *	Scan's the device setting table for a matching entry and returns
+ *	this or NULL if no entry is found. The caller must hold the
+ *	setting semaphore
+ */
+
+static ide_settings_t *ide_find_setting_by_name(ide_drive_t *drive, char *name)
+{
+	ide_settings_t *setting = drive->settings;
+
+	while (setting) {
+		if (strcmp(setting->name, name) == 0)
+			break;
+		setting = setting->next;
+	}
+	return setting;
+}
+
+/**
+ *	ide_read_setting	-	read an IDE setting
+ *	@drive: drive to read from
+ *	@setting: drive setting
+ *
+ *	Read a drive setting and return the value. The caller
+ *	must hold the ide_setting_sem when making this call.
+ *
+ *	BUGS: the data return and error are the same return value
+ *	so an error -EINVAL and true return of the same value cannot
+ *	be told apart
+ */
+
+static int ide_read_setting(ide_drive_t *drive, ide_settings_t *setting)
+{
+	int		val = -EINVAL;
+	unsigned long	flags;
+
+	if ((setting->rw & SETTING_READ)) {
+		spin_lock_irqsave(&ide_lock, flags);
+		switch(setting->data_type) {
+			case TYPE_BYTE:
+				val = *((u8 *) setting->data);
+				break;
+			case TYPE_SHORT:
+				val = *((u16 *) setting->data);
+				break;
+			case TYPE_INT:
+				val = *((u32 *) setting->data);
+				break;
+		}
+		spin_unlock_irqrestore(&ide_lock, flags);
+	}
+	return val;
+}
+
+/**
+ *	ide_write_setting	-	read an IDE setting
+ *	@drive: drive to read from
+ *	@setting: drive setting
+ *	@val: value
+ *
+ *	Write a drive setting if it is possible. The caller
+ *	must hold the ide_setting_sem when making this call.
+ *
+ *	BUGS: the data return and error are the same return value
+ *	so an error -EINVAL and true return of the same value cannot
+ *	be told apart
+ *
+ *	FIXME:  This should be changed to enqueue a special request
+ *	to the driver to change settings, and then wait on a sema for completion.
+ *	The current scheme of polling is kludgy, though safe enough.
+ */
+
+static int ide_write_setting(ide_drive_t *drive, ide_settings_t *setting, int val)
+{
+	if (!capable(CAP_SYS_ADMIN))
+		return -EACCES;
+	if (setting->set)
+		return setting->set(drive, val);
+	if (!(setting->rw & SETTING_WRITE))
+		return -EPERM;
+	if (val < setting->min || val > setting->max)
+		return -EINVAL;
+	if (ide_spin_wait_hwgroup(drive))
+		return -EBUSY;
+	switch (setting->data_type) {
+		case TYPE_BYTE:
+			*((u8 *) setting->data) = val;
+			break;
+		case TYPE_SHORT:
+			*((u16 *) setting->data) = val;
+			break;
+		case TYPE_INT:
+			*((u32 *) setting->data) = val;
+			break;
+	}
+	spin_unlock_irq(&ide_lock);
+	return 0;
+}
+
+static int set_xfer_rate (ide_drive_t *drive, int arg)
+{
+	int err;
+
+	if (arg < 0 || arg > 70)
+		return -EINVAL;
+
+	err = ide_wait_cmd(drive,
+			WIN_SETFEATURES, (u8) arg,
+			SETFEATURES_XFER, 0, NULL);
+
+	if (!err && arg) {
+		ide_set_xfer_rate(drive, (u8) arg);
+		ide_driveid_update(drive);
+	}
+	return err;
+}
+
+/**
+ *	ide_add_generic_settings	-	generic ide settings
+ *	@drive: drive being configured
+ *
+ *	Add the generic parts of the system settings to the /proc files.
+ *	The caller must not be holding the ide_setting_sem.
+ */
+
+void ide_add_generic_settings (ide_drive_t *drive)
+{
+/*
+ *			  drive		setting name		read/write access				data type	min	max				mul_factor	div_factor	data pointer			set function
+ */
+	__ide_add_setting(drive,	"io_32bit",		drive->no_io_32bit ? SETTING_READ : SETTING_RW,	TYPE_BYTE,	0,	1 + (SUPPORT_VLB_SYNC << 1),	1,		1,		&drive->io_32bit,		set_io_32bit,	0);
+	__ide_add_setting(drive,	"keepsettings",		SETTING_RW,					TYPE_BYTE,	0,	1,				1,		1,		&drive->keep_settings,		NULL,		0);
+	__ide_add_setting(drive,	"nice1",		SETTING_RW,					TYPE_BYTE,	0,	1,				1,		1,		&drive->nice1,			NULL,		0);
+	__ide_add_setting(drive,	"pio_mode",		SETTING_WRITE,					TYPE_BYTE,	0,	255,				1,		1,		NULL,				set_pio_mode,	0);
+	__ide_add_setting(drive,	"unmaskirq",		drive->no_unmask ? SETTING_READ : SETTING_RW,	TYPE_BYTE,	0,	1,				1,		1,		&drive->unmask,			NULL,		0);
+	__ide_add_setting(drive,	"using_dma",		SETTING_RW,					TYPE_BYTE,	0,	1,				1,		1,		&drive->using_dma,		set_using_dma,	0);
+	__ide_add_setting(drive,	"init_speed",		SETTING_RW,					TYPE_BYTE,	0,	70,				1,		1,		&drive->init_speed,		NULL,		0);
+	__ide_add_setting(drive,	"current_speed",	SETTING_RW,					TYPE_BYTE,	0,	70,				1,		1,		&drive->current_speed,		set_xfer_rate,	0);
+	__ide_add_setting(drive,	"number",		SETTING_RW,					TYPE_BYTE,	0,	3,				1,		1,		&drive->dn,			NULL,		0);
+}
+
 static void proc_ide_settings_warn(void)
 {
 	static int warned = 0;
@@ -399,7 +662,7 @@ static ide_proc_entry_t generic_drive_entries[] = {
 	{ NULL,	0, NULL, NULL }
 };
 
-void ide_add_proc_entries(struct proc_dir_entry *dir, ide_proc_entry_t *p, void *data)
+static void ide_add_proc_entries(struct proc_dir_entry *dir, ide_proc_entry_t *p, void *data)
 {
 	struct proc_dir_entry *ent;
 
@@ -415,7 +678,7 @@ void ide_add_proc_entries(struct proc_dir_entry *dir, ide_proc_entry_t *p, void
 	}
 }
 
-void ide_remove_proc_entries(struct proc_dir_entry *dir, ide_proc_entry_t *p)
+static void ide_remove_proc_entries(struct proc_dir_entry *dir, ide_proc_entry_t *p)
 {
 	if (!dir || !p)
 		return;
@@ -425,6 +688,51 @@ void ide_remove_proc_entries(struct proc_dir_entry *dir, ide_proc_entry_t *p)
 	}
 }
 
+void ide_proc_register_driver(ide_drive_t *drive, ide_driver_t *driver)
+{
+	ide_add_proc_entries(drive->proc, driver->proc, drive);
+}
+
+EXPORT_SYMBOL(ide_proc_register_driver);
+
+/**
+ *	ide_proc_unregister_driver	-	remove driver specific data
+ *	@drive: drive
+ *	@driver: driver
+ *
+ *	Clean up the driver specific /proc files and IDE settings
+ *	for a given drive.
+ *
+ *	Takes ide_setting_sem and ide_lock.
+ *	Caller must hold none of the locks.
+ */
+
+void ide_proc_unregister_driver(ide_drive_t *drive, ide_driver_t *driver)
+{
+	unsigned long flags;
+
+	ide_remove_proc_entries(drive->proc, driver->proc);
+
+	down(&ide_setting_sem);
+	spin_lock_irqsave(&ide_lock, flags);
+	/*
+	 * ide_setting_sem protects the settings list
+	 * ide_lock protects the use of settings
+	 *
+	 * so we need to hold both, ide_settings_sem because we want to
+	 * modify the settings list, and ide_lock because we cannot take
+	 * a setting out that is being used.
+	 *
+	 * OTOH both ide_{read,write}_setting are only ever used under
+	 * ide_setting_sem.
+	 */
+	auto_remove_settings(drive);
+	spin_unlock_irqrestore(&ide_lock, flags);
+	up(&ide_setting_sem);
+}
+
+EXPORT_SYMBOL(ide_proc_unregister_driver);
+
 static void create_proc_ide_drives(ide_hwif_t *hwif)
 {
 	int	d;
@@ -477,26 +785,24 @@ static ide_proc_entry_t hwif_entries[] = {
 	{ NULL,	0, NULL, NULL }
 };
 
-void create_proc_ide_interfaces(void)
+void ide_proc_register_port(ide_hwif_t *hwif)
 {
-	int	h;
+	if (!hwif->present)
+		return;
 
-	for (h = 0; h < MAX_HWIFS; h++) {
-		ide_hwif_t *hwif = &ide_hwifs[h];
+	if (!hwif->proc) {
+		hwif->proc = proc_mkdir(hwif->name, proc_ide_root);
 
-		if (!hwif->present)
-			continue;
-		if (!hwif->proc) {
-			hwif->proc = proc_mkdir(hwif->name, proc_ide_root);
-			if (!hwif->proc)
-				return;
-			ide_add_proc_entries(hwif->proc, hwif_entries, hwif);
-		}
-		create_proc_ide_drives(hwif);
+		if (!hwif->proc)
+			return;
+
+		ide_add_proc_entries(hwif->proc, hwif_entries, hwif);
 	}
+
+	create_proc_ide_drives(hwif);
 }
 
-EXPORT_SYMBOL(create_proc_ide_interfaces);
+EXPORT_SYMBOL_GPL(ide_proc_register_port);
 
 #ifdef CONFIG_BLK_DEV_IDEPCI
 void ide_pci_create_host_proc(const char *name, get_info_t *get_info)
@@ -507,7 +813,7 @@ void ide_pci_create_host_proc(const char *name, get_info_t *get_info)
 EXPORT_SYMBOL_GPL(ide_pci_create_host_proc);
 #endif
 
-void destroy_proc_ide_interface(ide_hwif_t *hwif)
+void ide_proc_unregister_port(ide_hwif_t *hwif)
 {
 	if (hwif->proc) {
 		destroy_proc_ide_drives(hwif);
@@ -554,11 +860,11 @@ void proc_ide_create(void)
 {
 	struct proc_dir_entry *entry;
 
+	proc_ide_root = proc_mkdir("ide", NULL);
+
 	if (!proc_ide_root)
 		return;
 
-	create_proc_ide_interfaces();
-
 	entry = create_proc_entry("drivers", 0, proc_ide_root);
 	if (entry)
 		entry->proc_fops = &ide_drivers_operations;
diff --git a/drivers/ide/ide-tape.c b/drivers/ide/ide-tape.c
index 4e59239fef75..e82bfa5e0ab8 100644
--- a/drivers/ide/ide-tape.c
+++ b/drivers/ide/ide-tape.c
@@ -4561,28 +4561,33 @@ static void idetape_get_blocksize_from_block_descriptor(ide_drive_t *drive)
 	printk(KERN_INFO "ide-tape: Adjusted block size - %d\n", tape->tape_block_size);
 #endif /* IDETAPE_DEBUG_INFO */
 }
+
+#ifdef CONFIG_IDE_PROC_FS
 static void idetape_add_settings (ide_drive_t *drive)
 {
 	idetape_tape_t *tape = drive->driver_data;
 
 /*
- *			drive	setting name	read/write	ioctl	ioctl		data type	min			max			mul_factor			div_factor			data pointer				set function
+ *			drive	setting name		read/write	data type	min			max			mul_factor			div_factor	data pointer				set function
  */
-	ide_add_setting(drive,	"buffer",	SETTING_READ,	-1,	-1,		TYPE_SHORT,	0,			0xffff,			1,				2,				&tape->capabilities.buffer_size,	NULL);
-	ide_add_setting(drive,	"pipeline_min",	SETTING_RW,	-1,	-1,		TYPE_INT,	1,			0xffff,			tape->stage_size / 1024,	1,				&tape->min_pipeline,			NULL);
-	ide_add_setting(drive,	"pipeline",	SETTING_RW,	-1,	-1,		TYPE_INT,	1,			0xffff,			tape->stage_size / 1024,	1,				&tape->max_stages,			NULL);
-	ide_add_setting(drive,	"pipeline_max",	SETTING_RW,	-1,	-1,		TYPE_INT,	1,			0xffff,			tape->stage_size / 1024,	1,				&tape->max_pipeline,			NULL);
-	ide_add_setting(drive,	"pipeline_used",SETTING_READ,	-1,	-1,		TYPE_INT,	0,			0xffff,			tape->stage_size / 1024,	1,				&tape->nr_stages,			NULL);
-	ide_add_setting(drive,	"pipeline_pending",SETTING_READ,-1,	-1,		TYPE_INT,	0,			0xffff,			tape->stage_size / 1024,	1,				&tape->nr_pending_stages,		NULL);
-	ide_add_setting(drive,	"speed",	SETTING_READ,	-1,	-1,		TYPE_SHORT,	0,			0xffff,			1,				1,				&tape->capabilities.speed,		NULL);
-	ide_add_setting(drive,	"stage",	SETTING_READ,	-1,	-1,		TYPE_INT,	0,			0xffff,			1,				1024,				&tape->stage_size,			NULL);
-	ide_add_setting(drive,	"tdsc",		SETTING_RW,	-1,	-1,		TYPE_INT,	IDETAPE_DSC_RW_MIN,	IDETAPE_DSC_RW_MAX,	1000,				HZ,				&tape->best_dsc_rw_frequency,		NULL);
-	ide_add_setting(drive,	"dsc_overlap",	SETTING_RW,	-1,	-1,		TYPE_BYTE,	0,			1,			1,				1,				&drive->dsc_overlap,			NULL);
-	ide_add_setting(drive,	"pipeline_head_speed_c",SETTING_READ,	-1,	-1,	TYPE_INT,	0,			0xffff,			1,				1,				&tape->controlled_pipeline_head_speed,	NULL);
-	ide_add_setting(drive,	"pipeline_head_speed_u",SETTING_READ,	-1,	-1,	TYPE_INT,	0,			0xffff,			1,				1,				&tape->uncontrolled_pipeline_head_speed,	NULL);
-	ide_add_setting(drive,	"avg_speed",	SETTING_READ,	-1,	-1,		TYPE_INT,	0,			0xffff,			1,				1,				&tape->avg_speed,		NULL);
-	ide_add_setting(drive,	"debug_level",SETTING_RW,	-1,	-1,		TYPE_INT,	0,			0xffff,			1,				1,				&tape->debug_level,		NULL);
+	ide_add_setting(drive,	"buffer",		SETTING_READ,	TYPE_SHORT,	0,			0xffff,			1,				2,		&tape->capabilities.buffer_size,	NULL);
+	ide_add_setting(drive,	"pipeline_min",		SETTING_RW,	TYPE_INT,	1,			0xffff,			tape->stage_size / 1024,	1,		&tape->min_pipeline,			NULL);
+	ide_add_setting(drive,	"pipeline",		SETTING_RW,	TYPE_INT,	1,			0xffff,			tape->stage_size / 1024,	1,		&tape->max_stages,			NULL);
+	ide_add_setting(drive,	"pipeline_max",		SETTING_RW,	TYPE_INT,	1,			0xffff,			tape->stage_size / 1024,	1,		&tape->max_pipeline,			NULL);
+	ide_add_setting(drive,	"pipeline_used",	SETTING_READ,	TYPE_INT,	0,			0xffff,			tape->stage_size / 1024,	1,		&tape->nr_stages,			NULL);
+	ide_add_setting(drive,	"pipeline_pending",	SETTING_READ,	TYPE_INT,	0,			0xffff,			tape->stage_size / 1024,	1,		&tape->nr_pending_stages,		NULL);
+	ide_add_setting(drive,	"speed",		SETTING_READ,	TYPE_SHORT,	0,			0xffff,			1,				1,		&tape->capabilities.speed,		NULL);
+	ide_add_setting(drive,	"stage",		SETTING_READ,	TYPE_INT,	0,			0xffff,			1,				1024,		&tape->stage_size,			NULL);
+	ide_add_setting(drive,	"tdsc",			SETTING_RW,	TYPE_INT,	IDETAPE_DSC_RW_MIN,	IDETAPE_DSC_RW_MAX,	1000,				HZ,		&tape->best_dsc_rw_frequency,		NULL);
+	ide_add_setting(drive,	"dsc_overlap",		SETTING_RW,	TYPE_BYTE,	0,			1,			1,				1,		&drive->dsc_overlap,			NULL);
+	ide_add_setting(drive,	"pipeline_head_speed_c",SETTING_READ,	TYPE_INT,	0,			0xffff,			1,				1,		&tape->controlled_pipeline_head_speed,	NULL);
+	ide_add_setting(drive,	"pipeline_head_speed_u",SETTING_READ,	TYPE_INT,	0,			0xffff,			1,				1,		&tape->uncontrolled_pipeline_head_speed,NULL);
+	ide_add_setting(drive,	"avg_speed",		SETTING_READ,	TYPE_INT,	0,			0xffff,			1,				1,		&tape->avg_speed,			NULL);
+	ide_add_setting(drive,	"debug_level",		SETTING_RW,	TYPE_INT,	0,			0xffff,			1,				1,		&tape->debug_level,			NULL);
 }
+#else
+static inline void idetape_add_settings(ide_drive_t *drive) { ; }
+#endif
 
 /*
  *	ide_setup is called to:
@@ -4703,7 +4708,7 @@ static void ide_tape_remove(ide_drive_t *drive)
 {
 	idetape_tape_t *tape = drive->driver_data;
 
-	ide_unregister_subdriver(drive, tape->driver);
+	ide_proc_unregister_driver(drive, tape->driver);
 
 	ide_unregister_region(tape->disk);
 
@@ -4730,8 +4735,7 @@ static void ide_tape_release(struct kref *kref)
 	kfree(tape);
 }
 
-#ifdef CONFIG_PROC_FS
-
+#ifdef CONFIG_IDE_PROC_FS
 static int proc_idetape_read_name
 	(char *page, char **start, off_t off, int count, int *eof, void *data)
 {
@@ -4749,11 +4753,6 @@ static ide_proc_entry_t idetape_proc[] = {
 	{ "name",	S_IFREG|S_IRUGO,	proc_idetape_read_name,	NULL },
 	{ NULL, 0, NULL, NULL }
 };
-
-#else
-
-#define	idetape_proc	NULL
-
 #endif
 
 static int ide_tape_probe(ide_drive_t *);
@@ -4773,7 +4772,9 @@ static ide_driver_t idetape_driver = {
 	.end_request		= idetape_end_request,
 	.error			= __ide_error,
 	.abort			= __ide_abort,
+#ifdef CONFIG_IDE_PROC_FS
 	.proc			= idetape_proc,
+#endif
 };
 
 /*
@@ -4864,7 +4865,7 @@ static int ide_tape_probe(ide_drive_t *drive)
 
 	ide_init_disk(g, drive);
 
-	ide_register_subdriver(drive, &idetape_driver);
+	ide_proc_register_driver(drive, &idetape_driver);
 
 	kref_init(&tape->kref);
 
diff --git a/drivers/ide/ide.c b/drivers/ide/ide.c
index ae5bf2be6f52..f2b547ff7722 100644
--- a/drivers/ide/ide.c
+++ b/drivers/ide/ide.c
@@ -168,12 +168,11 @@ static const u8 ide_hwif_to_major[] = { IDE0_MAJOR, IDE1_MAJOR,
 
 static int idebus_parameter;	/* holds the "idebus=" parameter */
 static int system_bus_speed;	/* holds what we think is VESA/PCI bus speed */
-static int initializing;	/* set while initializing built-in drivers */
 
 DECLARE_MUTEX(ide_cfg_sem);
  __cacheline_aligned_in_smp DEFINE_SPINLOCK(ide_lock);
 
-#ifdef CONFIG_BLK_DEV_IDEPCI
+#ifdef CONFIG_IDEPCI_PCIBUS_ORDER
 static int ide_scan_direction; /* THIS was formerly 2.2.x pci=reverse */
 #endif
 
@@ -216,9 +215,6 @@ static void init_hwif_data(ide_hwif_t *hwif, unsigned int index)
 	hwif->bus_state	= BUSSTATE_ON;
 
 	hwif->atapi_dma = 0;		/* disable all atapi dma */ 
-	hwif->ultra_mask = 0x80;	/* disable all ultra */
-	hwif->mwdma_mask = 0x80;	/* disable all mwdma */
-	hwif->swdma_mask = 0x80;	/* disable all swdma */
 
 	init_completion(&hwif->gendev_rel_comp);
 
@@ -305,9 +301,7 @@ static void __init init_ide_data (void)
 #endif
 	}
 #ifdef CONFIG_IDE_ARM
-	initializing = 1;
 	ide_arm_init();
-	initializing = 0;
 #endif
 }
 
@@ -353,10 +347,6 @@ static int ide_system_bus_speed(void)
 	return system_bus_speed;
 }
 
-#ifdef CONFIG_PROC_FS
-struct proc_dir_entry *proc_ide_root;
-#endif
-
 static struct resource* hwif_request_region(ide_hwif_t *hwif,
 					    unsigned long addr, int num)
 {
@@ -480,6 +470,7 @@ static void ide_hwif_restore(ide_hwif_t *hwif, ide_hwif_t *tmp_hwif)
 
 	hwif->tuneproc			= tmp_hwif->tuneproc;
 	hwif->speedproc			= tmp_hwif->speedproc;
+	hwif->udma_filter		= tmp_hwif->udma_filter;
 	hwif->selectproc		= tmp_hwif->selectproc;
 	hwif->reset_poll		= tmp_hwif->reset_poll;
 	hwif->pre_reset			= tmp_hwif->pre_reset;
@@ -599,7 +590,7 @@ void ide_unregister(unsigned int index)
 
 	spin_unlock_irq(&ide_lock);
 
-	destroy_proc_ide_interface(hwif);
+	ide_proc_unregister_port(hwif);
 
 	hwgroup = hwif->hwgroup;
 	/*
@@ -751,6 +742,7 @@ void ide_setup_ports (	hw_regs_t *hw,
 /**
  *	ide_register_hw_with_fixup	-	register IDE interface
  *	@hw: hardware registers
+ *	@initializing: set while initializing built-in drivers
  *	@hwifp: pointer to returned hwif
  *	@fixup: fixup function
  *
@@ -760,7 +752,9 @@ void ide_setup_ports (	hw_regs_t *hw,
  *	Returns -1 on error.
  */
 
-int ide_register_hw_with_fixup(hw_regs_t *hw, ide_hwif_t **hwifp, void(*fixup)(ide_hwif_t *hwif))
+int ide_register_hw_with_fixup(hw_regs_t *hw, int initializing,
+			       ide_hwif_t **hwifp,
+			       void(*fixup)(ide_hwif_t *hwif))
 {
 	int index, retry = 1;
 	ide_hwif_t *hwif;
@@ -801,7 +795,7 @@ found:
 
 	if (!initializing) {
 		probe_hwif_init_with_fixup(hwif, fixup);
-		create_proc_ide_interfaces();
+		ide_proc_register_port(hwif);
 	}
 
 	if (hwifp)
@@ -812,9 +806,9 @@ found:
 
 EXPORT_SYMBOL(ide_register_hw_with_fixup);
 
-int ide_register_hw(hw_regs_t *hw, ide_hwif_t **hwifp)
+int ide_register_hw(hw_regs_t *hw, int initializing, ide_hwif_t **hwifp)
 {
-	return ide_register_hw_with_fixup(hw, hwifp, NULL);
+	return ide_register_hw_with_fixup(hw, initializing, hwifp, NULL);
 }
 
 EXPORT_SYMBOL(ide_register_hw);
@@ -825,205 +819,7 @@ EXPORT_SYMBOL(ide_register_hw);
 
 DECLARE_MUTEX(ide_setting_sem);
 
-/**
- *	__ide_add_setting	-	add an ide setting option
- *	@drive: drive to use
- *	@name: setting name
- *	@rw: true if the function is read write
- *	@read_ioctl: function to call on read
- *	@write_ioctl: function to call on write
- *	@data_type: type of data
- *	@min: range minimum
- *	@max: range maximum
- *	@mul_factor: multiplication scale
- *	@div_factor: divison scale
- *	@data: private data field
- *	@set: setting
- *	@auto_remove: setting auto removal flag
- *
- *	Removes the setting named from the device if it is present.
- *	The function takes the settings_lock to protect against 
- *	parallel changes. This function must not be called from IRQ
- *	context. Returns 0 on success or -1 on failure.
- *
- *	BUGS: This code is seriously over-engineered. There is also
- *	magic about how the driver specific features are setup. If
- *	a driver is attached we assume the driver settings are auto
- *	remove.
- */
-
-static int __ide_add_setting(ide_drive_t *drive, const char *name, int rw, int read_ioctl, int write_ioctl, int data_type, int min, int max, int mul_factor, int div_factor, void *data, ide_procset_t *set, int auto_remove)
-{
-	ide_settings_t **p = (ide_settings_t **) &drive->settings, *setting = NULL;
-
-	down(&ide_setting_sem);
-	while ((*p) && strcmp((*p)->name, name) < 0)
-		p = &((*p)->next);
-	if ((setting = kzalloc(sizeof(*setting), GFP_KERNEL)) == NULL)
-		goto abort;
-	if ((setting->name = kmalloc(strlen(name) + 1, GFP_KERNEL)) == NULL)
-		goto abort;
-	strcpy(setting->name, name);
-	setting->rw = rw;
-	setting->read_ioctl = read_ioctl;
-	setting->write_ioctl = write_ioctl;
-	setting->data_type = data_type;
-	setting->min = min;
-	setting->max = max;
-	setting->mul_factor = mul_factor;
-	setting->div_factor = div_factor;
-	setting->data = data;
-	setting->set = set;
-	
-	setting->next = *p;
-	if (auto_remove)
-		setting->auto_remove = 1;
-	*p = setting;
-	up(&ide_setting_sem);
-	return 0;
-abort:
-	up(&ide_setting_sem);
-	kfree(setting);
-	return -1;
-}
-
-int ide_add_setting(ide_drive_t *drive, const char *name, int rw, int read_ioctl, int write_ioctl, int data_type, int min, int max, int mul_factor, int div_factor, void *data, ide_procset_t *set)
-{
-	return __ide_add_setting(drive, name, rw, read_ioctl, write_ioctl, data_type, min, max, mul_factor, div_factor, data, set, 1);
-}
-
-EXPORT_SYMBOL(ide_add_setting);
-
-/**
- *	__ide_remove_setting	-	remove an ide setting option
- *	@drive: drive to use
- *	@name: setting name
- *
- *	Removes the setting named from the device if it is present.
- *	The caller must hold the setting semaphore.
- */
- 
-static void __ide_remove_setting (ide_drive_t *drive, char *name)
-{
-	ide_settings_t **p, *setting;
-
-	p = (ide_settings_t **) &drive->settings;
-
-	while ((*p) && strcmp((*p)->name, name))
-		p = &((*p)->next);
-	if ((setting = (*p)) == NULL)
-		return;
-
-	(*p) = setting->next;
-	
-	kfree(setting->name);
-	kfree(setting);
-}
-
-/**
- *	ide_find_setting_by_ioctl	-	find a drive specific ioctl
- *	@drive: drive to scan
- *	@cmd: ioctl command to handle
- *
- *	Scan's the device setting table for a matching entry and returns
- *	this or NULL if no entry is found. The caller must hold the
- *	setting semaphore
- */
- 
-static ide_settings_t *ide_find_setting_by_ioctl (ide_drive_t *drive, int cmd)
-{
-	ide_settings_t *setting = drive->settings;
-
-	while (setting) {
-		if (setting->read_ioctl == cmd || setting->write_ioctl == cmd)
-			break;
-		setting = setting->next;
-	}
-	
-	return setting;
-}
-
-/**
- *	ide_find_setting_by_name	-	find a drive specific setting
- *	@drive: drive to scan
- *	@name: setting name
- *
- *	Scan's the device setting table for a matching entry and returns
- *	this or NULL if no entry is found. The caller must hold the
- *	setting semaphore
- */
- 
-ide_settings_t *ide_find_setting_by_name (ide_drive_t *drive, char *name)
-{
-	ide_settings_t *setting = drive->settings;
-
-	while (setting) {
-		if (strcmp(setting->name, name) == 0)
-			break;
-		setting = setting->next;
-	}
-	return setting;
-}
-
-/**
- *	auto_remove_settings	-	remove driver specific settings
- *	@drive: drive
- *
- *	Automatically remove all the driver specific settings for this
- *	drive. This function may not be called from IRQ context. The
- *	caller must hold ide_setting_sem.
- */
- 
-static void auto_remove_settings (ide_drive_t *drive)
-{
-	ide_settings_t *setting;
-repeat:
-	setting = drive->settings;
-	while (setting) {
-		if (setting->auto_remove) {
-			__ide_remove_setting(drive, setting->name);
-			goto repeat;
-		}
-		setting = setting->next;
-	}
-}
-
-/**
- *	ide_read_setting	-	read an IDE setting
- *	@drive: drive to read from
- *	@setting: drive setting
- *
- *	Read a drive setting and return the value. The caller
- *	must hold the ide_setting_sem when making this call.
- *
- *	BUGS: the data return and error are the same return value
- *	so an error -EINVAL and true return of the same value cannot
- *	be told apart
- */
- 
-int ide_read_setting (ide_drive_t *drive, ide_settings_t *setting)
-{
-	int		val = -EINVAL;
-	unsigned long	flags;
-
-	if ((setting->rw & SETTING_READ)) {
-		spin_lock_irqsave(&ide_lock, flags);
-		switch(setting->data_type) {
-			case TYPE_BYTE:
-				val = *((u8 *) setting->data);
-				break;
-			case TYPE_SHORT:
-				val = *((u16 *) setting->data);
-				break;
-			case TYPE_INT:
-			case TYPE_INTA:
-				val = *((u32 *) setting->data);
-				break;
-		}
-		spin_unlock_irqrestore(&ide_lock, flags);
-	}
-	return val;
-}
+EXPORT_SYMBOL_GPL(ide_setting_sem);
 
 /**
  *	ide_spin_wait_hwgroup	-	wait for group
@@ -1058,61 +854,14 @@ int ide_spin_wait_hwgroup (ide_drive_t *drive)
 
 EXPORT_SYMBOL(ide_spin_wait_hwgroup);
 
-/**
- *	ide_write_setting	-	read an IDE setting
- *	@drive: drive to read from
- *	@setting: drive setting
- *	@val: value
- *
- *	Write a drive setting if it is possible. The caller
- *	must hold the ide_setting_sem when making this call.
- *
- *	BUGS: the data return and error are the same return value
- *	so an error -EINVAL and true return of the same value cannot
- *	be told apart
- *
- *	FIXME:  This should be changed to enqueue a special request
- *	to the driver to change settings, and then wait on a sema for completion.
- *	The current scheme of polling is kludgy, though safe enough.
- */
-
-int ide_write_setting (ide_drive_t *drive, ide_settings_t *setting, int val)
+int set_io_32bit(ide_drive_t *drive, int arg)
 {
-	int i;
-	u32 *p;
-
-	if (!capable(CAP_SYS_ADMIN))
-		return -EACCES;
-	if (!(setting->rw & SETTING_WRITE))
+	if (drive->no_io_32bit)
 		return -EPERM;
-	if (val < setting->min || val > setting->max)
+
+	if (arg < 0 || arg > 1 + (SUPPORT_VLB_SYNC << 1))
 		return -EINVAL;
-	if (setting->set)
-		return setting->set(drive, val);
-	if (ide_spin_wait_hwgroup(drive))
-		return -EBUSY;
-	switch (setting->data_type) {
-		case TYPE_BYTE:
-			*((u8 *) setting->data) = val;
-			break;
-		case TYPE_SHORT:
-			*((u16 *) setting->data) = val;
-			break;
-		case TYPE_INT:
-			*((u32 *) setting->data) = val;
-			break;
-		case TYPE_INTA:
-			p = (u32 *) setting->data;
-			for (i = 0; i < 1 << PARTN_BITS; i++, p++)
-				*p = val;
-			break;
-	}
-	spin_unlock_irq(&ide_lock);
-	return 0;
-}
 
-static int set_io_32bit(ide_drive_t *drive, int arg)
-{
 	drive->io_32bit = arg;
 #ifdef CONFIG_BLK_DEV_DTC2278
 	if (HWIF(drive)->chipset == ide_dtc2278)
@@ -1121,12 +870,28 @@ static int set_io_32bit(ide_drive_t *drive, int arg)
 	return 0;
 }
 
-static int set_using_dma (ide_drive_t *drive, int arg)
+static int set_ksettings(ide_drive_t *drive, int arg)
+{
+	if (arg < 0 || arg > 1)
+		return -EINVAL;
+
+	if (ide_spin_wait_hwgroup(drive))
+		return -EBUSY;
+	drive->keep_settings = arg;
+	spin_unlock_irq(&ide_lock);
+
+	return 0;
+}
+
+int set_using_dma(ide_drive_t *drive, int arg)
 {
 #ifdef CONFIG_BLK_DEV_IDEDMA
 	ide_hwif_t *hwif = drive->hwif;
 	int err = -EPERM;
 
+	if (arg < 0 || arg > 1)
+		return -EINVAL;
+
 	if (!drive->id || !(drive->id->capability & 1))
 		goto out;
 
@@ -1159,14 +924,20 @@ static int set_using_dma (ide_drive_t *drive, int arg)
 out:
 	return err;
 #else
+	if (arg < 0 || arg > 1)
+		return -EINVAL;
+
 	return -EPERM;
 #endif
 }
 
-static int set_pio_mode (ide_drive_t *drive, int arg)
+int set_pio_mode(ide_drive_t *drive, int arg)
 {
 	struct request rq;
 
+	if (arg < 0 || arg > 255)
+		return -EINVAL;
+
 	if (!HWIF(drive)->tuneproc)
 		return -ENOSYS;
 	if (drive->special.b.set_tune)
@@ -1178,42 +949,20 @@ static int set_pio_mode (ide_drive_t *drive, int arg)
 	return 0;
 }
 
-static int set_xfer_rate (ide_drive_t *drive, int arg)
+static int set_unmaskirq(ide_drive_t *drive, int arg)
 {
-	int err = ide_wait_cmd(drive,
-			WIN_SETFEATURES, (u8) arg,
-			SETFEATURES_XFER, 0, NULL);
+	if (drive->no_unmask)
+		return -EPERM;
 
-	if (!err && arg) {
-		ide_set_xfer_rate(drive, (u8) arg);
-		ide_driveid_update(drive);
-	}
-	return err;
-}
+	if (arg < 0 || arg > 1)
+		return -EINVAL;
 
-/**
- *	ide_add_generic_settings	-	generic ide settings
- *	@drive: drive being configured
- *
- *	Add the generic parts of the system settings to the /proc files and
- *	ioctls for this IDE device. The caller must not be holding the
- *	ide_setting_sem.
- */
+	if (ide_spin_wait_hwgroup(drive))
+		return -EBUSY;
+	drive->unmask = arg;
+	spin_unlock_irq(&ide_lock);
 
-void ide_add_generic_settings (ide_drive_t *drive)
-{
-/*
- *			  drive		setting name		read/write access				read ioctl		write ioctl		data type	min	max				mul_factor	div_factor	data pointer			set function
- */
-	__ide_add_setting(drive,	"io_32bit",		drive->no_io_32bit ? SETTING_READ : SETTING_RW,	HDIO_GET_32BIT,		HDIO_SET_32BIT,		TYPE_BYTE,	0,	1 + (SUPPORT_VLB_SYNC << 1),	1,		1,		&drive->io_32bit,		set_io_32bit,	0);
-	__ide_add_setting(drive,	"keepsettings",		SETTING_RW,					HDIO_GET_KEEPSETTINGS,	HDIO_SET_KEEPSETTINGS,	TYPE_BYTE,	0,	1,				1,		1,		&drive->keep_settings,		NULL,		0);
-	__ide_add_setting(drive,	"nice1",		SETTING_RW,					-1,			-1,			TYPE_BYTE,	0,	1,				1,		1,		&drive->nice1,			NULL,		0);
-	__ide_add_setting(drive,	"pio_mode",		SETTING_WRITE,					-1,			HDIO_SET_PIO_MODE,	TYPE_BYTE,	0,	255,				1,		1,		NULL,				set_pio_mode,	0);
-	__ide_add_setting(drive,	"unmaskirq",		drive->no_unmask ? SETTING_READ : SETTING_RW,	HDIO_GET_UNMASKINTR,	HDIO_SET_UNMASKINTR,	TYPE_BYTE,	0,	1,				1,		1,		&drive->unmask,			NULL,		0);
-	__ide_add_setting(drive,	"using_dma",		SETTING_RW,					HDIO_GET_DMA,		HDIO_SET_DMA,		TYPE_BYTE,	0,	1,				1,		1,		&drive->using_dma,		set_using_dma,	0);
-	__ide_add_setting(drive,	"init_speed",		SETTING_RW,					-1,			-1,			TYPE_BYTE,	0,	70,				1,		1,		&drive->init_speed,		NULL,		0);
-	__ide_add_setting(drive,	"current_speed",	SETTING_RW,					-1,			-1,			TYPE_BYTE,	0,	70,				1,		1,		&drive->current_speed,		set_xfer_rate,	0);
-	__ide_add_setting(drive,	"number",		SETTING_RW,					-1,			-1,			TYPE_BYTE,	0,	3,				1,		1,		&drive->dn,			NULL,		0);
+	return 0;
 }
 
 /**
@@ -1285,27 +1034,23 @@ static int generic_ide_resume(struct device *dev)
 int generic_ide_ioctl(ide_drive_t *drive, struct file *file, struct block_device *bdev,
 			unsigned int cmd, unsigned long arg)
 {
-	ide_settings_t *setting;
+	unsigned long flags;
 	ide_driver_t *drv;
-	int err = 0;
 	void __user *p = (void __user *)arg;
+	int err = 0, (*setfunc)(ide_drive_t *, int);
+	u8 *val;
 
-	down(&ide_setting_sem);
-	if ((setting = ide_find_setting_by_ioctl(drive, cmd)) != NULL) {
-		if (cmd == setting->read_ioctl) {
-			err = ide_read_setting(drive, setting);
-			up(&ide_setting_sem);
-			return err >= 0 ? put_user(err, (long __user *)arg) : err;
-		} else {
-			if (bdev != bdev->bd_contains)
-				err = -EINVAL;
-			else
-				err = ide_write_setting(drive, setting, arg);
-			up(&ide_setting_sem);
-			return err;
-		}
+	switch (cmd) {
+	case HDIO_GET_32BIT:	    val = &drive->io_32bit;	 goto read_val;
+	case HDIO_GET_KEEPSETTINGS: val = &drive->keep_settings; goto read_val;
+	case HDIO_GET_UNMASKINTR:   val = &drive->unmask;	 goto read_val;
+	case HDIO_GET_DMA:	    val = &drive->using_dma;	 goto read_val;
+	case HDIO_SET_32BIT:	    setfunc = set_io_32bit;	 goto set_val;
+	case HDIO_SET_KEEPSETTINGS: setfunc = set_ksettings;	 goto set_val;
+	case HDIO_SET_PIO_MODE:	    setfunc = set_pio_mode;	 goto set_val;
+	case HDIO_SET_UNMASKINTR:   setfunc = set_unmaskirq;	 goto set_val;
+	case HDIO_SET_DMA:	    setfunc = set_using_dma;	 goto set_val;
 	}
-	up(&ide_setting_sem);
 
 	switch (cmd) {
 		case HDIO_OBSOLETE_IDENTITY:
@@ -1359,7 +1104,7 @@ int generic_ide_ioctl(ide_drive_t *drive, struct file *file, struct block_device
 			ide_init_hwif_ports(&hw, (unsigned long) args[0],
 					    (unsigned long) args[1], NULL);
 			hw.irq = args[2];
-			if (ide_register_hw(&hw, NULL) == -1)
+			if (ide_register_hw(&hw, 0, NULL) == -1)
 				return -EIO;
 			return 0;
 		}
@@ -1434,6 +1179,28 @@ int generic_ide_ioctl(ide_drive_t *drive, struct file *file, struct block_device
 		default:
 			return -EINVAL;
 	}
+
+read_val:
+	down(&ide_setting_sem);
+	spin_lock_irqsave(&ide_lock, flags);
+	err = *val;
+	spin_unlock_irqrestore(&ide_lock, flags);
+	up(&ide_setting_sem);
+	return err >= 0 ? put_user(err, (long __user *)arg) : err;
+
+set_val:
+	if (bdev != bdev->bd_contains)
+		err = -EINVAL;
+	else {
+		if (!capable(CAP_SYS_ADMIN))
+			err = -EACCES;
+		else {
+			down(&ide_setting_sem);
+			err = setfunc(drive, arg);
+			up(&ide_setting_sem);
+		}
+	}
+	return err;
 }
 
 EXPORT_SYMBOL(generic_ide_ioctl);
@@ -1566,13 +1333,13 @@ static int __init ide_setup(char *s)
 		return 1;
 	}
 
-#ifdef CONFIG_BLK_DEV_IDEPCI
+#ifdef CONFIG_IDEPCI_PCIBUS_ORDER
 	if (!strcmp(s, "ide=reverse")) {
 		ide_scan_direction = 1;
 		printk(" : Enabled support for IDE inverse scan order.\n");
 		return 1;
 	}
-#endif /* CONFIG_BLK_DEV_IDEPCI */
+#endif
 
 #ifdef CONFIG_BLK_DEV_IDEACPI
 	if (!strcmp(s, "ide=noacpi")) {
@@ -1832,9 +1599,9 @@ extern void __init h8300_ide_init(void);
  */
 static void __init probe_for_hwifs (void)
 {
-#ifdef CONFIG_BLK_DEV_IDEPCI
+#ifdef CONFIG_IDEPCI_PCIBUS_ORDER
 	ide_scan_pcibus(ide_scan_direction);
-#endif /* CONFIG_BLK_DEV_IDEPCI */
+#endif
 
 #ifdef CONFIG_ETRAX_IDE
 	{
@@ -1892,54 +1659,6 @@ static void __init probe_for_hwifs (void)
 #endif
 }
 
-void ide_register_subdriver(ide_drive_t *drive, ide_driver_t *driver)
-{
-#ifdef CONFIG_PROC_FS
-	ide_add_proc_entries(drive->proc, driver->proc, drive);
-#endif
-}
-
-EXPORT_SYMBOL(ide_register_subdriver);
-
-/**
- *	ide_unregister_subdriver	-	disconnect drive from driver
- *	@drive: drive to unplug
- *	@driver: driver
- *
- *	Disconnect a drive from the driver it was attached to and then
- *	clean up the various proc files and other objects attached to it.
- *
- *	Takes ide_setting_sem and ide_lock.
- *	Caller must hold none of the locks.
- */
-
-void ide_unregister_subdriver(ide_drive_t *drive, ide_driver_t *driver)
-{
-	unsigned long flags;
-	
-#ifdef CONFIG_PROC_FS
-	ide_remove_proc_entries(drive->proc, driver->proc);
-#endif
-	down(&ide_setting_sem);
-	spin_lock_irqsave(&ide_lock, flags);
-	/*
-	 * ide_setting_sem protects the settings list
-	 * ide_lock protects the use of settings
-	 *
-	 * so we need to hold both, ide_settings_sem because we want to
-	 * modify the settings list, and ide_lock because we cannot take
-	 * a setting out that is being used.
-	 *
-	 * OTOH both ide_{read,write}_setting are only ever used under
-	 * ide_setting_sem.
-	 */
-	auto_remove_settings(drive);
-	spin_unlock_irqrestore(&ide_lock, flags);
-	up(&ide_setting_sem);
-}
-
-EXPORT_SYMBOL(ide_unregister_subdriver);
-
 /*
  * Probe module
  */
@@ -2071,9 +1790,7 @@ static int __init ide_init(void)
 
 	init_ide_data();
 
-#ifdef CONFIG_PROC_FS
-	proc_ide_root = proc_mkdir("ide", NULL);
-#endif
+	proc_ide_create();
 
 #ifdef CONFIG_BLK_DEV_ALI14XX
 	if (probe_ali14xx)
@@ -2096,14 +1813,9 @@ static int __init ide_init(void)
 		(void)qd65xx_init();
 #endif
 
-	initializing = 1;
 	/* Probe for special PCI and other "known" interface chipsets. */
 	probe_for_hwifs();
-	initializing = 0;
 
-#ifdef CONFIG_PROC_FS
-	proc_ide_create();
-#endif
 	return 0;
 }
 
@@ -2143,9 +1855,7 @@ void __exit cleanup_module (void)
 	pnpide_exit();
 #endif
 
-#ifdef CONFIG_PROC_FS
 	proc_ide_destroy();
-#endif
 
 	bus_unregister(&ide_bus_type);
 }
diff --git a/drivers/ide/legacy/ali14xx.c b/drivers/ide/legacy/ali14xx.c
index 91961aa03047..df17ed68c0bc 100644
--- a/drivers/ide/legacy/ali14xx.c
+++ b/drivers/ide/legacy/ali14xx.c
@@ -223,7 +223,8 @@ static int __init ali14xx_probe(void)
 	probe_hwif_init(hwif);
 	probe_hwif_init(mate);
 
-	create_proc_ide_interfaces();
+	ide_proc_register_port(hwif);
+	ide_proc_register_port(mate);
 
 	return 0;
 }
diff --git a/drivers/ide/legacy/buddha.c b/drivers/ide/legacy/buddha.c
index 1ed224a01f79..101aee1711c4 100644
--- a/drivers/ide/legacy/buddha.c
+++ b/drivers/ide/legacy/buddha.c
@@ -213,7 +213,7 @@ fail_base2:
 						IRQ_AMIGA_PORTS);
 			}	
 			
-			index = ide_register_hw(&hw, &hwif);
+			index = ide_register_hw(&hw, 1, &hwif);
 			if (index != -1) {
 				hwif->mmio = 1;
 				printk("ide%d: ", index);
diff --git a/drivers/ide/legacy/dtc2278.c b/drivers/ide/legacy/dtc2278.c
index 0219ffa64db6..36a3f0ac6162 100644
--- a/drivers/ide/legacy/dtc2278.c
+++ b/drivers/ide/legacy/dtc2278.c
@@ -138,7 +138,8 @@ static int __init dtc2278_probe(void)
 	probe_hwif_init(hwif);
 	probe_hwif_init(mate);
 
-	create_proc_ide_interfaces();
+	ide_proc_register_port(hwif);
+	ide_proc_register_port(mate);
 
 	return 0;
 }
diff --git a/drivers/ide/legacy/falconide.c b/drivers/ide/legacy/falconide.c
index a9f2cd5bb81e..e1e9d9d6893f 100644
--- a/drivers/ide/legacy/falconide.c
+++ b/drivers/ide/legacy/falconide.c
@@ -70,7 +70,7 @@ void __init falconide_init(void)
 			0, 0, NULL,
 //			falconide_iops,
 			IRQ_MFP_IDE);
-	index = ide_register_hw(&hw, NULL);
+	index = ide_register_hw(&hw, 1, NULL);
 
 	if (index != -1)
 	    printk("ide%d: Falcon IDE interface\n", index);
diff --git a/drivers/ide/legacy/gayle.c b/drivers/ide/legacy/gayle.c
index dcfadbbf55d8..0830a021bbb6 100644
--- a/drivers/ide/legacy/gayle.c
+++ b/drivers/ide/legacy/gayle.c
@@ -165,7 +165,7 @@ found:
 //			&gayle_iops,
 			IRQ_AMIGA_PORTS);
 
-	index = ide_register_hw(&hw, &hwif);
+	index = ide_register_hw(&hw, 1, &hwif);
 	if (index != -1) {
 	    hwif->mmio = 1;
 	    switch (i) {
diff --git a/drivers/ide/legacy/ht6560b.c b/drivers/ide/legacy/ht6560b.c
index a2832643c522..c8f353b1296f 100644
--- a/drivers/ide/legacy/ht6560b.c
+++ b/drivers/ide/legacy/ht6560b.c
@@ -357,7 +357,8 @@ int __init ht6560b_init(void)
 	probe_hwif_init(hwif);
 	probe_hwif_init(mate);
 
-	create_proc_ide_interfaces();
+	ide_proc_register_port(hwif);
+	ide_proc_register_port(mate);
 
 	return 0;
 
diff --git a/drivers/ide/legacy/ide-cs.c b/drivers/ide/legacy/ide-cs.c
index c6522a64d7ec..2f3977f195b7 100644
--- a/drivers/ide/legacy/ide-cs.c
+++ b/drivers/ide/legacy/ide-cs.c
@@ -153,7 +153,7 @@ static int idecs_register(unsigned long io, unsigned long ctl, unsigned long irq
     hw.irq = irq;
     hw.chipset = ide_pci;
     hw.dev = &handle->dev;
-    return ide_register_hw_with_fixup(&hw, NULL, ide_undecoded_slave);
+    return ide_register_hw_with_fixup(&hw, 0, NULL, ide_undecoded_slave);
 }
 
 /*======================================================================
diff --git a/drivers/ide/legacy/macide.c b/drivers/ide/legacy/macide.c
index 4c0079ad52ac..c211fc78345d 100644
--- a/drivers/ide/legacy/macide.c
+++ b/drivers/ide/legacy/macide.c
@@ -102,21 +102,21 @@ void macide_init(void)
 				0, 0, macide_ack_intr,
 //				quadra_ide_iops,
 				IRQ_NUBUS_F);
-		index = ide_register_hw(&hw, &hwif);
+		index = ide_register_hw(&hw, 1, &hwif);
 		break;
 	case MAC_IDE_PB:
 		ide_setup_ports(&hw, IDE_BASE, macide_offsets,
 				0, 0, macide_ack_intr,
 //				macide_pb_iops,
 				IRQ_NUBUS_C);
-		index = ide_register_hw(&hw, &hwif);
+		index = ide_register_hw(&hw, 1, &hwif);
 		break;
 	case MAC_IDE_BABOON:
 		ide_setup_ports(&hw, BABOON_BASE, macide_offsets,
 				0, 0, NULL,
 //				macide_baboon_iops,
 				IRQ_BABOON_1);
-		index = ide_register_hw(&hw, &hwif);
+		index = ide_register_hw(&hw, 1, &hwif);
 		if (index == -1) break;
 		if (macintosh_config->ident == MAC_MODEL_PB190) {
 
diff --git a/drivers/ide/legacy/q40ide.c b/drivers/ide/legacy/q40ide.c
index 74f08124eabb..e628a983ce33 100644
--- a/drivers/ide/legacy/q40ide.c
+++ b/drivers/ide/legacy/q40ide.c
@@ -142,7 +142,7 @@ void q40ide_init(void)
 			0, NULL,
 //			m68kide_iops,
 			q40ide_default_irq(pcide_bases[i]));
-	index = ide_register_hw(&hw, &hwif);
+	index = ide_register_hw(&hw, 1, &hwif);
 	// **FIXME**
 	if (index != -1)
 		hwif->mmio = 1;
diff --git a/drivers/ide/legacy/qd65xx.c b/drivers/ide/legacy/qd65xx.c
index 2fb8f50f1293..d1414a75b523 100644
--- a/drivers/ide/legacy/qd65xx.c
+++ b/drivers/ide/legacy/qd65xx.c
@@ -427,7 +427,7 @@ static int __init qd_probe(int base)
 		qd_setup(hwif, base, config, QD6500_DEF_DATA, QD6500_DEF_DATA,
 			 &qd6500_tune_drive);
 
-		create_proc_ide_interfaces();
+		ide_proc_register_port(hwif);
 
 		return 1;
 	}
@@ -459,7 +459,7 @@ static int __init qd_probe(int base)
 				 &qd6580_tune_drive);
 			qd_write_reg(QD_DEF_CONTR,QD_CONTROL_PORT);
 
-			create_proc_ide_interfaces();
+			ide_proc_register_port(hwif);
 
 			return 1;
 		} else {
@@ -479,7 +479,8 @@ static int __init qd_probe(int base)
 				 &qd6580_tune_drive);
 			qd_write_reg(QD_DEF_CONTR,QD_CONTROL_PORT);
 
-			create_proc_ide_interfaces();
+			ide_proc_register_port(hwif);
+			ide_proc_register_port(mate);
 
 			return 0; /* no other qd65xx possible */
 		}
diff --git a/drivers/ide/legacy/umc8672.c b/drivers/ide/legacy/umc8672.c
index ca7974455578..ddc403a0bd82 100644
--- a/drivers/ide/legacy/umc8672.c
+++ b/drivers/ide/legacy/umc8672.c
@@ -160,7 +160,8 @@ static int __init umc8672_probe(void)
 	probe_hwif_init(hwif);
 	probe_hwif_init(mate);
 
-	create_proc_ide_interfaces();
+	ide_proc_register_port(hwif);
+	ide_proc_register_port(mate);
 
 	return 0;
 }
diff --git a/drivers/ide/mips/au1xxx-ide.c b/drivers/ide/mips/au1xxx-ide.c
index d54d9fe92a7d..ca95e990862e 100644
--- a/drivers/ide/mips/au1xxx-ide.c
+++ b/drivers/ide/mips/au1xxx-ide.c
@@ -760,6 +760,9 @@ static int au_ide_probe(struct device *dev)
 #endif
 
 	probe_hwif_init(hwif);
+
+	ide_proc_register_port(hwif);
+
 	dev_set_drvdata(dev, hwif);
 
 	printk(KERN_INFO "Au1xxx IDE(builtin) configured for %s\n", mode );
diff --git a/drivers/ide/mips/swarm.c b/drivers/ide/mips/swarm.c
index 81fa06851b27..6e935d7c63fd 100644
--- a/drivers/ide/mips/swarm.c
+++ b/drivers/ide/mips/swarm.c
@@ -129,6 +129,9 @@ static int __devinit swarm_ide_probe(struct device *dev)
 	hwif->irq = hwif->hw.irq;
 
 	probe_hwif_init(hwif);
+
+	ide_proc_register_port(hwif);
+
 	dev_set_drvdata(dev, hwif);
 
 	return 0;
diff --git a/drivers/ide/pci/aec62xx.c b/drivers/ide/pci/aec62xx.c
index 73bdf64dbbfc..b173bc66ce1e 100644
--- a/drivers/ide/pci/aec62xx.c
+++ b/drivers/ide/pci/aec62xx.c
@@ -87,38 +87,12 @@ static u8 pci_bus_clock_list_ultra (u8 speed, struct chipset_bus_clock_list_entr
 	return chipset_table->ultra_settings;
 }
 
-static u8 aec62xx_ratemask (ide_drive_t *drive)
-{
-	ide_hwif_t *hwif	= HWIF(drive);
-	u8 mode;
-
-	switch(hwif->pci_dev->device) {
-		case PCI_DEVICE_ID_ARTOP_ATP865:
-		case PCI_DEVICE_ID_ARTOP_ATP865R:
-			mode = (inb(hwif->channel ?
-				    hwif->mate->dma_status :
-				    hwif->dma_status) & 0x10) ? 4 : 3;
-			break;
-		case PCI_DEVICE_ID_ARTOP_ATP860:
-		case PCI_DEVICE_ID_ARTOP_ATP860R:
-			mode = 2;
-			break;
-		case PCI_DEVICE_ID_ARTOP_ATP850UF:
-		default:
-			return 1;
-	}
-
-	if (!eighty_ninty_three(drive))
-		mode = min(mode, (u8)1);
-	return mode;
-}
-
 static int aec6210_tune_chipset (ide_drive_t *drive, u8 xferspeed)
 {
 	ide_hwif_t *hwif	= HWIF(drive);
 	struct pci_dev *dev	= hwif->pci_dev;
 	u16 d_conf		= 0;
-	u8 speed	= ide_rate_filter(aec62xx_ratemask(drive), xferspeed);
+	u8 speed		= ide_rate_filter(drive, xferspeed);
 	u8 ultra = 0, ultra_conf = 0;
 	u8 tmp0 = 0, tmp1 = 0, tmp2 = 0;
 	unsigned long flags;
@@ -145,7 +119,7 @@ static int aec6260_tune_chipset (ide_drive_t *drive, u8 xferspeed)
 {
 	ide_hwif_t *hwif	= HWIF(drive);
 	struct pci_dev *dev	= hwif->pci_dev;
-	u8 speed	= ide_rate_filter(aec62xx_ratemask(drive), xferspeed);
+	u8 speed	= ide_rate_filter(drive, xferspeed);
 	u8 unit		= (drive->select.b.unit & 0x01);
 	u8 tmp1 = 0, tmp2 = 0;
 	u8 ultra = 0, drive_conf = 0, ultra_conf = 0;
@@ -181,17 +155,6 @@ static int aec62xx_tune_chipset (ide_drive_t *drive, u8 speed)
 	}
 }
 
-static int config_chipset_for_dma (ide_drive_t *drive)
-{
-	u8 speed = ide_dma_speed(drive, aec62xx_ratemask(drive));	
-
-	if (!(speed))
-		return 0;
-
-	(void) aec62xx_tune_chipset(drive, speed);
-	return ide_dma_enable(drive);
-}
-
 static void aec62xx_tune_drive (ide_drive_t *drive, u8 pio)
 {
 	pio = ide_get_best_pio_mode(drive, pio, 4, NULL);
@@ -200,7 +163,7 @@ static void aec62xx_tune_drive (ide_drive_t *drive, u8 pio)
 
 static int aec62xx_config_drive_xfer_rate (ide_drive_t *drive)
 {
-	if (ide_use_dma(drive) && config_chipset_for_dma(drive))
+	if (ide_tune_dma(drive))
 		return 0;
 
 	if (ide_use_fast_pio(drive))
@@ -261,11 +224,13 @@ static unsigned int __devinit init_chipset_aec62xx(struct pci_dev *dev, const ch
 
 static void __devinit init_hwif_aec62xx(ide_hwif_t *hwif)
 {
+	struct pci_dev *dev = hwif->pci_dev;
+
 	hwif->autodma = 0;
 	hwif->tuneproc = &aec62xx_tune_drive;
 	hwif->speedproc = &aec62xx_tune_chipset;
 
-	if (hwif->pci_dev->device == PCI_DEVICE_ID_ARTOP_ATP850UF)
+	if (dev->device == PCI_DEVICE_ID_ARTOP_ATP850UF)
 		hwif->serialized = hwif->channel;
 
 	if (hwif->mate)
@@ -277,7 +242,15 @@ static void __devinit init_hwif_aec62xx(ide_hwif_t *hwif)
 		return;
 	}
 
-	hwif->ultra_mask = 0x7f;
+	hwif->ultra_mask = hwif->cds->udma_mask;
+
+	/* atp865 and atp865r */
+	if (hwif->ultra_mask == 0x3f) {
+		/* check bit 0x10 of DMA status register */
+		if (inb(pci_resource_start(dev, 4) + 2) & 0x10)
+ 			hwif->ultra_mask = 0x7f; /* udma0-6 */
+	}
+
 	hwif->mwdma_mask = 0x07;
 
 	hwif->ide_dma_check	= &aec62xx_config_drive_xfer_rate;
@@ -344,6 +317,7 @@ static ide_pci_device_t aec62xx_chipsets[] __devinitdata = {
 		.autodma	= AUTODMA,
 		.enablebits	= {{0x4a,0x02,0x02}, {0x4a,0x04,0x04}},
 		.bootable	= OFF_BOARD,
+		.udma_mask	= 0x07, /* udma0-2 */
 	},{	/* 1 */
 		.name		= "AEC6260",
 		.init_setup	= init_setup_aec62xx,
@@ -353,6 +327,7 @@ static ide_pci_device_t aec62xx_chipsets[] __devinitdata = {
 		.channels	= 2,
 		.autodma	= NOAUTODMA,
 		.bootable	= OFF_BOARD,
+		.udma_mask	= 0x1f, /* udma0-4 */
 	},{	/* 2 */
 		.name		= "AEC6260R",
 		.init_setup	= init_setup_aec62xx,
@@ -363,6 +338,7 @@ static ide_pci_device_t aec62xx_chipsets[] __devinitdata = {
 		.autodma	= AUTODMA,
 		.enablebits	= {{0x4a,0x02,0x02}, {0x4a,0x04,0x04}},
 		.bootable	= NEVER_BOARD,
+		.udma_mask	= 0x1f, /* udma0-4 */
 	},{	/* 3 */
 		.name		= "AEC6X80",
 		.init_setup	= init_setup_aec6x80,
@@ -372,6 +348,7 @@ static ide_pci_device_t aec62xx_chipsets[] __devinitdata = {
 		.channels	= 2,
 		.autodma	= AUTODMA,
 		.bootable	= OFF_BOARD,
+		.udma_mask	= 0x3f, /* udma0-5 */
 	},{	/* 4 */
 		.name		= "AEC6X80R",
 		.init_setup	= init_setup_aec6x80,
@@ -382,6 +359,7 @@ static ide_pci_device_t aec62xx_chipsets[] __devinitdata = {
 		.autodma	= AUTODMA,
 		.enablebits	= {{0x4a,0x02,0x02}, {0x4a,0x04,0x04}},
 		.bootable	= OFF_BOARD,
+		.udma_mask	= 0x3f, /* udma0-5 */
 	}
 };
 
diff --git a/drivers/ide/pci/alim15x3.c b/drivers/ide/pci/alim15x3.c
index 946a12746cb5..428efdae0c7b 100644
--- a/drivers/ide/pci/alim15x3.c
+++ b/drivers/ide/pci/alim15x3.c
@@ -50,7 +50,7 @@ static u8 m5229_revision;
 static u8 chip_is_1543c_e;
 static struct pci_dev *isa_dev;
 
-#if defined(DISPLAY_ALI_TIMINGS) && defined(CONFIG_PROC_FS)
+#if defined(DISPLAY_ALI_TIMINGS) && defined(CONFIG_IDE_PROC_FS)
 #include <linux/stat.h>
 #include <linux/proc_fs.h>
 
@@ -278,7 +278,7 @@ static int ali_get_info (char *buffer, char **addr, off_t offset, int count)
 
 	return p-buffer; /* => must be less than 4k! */
 }
-#endif  /* defined(DISPLAY_ALI_TIMINGS) && defined(CONFIG_PROC_FS) */
+#endif  /* defined(DISPLAY_ALI_TIMINGS) && defined(CONFIG_IDE_PROC_FS) */
 
 /**
  *	ali15x3_tune_pio	-	set up chipset for PIO mode
@@ -378,74 +378,31 @@ static void ali15x3_tune_drive (ide_drive_t *drive, u8 pio)
 }
 
 /**
- *	ali15x3_can_ultra	-	check for ultra DMA support
- *	@drive: drive to do the check
+ *	ali_udma_filter		-	compute UDMA mask
+ *	@drive: IDE device
  *
- *	Check the drive and controller revisions. Return 0 if UDMA is
- *	not available, or 1 if UDMA can be used. The actual rules for
- *	the ALi are
+ *	Return available UDMA modes.
+ *
+ *	The actual rules for the ALi are:
  *		No UDMA on revisions <= 0x20
  *		Disk only for revisions < 0xC2
  *		Not WDC drives for revisions < 0xC2
  *
  *	FIXME: WDC ifdef needs to die
  */
- 
-static u8 ali15x3_can_ultra (ide_drive_t *drive)
-{
-#ifndef CONFIG_WDC_ALI15X3
-	struct hd_driveid *id	= drive->id;
-#endif /* CONFIG_WDC_ALI15X3 */
 
-	if (m5229_revision <= 0x20) {
-		return 0;
-	} else if ((m5229_revision < 0xC2) &&
-#ifndef CONFIG_WDC_ALI15X3
-		   ((chip_is_1543c_e && strstr(id->model, "WDC ")) ||
-		    (drive->media!=ide_disk))) {
-#else /* CONFIG_WDC_ALI15X3 */
-		   (drive->media!=ide_disk)) {
-#endif /* CONFIG_WDC_ALI15X3 */
-		return 0;
-	} else {
-		return 1;
-	}
-}
-
-/**
- *	ali15x3_ratemask	-	generate DMA mode list
- *	@drive: drive to compute against
- *
- *	Generate a list of the available DMA modes for the drive. 
- *	FIXME: this function contains lots of bogus masking we can dump
- *
- *	Return the highest available mode (UDMA33, UDMA66, UDMA100,..)
- */
- 
-static u8 ali15x3_ratemask (ide_drive_t *drive)
+static u8 ali_udma_filter(ide_drive_t *drive)
 {
-	u8 mode = 0, can_ultra	= ali15x3_can_ultra(drive);
-
-	if (m5229_revision > 0xC4 && can_ultra) {
-		mode = 4;
-	} else if (m5229_revision == 0xC4 && can_ultra) {
-		mode = 3;
-	} else if (m5229_revision >= 0xC2 && can_ultra) {
-		mode = 2;
-	} else if (can_ultra) {
-		return 1;
-	} else {
-		return 0;
+	if (m5229_revision > 0x20 && m5229_revision < 0xC2) {
+		if (drive->media != ide_disk)
+			return 0;
+#ifndef CONFIG_WDC_ALI15X3
+		if (chip_is_1543c_e && strstr(drive->id->model, "WDC "))
+			return 0;
+#endif
 	}
 
-	/*
-	 *	If the drive sees no suitable cable then UDMA 33
-	 *	is the highest permitted mode
-	 */
-	 
-	if (!eighty_ninty_three(drive))
-		mode = min(mode, (u8)1);
-	return mode;
+	return drive->hwif->ultra_mask;
 }
 
 /**
@@ -461,7 +418,7 @@ static int ali15x3_tune_chipset (ide_drive_t *drive, u8 xferspeed)
 {
 	ide_hwif_t *hwif	= HWIF(drive);
 	struct pci_dev *dev	= hwif->pci_dev;
-	u8 speed		= ide_rate_filter(ali15x3_ratemask(drive), xferspeed);
+	u8 speed		= ide_rate_filter(drive, xferspeed);
 	u8 speed1		= speed;
 	u8 unit			= (drive->select.b.unit & 0x01);
 	u8 tmpbyte		= 0x00;
@@ -511,7 +468,7 @@ static int ali15x3_tune_chipset (ide_drive_t *drive, u8 xferspeed)
  
 static int config_chipset_for_dma (ide_drive_t *drive)
 {
-	u8 speed = ide_dma_speed(drive, ali15x3_ratemask(drive));
+	u8 speed = ide_max_dma_mode(drive);
 
 	if (!(speed))
 		return 0;
@@ -609,13 +566,13 @@ static unsigned int __devinit init_chipset_ali15x3 (struct pci_dev *dev, const c
 
 	isa_dev = pci_get_device(PCI_VENDOR_ID_AL, PCI_DEVICE_ID_AL_M1533, NULL);
 
-#if defined(DISPLAY_ALI_TIMINGS) && defined(CONFIG_PROC_FS)
+#if defined(DISPLAY_ALI_TIMINGS) && defined(CONFIG_IDE_PROC_FS)
 	if (!ali_proc) {
 		ali_proc = 1;
 		bmide_dev = dev;
 		ide_pci_create_host_proc("ali", ali_get_info);
 	}
-#endif  /* defined(DISPLAY_ALI_TIMINGS) && defined(CONFIG_PROC_FS) */
+#endif  /* defined(DISPLAY_ALI_TIMINGS) && defined(CONFIG_IDE_PROC_FS) */
 
 	local_irq_save(flags);
 
@@ -771,6 +728,7 @@ static void __devinit init_hwif_common_ali15x3 (ide_hwif_t *hwif)
 	hwif->autodma = 0;
 	hwif->tuneproc = &ali15x3_tune_drive;
 	hwif->speedproc = &ali15x3_tune_chipset;
+	hwif->udma_filter = &ali_udma_filter;
 
 	/* don't use LBA48 DMA on ALi devices before rev 0xC5 */
 	hwif->no_lba48_dma = (m5229_revision <= 0xC4) ? 1 : 0;
@@ -783,8 +741,17 @@ static void __devinit init_hwif_common_ali15x3 (ide_hwif_t *hwif)
 
 	hwif->atapi_dma = 1;
 
-	if (m5229_revision > 0x20)
-		hwif->ultra_mask = 0x7f;
+	if (m5229_revision <= 0x20)
+		hwif->ultra_mask = 0x00; /* no udma */
+	else if (m5229_revision < 0xC2)
+		hwif->ultra_mask = 0x07; /* udma0-2 */
+	else if (m5229_revision == 0xC2 || m5229_revision == 0xC3)
+		hwif->ultra_mask = 0x1f; /* udma0-4 */
+	else if (m5229_revision == 0xC4)
+		hwif->ultra_mask = 0x3f; /* udma0-5 */
+	else
+		hwif->ultra_mask = 0x7f; /* udma0-6 */
+
 	hwif->mwdma_mask = 0x07;
 	hwif->swdma_mask = 0x07;
 
diff --git a/drivers/ide/pci/amd74xx.c b/drivers/ide/pci/amd74xx.c
index 7989bdd842a2..becb1a5648b0 100644
--- a/drivers/ide/pci/amd74xx.c
+++ b/drivers/ide/pci/amd74xx.c
@@ -92,7 +92,7 @@ static unsigned char amd_cyc2udma[] = { 6, 6, 5, 4, 0, 1, 1, 2, 2, 3, 3, 3, 3, 3
  * AMD /proc entry.
  */
 
-#ifdef CONFIG_PROC_FS
+#ifdef CONFIG_IDE_PROC_FS
 
 #include <linux/stat.h>
 #include <linux/proc_fs.h>
@@ -402,14 +402,14 @@ static unsigned int __devinit init_chipset_amd74xx(struct pci_dev *dev, const ch
  * Register /proc/ide/amd74xx entry
  */
 
-#if defined(DISPLAY_AMD_TIMINGS) && defined(CONFIG_PROC_FS)
+#if defined(DISPLAY_AMD_TIMINGS) && defined(CONFIG_IDE_PROC_FS)
         if (!amd74xx_proc) {
                 amd_base = pci_resource_start(dev, 4);
                 bmide_dev = dev;
 		ide_pci_create_host_proc("amd74xx", amd74xx_get_info);
                 amd74xx_proc = 1;
         }
-#endif /* DISPLAY_AMD_TIMINGS && CONFIG_PROC_FS */
+#endif /* DISPLAY_AMD_TIMINGS && CONFIG_IDE_PROC_FS */
 
 	return dev->irq;
 }
diff --git a/drivers/ide/pci/atiixp.c b/drivers/ide/pci/atiixp.c
index 2d48af32e3f4..0e52ad722a72 100644
--- a/drivers/ide/pci/atiixp.c
+++ b/drivers/ide/pci/atiixp.c
@@ -49,22 +49,6 @@ static int save_mdma_mode[4];
 static DEFINE_SPINLOCK(atiixp_lock);
 
 /**
- *	atiixp_ratemask		-	compute rate mask for ATIIXP IDE
- *	@drive: IDE drive to compute for
- *
- *	Returns the available modes for the ATIIXP IDE controller.
- */
-
-static u8 atiixp_ratemask(ide_drive_t *drive)
-{
-	u8 mode = 3;
-
-	if (!eighty_ninty_three(drive))
-		mode = min(mode, (u8)1);
-	return mode;
-}
-
-/**
  *	atiixp_dma_2_pio		-	return the PIO mode matching DMA
  *	@xfer_rate: transfer speed
  *
@@ -189,7 +173,7 @@ static int atiixp_speedproc(ide_drive_t *drive, u8 xferspeed)
 	u16 tmp16;
 	u8 speed, pio;
 
-	speed = ide_rate_filter(atiixp_ratemask(drive), xferspeed);
+	speed = ide_rate_filter(drive, xferspeed);
 
 	spin_lock_irqsave(&atiixp_lock, flags);
 
@@ -223,26 +207,6 @@ static int atiixp_speedproc(ide_drive_t *drive, u8 xferspeed)
 }
 
 /**
- *	atiixp_config_drive_for_dma	-	configure drive for DMA
- *	@drive: IDE drive to configure
- *
- *	Set up a ATIIXP interface channel for the best available speed.
- *	We prefer UDMA if it is available and then MWDMA. If DMA is
- *	not available we switch to PIO and return 0.
- */
-
-static int atiixp_config_drive_for_dma(ide_drive_t *drive)
-{
-	u8 speed = ide_dma_speed(drive, atiixp_ratemask(drive));
-
-	if (!speed)
-		return 0;
-
-	(void) atiixp_speedproc(drive, speed);
-	return ide_dma_enable(drive);
-}
-
-/**
  *	atiixp_dma_check	-	set up an IDE device
  *	@drive: IDE drive to configure
  *
@@ -256,7 +220,7 @@ static int atiixp_dma_check(ide_drive_t *drive)
 
 	drive->init_speed = 0;
 
-	if (ide_use_dma(drive) && atiixp_config_drive_for_dma(drive))
+	if (ide_tune_dma(drive))
 		return 0;
 
 	if (ide_use_fast_pio(drive)) {
diff --git a/drivers/ide/pci/cmd64x.c b/drivers/ide/pci/cmd64x.c
index 77f51ab6d439..61ea96b5555c 100644
--- a/drivers/ide/pci/cmd64x.c
+++ b/drivers/ide/pci/cmd64x.c
@@ -74,7 +74,7 @@
 #define UDIDETCR1	0x7B
 #define DTPR1		0x7C
 
-#if defined(DISPLAY_CMD64X_TIMINGS) && defined(CONFIG_PROC_FS)
+#if defined(DISPLAY_CMD64X_TIMINGS) && defined(CONFIG_IDE_PROC_FS)
 #include <linux/stat.h>
 #include <linux/proc_fs.h>
 
@@ -165,7 +165,7 @@ static int cmd64x_get_info (char *buffer, char **addr, off_t offset, int count)
 	return p-buffer;	/* => must be less than 4k! */
 }
 
-#endif	/* defined(DISPLAY_CMD64X_TIMINGS) && defined(CONFIG_PROC_FS) */
+#endif	/* defined(DISPLAY_CMD64X_TIMINGS) && defined(CONFIG_IDE_PROC_FS) */
 
 static u8 quantize_timing(int timing, int quant)
 {
@@ -292,55 +292,6 @@ static void cmd64x_tune_drive (ide_drive_t *drive, u8 pio)
 	(void) ide_config_drive_speed(drive, XFER_PIO_0 + pio);
 }
 
-static u8 cmd64x_ratemask (ide_drive_t *drive)
-{
-	struct pci_dev *dev	= HWIF(drive)->pci_dev;
-	u8 mode = 0;
-
-	switch(dev->device) {
-		case PCI_DEVICE_ID_CMD_649:
-			mode = 3;
-			break;
-		case PCI_DEVICE_ID_CMD_648:
-			mode = 2;
-			break;
-		case PCI_DEVICE_ID_CMD_643:
-			return 0;
-
-		case PCI_DEVICE_ID_CMD_646:
-		{
-			unsigned int class_rev	= 0;
-			pci_read_config_dword(dev,
-				PCI_CLASS_REVISION, &class_rev);
-			class_rev &= 0xff;
-		/*
-		 * UltraDMA only supported on PCI646U and PCI646U2, which
-		 * correspond to revisions 0x03, 0x05 and 0x07 respectively.
-		 * Actually, although the CMD tech support people won't
-		 * tell me the details, the 0x03 revision cannot support
-		 * UDMA correctly without hardware modifications, and even
-		 * then it only works with Quantum disks due to some
-		 * hold time assumptions in the 646U part which are fixed
-		 * in the 646U2.
-		 *
-		 * So we only do UltraDMA on revision 0x05 and 0x07 chipsets.
-		 */
-			switch(class_rev) {
-				case 0x07:
-				case 0x05:
-					return 1;
-				case 0x03:
-				case 0x01:
-				default:
-					return 0;
-			}
-		}
-	}
-	if (!eighty_ninty_three(drive))
-		mode = min(mode, (u8)1);
-	return mode;
-}
-
 static int cmd64x_tune_chipset (ide_drive_t *drive, u8 speed)
 {
 	ide_hwif_t *hwif	= HWIF(drive);
@@ -348,7 +299,7 @@ static int cmd64x_tune_chipset (ide_drive_t *drive, u8 speed)
 	u8 unit			= drive->dn & 0x01;
 	u8 regU = 0, pciU	= hwif->channel ? UDIDETCR1 : UDIDETCR0;
 
-	speed = ide_rate_filter(cmd64x_ratemask(drive), speed);
+	speed = ide_rate_filter(drive, speed);
 
 	if (speed >= XFER_SW_DMA_0) {
 		(void) pci_read_config_byte(dev, pciU, &regU);
@@ -403,7 +354,7 @@ static int cmd64x_tune_chipset (ide_drive_t *drive, u8 speed)
 
 static int config_chipset_for_dma (ide_drive_t *drive)
 {
-	u8 speed	= ide_dma_speed(drive, cmd64x_ratemask(drive));
+	u8 speed = ide_max_dma_mode(drive);
 
 	if (!speed)
 		return 0;
@@ -597,7 +548,7 @@ static unsigned int __devinit init_chipset_cmd64x(struct pci_dev *dev, const cha
 	(void) pci_write_config_byte(dev, UDIDETCR0, 0xf0);
 #endif /* CONFIG_PPC */
 
-#if defined(DISPLAY_CMD64X_TIMINGS) && defined(CONFIG_PROC_FS)
+#if defined(DISPLAY_CMD64X_TIMINGS) && defined(CONFIG_IDE_PROC_FS)
 
 	cmd_devs[n_cmd_devs++] = dev;
 
@@ -605,7 +556,7 @@ static unsigned int __devinit init_chipset_cmd64x(struct pci_dev *dev, const cha
 		cmd64x_proc = 1;
 		ide_pci_create_host_proc("cmd64x", cmd64x_get_info);
 	}
-#endif /* DISPLAY_CMD64X_TIMINGS && CONFIG_PROC_FS */
+#endif /* DISPLAY_CMD64X_TIMINGS && CONFIG_IDE_PROC_FS */
 
 	return 0;
 }
@@ -644,15 +595,24 @@ static void __devinit init_hwif_cmd64x(ide_hwif_t *hwif)
 
 	hwif->atapi_dma = 1;
 
-	hwif->ultra_mask = 0x3f;
-	hwif->mwdma_mask = 0x07;
+	hwif->ultra_mask = hwif->cds->udma_mask;
+
+	/*
+	 * UltraDMA only supported on PCI646U and PCI646U2, which
+	 * correspond to revisions 0x03, 0x05 and 0x07 respectively.
+	 * Actually, although the CMD tech support people won't
+	 * tell me the details, the 0x03 revision cannot support
+	 * UDMA correctly without hardware modifications, and even
+	 * then it only works with Quantum disks due to some
+	 * hold time assumptions in the 646U part which are fixed
+	 * in the 646U2.
+	 *
+	 * So we only do UltraDMA on revision 0x05 and 0x07 chipsets.
+	 */
+	if (dev->device == PCI_DEVICE_ID_CMD_646 && class_rev < 5)
+		hwif->ultra_mask = 0x00;
 
-	if (dev->device == PCI_DEVICE_ID_CMD_643)
-		hwif->ultra_mask = 0x80;
-	if (dev->device == PCI_DEVICE_ID_CMD_646)
-		hwif->ultra_mask = (class_rev > 0x04) ? 0x07 : 0x80;
-	if (dev->device == PCI_DEVICE_ID_CMD_648)
-		hwif->ultra_mask = 0x1f;
+	hwif->mwdma_mask = 0x07;
 
 	hwif->ide_dma_check = &cmd64x_config_drive_for_dma;
 	if (!(hwif->udma_four))
@@ -716,6 +676,7 @@ static ide_pci_device_t cmd64x_chipsets[] __devinitdata = {
 		.autodma	= AUTODMA,
 		.enablebits	= {{0x00,0x00,0x00}, {0x51,0x08,0x08}},
 		.bootable	= ON_BOARD,
+		.udma_mask	= 0x00, /* no udma */
 	},{	/* 1 */
 		.name		= "CMD646",
 		.init_setup	= init_setup_cmd646,
@@ -725,6 +686,7 @@ static ide_pci_device_t cmd64x_chipsets[] __devinitdata = {
 		.autodma	= AUTODMA,
 		.enablebits	= {{0x51,0x04,0x04}, {0x51,0x08,0x08}},
 		.bootable	= ON_BOARD,
+		.udma_mask	= 0x07, /* udma0-2 */
 	},{	/* 2 */
 		.name		= "CMD648",
 		.init_setup	= init_setup_cmd64x,
@@ -734,6 +696,7 @@ static ide_pci_device_t cmd64x_chipsets[] __devinitdata = {
 		.autodma	= AUTODMA,
 		.enablebits	= {{0x51,0x04,0x04}, {0x51,0x08,0x08}},
 		.bootable	= ON_BOARD,
+		.udma_mask	= 0x1f, /* udma0-4 */
 	},{	/* 3 */
 		.name		= "CMD649",
 		.init_setup	= init_setup_cmd64x,
@@ -743,6 +706,7 @@ static ide_pci_device_t cmd64x_chipsets[] __devinitdata = {
 		.autodma	= AUTODMA,
 		.enablebits	= {{0x51,0x04,0x04}, {0x51,0x08,0x08}},
 		.bootable	= ON_BOARD,
+		.udma_mask	= 0x3f, /* udma0-5 */
 	}
 };
 
diff --git a/drivers/ide/pci/cs5520.c b/drivers/ide/pci/cs5520.c
index 400859a839f7..3b88a3a56116 100644
--- a/drivers/ide/pci/cs5520.c
+++ b/drivers/ide/pci/cs5520.c
@@ -213,6 +213,7 @@ static ide_pci_device_t cyrix_chipsets[] __devinitdata = {
  
 static int __devinit cs5520_init_one(struct pci_dev *dev, const struct pci_device_id *id)
 {
+	ide_hwif_t *hwif = NULL, *mate = NULL;
 	ata_index_t index;
 	ide_pci_device_t *d = &cyrix_chipsets[id->driver_data];
 
@@ -239,10 +240,21 @@ static int __devinit cs5520_init_one(struct pci_dev *dev, const struct pci_devic
 
 	ide_pci_setup_ports(dev, d, 14, &index);
 
-	if((index.b.low & 0xf0) != 0xf0)
-		probe_hwif_init(&ide_hwifs[index.b.low]);
-	if((index.b.high & 0xf0) != 0xf0)
-		probe_hwif_init(&ide_hwifs[index.b.high]);
+	if ((index.b.low & 0xf0) != 0xf0)
+		hwif = &ide_hwifs[index.b.low];
+	if ((index.b.high & 0xf0) != 0xf0)
+		mate = &ide_hwifs[index.b.high];
+
+	if (hwif)
+		probe_hwif_init(hwif);
+	if (mate)
+		probe_hwif_init(mate);
+
+	if (hwif)
+		ide_proc_register_port(hwif);
+	if (mate)
+		ide_proc_register_port(mate);
+
 	return 0;
 }
 
diff --git a/drivers/ide/pci/cs5535.c b/drivers/ide/pci/cs5535.c
index 45f43efbf92c..41925c47ef05 100644
--- a/drivers/ide/pci/cs5535.c
+++ b/drivers/ide/pci/cs5535.c
@@ -127,20 +127,6 @@ static void cs5535_set_speed(ide_drive_t *drive, u8 speed)
 	}
 }
 
-static u8 cs5535_ratemask(ide_drive_t *drive)
-{
-	/* eighty93 will return 1 if it's 80core and capable of
-	exceeding udma2, 0 otherwise. we need ratemask to set
-	the max speed and if we can > udma2 then we return 2
-	which selects speed_max as udma4 which is the 5535's max
-	speed, and 1 selects udma2 which is the max for 40c */
-	if (!eighty_ninty_three(drive))
-		return 1;
-
-	return 2;
-}
-
-
 /****
  *	cs5535_set_drive         -     Configure the drive to the new speed
  *	@drive: Drive to set up
@@ -151,7 +137,7 @@ static u8 cs5535_ratemask(ide_drive_t *drive)
  */
 static int cs5535_set_drive(ide_drive_t *drive, u8 speed)
 {
-	speed = ide_rate_filter(cs5535_ratemask(drive), speed);
+	speed = ide_rate_filter(drive, speed);
 	ide_config_drive_speed(drive, speed);
 	cs5535_set_speed(drive, speed);
 
@@ -178,28 +164,13 @@ static void cs5535_tuneproc(ide_drive_t *drive, u8 xferspeed)
 	cs5535_set_speed(drive, xferspeed);
 }
 
-static int cs5535_config_drive_for_dma(ide_drive_t *drive)
-{
-	u8 speed;
-
-	speed = ide_dma_speed(drive, cs5535_ratemask(drive));
-
-	/* If no DMA speed was available then let dma_check hit pio */
-	if (!speed) {
-		return 0;
-	}
-
-	cs5535_set_drive(drive, speed);
-	return ide_dma_enable(drive);
-}
-
 static int cs5535_dma_check(ide_drive_t *drive)
 {
 	u8 speed;
 
 	drive->init_speed = 0;
 
-	if (ide_use_dma(drive) && cs5535_config_drive_for_dma(drive))
+	if (ide_tune_dma(drive))
 		return 0;
 
 	if (ide_use_fast_pio(drive)) {
diff --git a/drivers/ide/pci/delkin_cb.c b/drivers/ide/pci/delkin_cb.c
index dd7ec37fdeab..46f4a888c037 100644
--- a/drivers/ide/pci/delkin_cb.c
+++ b/drivers/ide/pci/delkin_cb.c
@@ -80,7 +80,7 @@ delkin_cb_probe (struct pci_dev *dev, const struct pci_device_id *id)
 	hw.irq = dev->irq;
 	hw.chipset = ide_pci;		/* this enables IRQ sharing */
 
-	rc = ide_register_hw_with_fixup(&hw, &hwif, ide_undecoded_slave);
+	rc = ide_register_hw_with_fixup(&hw, 0, &hwif, ide_undecoded_slave);
 	if (rc < 0) {
 		printk(KERN_ERR "delkin_cb: ide_register_hw failed (%d)\n", rc);
 		pci_disable_device(dev);
diff --git a/drivers/ide/pci/hpt34x.c b/drivers/ide/pci/hpt34x.c
index 924eaa3a5708..2c24c3de8846 100644
--- a/drivers/ide/pci/hpt34x.c
+++ b/drivers/ide/pci/hpt34x.c
@@ -43,15 +43,10 @@
 
 #define HPT343_DEBUG_DRIVE_INFO		0
 
-static u8 hpt34x_ratemask (ide_drive_t *drive)
-{
-	return 1;
-}
-
 static int hpt34x_tune_chipset (ide_drive_t *drive, u8 xferspeed)
 {
 	struct pci_dev *dev	= HWIF(drive)->pci_dev;
-	u8 speed	= ide_rate_filter(hpt34x_ratemask(drive), xferspeed);
+	u8 speed = ide_rate_filter(drive, xferspeed);
 	u32 reg1= 0, tmp1 = 0, reg2 = 0, tmp2 = 0;
 	u8			hi_speed, lo_speed;
 
@@ -89,29 +84,11 @@ static void hpt34x_tune_drive (ide_drive_t *drive, u8 pio)
 	(void) hpt34x_tune_chipset(drive, (XFER_PIO_0 + pio));
 }
 
-/*
- * This allows the configuration of ide_pci chipset registers
- * for cards that learn about the drive's UDMA, DMA, PIO capabilities
- * after the drive is reported by the OS.  Initially for designed for
- * HPT343 UDMA chipset by HighPoint|Triones Technologies, Inc.
- */
-
-static int config_chipset_for_dma (ide_drive_t *drive)
-{
-	u8 speed = ide_dma_speed(drive, hpt34x_ratemask(drive));
-
-	if (!(speed))
-		return 0;
-
-	(void) hpt34x_tune_chipset(drive, speed);
-	return ide_dma_enable(drive);
-}
-
 static int hpt34x_config_drive_xfer_rate (ide_drive_t *drive)
 {
 	drive->init_speed = 0;
 
-	if (ide_use_dma(drive) && config_chipset_for_dma(drive))
+	if (ide_tune_dma(drive))
 #ifndef CONFIG_HPT34X_AUTODMA
 		return -1;
 #else
diff --git a/drivers/ide/pci/hpt366.c b/drivers/ide/pci/hpt366.c
index cf9d344d19f8..fcbc5605b38e 100644
--- a/drivers/ide/pci/hpt366.c
+++ b/drivers/ide/pci/hpt366.c
@@ -514,43 +514,31 @@ static int check_in_drive_list(ide_drive_t *drive, const char **list)
 	return 0;
 }
 
-static u8 hpt3xx_ratemask(ide_drive_t *drive)
-{
-	struct hpt_info *info	= pci_get_drvdata(HWIF(drive)->pci_dev);
-	u8 mode			= info->max_mode;
-
-	if (!eighty_ninty_three(drive) && mode)
-		mode = min(mode, (u8)1);
-	return mode;
-}
-
 /*
  *	Note for the future; the SATA hpt37x we must set
  *	either PIO or UDMA modes 0,4,5
  */
- 
-static u8 hpt3xx_ratefilter(ide_drive_t *drive, u8 speed)
+
+static u8 hpt3xx_udma_filter(ide_drive_t *drive)
 {
 	struct hpt_info *info	= pci_get_drvdata(HWIF(drive)->pci_dev);
 	u8 chip_type		= info->chip_type;
-	u8 mode			= hpt3xx_ratemask(drive);
-
-	if (drive->media != ide_disk)
-		return min(speed, (u8)XFER_PIO_4);
+	u8 mode			= info->max_mode;
+	u8 mask;
 
 	switch (mode) {
 		case 0x04:
-			speed = min_t(u8, speed, XFER_UDMA_6);
+			mask = 0x7f;
 			break;
 		case 0x03:
-			speed = min_t(u8, speed, XFER_UDMA_5);
+			mask = 0x3f;
 			if (chip_type >= HPT374)
 				break;
 			if (!check_in_drive_list(drive, bad_ata100_5))
 				goto check_bad_ata33;
 			/* fall thru */
 		case 0x02:
-			speed = min_t(u8, speed, XFER_UDMA_4);
+			mask = 0x1f;
 
 			/*
 			 * CHECK ME, Does this need to be changed to HPT374 ??
@@ -561,13 +549,13 @@ static u8 hpt3xx_ratefilter(ide_drive_t *drive, u8 speed)
 			    !check_in_drive_list(drive, bad_ata66_4))
 				goto check_bad_ata33;
 
-			speed = min_t(u8, speed, XFER_UDMA_3);
+			mask = 0x0f;
 			if (HPT366_ALLOW_ATA66_3 &&
 			    !check_in_drive_list(drive, bad_ata66_3))
 				goto check_bad_ata33;
 			/* fall thru */
 		case 0x01:
-			speed = min_t(u8, speed, XFER_UDMA_2);
+			mask = 0x07;
 
 		check_bad_ata33:
 			if (chip_type >= HPT370A)
@@ -577,10 +565,10 @@ static u8 hpt3xx_ratefilter(ide_drive_t *drive, u8 speed)
 			/* fall thru */
 		case 0x00:
 		default:
-			speed = min_t(u8, speed, XFER_MW_DMA_2);
+			mask = 0x00;
 			break;
 	}
-	return speed;
+	return mask;
 }
 
 static u32 get_speed_setting(u8 speed, struct hpt_info *info)
@@ -608,12 +596,19 @@ static int hpt36x_tune_chipset(ide_drive_t *drive, u8 xferspeed)
 	ide_hwif_t *hwif	= HWIF(drive);
 	struct pci_dev  *dev	= hwif->pci_dev;
 	struct hpt_info	*info	= pci_get_drvdata(dev);
-	u8  speed		= hpt3xx_ratefilter(drive, xferspeed);
+	u8  speed		= ide_rate_filter(drive, xferspeed);
 	u8  itr_addr		= drive->dn ? 0x44 : 0x40;
-	u32 itr_mask		= speed < XFER_MW_DMA_0 ? 0x30070000 :
-				 (speed < XFER_UDMA_0   ? 0xc0070000 : 0xc03800ff);
-	u32 new_itr		= get_speed_setting(speed, info);
 	u32 old_itr		= 0;
+	u32 itr_mask, new_itr;
+
+	/* TODO: move this to ide_rate_filter() [ check ->atapi_dma ] */
+	if (drive->media != ide_disk)
+		speed = min_t(u8, speed, XFER_PIO_4);
+
+	itr_mask = speed < XFER_MW_DMA_0 ? 0x30070000 :
+		  (speed < XFER_UDMA_0   ? 0xc0070000 : 0xc03800ff);
+
+	new_itr = get_speed_setting(speed, info);
 
 	/*
 	 * Disable on-chip PIO FIFO/buffer (and PIO MST mode as well)
@@ -633,12 +628,19 @@ static int hpt37x_tune_chipset(ide_drive_t *drive, u8 xferspeed)
 	ide_hwif_t *hwif	= HWIF(drive);
 	struct pci_dev  *dev	= hwif->pci_dev;
 	struct hpt_info	*info	= pci_get_drvdata(dev);
-	u8  speed		= hpt3xx_ratefilter(drive, xferspeed);
+	u8  speed		= ide_rate_filter(drive, xferspeed);
 	u8  itr_addr		= 0x40 + (drive->dn * 4);
-	u32 itr_mask		= speed < XFER_MW_DMA_0 ? 0x303c0000 :
-				 (speed < XFER_UDMA_0   ? 0xc03c0000 : 0xc1c001ff);
-	u32 new_itr		= get_speed_setting(speed, info);
 	u32 old_itr		= 0;
+	u32 itr_mask, new_itr;
+
+	/* TODO: move this to ide_rate_filter() [ check ->atapi_dma ] */
+	if (drive->media != ide_disk)
+		speed = min_t(u8, speed, XFER_PIO_4);
+
+	itr_mask = speed < XFER_MW_DMA_0 ? 0x303c0000 :
+		  (speed < XFER_UDMA_0   ? 0xc03c0000 : 0xc1c001ff);
+
+	new_itr = get_speed_setting(speed, info);
 
 	pci_read_config_dword(dev, itr_addr, &old_itr);
 	new_itr = (new_itr & ~itr_mask) | (old_itr & itr_mask);
@@ -667,24 +669,6 @@ static void hpt3xx_tune_drive(ide_drive_t *drive, u8 pio)
 	(void) hpt3xx_tune_chipset (drive, XFER_PIO_0 + pio);
 }
 
-/*
- * This allows the configuration of ide_pci chipset registers
- * for cards that learn about the drive's UDMA, DMA, PIO capabilities
- * after the drive is reported by the OS.  Initially designed for
- * HPT366 UDMA chipset by HighPoint|Triones Technologies, Inc.
- *
- */
-static int config_chipset_for_dma(ide_drive_t *drive)
-{
-	u8 speed = ide_dma_speed(drive, hpt3xx_ratemask(drive));
-
-	if (!speed)
-		return 0;
-
-	(void) hpt3xx_tune_chipset(drive, speed);
-	return ide_dma_enable(drive);
-}
-
 static int hpt3xx_quirkproc(ide_drive_t *drive)
 {
 	struct hd_driveid *id	= drive->id;
@@ -739,7 +723,7 @@ static int hpt366_config_drive_xfer_rate(ide_drive_t *drive)
 {
 	drive->init_speed = 0;
 
-	if (ide_use_dma(drive) && config_chipset_for_dma(drive))
+	if (ide_tune_dma(drive))
 		return 0;
 
 	if (ide_use_fast_pio(drive))
@@ -1271,6 +1255,7 @@ static void __devinit init_hwif_hpt366(ide_hwif_t *hwif)
 	hwif->intrproc			= &hpt3xx_intrproc;
 	hwif->maskproc			= &hpt3xx_maskproc;
 	hwif->busproc			= &hpt3xx_busproc;
+	hwif->udma_filter		= &hpt3xx_udma_filter;
 
 	/*
 	 * HPT3xxN chips have some complications:
diff --git a/drivers/ide/pci/it8213.c b/drivers/ide/pci/it8213.c
index 424f00bb160d..c04a02687b95 100644
--- a/drivers/ide/pci/it8213.c
+++ b/drivers/ide/pci/it8213.c
@@ -17,22 +17,6 @@
 
 #include <asm/io.h>
 
-/*
- *	it8213_ratemask	-	Compute available modes
- *	@drive: IDE drive
- *
- *	Compute the available speeds for the devices on the interface. This
- *	is all modes to ATA133 clipped by drive cable setup.
- */
-
-static u8 it8213_ratemask (ide_drive_t *drive)
-{
-	u8 mode	= 4;
-	if (!eighty_ninty_three(drive))
-		mode = min_t(u8, mode, 1);
-	return mode;
-}
-
 /**
  *	it8213_dma_2_pio		-	return the PIO mode matching DMA
  *	@xfer_rate: transfer speed
@@ -145,7 +129,7 @@ static int it8213_tune_chipset (ide_drive_t *drive, u8 xferspeed)
 	ide_hwif_t *hwif	= HWIF(drive);
 	struct pci_dev *dev	= hwif->pci_dev;
 	u8 maslave		= 0x40;
-	u8 speed		= ide_rate_filter(it8213_ratemask(drive), xferspeed);
+	u8 speed		= ide_rate_filter(drive, xferspeed);
 	int a_speed		= 3 << (drive->dn * 4);
 	int u_flag		= 1 << drive->dn;
 	int v_flag		= 0x01 << drive->dn;
@@ -213,25 +197,6 @@ static int it8213_tune_chipset (ide_drive_t *drive, u8 xferspeed)
 	return ide_config_drive_speed(drive, speed);
 }
 
-/*
- *	config_chipset_for_dma	-	configure for DMA
- *	@drive: drive to configure
- *
- *	Called by the IDE layer when it wants the timings set up.
- */
-
-static int config_chipset_for_dma (ide_drive_t *drive)
-{
-	u8 speed = ide_dma_speed(drive, it8213_ratemask(drive));
-
-	if (!speed)
-		return 0;
-
-	it8213_tune_chipset(drive, speed);
-
-	return ide_dma_enable(drive);
-}
-
 /**
  *	it8213_configure_drive_for_dma	-	set up for DMA transfers
  *	@drive: drive we are going to set up
@@ -246,7 +211,7 @@ static int it8213_config_drive_for_dma (ide_drive_t *drive)
 {
 	u8 pio;
 
-	if (ide_use_dma(drive) && config_chipset_for_dma(drive))
+	if (ide_tune_dma(drive))
 		return 0;
 
 	pio = ide_get_best_pio_mode(drive, 255, 4, NULL);
diff --git a/drivers/ide/pci/it821x.c b/drivers/ide/pci/it821x.c
index 4e1254813ee0..442f658c6ae7 100644
--- a/drivers/ide/pci/it821x.c
+++ b/drivers/ide/pci/it821x.c
@@ -229,22 +229,6 @@ static void it821x_clock_strategy(ide_drive_t *drive)
 }
 
 /**
- *	it821x_ratemask	-	Compute available modes
- *	@drive: IDE drive
- *
- *	Compute the available speeds for the devices on the interface. This
- *	is all modes to ATA133 clipped by drive cable setup.
- */
-
-static u8 it821x_ratemask (ide_drive_t *drive)
-{
-	u8 mode	= 4;
-	if (!eighty_ninty_three(drive))
-		mode = min(mode, (u8)1);
-	return mode;
-}
-
-/**
  *	it821x_tunepio	-	tune a drive
  *	@drive: drive to tune
  *	@pio: the desired PIO mode
@@ -438,7 +422,7 @@ static int it821x_tune_chipset (ide_drive_t *drive, byte xferspeed)
 
 	ide_hwif_t *hwif	= drive->hwif;
 	struct it821x_dev *itdev = ide_get_hwifdata(hwif);
-	u8 speed		= ide_rate_filter(it821x_ratemask(drive), xferspeed);
+	u8 speed		= ide_rate_filter(drive, xferspeed);
 
 	switch (speed) {
 	case XFER_PIO_4:
@@ -488,7 +472,7 @@ static int it821x_tune_chipset (ide_drive_t *drive, byte xferspeed)
 
 static int config_chipset_for_dma (ide_drive_t *drive)
 {
-	u8 speed	= ide_dma_speed(drive, it821x_ratemask(drive));
+	u8 speed = ide_max_dma_mode(drive);
 
 	if (speed == 0)
 		return 0;
diff --git a/drivers/ide/pci/jmicron.c b/drivers/ide/pci/jmicron.c
index be4fc96c29e0..76ed25147229 100644
--- a/drivers/ide/pci/jmicron.c
+++ b/drivers/ide/pci/jmicron.c
@@ -22,22 +22,6 @@ typedef enum {
 } port_type;
 
 /**
- *	jmicron_ratemask	-	Compute available modes
- *	@drive: IDE drive
- *
- *	Compute the available speeds for the devices on the interface. This
- *	is all modes to ATA133 clipped by drive cable setup.
- */
-
-static u8 jmicron_ratemask(ide_drive_t *drive)
-{
-	u8 mode	= 4;
-	if (!eighty_ninty_three(drive))
-		mode = min(mode, (u8)1);
-	return mode;
-}
-
-/**
  *	ata66_jmicron		-	Cable check
  *	@hwif: IDE port
  *
@@ -129,32 +113,12 @@ static void config_jmicron_chipset_for_pio (ide_drive_t *drive, byte set_speed)
 
 static int jmicron_tune_chipset (ide_drive_t *drive, byte xferspeed)
 {
-
-	u8 speed		= ide_rate_filter(jmicron_ratemask(drive), xferspeed);
+	u8 speed = ide_rate_filter(drive, xferspeed);
 
 	return ide_config_drive_speed(drive, speed);
 }
 
 /**
- *	config_chipset_for_dma	-	configure for DMA
- *	@drive: drive to configure
- *
- *	As the JMicron snoops for timings all we actually need to do is
- *	make sure we don't set an invalid mode.
- */
-
-static int config_chipset_for_dma (ide_drive_t *drive)
-{
-	u8 speed	= ide_dma_speed(drive, jmicron_ratemask(drive));
-
-	if (!speed)
-		return 0;
-
-	jmicron_tune_chipset(drive, speed);
-	return ide_dma_enable(drive);
-}
-
-/**
  *	jmicron_configure_drive_for_dma	-	set up for DMA transfers
  *	@drive: drive we are going to set up
  *
@@ -164,7 +128,7 @@ static int config_chipset_for_dma (ide_drive_t *drive)
 
 static int jmicron_config_drive_for_dma (ide_drive_t *drive)
 {
-	if (ide_use_dma(drive) && config_chipset_for_dma(drive))
+	if (ide_tune_dma(drive))
 		return 0;
 
 	config_jmicron_chipset_for_pio(drive, 1);
diff --git a/drivers/ide/pci/pdc202xx_new.c b/drivers/ide/pci/pdc202xx_new.c
index 2cdd629c653d..65b1e124edf7 100644
--- a/drivers/ide/pci/pdc202xx_new.c
+++ b/drivers/ide/pci/pdc202xx_new.c
@@ -37,8 +37,6 @@
 #include <asm/pci-bridge.h>
 #endif
 
-#define PDC202_DEBUG_CABLE	0
-
 #undef DEBUG
 
 #ifdef DEBUG
@@ -82,16 +80,6 @@ static u8 max_dma_rate(struct pci_dev *pdev)
 	return mode;
 }
 
-static u8 pdcnew_ratemask(ide_drive_t *drive)
-{
-	u8 mode = max_dma_rate(HWIF(drive)->pci_dev);
-
-	if (!eighty_ninty_three(drive))
-		mode = min_t(u8, mode, 1);
-
-	return	mode;
-}
-
 /**
  * get_indexed_reg - Get indexed register
  * @hwif: for the port address
@@ -164,7 +152,7 @@ static int pdcnew_tune_chipset(ide_drive_t *drive, u8 speed)
 	u8 adj			= (drive->dn & 1) ? 0x08 : 0x00;
 	int			err;
 
-	speed = ide_rate_filter(pdcnew_ratemask(drive), speed);
+	speed = ide_rate_filter(drive, speed);
 
 	/*
 	 * Issue SETFEATURES_XFER to the drive first. PDC202xx hardware will
@@ -244,17 +232,8 @@ static int config_chipset_for_dma(ide_drive_t *drive)
 {
 	struct hd_driveid *id	= drive->id;
 	ide_hwif_t *hwif	= HWIF(drive);
-	u8 ultra_66		= (id->dma_ultra & 0x0078) ? 1 : 0;
-	u8 cable		= pdcnew_cable_detect(hwif);
 	u8 speed;
 
-	if (ultra_66 && cable) {
-		printk(KERN_WARNING "Warning: %s channel "
-		       "requires an 80-pin cable for operation.\n",
-		       hwif->channel ? "Secondary" : "Primary");
-		printk(KERN_WARNING "%s reduced to Ultra33 mode.\n", drive->name);
-	}
-
 	if (id->capability & 4) {
 		/*
 		 * Set IORDY_EN & PREFETCH_EN (this seems to have
@@ -267,7 +246,7 @@ static int config_chipset_for_dma(ide_drive_t *drive)
 		set_indexed_reg(hwif, 0x13 + adj, tmp | 0x03);
 	}
 
-	speed = ide_dma_speed(drive, pdcnew_ratemask(drive));
+	speed = ide_max_dma_mode(drive);
 
 	if (!speed)
 		return 0;
@@ -543,7 +522,8 @@ static void __devinit init_hwif_pdc202new(ide_hwif_t *hwif)
 	hwif->drives[0].autotune = hwif->drives[1].autotune = 1;
 
 	hwif->atapi_dma  = 1;
-	hwif->ultra_mask = 0x7f;
+
+	hwif->ultra_mask = hwif->cds->udma_mask;
 	hwif->mwdma_mask = 0x07;
 
 	hwif->err_stops_fifo = 1;
@@ -556,11 +536,6 @@ static void __devinit init_hwif_pdc202new(ide_hwif_t *hwif)
 	if (!noautodma)
 		hwif->autodma = 1;
 	hwif->drives[0].autodma = hwif->drives[1].autodma = hwif->autodma;
-
-#if PDC202_DEBUG_CABLE
-	printk(KERN_DEBUG "%s: %s-pin cable\n",
-		hwif->name, hwif->udma_four ? "80" : "40");
-#endif /* PDC202_DEBUG_CABLE */
 }
 
 static int __devinit init_setup_pdcnew(struct pci_dev *dev, ide_pci_device_t *d)
@@ -619,6 +594,7 @@ static ide_pci_device_t pdcnew_chipsets[] __devinitdata = {
 		.channels	= 2,
 		.autodma	= AUTODMA,
 		.bootable	= OFF_BOARD,
+		.udma_mask	= 0x3f, /* udma0-5 */
 	},{	/* 1 */
 		.name		= "PDC20269",
 		.init_setup	= init_setup_pdcnew,
@@ -627,6 +603,7 @@ static ide_pci_device_t pdcnew_chipsets[] __devinitdata = {
 		.channels	= 2,
 		.autodma	= AUTODMA,
 		.bootable	= OFF_BOARD,
+		.udma_mask	= 0x7f, /* udma0-6*/
 	},{	/* 2 */
 		.name		= "PDC20270",
 		.init_setup	= init_setup_pdc20270,
@@ -635,6 +612,7 @@ static ide_pci_device_t pdcnew_chipsets[] __devinitdata = {
 		.channels	= 2,
 		.autodma	= AUTODMA,
 		.bootable	= OFF_BOARD,
+		.udma_mask	= 0x3f, /* udma0-5 */
 	},{	/* 3 */
 		.name		= "PDC20271",
 		.init_setup	= init_setup_pdcnew,
@@ -643,6 +621,7 @@ static ide_pci_device_t pdcnew_chipsets[] __devinitdata = {
 		.channels	= 2,
 		.autodma	= AUTODMA,
 		.bootable	= OFF_BOARD,
+		.udma_mask	= 0x7f, /* udma0-6*/
 	},{	/* 4 */
 		.name		= "PDC20275",
 		.init_setup	= init_setup_pdcnew,
@@ -651,6 +630,7 @@ static ide_pci_device_t pdcnew_chipsets[] __devinitdata = {
 		.channels	= 2,
 		.autodma	= AUTODMA,
 		.bootable	= OFF_BOARD,
+		.udma_mask	= 0x7f, /* udma0-6*/
 	},{	/* 5 */
 		.name		= "PDC20276",
 		.init_setup	= init_setup_pdc20276,
@@ -659,6 +639,7 @@ static ide_pci_device_t pdcnew_chipsets[] __devinitdata = {
 		.channels	= 2,
 		.autodma	= AUTODMA,
 		.bootable	= OFF_BOARD,
+		.udma_mask	= 0x7f, /* udma0-6*/
 	},{	/* 6 */
 		.name		= "PDC20277",
 		.init_setup	= init_setup_pdcnew,
@@ -667,6 +648,7 @@ static ide_pci_device_t pdcnew_chipsets[] __devinitdata = {
 		.channels	= 2,
 		.autodma	= AUTODMA,
 		.bootable	= OFF_BOARD,
+		.udma_mask	= 0x7f, /* udma0-6*/
 	}
 };
 
diff --git a/drivers/ide/pci/pdc202xx_old.c b/drivers/ide/pci/pdc202xx_old.c
index a7a639fe1eaf..7146fe3f6ba7 100644
--- a/drivers/ide/pci/pdc202xx_old.c
+++ b/drivers/ide/pci/pdc202xx_old.c
@@ -46,7 +46,6 @@
 #include <asm/io.h>
 #include <asm/irq.h>
 
-#define PDC202_DEBUG_CABLE		0
 #define PDC202XX_DEBUG_DRIVE_INFO	0
 
 static const char *pdc_quirk_drives[] = {
@@ -101,35 +100,12 @@ static const char *pdc_quirk_drives[] = {
 #define	MC1		0x02	/* DMA"C" timing */
 #define	MC0		0x01	/* DMA"C" timing */
 
-static u8 pdc202xx_ratemask (ide_drive_t *drive)
-{
-	u8 mode;
-
-	switch(HWIF(drive)->pci_dev->device) {
-		case PCI_DEVICE_ID_PROMISE_20267:
-		case PCI_DEVICE_ID_PROMISE_20265:
-			mode = 3;
-			break;
-		case PCI_DEVICE_ID_PROMISE_20263:
-		case PCI_DEVICE_ID_PROMISE_20262:
-			mode = 2;
-			break;
-		case PCI_DEVICE_ID_PROMISE_20246:
-			return 1;
-		default:
-			return 0;
-	}
-	if (!eighty_ninty_three(drive))
-		mode = min(mode, (u8)1);
-	return mode;
-}
-
 static int pdc202xx_tune_chipset (ide_drive_t *drive, u8 xferspeed)
 {
 	ide_hwif_t *hwif	= HWIF(drive);
 	struct pci_dev *dev	= hwif->pci_dev;
 	u8 drive_pci		= 0x60 + (drive->dn << 2);
-	u8 speed	= ide_rate_filter(pdc202xx_ratemask(drive), xferspeed);
+	u8 speed		= ide_rate_filter(drive, xferspeed);
 
 	u32			drive_conf;
 	u8			AP, BP, CP, DP;
@@ -261,20 +237,7 @@ static int config_chipset_for_dma (ide_drive_t *drive)
 	u32 drive_conf		= 0;
 	u8 drive_pci		= 0x60 + (drive->dn << 2);
 	u8 test1 = 0, test2 = 0, speed = -1;
-	u8 AP = 0, cable = 0;
-
-	u8 ultra_66		= ((id->dma_ultra & 0x0010) ||
-				   (id->dma_ultra & 0x0008)) ? 1 : 0;
-
-	if (dev->device != PCI_DEVICE_ID_PROMISE_20246)
-		cable = pdc202xx_old_cable_detect(hwif);
-	else
-		ultra_66 = 0;
-
-	if (ultra_66 && cable) {
-		printk(KERN_WARNING "Warning: %s channel requires an 80-pin cable for operation.\n", hwif->channel ? "Secondary":"Primary");
-		printk(KERN_WARNING "%s reduced to Ultra33 mode.\n", drive->name);
-	}
+	u8 AP = 0;
 
 	if (dev->device != PCI_DEVICE_ID_PROMISE_20246)
 		pdc_old_disable_66MHz_clock(drive->hwif);
@@ -308,7 +271,7 @@ chipset_is_set:
 	if (drive->media == ide_disk)	/* PREFETCH_EN */
 		pci_write_config_byte(dev, (drive_pci), AP|PREFETCH_EN);
 
-	speed = ide_dma_speed(drive, pdc202xx_ratemask(drive));
+	speed = ide_max_dma_mode(drive);
 
 	if (!(speed)) {
 		/* restore original pci-config space */
@@ -478,7 +441,7 @@ static void __devinit init_hwif_pdc202xx(ide_hwif_t *hwif)
 
 	hwif->drives[0].autotune = hwif->drives[1].autotune = 1;
 
-	hwif->ultra_mask = 0x3f;
+	hwif->ultra_mask = hwif->cds->udma_mask;
 	hwif->mwdma_mask = 0x07;
 	hwif->swdma_mask = 0x07;
 	hwif->atapi_dma = 1;
@@ -500,10 +463,6 @@ static void __devinit init_hwif_pdc202xx(ide_hwif_t *hwif)
 	if (!noautodma)
 		hwif->autodma = 1;
 	hwif->drives[0].autodma = hwif->drives[1].autodma = hwif->autodma;
-#if PDC202_DEBUG_CABLE
-	printk(KERN_DEBUG "%s: %s-pin cable\n",
-		hwif->name, hwif->udma_four ? "80" : "40");
-#endif /* PDC202_DEBUG_CABLE */	
 }
 
 static void __devinit init_dma_pdc202xx(ide_hwif_t *hwif, unsigned long dmabase)
@@ -587,6 +546,7 @@ static ide_pci_device_t pdc202xx_chipsets[] __devinitdata = {
 		.autodma	= AUTODMA,
 		.bootable	= OFF_BOARD,
 		.extra		= 16,
+		.udma_mask	= 0x07, /* udma0-2 */
 	},{	/* 1 */
 		.name		= "PDC20262",
 		.init_setup	= init_setup_pdc202ata4,
@@ -597,6 +557,7 @@ static ide_pci_device_t pdc202xx_chipsets[] __devinitdata = {
 		.autodma	= AUTODMA,
 		.bootable	= OFF_BOARD,
 		.extra		= 48,
+		.udma_mask	= 0x1f, /* udma0-4 */
 	},{	/* 2 */
 		.name		= "PDC20263",
 		.init_setup	= init_setup_pdc202ata4,
@@ -607,6 +568,7 @@ static ide_pci_device_t pdc202xx_chipsets[] __devinitdata = {
 		.autodma	= AUTODMA,
 		.bootable	= OFF_BOARD,
 		.extra		= 48,
+		.udma_mask	= 0x1f, /* udma0-4 */
 	},{	/* 3 */
 		.name		= "PDC20265",
 		.init_setup	= init_setup_pdc20265,
@@ -617,6 +579,7 @@ static ide_pci_device_t pdc202xx_chipsets[] __devinitdata = {
 		.autodma	= AUTODMA,
 		.bootable	= OFF_BOARD,
 		.extra		= 48,
+		.udma_mask	= 0x3f, /* udma0-5 */
 	},{	/* 4 */
 		.name		= "PDC20267",
 		.init_setup	= init_setup_pdc202xx,
@@ -627,6 +590,7 @@ static ide_pci_device_t pdc202xx_chipsets[] __devinitdata = {
 		.autodma	= AUTODMA,
 		.bootable	= OFF_BOARD,
 		.extra		= 48,
+		.udma_mask	= 0x3f, /* udma0-5 */
 	}
 };
 
diff --git a/drivers/ide/pci/piix.c b/drivers/ide/pci/piix.c
index 061d300ab8be..8b219dd63024 100644
--- a/drivers/ide/pci/piix.c
+++ b/drivers/ide/pci/piix.c
@@ -106,68 +106,6 @@
 static int no_piix_dma;
 
 /**
- *	piix_ratemask		-	compute rate mask for PIIX IDE
- *	@drive: IDE drive to compute for
- *
- *	Returns the available modes for the PIIX IDE controller.
- */
- 
-static u8 piix_ratemask (ide_drive_t *drive)
-{
-	struct pci_dev *dev	= HWIF(drive)->pci_dev;
-	u8 mode;
-
-	switch(dev->device) {
-		case PCI_DEVICE_ID_INTEL_82801EB_1:
-			mode = 3;
-			break;
-		/* UDMA 100 capable */
-		case PCI_DEVICE_ID_INTEL_82801BA_8:
-		case PCI_DEVICE_ID_INTEL_82801BA_9:
-		case PCI_DEVICE_ID_INTEL_82801CA_10:
-		case PCI_DEVICE_ID_INTEL_82801CA_11:
-		case PCI_DEVICE_ID_INTEL_82801E_11:
-		case PCI_DEVICE_ID_INTEL_82801DB_1:
-		case PCI_DEVICE_ID_INTEL_82801DB_10:
-		case PCI_DEVICE_ID_INTEL_82801DB_11:
-		case PCI_DEVICE_ID_INTEL_82801EB_11:
-		case PCI_DEVICE_ID_INTEL_ESB_2:
-		case PCI_DEVICE_ID_INTEL_ICH6_19:
-		case PCI_DEVICE_ID_INTEL_ICH7_21:
-		case PCI_DEVICE_ID_INTEL_ESB2_18:
-		case PCI_DEVICE_ID_INTEL_ICH8_6:
-			mode = 3;
-			break;
-		/* UDMA 66 capable */
-		case PCI_DEVICE_ID_INTEL_82801AA_1:
-		case PCI_DEVICE_ID_INTEL_82372FB_1:
-			mode = 2;
-			break;
-		/* UDMA 33 capable */
-		case PCI_DEVICE_ID_INTEL_82371AB:
-		case PCI_DEVICE_ID_INTEL_82443MX_1:
-		case PCI_DEVICE_ID_INTEL_82451NX:
-		case PCI_DEVICE_ID_INTEL_82801AB_1:
-			return 1;
-		/* Non UDMA capable (MWDMA2) */
-		case PCI_DEVICE_ID_INTEL_82371SB_1:
-		case PCI_DEVICE_ID_INTEL_82371FB_1:
-		case PCI_DEVICE_ID_INTEL_82371FB_0:
-		case PCI_DEVICE_ID_INTEL_82371MX:
-		default:
-			return 0;
-	}
-	
-	/*
-	 *	If we are UDMA66 capable fall back to UDMA33 
-	 *	if the drive cannot see an 80pin cable.
-	 */
-	if (!eighty_ninty_three(drive))
-		mode = min_t(u8, mode, 1);
-	return mode;
-}
-
-/**
  *	piix_dma_2_pio		-	return the PIO mode matching DMA
  *	@xfer_rate: transfer speed
  *
@@ -301,7 +239,7 @@ static int piix_tune_chipset (ide_drive_t *drive, u8 xferspeed)
 	ide_hwif_t *hwif	= HWIF(drive);
 	struct pci_dev *dev	= hwif->pci_dev;
 	u8 maslave		= hwif->channel ? 0x42 : 0x40;
-	u8 speed		= ide_rate_filter(piix_ratemask(drive), xferspeed);
+	u8 speed		= ide_rate_filter(drive, xferspeed);
 	int a_speed		= 3 << (drive->dn * 4);
 	int u_flag		= 1 << drive->dn;
 	int v_flag		= 0x01 << drive->dn;
@@ -366,30 +304,6 @@ static int piix_tune_chipset (ide_drive_t *drive, u8 xferspeed)
 }
 
 /**
- *	piix_config_drive_for_dma	-	configure drive for DMA
- *	@drive: IDE drive to configure
- *
- *	Set up a PIIX interface channel for the best available speed.
- *	We prefer UDMA if it is available and then MWDMA.  If DMA is
- *	not available we switch to PIO and return 0.
- */
- 
-static int piix_config_drive_for_dma (ide_drive_t *drive)
-{
-	u8 speed = ide_dma_speed(drive, piix_ratemask(drive));
-
-	/*
-	 * If no DMA speed was available or the chipset has DMA bugs
-	 * then disable DMA and use PIO
-	 */
-	if (!speed)
-		return 0;
-
-	(void) piix_tune_chipset(drive, speed);
-	return ide_dma_enable(drive);
-}
-
-/**
  *	piix_config_drive_xfer_rate	-	set up an IDE device
  *	@drive: IDE drive to configure
  *
@@ -401,7 +315,7 @@ static int piix_config_drive_xfer_rate (ide_drive_t *drive)
 {
 	drive->init_speed = 0;
 
-	if (ide_use_dma(drive) && piix_config_drive_for_dma(drive))
+	if (ide_tune_dma(drive))
 		return 0;
 
 	if (ide_use_fast_pio(drive))
@@ -524,26 +438,14 @@ static void __devinit init_hwif_piix(ide_hwif_t *hwif)
 		hwif->ide_dma_clear_irq = &piix_dma_clear_irq;
 
 	hwif->atapi_dma = 1;
-	hwif->ultra_mask = 0x3f;
+
+	hwif->ultra_mask = hwif->cds->udma_mask;
 	hwif->mwdma_mask = 0x06;
 	hwif->swdma_mask = 0x04;
 
-	switch(hwif->pci_dev->device) {
-		case PCI_DEVICE_ID_INTEL_82371FB_0:
-		case PCI_DEVICE_ID_INTEL_82371FB_1:
-		case PCI_DEVICE_ID_INTEL_82371SB_1:
-			hwif->ultra_mask = 0x80;
-			break;
-		case PCI_DEVICE_ID_INTEL_82371AB:
-		case PCI_DEVICE_ID_INTEL_82443MX_1:
-		case PCI_DEVICE_ID_INTEL_82451NX:
-		case PCI_DEVICE_ID_INTEL_82801AB_1:
-			hwif->ultra_mask = 0x07;
-			break;
-		default:
-			if (!hwif->udma_four)
-				hwif->udma_four = piix_cable_detect(hwif);
-			break;
+	if (hwif->ultra_mask & 0x78) {
+		if (!hwif->udma_four)
+			hwif->udma_four = piix_cable_detect(hwif);
 	}
 
 	if (no_piix_dma)
@@ -557,7 +459,7 @@ static void __devinit init_hwif_piix(ide_hwif_t *hwif)
 	hwif->drives[0].autodma = hwif->autodma;
 }
 
-#define DECLARE_PIIX_DEV(name_str) \
+#define DECLARE_PIIX_DEV(name_str, udma) \
 	{						\
 		.name		= name_str,		\
 		.init_chipset	= init_chipset_piix,	\
@@ -566,11 +468,12 @@ static void __devinit init_hwif_piix(ide_hwif_t *hwif)
 		.autodma	= AUTODMA,		\
 		.enablebits	= {{0x41,0x80,0x80}, {0x43,0x80,0x80}}, \
 		.bootable	= ON_BOARD,		\
+		.udma_mask	= udma,			\
 	}
 
 static ide_pci_device_t piix_pci_info[] __devinitdata = {
-	/*  0 */ DECLARE_PIIX_DEV("PIIXa"),
-	/*  1 */ DECLARE_PIIX_DEV("PIIXb"),
+	/*  0 */ DECLARE_PIIX_DEV("PIIXa", 0x00),	/* no udma */
+	/*  1 */ DECLARE_PIIX_DEV("PIIXb", 0x00),	/* no udma */
 
 	/*  2 */
 	{	/*
@@ -587,28 +490,28 @@ static ide_pci_device_t piix_pci_info[] __devinitdata = {
 		.flags		= IDEPCI_FLAG_ISA_PORTS
 	},
 
-	/*  3 */ DECLARE_PIIX_DEV("PIIX3"),
-	/*  4 */ DECLARE_PIIX_DEV("PIIX4"),
-	/*  5 */ DECLARE_PIIX_DEV("ICH0"),
-	/*  6 */ DECLARE_PIIX_DEV("PIIX4"),
-	/*  7 */ DECLARE_PIIX_DEV("ICH"),
-	/*  8 */ DECLARE_PIIX_DEV("PIIX4"),
-	/*  9 */ DECLARE_PIIX_DEV("PIIX4"),
-	/* 10 */ DECLARE_PIIX_DEV("ICH2"),
-	/* 11 */ DECLARE_PIIX_DEV("ICH2M"),
-	/* 12 */ DECLARE_PIIX_DEV("ICH3M"),
-	/* 13 */ DECLARE_PIIX_DEV("ICH3"),
-	/* 14 */ DECLARE_PIIX_DEV("ICH4"),
-	/* 15 */ DECLARE_PIIX_DEV("ICH5"),
-	/* 16 */ DECLARE_PIIX_DEV("C-ICH"),
-	/* 17 */ DECLARE_PIIX_DEV("ICH4"),
-	/* 18 */ DECLARE_PIIX_DEV("ICH5-SATA"),
-	/* 19 */ DECLARE_PIIX_DEV("ICH5"),
-	/* 20 */ DECLARE_PIIX_DEV("ICH6"),
-	/* 21 */ DECLARE_PIIX_DEV("ICH7"),
-	/* 22 */ DECLARE_PIIX_DEV("ICH4"),
-	/* 23 */ DECLARE_PIIX_DEV("ESB2"),
-	/* 24 */ DECLARE_PIIX_DEV("ICH8M"),
+	/*  3 */ DECLARE_PIIX_DEV("PIIX3", 0x00),	/* no udma */
+	/*  4 */ DECLARE_PIIX_DEV("PIIX4", 0x07),	/* udma0-2 */
+	/*  5 */ DECLARE_PIIX_DEV("ICH0",  0x07),	/* udma0-2 */
+	/*  6 */ DECLARE_PIIX_DEV("PIIX4", 0x07),	/* udma0-2 */
+	/*  7 */ DECLARE_PIIX_DEV("ICH",   0x1f),	/* udma0-4 */
+	/*  8 */ DECLARE_PIIX_DEV("PIIX4", 0x1f),	/* udma0-4 */
+	/*  9 */ DECLARE_PIIX_DEV("PIIX4", 0x07),	/* udma0-2 */
+	/* 10 */ DECLARE_PIIX_DEV("ICH2",  0x3f),	/* udma0-5 */
+	/* 11 */ DECLARE_PIIX_DEV("ICH2M", 0x3f),	/* udma0-5 */
+	/* 12 */ DECLARE_PIIX_DEV("ICH3M", 0x3f),	/* udma0-5 */
+	/* 13 */ DECLARE_PIIX_DEV("ICH3",  0x3f),	/* udma0-5 */
+	/* 14 */ DECLARE_PIIX_DEV("ICH4",  0x3f),	/* udma0-5 */
+	/* 15 */ DECLARE_PIIX_DEV("ICH5",  0x3f),	/* udma0-5 */
+	/* 16 */ DECLARE_PIIX_DEV("C-ICH", 0x3f),	/* udma0-5 */
+	/* 17 */ DECLARE_PIIX_DEV("ICH4",  0x3f),	/* udma0-5 */
+	/* 18 */ DECLARE_PIIX_DEV("ICH5-SATA", 0x3f),	/* udma0-5 */
+	/* 19 */ DECLARE_PIIX_DEV("ICH5",  0x3f),	/* udma0-5 */
+	/* 20 */ DECLARE_PIIX_DEV("ICH6",  0x3f),	/* udma0-5 */
+	/* 21 */ DECLARE_PIIX_DEV("ICH7",  0x3f),	/* udma0-5 */
+	/* 22 */ DECLARE_PIIX_DEV("ICH4",  0x3f),	/* udma0-5 */
+	/* 23 */ DECLARE_PIIX_DEV("ESB2",  0x3f),	/* udma0-5 */
+	/* 24 */ DECLARE_PIIX_DEV("ICH8M", 0x3f),	/* udma0-5 */
 };
 
 /**
diff --git a/drivers/ide/pci/scc_pata.c b/drivers/ide/pci/scc_pata.c
index f84bf791f72e..cbf936325355 100644
--- a/drivers/ide/pci/scc_pata.c
+++ b/drivers/ide/pci/scc_pata.c
@@ -190,23 +190,6 @@ scc_ide_outsl(unsigned long port, void *addr, u32 count)
 }
 
 /**
- *	scc_ratemask	-	Compute available modes
- *	@drive: IDE drive
- *
- *	Compute the available speeds for the devices on the interface.
- *	Enforce UDMA33 as a limit if there is no 80pin cable present.
- */
-
-static u8 scc_ratemask(ide_drive_t *drive)
-{
-	u8 mode = 4;
-
-	if (!eighty_ninty_three(drive))
-		mode = min(mode, (u8)1);
-	return mode;
-}
-
-/**
  *	scc_tuneproc	-	tune a drive PIO mode
  *	@drive: drive to tune
  *	@mode_wanted: the target operating mode
@@ -273,7 +256,7 @@ static void scc_tuneproc(ide_drive_t *drive, byte mode_wanted)
 static int scc_tune_chipset(ide_drive_t *drive, byte xferspeed)
 {
 	ide_hwif_t *hwif = HWIF(drive);
-	u8 speed = ide_rate_filter(scc_ratemask(drive), xferspeed);
+	u8 speed = ide_rate_filter(drive, xferspeed);
 	struct scc_ports *ports = ide_get_hwifdata(hwif);
 	unsigned long ctl_base = ports->ctl;
 	unsigned long cckctrl_port = ctl_base + 0xff0;
@@ -347,7 +330,7 @@ static int scc_tune_chipset(ide_drive_t *drive, byte xferspeed)
 
 static int scc_config_chipset_for_dma(ide_drive_t *drive)
 {
-	u8 speed = ide_dma_speed(drive, scc_ratemask(drive));
+	u8 speed = ide_max_dma_mode(drive);
 
 	if (!speed)
 		return 0;
diff --git a/drivers/ide/pci/serverworks.c b/drivers/ide/pci/serverworks.c
index dbcd37a0c652..2fa6d92d16cc 100644
--- a/drivers/ide/pci/serverworks.c
+++ b/drivers/ide/pci/serverworks.c
@@ -65,16 +65,16 @@ static int check_in_drive_lists (ide_drive_t *drive, const char **list)
 	return 0;
 }
 
-static u8 svwks_ratemask (ide_drive_t *drive)
+static u8 svwks_udma_filter(ide_drive_t *drive)
 {
 	struct pci_dev *dev     = HWIF(drive)->pci_dev;
-	u8 mode = 0;
+	u8 mask = 0;
 
 	if (!svwks_revision)
 		pci_read_config_byte(dev, PCI_REVISION_ID, &svwks_revision);
 
 	if (dev->device == PCI_DEVICE_ID_SERVERWORKS_HT1000IDE)
-		return 2;
+		return 0x1f;
 	if (dev->device == PCI_DEVICE_ID_SERVERWORKS_OSB4IDE) {
 		u32 reg = 0;
 		if (isa_dev)
@@ -86,25 +86,31 @@ static u8 svwks_ratemask (ide_drive_t *drive)
 		if(drive->media == ide_disk)
 			return 0;
 		/* Check the OSB4 DMA33 enable bit */
-		return ((reg & 0x00004000) == 0x00004000) ? 1 : 0;
+		return ((reg & 0x00004000) == 0x00004000) ? 0x07 : 0;
 	} else if (svwks_revision < SVWKS_CSB5_REVISION_NEW) {
-		return 1;
+		return 0x07;
 	} else if (svwks_revision >= SVWKS_CSB5_REVISION_NEW) {
-		u8 btr = 0;
+		u8 btr = 0, mode;
 		pci_read_config_byte(dev, 0x5A, &btr);
 		mode = btr & 0x3;
-		if (!eighty_ninty_three(drive))
-			mode = min(mode, (u8)1);
+
 		/* If someone decides to do UDMA133 on CSB5 the same
 		   issue will bite so be inclusive */
 		if (mode > 2 && check_in_drive_lists(drive, svwks_bad_ata100))
 			mode = 2;
+
+		switch(mode) {
+		case 2:	 mask = 0x1f; break;
+		case 1:	 mask = 0x07; break;
+		default: mask = 0x00; break;
+		}
 	}
 	if (((dev->device == PCI_DEVICE_ID_SERVERWORKS_CSB6IDE) ||
 	     (dev->device == PCI_DEVICE_ID_SERVERWORKS_CSB6IDE2)) &&
 	    (!(PCI_FUNC(dev->devfn) & 1)))
-		mode = 2;
-	return mode;
+		mask = 0x1f;
+
+	return mask;
 }
 
 static u8 svwks_csb_check (struct pci_dev *dev)
@@ -141,7 +147,7 @@ static int svwks_tune_chipset (ide_drive_t *drive, u8 xferspeed)
 	if (xferspeed == 255)	/* PIO auto-tuning */
 		speed = XFER_PIO_0 + pio;
 	else
-		speed = ide_rate_filter(svwks_ratemask(drive), xferspeed);
+		speed = ide_rate_filter(drive, xferspeed);
 
 	/* If we are about to put a disk into UDMA mode we screwed up.
 	   Our code assumes we never _ever_ do this on an OSB4 */
@@ -304,7 +310,7 @@ static void svwks_tune_drive (ide_drive_t *drive, u8 pio)
 
 static int config_chipset_for_dma (ide_drive_t *drive)
 {
-	u8 speed = ide_dma_speed(drive, svwks_ratemask(drive));
+	u8 speed = ide_max_dma_mode(drive);
 
 	if (!(speed))
 		speed = XFER_PIO_0 + ide_get_best_pio_mode(drive, 255, 5, NULL);
@@ -500,6 +506,7 @@ static void __devinit init_hwif_svwks (ide_hwif_t *hwif)
 
 	hwif->tuneproc = &svwks_tune_drive;
 	hwif->speedproc = &svwks_tune_chipset;
+	hwif->udma_filter = &svwks_udma_filter;
 
 	hwif->atapi_dma = 1;
 
diff --git a/drivers/ide/pci/sgiioc4.c b/drivers/ide/pci/sgiioc4.c
index fd09b295a69d..d3185e29a38e 100644
--- a/drivers/ide/pci/sgiioc4.c
+++ b/drivers/ide/pci/sgiioc4.c
@@ -692,7 +692,7 @@ sgiioc4_ide_setup_pci_device(struct pci_dev *dev, ide_pci_device_t * d)
 		return -EIO;
 
 	/* Create /proc/ide entries */
-	create_proc_ide_interfaces();
+	ide_proc_register_port(hwif);
 
 	return 0;
 }
diff --git a/drivers/ide/pci/siimage.c b/drivers/ide/pci/siimage.c
index c0188de3cc66..d09e74c2996e 100644
--- a/drivers/ide/pci/siimage.c
+++ b/drivers/ide/pci/siimage.c
@@ -122,45 +122,41 @@ static inline unsigned long siimage_seldev(ide_drive_t *drive, int r)
 }
 
 /**
- *	siimage_ratemask	-	Compute available modes
- *	@drive: IDE drive
+ *	sil_udma_filter		-	compute UDMA mask
+ *	@drive: IDE device
+ *
+ *	Compute the available UDMA speeds for the device on the interface.
  *
- *	Compute the available speeds for the devices on the interface.
  *	For the CMD680 this depends on the clocking mode (scsc), for the
- *	SI3312 SATA controller life is a bit simpler. Enforce UDMA33
- *	as a limit if there is no 80pin cable present.
+ *	SI3112 SATA controller life is a bit simpler.
  */
- 
-static byte siimage_ratemask (ide_drive_t *drive)
+
+static u8 sil_udma_filter(ide_drive_t *drive)
 {
-	ide_hwif_t *hwif	= HWIF(drive);
-	u8 mode	= 0, scsc = 0;
+	ide_hwif_t *hwif = drive->hwif;
 	unsigned long base = (unsigned long) hwif->hwif_data;
+	u8 mask = 0, scsc = 0;
 
 	if (hwif->mmio)
 		scsc = hwif->INB(base + 0x4A);
 	else
 		pci_read_config_byte(hwif->pci_dev, 0x8A, &scsc);
 
-	if(is_sata(hwif))
-	{
-		if(strstr(drive->id->model, "Maxtor"))
-			return 3;
-		return 4;
+	if (is_sata(hwif)) {
+		mask = strstr(drive->id->model, "Maxtor") ? 0x3f : 0x7f;
+		goto out;
 	}
-	
+
 	if ((scsc & 0x30) == 0x10)	/* 133 */
-		mode = 4;
+		mask = 0x7f;
 	else if ((scsc & 0x30) == 0x20)	/* 2xPCI */
-		mode = 4;
+		mask = 0x7f;
 	else if ((scsc & 0x30) == 0x00)	/* 100 */
-		mode = 3;
+		mask = 0x3f;
 	else 	/* Disabled ? */
 		BUG();
-
-	if (!eighty_ninty_three(drive))
-		mode = min(mode, (u8)1);
-	return mode;
+out:
+	return mask;
 }
 
 /**
@@ -306,7 +302,7 @@ static int siimage_tune_chipset (ide_drive_t *drive, byte xferspeed)
 	ide_hwif_t *hwif	= HWIF(drive);
 	u16 ultra = 0, multi	= 0;
 	u8 mode = 0, unit	= drive->select.b.unit;
-	u8 speed		= ide_rate_filter(siimage_ratemask(drive), xferspeed);
+	u8 speed		= ide_rate_filter(drive, xferspeed);
 	unsigned long base	= (unsigned long)hwif->hwif_data;
 	u8 scsc = 0, addr_mask	= ((hwif->channel) ?
 				    ((hwif->mmio) ? 0xF4 : 0x84) :
@@ -389,7 +385,7 @@ static int siimage_tune_chipset (ide_drive_t *drive, byte xferspeed)
  
 static int config_chipset_for_dma (ide_drive_t *drive)
 {
-	u8 speed	= ide_dma_speed(drive, siimage_ratemask(drive));
+	u8 speed = ide_max_dma_mode(drive);
 
 	if (!speed)
 		return 0;
@@ -831,7 +827,7 @@ static void __devinit init_mmio_iops_siimage(ide_hwif_t *hwif)
 
 	/*
 	 *	Now set up the hw. We have to do this ourselves as
-	 *	the MMIO layout isnt the same as the the standard port
+	 *	the MMIO layout isnt the same as the standard port
 	 *	based I/O
 	 */
 
@@ -989,6 +985,7 @@ static void __devinit init_hwif_siimage(ide_hwif_t *hwif)
 	hwif->tuneproc	= &siimage_tuneproc;
 	hwif->reset_poll = &siimage_reset_poll;
 	hwif->pre_reset = &siimage_pre_reset;
+	hwif->udma_filter = &sil_udma_filter;
 
 	if(is_sata(hwif)) {
 		static int first = 1;
diff --git a/drivers/ide/pci/sis5513.c b/drivers/ide/pci/sis5513.c
index 2ba0669f36a1..2bde1b92784a 100644
--- a/drivers/ide/pci/sis5513.c
+++ b/drivers/ide/pci/sis5513.c
@@ -191,7 +191,7 @@ static char* chipset_capability[] = {
 	"ATA 133 (1st gen)", "ATA 133 (2nd gen)"
 };
 
-#if defined(DISPLAY_SIS_TIMINGS) && defined(CONFIG_PROC_FS)
+#if defined(DISPLAY_SIS_TIMINGS) && defined(CONFIG_IDE_PROC_FS)
 #include <linux/stat.h>
 #include <linux/proc_fs.h>
 
@@ -426,17 +426,7 @@ static int sis_get_info (char *buffer, char **addr, off_t offset, int count)
 
 	return len > count ? count : len;
 }
-#endif /* defined(DISPLAY_SIS_TIMINGS) && defined(CONFIG_PROC_FS) */
-
-static u8 sis5513_ratemask (ide_drive_t *drive)
-{
-	u8 rates[] = { 0, 0, 1, 2, 3, 3, 4, 4 };
-	u8 mode = rates[chipset_family];
-
-	if (!eighty_ninty_three(drive))
-		mode = min(mode, (u8)1);
-	return mode;
-}
+#endif /* defined(DISPLAY_SIS_TIMINGS) && defined(CONFIG_IDE_PROC_FS) */
 
 /*
  * Configuration functions
@@ -563,7 +553,7 @@ static int sis5513_tune_chipset (ide_drive_t *drive, u8 xferspeed)
 	u8 drive_pci, reg, speed;
 	u32 regdw;
 
-	speed = ide_rate_filter(sis5513_ratemask(drive), xferspeed);
+	speed = ide_rate_filter(drive, xferspeed);
 
 	/* See config_art_rwp_pio for drive pci config registers */
 	drive_pci = 0x40;
@@ -648,32 +638,13 @@ static void sis5513_tune_drive (ide_drive_t *drive, u8 pio)
 	(void) config_chipset_for_pio(drive, pio);
 }
 
-/*
- * ((id->hw_config & 0x4000|0x2000) && (HWIF(drive)->udma_four))
- */
-static int config_chipset_for_dma (ide_drive_t *drive)
-{
-	u8 speed	= ide_dma_speed(drive, sis5513_ratemask(drive));
-
-#ifdef DEBUG
-	printk("SIS5513: config_chipset_for_dma, drive %d, ultra %x\n",
-	       drive->dn, drive->id->dma_ultra);
-#endif
-
-	if (!(speed))
-		return 0;
-
-	sis5513_tune_chipset(drive, speed);
-	return ide_dma_enable(drive);
-}
-
 static int sis5513_config_xfer_rate(ide_drive_t *drive)
 {
 	config_art_rwp_pio(drive, 5);
 
 	drive->init_speed = 0;
 
-	if (ide_use_dma(drive) && config_chipset_for_dma(drive))
+	if (ide_tune_dma(drive))
 		return 0;
 
 	if (ide_use_fast_pio(drive))
@@ -826,7 +797,7 @@ static unsigned int __devinit init_chipset_sis5513 (struct pci_dev *dev, const c
 				break;
 		}
 
-#if defined(DISPLAY_SIS_TIMINGS) && defined(CONFIG_PROC_FS)
+#if defined(DISPLAY_SIS_TIMINGS) && defined(CONFIG_IDE_PROC_FS)
 		if (!sis_proc) {
 			sis_proc = 1;
 			bmide_dev = dev;
@@ -858,6 +829,8 @@ static unsigned int __devinit ata66_sis5513 (ide_hwif_t *hwif)
 
 static void __devinit init_hwif_sis5513 (ide_hwif_t *hwif)
 {
+	u8 udma_rates[] = { 0x00, 0x00, 0x07, 0x1f, 0x3f, 0x3f, 0x7f, 0x7f };
+
 	hwif->autodma = 0;
 
 	if (!hwif->irq)
@@ -873,7 +846,8 @@ static void __devinit init_hwif_sis5513 (ide_hwif_t *hwif)
 	}
 
 	hwif->atapi_dma = 1;
-	hwif->ultra_mask = 0x7f;
+
+	hwif->ultra_mask = udma_rates[chipset_family];
 	hwif->mwdma_mask = 0x07;
 	hwif->swdma_mask = 0x07;
 
diff --git a/drivers/ide/pci/slc90e66.c b/drivers/ide/pci/slc90e66.c
index 852ccb36da1d..c40f291f91e0 100644
--- a/drivers/ide/pci/slc90e66.c
+++ b/drivers/ide/pci/slc90e66.c
@@ -21,15 +21,6 @@
 
 #include <asm/io.h>
 
-static u8 slc90e66_ratemask (ide_drive_t *drive)
-{
-	u8 mode	= 2;
-
-	if (!eighty_ninty_three(drive))
-		mode = min_t(u8, mode, 1);
-	return mode;
-}
-
 static u8 slc90e66_dma_2_pio (u8 xfer_rate) {
 	switch(xfer_rate) {
 		case XFER_UDMA_4:
@@ -122,7 +113,7 @@ static int slc90e66_tune_chipset (ide_drive_t *drive, u8 xferspeed)
 	ide_hwif_t *hwif	= HWIF(drive);
 	struct pci_dev *dev	= hwif->pci_dev;
 	u8 maslave		= hwif->channel ? 0x42 : 0x40;
-	u8 speed	= ide_rate_filter(slc90e66_ratemask(drive), xferspeed);
+	u8 speed		= ide_rate_filter(drive, xferspeed);
 	int sitre = 0, a_speed	= 7 << (drive->dn * 4);
 	int u_speed = 0, u_flag = 1 << drive->dn;
 	u16			reg4042, reg44, reg48, reg4a;
@@ -169,22 +160,11 @@ static int slc90e66_tune_chipset (ide_drive_t *drive, u8 xferspeed)
 	return ide_config_drive_speed(drive, speed);
 }
 
-static int slc90e66_config_drive_for_dma (ide_drive_t *drive)
-{
-	u8 speed = ide_dma_speed(drive, slc90e66_ratemask(drive));
-
-	if (!speed)
-		return 0;
-
-	(void) slc90e66_tune_chipset(drive, speed);
-	return ide_dma_enable(drive);
-}
-
 static int slc90e66_config_drive_xfer_rate (ide_drive_t *drive)
 {
 	drive->init_speed = 0;
 
-	if (ide_use_dma(drive) && slc90e66_config_drive_for_dma(drive))
+	if (ide_tune_dma(drive))
 		return 0;
 
 	if (ide_use_fast_pio(drive))
diff --git a/drivers/ide/pci/tc86c001.c b/drivers/ide/pci/tc86c001.c
index 0b6d81d6ce48..cee619bb2eaf 100644
--- a/drivers/ide/pci/tc86c001.c
+++ b/drivers/ide/pci/tc86c001.c
@@ -13,18 +13,13 @@
 #include <linux/pci.h>
 #include <linux/ide.h>
 
-static inline u8 tc86c001_ratemask(ide_drive_t *drive)
-{
-	return eighty_ninty_three(drive) ? 2 : 1;
-}
-
 static int tc86c001_tune_chipset(ide_drive_t *drive, u8 speed)
 {
 	ide_hwif_t *hwif	= HWIF(drive);
 	unsigned long scr_port	= hwif->config_data + (drive->dn ? 0x02 : 0x00);
 	u16 mode, scr		= hwif->INW(scr_port);
 
-	speed = ide_rate_filter(tc86c001_ratemask(drive), speed);
+	speed = ide_rate_filter(drive, speed);
 
 	switch (speed) {
 		case XFER_UDMA_4:	mode = 0x00c0; break;
@@ -172,20 +167,9 @@ static int tc86c001_busproc(ide_drive_t *drive, int state)
 	return 0;
 }
 
-static int config_chipset_for_dma(ide_drive_t *drive)
-{
-	u8 speed = ide_dma_speed(drive, tc86c001_ratemask(drive));
-
-	if (!speed)
-		return 0;
-
-	(void) tc86c001_tune_chipset(drive, speed);
-	return ide_dma_enable(drive);
-}
-
 static int tc86c001_config_drive_xfer_rate(ide_drive_t *drive)
 {
-	if (ide_use_dma(drive) && config_chipset_for_dma(drive))
+	if (ide_tune_dma(drive))
 		return 0;
 
 	if (ide_use_fast_pio(drive))
diff --git a/drivers/ide/pci/triflex.c b/drivers/ide/pci/triflex.c
index 5e06179c3469..35e8c612638f 100644
--- a/drivers/ide/pci/triflex.c
+++ b/drivers/ide/pci/triflex.c
@@ -48,7 +48,7 @@ static int triflex_tune_chipset(ide_drive_t *drive, u8 xferspeed)
 	u16 timing = 0;
 	u32 triflex_timings = 0;
 	u8 unit = (drive->select.b.unit & 0x01);
-	u8 speed = ide_rate_filter(0, xferspeed);
+	u8 speed = ide_rate_filter(drive, xferspeed);
 	
 	pci_read_config_dword(dev, channel_offset, &triflex_timings);
 	
@@ -100,20 +100,9 @@ static void triflex_tune_drive(ide_drive_t *drive, u8 pio)
 	(void) triflex_tune_chipset(drive, (XFER_PIO_0 + use_pio));
 }
 
-static int triflex_config_drive_for_dma(ide_drive_t *drive)
-{
-	int speed = ide_dma_speed(drive, 0); /* No ultra speeds */
-
-	if (!speed)
-		return 0;
-
-	(void) triflex_tune_chipset(drive, speed);
-	 return ide_dma_enable(drive);
-}
-
 static int triflex_config_drive_xfer_rate(ide_drive_t *drive)
 {
-	if (ide_use_dma(drive) && triflex_config_drive_for_dma(drive))
+	if (ide_tune_dma(drive))
 		return 0;
 
 	triflex_tune_drive(drive, 255);
diff --git a/drivers/ide/ppc/pmac.c b/drivers/ide/ppc/pmac.c
index a49ebe44babd..45fc36f0f219 100644
--- a/drivers/ide/ppc/pmac.c
+++ b/drivers/ide/ppc/pmac.c
@@ -1276,6 +1276,8 @@ pmac_ide_setup_device(pmac_ide_hwif_t *pmif, ide_hwif_t *hwif)
 	/* We probe the hwif now */
 	probe_hwif_init(hwif);
 
+	ide_proc_register_port(hwif);
+
 	return 0;
 }
 
diff --git a/drivers/ide/setup-pci.c b/drivers/ide/setup-pci.c
index 118fb3205ca8..67035ba4bf5e 100644
--- a/drivers/ide/setup-pci.c
+++ b/drivers/ide/setup-pci.c
@@ -702,6 +702,7 @@ out:
 
 int ide_setup_pci_device(struct pci_dev *dev, ide_pci_device_t *d)
 {
+	ide_hwif_t *hwif = NULL, *mate = NULL;
 	ata_index_t index_list;
 	int ret;
 
@@ -710,11 +711,19 @@ int ide_setup_pci_device(struct pci_dev *dev, ide_pci_device_t *d)
 		goto out;
 
 	if ((index_list.b.low & 0xf0) != 0xf0)
-		probe_hwif_init_with_fixup(&ide_hwifs[index_list.b.low], d->fixup);
+		hwif = &ide_hwifs[index_list.b.low];
 	if ((index_list.b.high & 0xf0) != 0xf0)
-		probe_hwif_init_with_fixup(&ide_hwifs[index_list.b.high], d->fixup);
+		mate = &ide_hwifs[index_list.b.high];
 
-	create_proc_ide_interfaces();
+	if (hwif)
+		probe_hwif_init_with_fixup(hwif, d->fixup);
+	if (mate)
+		probe_hwif_init_with_fixup(mate, d->fixup);
+
+	if (hwif)
+		ide_proc_register_port(hwif);
+	if (mate)
+		ide_proc_register_port(mate);
 out:
 	return ret;
 }
@@ -748,13 +757,22 @@ int ide_setup_pci_devices(struct pci_dev *dev1, struct pci_dev *dev2,
 		}
 	}
 
-	create_proc_ide_interfaces();
+	for (i = 0; i < 2; i++) {
+		u8 idx[2] = { index_list[i].b.low, index_list[i].b.high };
+		int j;
+
+		for (j = 0; j < 2; j++) {
+			if ((idx[j] & 0xf0) != 0xf0)
+				ide_proc_register_port(ide_hwifs + idx[j]);
+		}
+	}
 out:
 	return ret;
 }
 
 EXPORT_SYMBOL_GPL(ide_setup_pci_devices);
 
+#ifdef CONFIG_IDEPCI_PCIBUS_ORDER
 /*
  *	Module interfaces
  */
@@ -861,3 +879,4 @@ void __init ide_scan_pcibus (int scan_direction)
 		__pci_register_driver(d, d->driver.owner, d->driver.mod_name);
 	}
 }
+#endif
diff --git a/drivers/ieee1394/nodemgr.c b/drivers/ieee1394/nodemgr.c
index 6a1a0572275e..835937e38529 100644
--- a/drivers/ieee1394/nodemgr.c
+++ b/drivers/ieee1394/nodemgr.c
@@ -1702,7 +1702,7 @@ static int nodemgr_host_thread(void *__hi)
 			generation = get_hpsb_generation(host);
 
 			/* If we get a reset before we are done waiting, then
-			 * start the the waiting over again */
+			 * start the waiting over again */
 			if (generation != g)
 				g = generation, i = 0;
 		}
diff --git a/drivers/infiniband/Kconfig b/drivers/infiniband/Kconfig
index 66b36de9fa6f..37deaae49190 100644
--- a/drivers/infiniband/Kconfig
+++ b/drivers/infiniband/Kconfig
@@ -29,6 +29,11 @@ config INFINIBAND_USER_ACCESS
 	  libibverbs, libibcm and a hardware driver library from
 	  <http://www.openib.org>.
 
+config INFINIBAND_USER_MEM
+	bool
+	depends on INFINIBAND_USER_ACCESS != n
+	default y
+
 config INFINIBAND_ADDR_TRANS
 	bool
 	depends on INFINIBAND && INET
@@ -40,6 +45,8 @@ source "drivers/infiniband/hw/ehca/Kconfig"
 source "drivers/infiniband/hw/amso1100/Kconfig"
 source "drivers/infiniband/hw/cxgb3/Kconfig"
 
+source "drivers/infiniband/hw/mlx4/Kconfig"
+
 source "drivers/infiniband/ulp/ipoib/Kconfig"
 
 source "drivers/infiniband/ulp/srp/Kconfig"
diff --git a/drivers/infiniband/Makefile b/drivers/infiniband/Makefile
index da2066c4f22c..75f325e40b54 100644
--- a/drivers/infiniband/Makefile
+++ b/drivers/infiniband/Makefile
@@ -4,6 +4,7 @@ obj-$(CONFIG_INFINIBAND_IPATH)		+= hw/ipath/
 obj-$(CONFIG_INFINIBAND_EHCA)		+= hw/ehca/
 obj-$(CONFIG_INFINIBAND_AMSO1100)	+= hw/amso1100/
 obj-$(CONFIG_INFINIBAND_CXGB3)		+= hw/cxgb3/
+obj-$(CONFIG_MLX4_INFINIBAND)		+= hw/mlx4/
 obj-$(CONFIG_INFINIBAND_IPOIB)		+= ulp/ipoib/
 obj-$(CONFIG_INFINIBAND_SRP)		+= ulp/srp/
 obj-$(CONFIG_INFINIBAND_ISER)		+= ulp/iser/
diff --git a/drivers/infiniband/core/Makefile b/drivers/infiniband/core/Makefile
index 189e5d4b9b17..cb1ab3ea4998 100644
--- a/drivers/infiniband/core/Makefile
+++ b/drivers/infiniband/core/Makefile
@@ -9,6 +9,7 @@ obj-$(CONFIG_INFINIBAND_USER_ACCESS) +=	ib_uverbs.o ib_ucm.o \
 
 ib_core-y :=			packer.o ud_header.o verbs.o sysfs.o \
 				device.o fmr_pool.o cache.o
+ib_core-$(CONFIG_INFINIBAND_USER_MEM) += umem.o
 
 ib_mad-y :=			mad.o smi.o agent.o mad_rmpp.o
 
@@ -28,5 +29,4 @@ ib_umad-y :=			user_mad.o
 
 ib_ucm-y :=			ucm.o
 
-ib_uverbs-y :=			uverbs_main.o uverbs_cmd.o uverbs_mem.o \
-				uverbs_marshall.o
+ib_uverbs-y :=			uverbs_main.o uverbs_cmd.o uverbs_marshall.o
diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index 7fabb425b033..592c90aa3183 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -613,6 +613,8 @@ static void __exit ib_core_cleanup(void)
 {
 	ib_cache_cleanup();
 	ib_sysfs_cleanup();
+	/* Make sure that any pending umem accounting work is done. */
+	flush_scheduled_work();
 }
 
 module_init(ib_core_init);
diff --git a/drivers/infiniband/core/uverbs_mem.c b/drivers/infiniband/core/umem.c
index c95fe952abd5..f32ca5fbb26b 100644
--- a/drivers/infiniband/core/uverbs_mem.c
+++ b/drivers/infiniband/core/umem.c
@@ -39,13 +39,6 @@
 
 #include "uverbs.h"
 
-struct ib_umem_account_work {
-	struct work_struct work;
-	struct mm_struct  *mm;
-	unsigned long      diff;
-};
-
-
 static void __ib_umem_release(struct ib_device *dev, struct ib_umem *umem, int dirty)
 {
 	struct ib_umem_chunk *chunk, *tmp;
@@ -64,35 +57,56 @@ static void __ib_umem_release(struct ib_device *dev, struct ib_umem *umem, int d
 	}
 }
 
-int ib_umem_get(struct ib_device *dev, struct ib_umem *mem,
-		void *addr, size_t size, int write)
+/**
+ * ib_umem_get - Pin and DMA map userspace memory.
+ * @context: userspace context to pin memory for
+ * @addr: userspace virtual address to start at
+ * @size: length of region to pin
+ * @access: IB_ACCESS_xxx flags for memory being pinned
+ */
+struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr,
+			    size_t size, int access)
 {
+	struct ib_umem *umem;
 	struct page **page_list;
 	struct ib_umem_chunk *chunk;
 	unsigned long locked;
 	unsigned long lock_limit;
 	unsigned long cur_base;
 	unsigned long npages;
-	int ret = 0;
+	int ret;
 	int off;
 	int i;
 
 	if (!can_do_mlock())
-		return -EPERM;
+		return ERR_PTR(-EPERM);
 
-	page_list = (struct page **) __get_free_page(GFP_KERNEL);
-	if (!page_list)
-		return -ENOMEM;
+	umem = kmalloc(sizeof *umem, GFP_KERNEL);
+	if (!umem)
+		return ERR_PTR(-ENOMEM);
+
+	umem->context   = context;
+	umem->length    = size;
+	umem->offset    = addr & ~PAGE_MASK;
+	umem->page_size = PAGE_SIZE;
+	/*
+	 * We ask for writable memory if any access flags other than
+	 * "remote read" are set.  "Local write" and "remote write"
+	 * obviously require write access.  "Remote atomic" can do
+	 * things like fetch and add, which will modify memory, and
+	 * "MW bind" can change permissions by binding a window.
+	 */
+	umem->writable  = !!(access & ~IB_ACCESS_REMOTE_READ);
 
-	mem->user_base = (unsigned long) addr;
-	mem->length    = size;
-	mem->offset    = (unsigned long) addr & ~PAGE_MASK;
-	mem->page_size = PAGE_SIZE;
-	mem->writable  = write;
+	INIT_LIST_HEAD(&umem->chunk_list);
 
-	INIT_LIST_HEAD(&mem->chunk_list);
+	page_list = (struct page **) __get_free_page(GFP_KERNEL);
+	if (!page_list) {
+		kfree(umem);
+		return ERR_PTR(-ENOMEM);
+	}
 
-	npages = PAGE_ALIGN(size + mem->offset) >> PAGE_SHIFT;
+	npages = PAGE_ALIGN(size + umem->offset) >> PAGE_SHIFT;
 
 	down_write(&current->mm->mmap_sem);
 
@@ -104,13 +118,13 @@ int ib_umem_get(struct ib_device *dev, struct ib_umem *mem,
 		goto out;
 	}
 
-	cur_base = (unsigned long) addr & PAGE_MASK;
+	cur_base = addr & PAGE_MASK;
 
 	while (npages) {
 		ret = get_user_pages(current, current->mm, cur_base,
 				     min_t(int, npages,
 					   PAGE_SIZE / sizeof (struct page *)),
-				     1, !write, page_list, NULL);
+				     1, !umem->writable, page_list, NULL);
 
 		if (ret < 0)
 			goto out;
@@ -136,7 +150,7 @@ int ib_umem_get(struct ib_device *dev, struct ib_umem *mem,
 				chunk->page_list[i].length = PAGE_SIZE;
 			}
 
-			chunk->nmap = ib_dma_map_sg(dev,
+			chunk->nmap = ib_dma_map_sg(context->device,
 						    &chunk->page_list[0],
 						    chunk->nents,
 						    DMA_BIDIRECTIONAL);
@@ -151,75 +165,94 @@ int ib_umem_get(struct ib_device *dev, struct ib_umem *mem,
 
 			ret -= chunk->nents;
 			off += chunk->nents;
-			list_add_tail(&chunk->list, &mem->chunk_list);
+			list_add_tail(&chunk->list, &umem->chunk_list);
 		}
 
 		ret = 0;
 	}
 
 out:
-	if (ret < 0)
-		__ib_umem_release(dev, mem, 0);
-	else
+	if (ret < 0) {
+		__ib_umem_release(context->device, umem, 0);
+		kfree(umem);
+	} else
 		current->mm->locked_vm = locked;
 
 	up_write(&current->mm->mmap_sem);
 	free_page((unsigned long) page_list);
 
-	return ret;
+	return ret < 0 ? ERR_PTR(ret) : umem;
 }
+EXPORT_SYMBOL(ib_umem_get);
 
-void ib_umem_release(struct ib_device *dev, struct ib_umem *umem)
+static void ib_umem_account(struct work_struct *work)
 {
-	__ib_umem_release(dev, umem, 1);
-
-	down_write(&current->mm->mmap_sem);
-	current->mm->locked_vm -=
-		PAGE_ALIGN(umem->length + umem->offset) >> PAGE_SHIFT;
-	up_write(&current->mm->mmap_sem);
-}
+	struct ib_umem *umem = container_of(work, struct ib_umem, work);
 
-static void ib_umem_account(struct work_struct *_work)
-{
-	struct ib_umem_account_work *work =
-		container_of(_work, struct ib_umem_account_work, work);
-
-	down_write(&work->mm->mmap_sem);
-	work->mm->locked_vm -= work->diff;
-	up_write(&work->mm->mmap_sem);
-	mmput(work->mm);
-	kfree(work);
+	down_write(&umem->mm->mmap_sem);
+	umem->mm->locked_vm -= umem->diff;
+	up_write(&umem->mm->mmap_sem);
+	mmput(umem->mm);
+	kfree(umem);
 }
 
-void ib_umem_release_on_close(struct ib_device *dev, struct ib_umem *umem)
+/**
+ * ib_umem_release - release memory pinned with ib_umem_get
+ * @umem: umem struct to release
+ */
+void ib_umem_release(struct ib_umem *umem)
 {
-	struct ib_umem_account_work *work;
+	struct ib_ucontext *context = umem->context;
 	struct mm_struct *mm;
+	unsigned long diff;
 
-	__ib_umem_release(dev, umem, 1);
+	__ib_umem_release(umem->context->device, umem, 1);
 
 	mm = get_task_mm(current);
 	if (!mm)
 		return;
 
+	diff = PAGE_ALIGN(umem->length + umem->offset) >> PAGE_SHIFT;
+
 	/*
 	 * We may be called with the mm's mmap_sem already held.  This
 	 * can happen when a userspace munmap() is the call that drops
 	 * the last reference to our file and calls our release
 	 * method.  If there are memory regions to destroy, we'll end
-	 * up here and not be able to take the mmap_sem.  Therefore we
-	 * defer the vm_locked accounting to the system workqueue.
+	 * up here and not be able to take the mmap_sem.  In that case
+	 * we defer the vm_locked accounting to the system workqueue.
 	 */
+	if (context->closing && !down_write_trylock(&mm->mmap_sem)) {
+		INIT_WORK(&umem->work, ib_umem_account);
+		umem->mm   = mm;
+		umem->diff = diff;
 
-	work = kmalloc(sizeof *work, GFP_KERNEL);
-	if (!work) {
-		mmput(mm);
+		schedule_work(&umem->work);
 		return;
-	}
+	} else
+		down_write(&mm->mmap_sem);
+
+	current->mm->locked_vm -= diff;
+	up_write(&mm->mmap_sem);
+	mmput(mm);
+	kfree(umem);
+}
+EXPORT_SYMBOL(ib_umem_release);
+
+int ib_umem_page_count(struct ib_umem *umem)
+{
+	struct ib_umem_chunk *chunk;
+	int shift;
+	int i;
+	int n;
+
+	shift = ilog2(umem->page_size);
 
-	INIT_WORK(&work->work, ib_umem_account);
-	work->mm   = mm;
-	work->diff = PAGE_ALIGN(umem->length + umem->offset) >> PAGE_SHIFT;
+	n = 0;
+	list_for_each_entry(chunk, &umem->chunk_list, list)
+		for (i = 0; i < chunk->nmap; ++i)
+			n += sg_dma_len(&chunk->page_list[i]) >> shift;
 
-	schedule_work(&work->work);
+	return n;
 }
+EXPORT_SYMBOL(ib_umem_page_count);
diff --git a/drivers/infiniband/core/uverbs.h b/drivers/infiniband/core/uverbs.h
index 102a59c033ff..c33546f9e961 100644
--- a/drivers/infiniband/core/uverbs.h
+++ b/drivers/infiniband/core/uverbs.h
@@ -45,6 +45,7 @@
 #include <linux/completion.h>
 
 #include <rdma/ib_verbs.h>
+#include <rdma/ib_umem.h>
 #include <rdma/ib_user_verbs.h>
 
 /*
@@ -163,11 +164,6 @@ void ib_uverbs_srq_event_handler(struct ib_event *event, void *context_ptr);
 void ib_uverbs_event_handler(struct ib_event_handler *handler,
 			     struct ib_event *event);
 
-int ib_umem_get(struct ib_device *dev, struct ib_umem *mem,
-		void *addr, size_t size, int write);
-void ib_umem_release(struct ib_device *dev, struct ib_umem *umem);
-void ib_umem_release_on_close(struct ib_device *dev, struct ib_umem *umem);
-
 #define IB_UVERBS_DECLARE_CMD(name)					\
 	ssize_t ib_uverbs_##name(struct ib_uverbs_file *file,		\
 				 const char __user *buf, int in_len,	\
diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index bab66769be14..01d70084aebe 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2005 Topspin Communications.  All rights reserved.
- * Copyright (c) 2005, 2006 Cisco Systems.  All rights reserved.
+ * Copyright (c) 2005, 2006, 2007 Cisco Systems.  All rights reserved.
  * Copyright (c) 2005 PathScale, Inc.  All rights reserved.
  * Copyright (c) 2006 Mellanox Technologies.  All rights reserved.
  *
@@ -295,6 +295,7 @@ ssize_t ib_uverbs_get_context(struct ib_uverbs_file *file,
 	INIT_LIST_HEAD(&ucontext->qp_list);
 	INIT_LIST_HEAD(&ucontext->srq_list);
 	INIT_LIST_HEAD(&ucontext->ah_list);
+	ucontext->closing = 0;
 
 	resp.num_comp_vectors = file->device->num_comp_vectors;
 
@@ -573,7 +574,7 @@ ssize_t ib_uverbs_reg_mr(struct ib_uverbs_file *file,
 	struct ib_uverbs_reg_mr      cmd;
 	struct ib_uverbs_reg_mr_resp resp;
 	struct ib_udata              udata;
-	struct ib_umem_object       *obj;
+	struct ib_uobject           *uobj;
 	struct ib_pd                *pd;
 	struct ib_mr                *mr;
 	int                          ret;
@@ -599,35 +600,21 @@ ssize_t ib_uverbs_reg_mr(struct ib_uverbs_file *file,
 	    !(cmd.access_flags & IB_ACCESS_LOCAL_WRITE))
 		return -EINVAL;
 
-	obj = kmalloc(sizeof *obj, GFP_KERNEL);
-	if (!obj)
+	uobj = kmalloc(sizeof *uobj, GFP_KERNEL);
+	if (!uobj)
 		return -ENOMEM;
 
-	init_uobj(&obj->uobject, 0, file->ucontext, &mr_lock_key);
-	down_write(&obj->uobject.mutex);
-
-	/*
-	 * We ask for writable memory if any access flags other than
-	 * "remote read" are set.  "Local write" and "remote write"
-	 * obviously require write access.  "Remote atomic" can do
-	 * things like fetch and add, which will modify memory, and
-	 * "MW bind" can change permissions by binding a window.
-	 */
-	ret = ib_umem_get(file->device->ib_dev, &obj->umem,
-			  (void *) (unsigned long) cmd.start, cmd.length,
-			  !!(cmd.access_flags & ~IB_ACCESS_REMOTE_READ));
-	if (ret)
-		goto err_free;
-
-	obj->umem.virt_base = cmd.hca_va;
+	init_uobj(uobj, 0, file->ucontext, &mr_lock_key);
+	down_write(&uobj->mutex);
 
 	pd = idr_read_pd(cmd.pd_handle, file->ucontext);
 	if (!pd) {
 		ret = -EINVAL;
-		goto err_release;
+		goto err_free;
 	}
 
-	mr = pd->device->reg_user_mr(pd, &obj->umem, cmd.access_flags, &udata);
+	mr = pd->device->reg_user_mr(pd, cmd.start, cmd.length, cmd.hca_va,
+				     cmd.access_flags, &udata);
 	if (IS_ERR(mr)) {
 		ret = PTR_ERR(mr);
 		goto err_put;
@@ -635,19 +622,19 @@ ssize_t ib_uverbs_reg_mr(struct ib_uverbs_file *file,
 
 	mr->device  = pd->device;
 	mr->pd      = pd;
-	mr->uobject = &obj->uobject;
+	mr->uobject = uobj;
 	atomic_inc(&pd->usecnt);
 	atomic_set(&mr->usecnt, 0);
 
-	obj->uobject.object = mr;
-	ret = idr_add_uobj(&ib_uverbs_mr_idr, &obj->uobject);
+	uobj->object = mr;
+	ret = idr_add_uobj(&ib_uverbs_mr_idr, uobj);
 	if (ret)
 		goto err_unreg;
 
 	memset(&resp, 0, sizeof resp);
 	resp.lkey      = mr->lkey;
 	resp.rkey      = mr->rkey;
-	resp.mr_handle = obj->uobject.id;
+	resp.mr_handle = uobj->id;
 
 	if (copy_to_user((void __user *) (unsigned long) cmd.response,
 			 &resp, sizeof resp)) {
@@ -658,17 +645,17 @@ ssize_t ib_uverbs_reg_mr(struct ib_uverbs_file *file,
 	put_pd_read(pd);
 
 	mutex_lock(&file->mutex);
-	list_add_tail(&obj->uobject.list, &file->ucontext->mr_list);
+	list_add_tail(&uobj->list, &file->ucontext->mr_list);
 	mutex_unlock(&file->mutex);
 
-	obj->uobject.live = 1;
+	uobj->live = 1;
 
-	up_write(&obj->uobject.mutex);
+	up_write(&uobj->mutex);
 
 	return in_len;
 
 err_copy:
-	idr_remove_uobj(&ib_uverbs_mr_idr, &obj->uobject);
+	idr_remove_uobj(&ib_uverbs_mr_idr, uobj);
 
 err_unreg:
 	ib_dereg_mr(mr);
@@ -676,11 +663,8 @@ err_unreg:
 err_put:
 	put_pd_read(pd);
 
-err_release:
-	ib_umem_release(file->device->ib_dev, &obj->umem);
-
 err_free:
-	put_uobj_write(&obj->uobject);
+	put_uobj_write(uobj);
 	return ret;
 }
 
@@ -691,7 +675,6 @@ ssize_t ib_uverbs_dereg_mr(struct ib_uverbs_file *file,
 	struct ib_uverbs_dereg_mr cmd;
 	struct ib_mr             *mr;
 	struct ib_uobject	 *uobj;
-	struct ib_umem_object    *memobj;
 	int                       ret = -EINVAL;
 
 	if (copy_from_user(&cmd, buf, sizeof cmd))
@@ -701,8 +684,7 @@ ssize_t ib_uverbs_dereg_mr(struct ib_uverbs_file *file,
 	if (!uobj)
 		return -EINVAL;
 
-	memobj = container_of(uobj, struct ib_umem_object, uobject);
-	mr     = uobj->object;
+	mr = uobj->object;
 
 	ret = ib_dereg_mr(mr);
 	if (!ret)
@@ -719,8 +701,6 @@ ssize_t ib_uverbs_dereg_mr(struct ib_uverbs_file *file,
 	list_del(&uobj->list);
 	mutex_unlock(&file->mutex);
 
-	ib_umem_release(file->device->ib_dev, &memobj->umem);
-
 	put_uobj(uobj);
 
 	return in_len;
diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c
index d44e54799651..14d7ccd89195 100644
--- a/drivers/infiniband/core/uverbs_main.c
+++ b/drivers/infiniband/core/uverbs_main.c
@@ -183,6 +183,8 @@ static int ib_uverbs_cleanup_ucontext(struct ib_uverbs_file *file,
 	if (!context)
 		return 0;
 
+	context->closing = 1;
+
 	list_for_each_entry_safe(uobj, tmp, &context->ah_list, list) {
 		struct ib_ah *ah = uobj->object;
 
@@ -230,16 +232,10 @@ static int ib_uverbs_cleanup_ucontext(struct ib_uverbs_file *file,
 
 	list_for_each_entry_safe(uobj, tmp, &context->mr_list, list) {
 		struct ib_mr *mr = uobj->object;
-		struct ib_device *mrdev = mr->device;
-		struct ib_umem_object *memobj;
 
 		idr_remove_uobj(&ib_uverbs_mr_idr, uobj);
 		ib_dereg_mr(mr);
-
-		memobj = container_of(uobj, struct ib_umem_object, uobject);
-		ib_umem_release_on_close(mrdev, &memobj->umem);
-
-		kfree(memobj);
+		kfree(uobj);
 	}
 
 	list_for_each_entry_safe(uobj, tmp, &context->pd_list, list) {
@@ -906,7 +902,6 @@ static void __exit ib_uverbs_cleanup(void)
 	unregister_filesystem(&uverbs_event_fs);
 	class_destroy(uverbs_class);
 	unregister_chrdev_region(IB_UVERBS_BASE_DEV, IB_UVERBS_MAX_DEVICES);
-	flush_scheduled_work();
 	idr_destroy(&ib_uverbs_pd_idr);
 	idr_destroy(&ib_uverbs_mr_idr);
 	idr_destroy(&ib_uverbs_mw_idr);
diff --git a/drivers/infiniband/hw/amso1100/c2_provider.c b/drivers/infiniband/hw/amso1100/c2_provider.c
index 109166223c09..997cf1530762 100644
--- a/drivers/infiniband/hw/amso1100/c2_provider.c
+++ b/drivers/infiniband/hw/amso1100/c2_provider.c
@@ -56,6 +56,7 @@
 #include <asm/byteorder.h>
 
 #include <rdma/ib_smi.h>
+#include <rdma/ib_umem.h>
 #include <rdma/ib_user_verbs.h>
 #include "c2.h"
 #include "c2_provider.h"
@@ -396,6 +397,7 @@ static struct ib_mr *c2_reg_phys_mr(struct ib_pd *ib_pd,
 	}
 
 	mr->pd = to_c2pd(ib_pd);
+	mr->umem = NULL;
 	pr_debug("%s - page shift %d, pbl_depth %d, total_len %u, "
 		"*iova_start %llx, first pa %llx, last pa %llx\n",
 		__FUNCTION__, page_shift, pbl_depth, total_len,
@@ -428,8 +430,8 @@ static struct ib_mr *c2_get_dma_mr(struct ib_pd *pd, int acc)
 	return c2_reg_phys_mr(pd, &bl, 1, acc, &kva);
 }
 
-static struct ib_mr *c2_reg_user_mr(struct ib_pd *pd, struct ib_umem *region,
-				    int acc, struct ib_udata *udata)
+static struct ib_mr *c2_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
+				    u64 virt, int acc, struct ib_udata *udata)
 {
 	u64 *pages;
 	u64 kva = 0;
@@ -441,15 +443,23 @@ static struct ib_mr *c2_reg_user_mr(struct ib_pd *pd, struct ib_umem *region,
 	struct c2_mr *c2mr;
 
 	pr_debug("%s:%u\n", __FUNCTION__, __LINE__);
-	shift = ffs(region->page_size) - 1;
 
 	c2mr = kmalloc(sizeof(*c2mr), GFP_KERNEL);
 	if (!c2mr)
 		return ERR_PTR(-ENOMEM);
 	c2mr->pd = c2pd;
 
+	c2mr->umem = ib_umem_get(pd->uobject->context, start, length, acc);
+	if (IS_ERR(c2mr->umem)) {
+		err = PTR_ERR(c2mr->umem);
+		kfree(c2mr);
+		return ERR_PTR(err);
+	}
+
+	shift = ffs(c2mr->umem->page_size) - 1;
+
 	n = 0;
-	list_for_each_entry(chunk, &region->chunk_list, list)
+	list_for_each_entry(chunk, &c2mr->umem->chunk_list, list)
 		n += chunk->nents;
 
 	pages = kmalloc(n * sizeof(u64), GFP_KERNEL);
@@ -459,35 +469,34 @@ static struct ib_mr *c2_reg_user_mr(struct ib_pd *pd, struct ib_umem *region,
 	}
 
 	i = 0;
-	list_for_each_entry(chunk, &region->chunk_list, list) {
+	list_for_each_entry(chunk, &c2mr->umem->chunk_list, list) {
 		for (j = 0; j < chunk->nmap; ++j) {
 			len = sg_dma_len(&chunk->page_list[j]) >> shift;
 			for (k = 0; k < len; ++k) {
 				pages[i++] =
 					sg_dma_address(&chunk->page_list[j]) +
-					(region->page_size * k);
+					(c2mr->umem->page_size * k);
 			}
 		}
 	}
 
-	kva = (u64)region->virt_base;
+	kva = virt;
   	err = c2_nsmr_register_phys_kern(to_c2dev(pd->device),
 					 pages,
- 					 region->page_size,
+					 c2mr->umem->page_size,
 					 i,
-					 region->length,
-					 region->offset,
+					 length,
+					 c2mr->umem->offset,
 					 &kva,
 					 c2_convert_access(acc),
 					 c2mr);
 	kfree(pages);
-	if (err) {
-		kfree(c2mr);
-		return ERR_PTR(err);
-	}
+	if (err)
+		goto err;
 	return &c2mr->ibmr;
 
 err:
+	ib_umem_release(c2mr->umem);
 	kfree(c2mr);
 	return ERR_PTR(err);
 }
@@ -502,8 +511,11 @@ static int c2_dereg_mr(struct ib_mr *ib_mr)
 	err = c2_stag_dealloc(to_c2dev(ib_mr->device), ib_mr->lkey);
 	if (err)
 		pr_debug("c2_stag_dealloc failed: %d\n", err);
-	else
+	else {
+		if (mr->umem)
+			ib_umem_release(mr->umem);
 		kfree(mr);
+	}
 
 	return err;
 }
diff --git a/drivers/infiniband/hw/amso1100/c2_provider.h b/drivers/infiniband/hw/amso1100/c2_provider.h
index fc906223220f..1076df2ee96a 100644
--- a/drivers/infiniband/hw/amso1100/c2_provider.h
+++ b/drivers/infiniband/hw/amso1100/c2_provider.h
@@ -73,6 +73,7 @@ struct c2_pd {
 struct c2_mr {
 	struct ib_mr ibmr;
 	struct c2_pd *pd;
+	struct ib_umem *umem;
 };
 
 struct c2_av;
diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c
index a891493fd340..e7c2c3948037 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_provider.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c
@@ -47,6 +47,7 @@
 #include <rdma/iw_cm.h>
 #include <rdma/ib_verbs.h>
 #include <rdma/ib_smi.h>
+#include <rdma/ib_umem.h>
 #include <rdma/ib_user_verbs.h>
 
 #include "cxio_hal.h"
@@ -443,6 +444,8 @@ static int iwch_dereg_mr(struct ib_mr *ib_mr)
 	remove_handle(rhp, &rhp->mmidr, mmid);
 	if (mhp->kva)
 		kfree((void *) (unsigned long) mhp->kva);
+	if (mhp->umem)
+		ib_umem_release(mhp->umem);
 	PDBG("%s mmid 0x%x ptr %p\n", __FUNCTION__, mmid, mhp);
 	kfree(mhp);
 	return 0;
@@ -577,8 +580,8 @@ static int iwch_reregister_phys_mem(struct ib_mr *mr,
 }
 
 
-static struct ib_mr *iwch_reg_user_mr(struct ib_pd *pd, struct ib_umem *region,
-				      int acc, struct ib_udata *udata)
+static struct ib_mr *iwch_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
+				      u64 virt, int acc, struct ib_udata *udata)
 {
 	__be64 *pages;
 	int shift, n, len;
@@ -591,7 +594,6 @@ static struct ib_mr *iwch_reg_user_mr(struct ib_pd *pd, struct ib_umem *region,
 	struct iwch_reg_user_mr_resp uresp;
 
 	PDBG("%s ib_pd %p\n", __FUNCTION__, pd);
-	shift = ffs(region->page_size) - 1;
 
 	php = to_iwch_pd(pd);
 	rhp = php->rhp;
@@ -599,8 +601,17 @@ static struct ib_mr *iwch_reg_user_mr(struct ib_pd *pd, struct ib_umem *region,
 	if (!mhp)
 		return ERR_PTR(-ENOMEM);
 
+	mhp->umem = ib_umem_get(pd->uobject->context, start, length, acc);
+	if (IS_ERR(mhp->umem)) {
+		err = PTR_ERR(mhp->umem);
+		kfree(mhp);
+		return ERR_PTR(err);
+	}
+
+	shift = ffs(mhp->umem->page_size) - 1;
+
 	n = 0;
-	list_for_each_entry(chunk, &region->chunk_list, list)
+	list_for_each_entry(chunk, &mhp->umem->chunk_list, list)
 		n += chunk->nents;
 
 	pages = kmalloc(n * sizeof(u64), GFP_KERNEL);
@@ -611,13 +622,13 @@ static struct ib_mr *iwch_reg_user_mr(struct ib_pd *pd, struct ib_umem *region,
 
 	i = n = 0;
 
-	list_for_each_entry(chunk, &region->chunk_list, list)
+	list_for_each_entry(chunk, &mhp->umem->chunk_list, list)
 		for (j = 0; j < chunk->nmap; ++j) {
 			len = sg_dma_len(&chunk->page_list[j]) >> shift;
 			for (k = 0; k < len; ++k) {
 				pages[i++] = cpu_to_be64(sg_dma_address(
 					&chunk->page_list[j]) +
-					region->page_size * k);
+					mhp->umem->page_size * k);
 			}
 		}
 
@@ -625,9 +636,9 @@ static struct ib_mr *iwch_reg_user_mr(struct ib_pd *pd, struct ib_umem *region,
 	mhp->attr.pdid = php->pdid;
 	mhp->attr.zbva = 0;
 	mhp->attr.perms = iwch_ib_to_tpt_access(acc);
-	mhp->attr.va_fbo = region->virt_base;
+	mhp->attr.va_fbo = virt;
 	mhp->attr.page_size = shift - 12;
-	mhp->attr.len = (u32) region->length;
+	mhp->attr.len = (u32) length;
 	mhp->attr.pbl_size = i;
 	err = iwch_register_mem(rhp, php, mhp, shift, pages);
 	kfree(pages);
@@ -650,6 +661,7 @@ static struct ib_mr *iwch_reg_user_mr(struct ib_pd *pd, struct ib_umem *region,
 	return &mhp->ibmr;
 
 err:
+	ib_umem_release(mhp->umem);
 	kfree(mhp);
 	return ERR_PTR(err);
 }
diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.h b/drivers/infiniband/hw/cxgb3/iwch_provider.h
index 93bcc56756bd..48833f3f3bd0 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_provider.h
+++ b/drivers/infiniband/hw/cxgb3/iwch_provider.h
@@ -73,6 +73,7 @@ struct tpt_attributes {
 
 struct iwch_mr {
 	struct ib_mr ibmr;
+	struct ib_umem *umem;
 	struct iwch_dev *rhp;
 	u64 kva;
 	struct tpt_attributes attr;
diff --git a/drivers/infiniband/hw/ehca/ehca_classes.h b/drivers/infiniband/hw/ehca/ehca_classes.h
index 10fb8fbafa0c..f64d42b08674 100644
--- a/drivers/infiniband/hw/ehca/ehca_classes.h
+++ b/drivers/infiniband/hw/ehca/ehca_classes.h
@@ -176,6 +176,7 @@ struct ehca_mr {
 		struct ib_mr ib_mr;	/* must always be first in ehca_mr */
 		struct ib_fmr ib_fmr;	/* must always be first in ehca_mr */
 	} ib;
+	struct ib_umem *umem;
 	spinlock_t mrlock;
 
 	enum ehca_mr_flag flags;
diff --git a/drivers/infiniband/hw/ehca/ehca_irq.c b/drivers/infiniband/hw/ehca/ehca_irq.c
index f284be1c9166..82dda2faf4d0 100644
--- a/drivers/infiniband/hw/ehca/ehca_irq.c
+++ b/drivers/infiniband/hw/ehca/ehca_irq.c
@@ -745,6 +745,7 @@ static int comp_pool_callback(struct notifier_block *nfb,
 
 	switch (action) {
 	case CPU_UP_PREPARE:
+	case CPU_UP_PREPARE_FROZEN:
 		ehca_gen_dbg("CPU: %x (CPU_PREPARE)", cpu);
 		if(!create_comp_task(pool, cpu)) {
 			ehca_gen_err("Can't create comp_task for cpu: %x", cpu);
@@ -752,24 +753,29 @@ static int comp_pool_callback(struct notifier_block *nfb,
 		}
 		break;
 	case CPU_UP_CANCELED:
+	case CPU_UP_CANCELED_FROZEN:
 		ehca_gen_dbg("CPU: %x (CPU_CANCELED)", cpu);
 		cct = per_cpu_ptr(pool->cpu_comp_tasks, cpu);
 		kthread_bind(cct->task, any_online_cpu(cpu_online_map));
 		destroy_comp_task(pool, cpu);
 		break;
 	case CPU_ONLINE:
+	case CPU_ONLINE_FROZEN:
 		ehca_gen_dbg("CPU: %x (CPU_ONLINE)", cpu);
 		cct = per_cpu_ptr(pool->cpu_comp_tasks, cpu);
 		kthread_bind(cct->task, cpu);
 		wake_up_process(cct->task);
 		break;
 	case CPU_DOWN_PREPARE:
+	case CPU_DOWN_PREPARE_FROZEN:
 		ehca_gen_dbg("CPU: %x (CPU_DOWN_PREPARE)", cpu);
 		break;
 	case CPU_DOWN_FAILED:
+	case CPU_DOWN_FAILED_FROZEN:
 		ehca_gen_dbg("CPU: %x (CPU_DOWN_FAILED)", cpu);
 		break;
 	case CPU_DEAD:
+	case CPU_DEAD_FROZEN:
 		ehca_gen_dbg("CPU: %x (CPU_DEAD)", cpu);
 		destroy_comp_task(pool, cpu);
 		take_over_work(pool, cpu);
diff --git a/drivers/infiniband/hw/ehca/ehca_iverbs.h b/drivers/infiniband/hw/ehca/ehca_iverbs.h
index e14b029332c8..37e7fe0908cf 100644
--- a/drivers/infiniband/hw/ehca/ehca_iverbs.h
+++ b/drivers/infiniband/hw/ehca/ehca_iverbs.h
@@ -78,8 +78,7 @@ struct ib_mr *ehca_reg_phys_mr(struct ib_pd *pd,
 			       int num_phys_buf,
 			       int mr_access_flags, u64 *iova_start);
 
-struct ib_mr *ehca_reg_user_mr(struct ib_pd *pd,
-			       struct ib_umem *region,
+struct ib_mr *ehca_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, u64 virt,
 			       int mr_access_flags, struct ib_udata *udata);
 
 int ehca_rereg_phys_mr(struct ib_mr *mr,
diff --git a/drivers/infiniband/hw/ehca/ehca_mrmw.c b/drivers/infiniband/hw/ehca/ehca_mrmw.c
index d22ab563633f..84c5bb498563 100644
--- a/drivers/infiniband/hw/ehca/ehca_mrmw.c
+++ b/drivers/infiniband/hw/ehca/ehca_mrmw.c
@@ -39,6 +39,8 @@
  * POSSIBILITY OF SUCH DAMAGE.
  */
 
+#include <rdma/ib_umem.h>
+
 #include <asm/current.h>
 
 #include "ehca_iverbs.h"
@@ -238,10 +240,8 @@ reg_phys_mr_exit0:
 
 /*----------------------------------------------------------------------*/
 
-struct ib_mr *ehca_reg_user_mr(struct ib_pd *pd,
-			       struct ib_umem *region,
-			       int mr_access_flags,
-			       struct ib_udata *udata)
+struct ib_mr *ehca_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, u64 virt,
+			       int mr_access_flags, struct ib_udata *udata)
 {
 	struct ib_mr *ib_mr;
 	struct ehca_mr *e_mr;
@@ -257,11 +257,7 @@ struct ib_mr *ehca_reg_user_mr(struct ib_pd *pd,
 		ehca_gen_err("bad pd=%p", pd);
 		return ERR_PTR(-EFAULT);
 	}
-	if (!region) {
-		ehca_err(pd->device, "bad input values: region=%p", region);
-		ib_mr = ERR_PTR(-EINVAL);
-		goto reg_user_mr_exit0;
-	}
+
 	if (((mr_access_flags & IB_ACCESS_REMOTE_WRITE) &&
 	     !(mr_access_flags & IB_ACCESS_LOCAL_WRITE)) ||
 	    ((mr_access_flags & IB_ACCESS_REMOTE_ATOMIC) &&
@@ -275,17 +271,10 @@ struct ib_mr *ehca_reg_user_mr(struct ib_pd *pd,
 		ib_mr = ERR_PTR(-EINVAL);
 		goto reg_user_mr_exit0;
 	}
-	if (region->page_size != PAGE_SIZE) {
-		ehca_err(pd->device, "page size not supported, "
-			 "region->page_size=%x", region->page_size);
-		ib_mr = ERR_PTR(-EINVAL);
-		goto reg_user_mr_exit0;
-	}
 
-	if ((region->length == 0) ||
-	    ((region->virt_base + region->length) < region->virt_base)) {
+	if (length == 0 || virt + length < virt) {
 		ehca_err(pd->device, "bad input values: length=%lx "
-			 "virt_base=%lx", region->length, region->virt_base);
+			 "virt_base=%lx", length, virt);
 		ib_mr = ERR_PTR(-EINVAL);
 		goto reg_user_mr_exit0;
 	}
@@ -297,40 +286,55 @@ struct ib_mr *ehca_reg_user_mr(struct ib_pd *pd,
 		goto reg_user_mr_exit0;
 	}
 
+	e_mr->umem = ib_umem_get(pd->uobject->context, start, length,
+				 mr_access_flags);
+	if (IS_ERR(e_mr->umem)) {
+		ib_mr = (void *) e_mr->umem;
+		goto reg_user_mr_exit1;
+	}
+
+	if (e_mr->umem->page_size != PAGE_SIZE) {
+		ehca_err(pd->device, "page size not supported, "
+			 "e_mr->umem->page_size=%x", e_mr->umem->page_size);
+		ib_mr = ERR_PTR(-EINVAL);
+		goto reg_user_mr_exit2;
+	}
+
 	/* determine number of MR pages */
-	num_pages_mr = (((region->virt_base % PAGE_SIZE) + region->length +
-			 PAGE_SIZE - 1) / PAGE_SIZE);
-	num_pages_4k = (((region->virt_base % EHCA_PAGESIZE) + region->length +
-			 EHCA_PAGESIZE - 1) / EHCA_PAGESIZE);
+	num_pages_mr = (((virt % PAGE_SIZE) + length + PAGE_SIZE - 1) /
+			PAGE_SIZE);
+	num_pages_4k = (((virt % EHCA_PAGESIZE) + length + EHCA_PAGESIZE - 1) /
+			EHCA_PAGESIZE);
 
 	/* register MR on HCA */
 	pginfo.type       = EHCA_MR_PGI_USER;
 	pginfo.num_pages  = num_pages_mr;
 	pginfo.num_4k     = num_pages_4k;
-	pginfo.region     = region;
-	pginfo.next_4k	  = region->offset / EHCA_PAGESIZE;
+	pginfo.region     = e_mr->umem;
+	pginfo.next_4k	  = e_mr->umem->offset / EHCA_PAGESIZE;
 	pginfo.next_chunk = list_prepare_entry(pginfo.next_chunk,
-					       (&region->chunk_list),
+					       (&e_mr->umem->chunk_list),
 					       list);
 
-	ret = ehca_reg_mr(shca, e_mr, (u64*)region->virt_base,
-			  region->length, mr_access_flags, e_pd, &pginfo,
-			  &e_mr->ib.ib_mr.lkey, &e_mr->ib.ib_mr.rkey);
+	ret = ehca_reg_mr(shca, e_mr, (u64*) virt, length, mr_access_flags, e_pd,
+			  &pginfo, &e_mr->ib.ib_mr.lkey, &e_mr->ib.ib_mr.rkey);
 	if (ret) {
 		ib_mr = ERR_PTR(ret);
-		goto reg_user_mr_exit1;
+		goto reg_user_mr_exit2;
 	}
 
 	/* successful registration of all pages */
 	return &e_mr->ib.ib_mr;
 
+reg_user_mr_exit2:
+	ib_umem_release(e_mr->umem);
 reg_user_mr_exit1:
 	ehca_mr_delete(e_mr);
 reg_user_mr_exit0:
 	if (IS_ERR(ib_mr))
-		ehca_err(pd->device, "rc=%lx pd=%p region=%p mr_access_flags=%x"
+		ehca_err(pd->device, "rc=%lx pd=%p mr_access_flags=%x"
 			 " udata=%p",
-			 PTR_ERR(ib_mr), pd, region, mr_access_flags, udata);
+			 PTR_ERR(ib_mr), pd, mr_access_flags, udata);
 	return ib_mr;
 } /* end ehca_reg_user_mr() */
 
@@ -596,6 +600,9 @@ int ehca_dereg_mr(struct ib_mr *mr)
 		goto dereg_mr_exit0;
 	}
 
+	if (e_mr->umem)
+		ib_umem_release(e_mr->umem);
+
 	/* successful deregistration */
 	ehca_mr_delete(e_mr);
 
diff --git a/drivers/infiniband/hw/ipath/ipath_mr.c b/drivers/infiniband/hw/ipath/ipath_mr.c
index 31e70732e369..bdeef8d4f279 100644
--- a/drivers/infiniband/hw/ipath/ipath_mr.c
+++ b/drivers/infiniband/hw/ipath/ipath_mr.c
@@ -31,6 +31,7 @@
  * SOFTWARE.
  */
 
+#include <rdma/ib_umem.h>
 #include <rdma/ib_pack.h>
 #include <rdma/ib_smi.h>
 
@@ -147,6 +148,7 @@ struct ib_mr *ipath_reg_phys_mr(struct ib_pd *pd,
 	mr->mr.offset = 0;
 	mr->mr.access_flags = acc;
 	mr->mr.max_segs = num_phys_buf;
+	mr->umem = NULL;
 
 	m = 0;
 	n = 0;
@@ -170,46 +172,56 @@ bail:
 /**
  * ipath_reg_user_mr - register a userspace memory region
  * @pd: protection domain for this memory region
- * @region: the user memory region
+ * @start: starting userspace address
+ * @length: length of region to register
+ * @virt_addr: virtual address to use (from HCA's point of view)
  * @mr_access_flags: access flags for this memory region
  * @udata: unused by the InfiniPath driver
  *
  * Returns the memory region on success, otherwise returns an errno.
  */
-struct ib_mr *ipath_reg_user_mr(struct ib_pd *pd, struct ib_umem *region,
-				int mr_access_flags, struct ib_udata *udata)
+struct ib_mr *ipath_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
+				u64 virt_addr, int mr_access_flags,
+				struct ib_udata *udata)
 {
 	struct ipath_mr *mr;
+	struct ib_umem *umem;
 	struct ib_umem_chunk *chunk;
 	int n, m, i;
 	struct ib_mr *ret;
 
-	if (region->length == 0) {
+	if (length == 0) {
 		ret = ERR_PTR(-EINVAL);
 		goto bail;
 	}
 
+	umem = ib_umem_get(pd->uobject->context, start, length, mr_access_flags);
+	if (IS_ERR(umem))
+		return (void *) umem;
+
 	n = 0;
-	list_for_each_entry(chunk, &region->chunk_list, list)
+	list_for_each_entry(chunk, &umem->chunk_list, list)
 		n += chunk->nents;
 
 	mr = alloc_mr(n, &to_idev(pd->device)->lk_table);
 	if (!mr) {
 		ret = ERR_PTR(-ENOMEM);
+		ib_umem_release(umem);
 		goto bail;
 	}
 
 	mr->mr.pd = pd;
-	mr->mr.user_base = region->user_base;
-	mr->mr.iova = region->virt_base;
-	mr->mr.length = region->length;
-	mr->mr.offset = region->offset;
+	mr->mr.user_base = start;
+	mr->mr.iova = virt_addr;
+	mr->mr.length = length;
+	mr->mr.offset = umem->offset;
 	mr->mr.access_flags = mr_access_flags;
 	mr->mr.max_segs = n;
+	mr->umem = umem;
 
 	m = 0;
 	n = 0;
-	list_for_each_entry(chunk, &region->chunk_list, list) {
+	list_for_each_entry(chunk, &umem->chunk_list, list) {
 		for (i = 0; i < chunk->nents; i++) {
 			void *vaddr;
 
@@ -219,7 +231,7 @@ struct ib_mr *ipath_reg_user_mr(struct ib_pd *pd, struct ib_umem *region,
 				goto bail;
 			}
 			mr->mr.map[m]->segs[n].vaddr = vaddr;
-			mr->mr.map[m]->segs[n].length = region->page_size;
+			mr->mr.map[m]->segs[n].length = umem->page_size;
 			n++;
 			if (n == IPATH_SEGSZ) {
 				m++;
@@ -253,6 +265,10 @@ int ipath_dereg_mr(struct ib_mr *ibmr)
 		i--;
 		kfree(mr->mr.map[i]);
 	}
+
+	if (mr->umem)
+		ib_umem_release(mr->umem);
+
 	kfree(mr);
 	return 0;
 }
diff --git a/drivers/infiniband/hw/ipath/ipath_verbs.h b/drivers/infiniband/hw/ipath/ipath_verbs.h
index 7064fc222727..088b837ebea8 100644
--- a/drivers/infiniband/hw/ipath/ipath_verbs.h
+++ b/drivers/infiniband/hw/ipath/ipath_verbs.h
@@ -251,6 +251,7 @@ struct ipath_sge {
 /* Memory region */
 struct ipath_mr {
 	struct ib_mr ibmr;
+	struct ib_umem *umem;
 	struct ipath_mregion mr;	/* must be last */
 };
 
@@ -751,8 +752,8 @@ struct ib_mr *ipath_reg_phys_mr(struct ib_pd *pd,
 				struct ib_phys_buf *buffer_list,
 				int num_phys_buf, int acc, u64 *iova_start);
 
-struct ib_mr *ipath_reg_user_mr(struct ib_pd *pd, struct ib_umem *region,
-				int mr_access_flags,
+struct ib_mr *ipath_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
+				u64 virt_addr, int mr_access_flags,
 				struct ib_udata *udata);
 
 int ipath_dereg_mr(struct ib_mr *ibmr);
diff --git a/drivers/infiniband/hw/mlx4/Kconfig b/drivers/infiniband/hw/mlx4/Kconfig
new file mode 100644
index 000000000000..b8912cdb9663
--- /dev/null
+++ b/drivers/infiniband/hw/mlx4/Kconfig
@@ -0,0 +1,9 @@
+config MLX4_INFINIBAND
+	tristate "Mellanox ConnectX HCA support"
+	depends on INFINIBAND
+	select MLX4_CORE
+	---help---
+	  This driver provides low-level InfiniBand support for
+	  Mellanox ConnectX PCI Express host channel adapters (HCAs).
+	  This is required to use InfiniBand protocols such as
+	  IP-over-IB or SRP with these devices.
diff --git a/drivers/infiniband/hw/mlx4/Makefile b/drivers/infiniband/hw/mlx4/Makefile
new file mode 100644
index 000000000000..70f09c7826da
--- /dev/null
+++ b/drivers/infiniband/hw/mlx4/Makefile
@@ -0,0 +1,3 @@
+obj-$(CONFIG_MLX4_INFINIBAND)	+= mlx4_ib.o
+
+mlx4_ib-y :=	ah.o cq.o doorbell.o mad.o main.o mr.o qp.o srq.o
diff --git a/drivers/infiniband/hw/mlx4/ah.c b/drivers/infiniband/hw/mlx4/ah.c
new file mode 100644
index 000000000000..c75ac9463e20
--- /dev/null
+++ b/drivers/infiniband/hw/mlx4/ah.c
@@ -0,0 +1,100 @@
+/*
+ * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "mlx4_ib.h"
+
+struct ib_ah *mlx4_ib_create_ah(struct ib_pd *pd, struct ib_ah_attr *ah_attr)
+{
+	struct mlx4_dev *dev = to_mdev(pd->device)->dev;
+	struct mlx4_ib_ah *ah;
+
+	ah = kmalloc(sizeof *ah, GFP_ATOMIC);
+	if (!ah)
+		return ERR_PTR(-ENOMEM);
+
+	memset(&ah->av, 0, sizeof ah->av);
+
+	ah->av.port_pd = cpu_to_be32(to_mpd(pd)->pdn | (ah_attr->port_num << 24));
+	ah->av.g_slid  = ah_attr->src_path_bits;
+	ah->av.dlid    = cpu_to_be16(ah_attr->dlid);
+	if (ah_attr->static_rate) {
+		ah->av.stat_rate = ah_attr->static_rate + MLX4_STAT_RATE_OFFSET;
+		while (ah->av.stat_rate > IB_RATE_2_5_GBPS + MLX4_STAT_RATE_OFFSET &&
+		       !(1 << ah->av.stat_rate & dev->caps.stat_rate_support))
+			--ah->av.stat_rate;
+	}
+	ah->av.sl_tclass_flowlabel = cpu_to_be32(ah_attr->sl << 28);
+	if (ah_attr->ah_flags & IB_AH_GRH) {
+		ah->av.g_slid   |= 0x80;
+		ah->av.gid_index = ah_attr->grh.sgid_index;
+		ah->av.hop_limit = ah_attr->grh.hop_limit;
+		ah->av.sl_tclass_flowlabel |=
+			cpu_to_be32((ah_attr->grh.traffic_class << 20) |
+				    ah_attr->grh.flow_label);
+		memcpy(ah->av.dgid, ah_attr->grh.dgid.raw, 16);
+	}
+
+	return &ah->ibah;
+}
+
+int mlx4_ib_query_ah(struct ib_ah *ibah, struct ib_ah_attr *ah_attr)
+{
+	struct mlx4_ib_ah *ah = to_mah(ibah);
+
+	memset(ah_attr, 0, sizeof *ah_attr);
+	ah_attr->dlid	       = be16_to_cpu(ah->av.dlid);
+	ah_attr->sl	       = be32_to_cpu(ah->av.sl_tclass_flowlabel) >> 28;
+	ah_attr->port_num      = be32_to_cpu(ah->av.port_pd) >> 24;
+	if (ah->av.stat_rate)
+		ah_attr->static_rate = ah->av.stat_rate - MLX4_STAT_RATE_OFFSET;
+	ah_attr->src_path_bits = ah->av.g_slid & 0x7F;
+
+	if (mlx4_ib_ah_grh_present(ah)) {
+		ah_attr->ah_flags = IB_AH_GRH;
+
+		ah_attr->grh.traffic_class =
+			be32_to_cpu(ah->av.sl_tclass_flowlabel) >> 20;
+		ah_attr->grh.flow_label =
+			be32_to_cpu(ah->av.sl_tclass_flowlabel) & 0xfffff;
+		ah_attr->grh.hop_limit  = ah->av.hop_limit;
+		ah_attr->grh.sgid_index = ah->av.gid_index;
+		memcpy(ah_attr->grh.dgid.raw, ah->av.dgid, 16);
+	}
+
+	return 0;
+}
+
+int mlx4_ib_destroy_ah(struct ib_ah *ah)
+{
+	kfree(to_mah(ah));
+	return 0;
+}
diff --git a/drivers/infiniband/hw/mlx4/cq.c b/drivers/infiniband/hw/mlx4/cq.c
new file mode 100644
index 000000000000..b2a290c6703a
--- /dev/null
+++ b/drivers/infiniband/hw/mlx4/cq.c
@@ -0,0 +1,525 @@
+/*
+ * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx4/cq.h>
+#include <linux/mlx4/qp.h>
+
+#include "mlx4_ib.h"
+#include "user.h"
+
+static void mlx4_ib_cq_comp(struct mlx4_cq *cq)
+{
+	struct ib_cq *ibcq = &to_mibcq(cq)->ibcq;
+	ibcq->comp_handler(ibcq, ibcq->cq_context);
+}
+
+static void mlx4_ib_cq_event(struct mlx4_cq *cq, enum mlx4_event type)
+{
+	struct ib_event event;
+	struct ib_cq *ibcq;
+
+	if (type != MLX4_EVENT_TYPE_CQ_ERROR) {
+		printk(KERN_WARNING "mlx4_ib: Unexpected event type %d "
+		       "on CQ %06x\n", type, cq->cqn);
+		return;
+	}
+
+	ibcq = &to_mibcq(cq)->ibcq;
+	if (ibcq->event_handler) {
+		event.device     = ibcq->device;
+		event.event      = IB_EVENT_CQ_ERR;
+		event.element.cq = ibcq;
+		ibcq->event_handler(&event, ibcq->cq_context);
+	}
+}
+
+static void *get_cqe_from_buf(struct mlx4_ib_cq_buf *buf, int n)
+{
+	int offset = n * sizeof (struct mlx4_cqe);
+
+	if (buf->buf.nbufs == 1)
+		return buf->buf.u.direct.buf + offset;
+	else
+		return buf->buf.u.page_list[offset >> PAGE_SHIFT].buf +
+			(offset & (PAGE_SIZE - 1));
+}
+
+static void *get_cqe(struct mlx4_ib_cq *cq, int n)
+{
+	return get_cqe_from_buf(&cq->buf, n);
+}
+
+static void *get_sw_cqe(struct mlx4_ib_cq *cq, int n)
+{
+	struct mlx4_cqe *cqe = get_cqe(cq, n & cq->ibcq.cqe);
+
+	return (!!(cqe->owner_sr_opcode & MLX4_CQE_OWNER_MASK) ^
+		!!(n & (cq->ibcq.cqe + 1))) ? NULL : cqe;
+}
+
+static struct mlx4_cqe *next_cqe_sw(struct mlx4_ib_cq *cq)
+{
+	return get_sw_cqe(cq, cq->mcq.cons_index);
+}
+
+struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector,
+				struct ib_ucontext *context,
+				struct ib_udata *udata)
+{
+	struct mlx4_ib_dev *dev = to_mdev(ibdev);
+	struct mlx4_ib_cq *cq;
+	struct mlx4_uar *uar;
+	int buf_size;
+	int err;
+
+	if (entries < 1 || entries > dev->dev->caps.max_cqes)
+		return ERR_PTR(-EINVAL);
+
+	cq = kmalloc(sizeof *cq, GFP_KERNEL);
+	if (!cq)
+		return ERR_PTR(-ENOMEM);
+
+	entries      = roundup_pow_of_two(entries + 1);
+	cq->ibcq.cqe = entries - 1;
+	buf_size     = entries * sizeof (struct mlx4_cqe);
+	spin_lock_init(&cq->lock);
+
+	if (context) {
+		struct mlx4_ib_create_cq ucmd;
+
+		if (ib_copy_from_udata(&ucmd, udata, sizeof ucmd)) {
+			err = -EFAULT;
+			goto err_cq;
+		}
+
+		cq->umem = ib_umem_get(context, ucmd.buf_addr, buf_size,
+				       IB_ACCESS_LOCAL_WRITE);
+		if (IS_ERR(cq->umem)) {
+			err = PTR_ERR(cq->umem);
+			goto err_cq;
+		}
+
+		err = mlx4_mtt_init(dev->dev, ib_umem_page_count(cq->umem),
+				    ilog2(cq->umem->page_size), &cq->buf.mtt);
+		if (err)
+			goto err_buf;
+
+		err = mlx4_ib_umem_write_mtt(dev, &cq->buf.mtt, cq->umem);
+		if (err)
+			goto err_mtt;
+
+		err = mlx4_ib_db_map_user(to_mucontext(context), ucmd.db_addr,
+					  &cq->db);
+		if (err)
+			goto err_mtt;
+
+		uar = &to_mucontext(context)->uar;
+	} else {
+		err = mlx4_ib_db_alloc(dev, &cq->db, 1);
+		if (err)
+			goto err_cq;
+
+		cq->mcq.set_ci_db  = cq->db.db;
+		cq->mcq.arm_db     = cq->db.db + 1;
+		*cq->mcq.set_ci_db = 0;
+		*cq->mcq.arm_db    = 0;
+
+		if (mlx4_buf_alloc(dev->dev, buf_size, PAGE_SIZE * 2, &cq->buf.buf)) {
+			err = -ENOMEM;
+			goto err_db;
+		}
+
+		err = mlx4_mtt_init(dev->dev, cq->buf.buf.npages, cq->buf.buf.page_shift,
+				    &cq->buf.mtt);
+		if (err)
+			goto err_buf;
+
+		err = mlx4_buf_write_mtt(dev->dev, &cq->buf.mtt, &cq->buf.buf);
+		if (err)
+			goto err_mtt;
+
+		uar = &dev->priv_uar;
+	}
+
+	err = mlx4_cq_alloc(dev->dev, entries, &cq->buf.mtt, uar,
+			    cq->db.dma, &cq->mcq);
+	if (err)
+		goto err_dbmap;
+
+	cq->mcq.comp  = mlx4_ib_cq_comp;
+	cq->mcq.event = mlx4_ib_cq_event;
+
+	if (context)
+		if (ib_copy_to_udata(udata, &cq->mcq.cqn, sizeof (__u32))) {
+			err = -EFAULT;
+			goto err_dbmap;
+		}
+
+	return &cq->ibcq;
+
+err_dbmap:
+	if (context)
+		mlx4_ib_db_unmap_user(to_mucontext(context), &cq->db);
+
+err_mtt:
+	mlx4_mtt_cleanup(dev->dev, &cq->buf.mtt);
+
+err_buf:
+	if (context)
+		ib_umem_release(cq->umem);
+	else
+		mlx4_buf_free(dev->dev, entries * sizeof (struct mlx4_cqe),
+			      &cq->buf.buf);
+
+err_db:
+	if (!context)
+		mlx4_ib_db_free(dev, &cq->db);
+
+err_cq:
+	kfree(cq);
+
+	return ERR_PTR(err);
+}
+
+int mlx4_ib_destroy_cq(struct ib_cq *cq)
+{
+	struct mlx4_ib_dev *dev = to_mdev(cq->device);
+	struct mlx4_ib_cq *mcq = to_mcq(cq);
+
+	mlx4_cq_free(dev->dev, &mcq->mcq);
+	mlx4_mtt_cleanup(dev->dev, &mcq->buf.mtt);
+
+	if (cq->uobject) {
+		mlx4_ib_db_unmap_user(to_mucontext(cq->uobject->context), &mcq->db);
+		ib_umem_release(mcq->umem);
+	} else {
+		mlx4_buf_free(dev->dev, (cq->cqe + 1) * sizeof (struct mlx4_cqe),
+			      &mcq->buf.buf);
+		mlx4_ib_db_free(dev, &mcq->db);
+	}
+
+	kfree(mcq);
+
+	return 0;
+}
+
+static void dump_cqe(void *cqe)
+{
+	__be32 *buf = cqe;
+
+	printk(KERN_DEBUG "CQE contents %08x %08x %08x %08x %08x %08x %08x %08x\n",
+	       be32_to_cpu(buf[0]), be32_to_cpu(buf[1]), be32_to_cpu(buf[2]),
+	       be32_to_cpu(buf[3]), be32_to_cpu(buf[4]), be32_to_cpu(buf[5]),
+	       be32_to_cpu(buf[6]), be32_to_cpu(buf[7]));
+}
+
+static void mlx4_ib_handle_error_cqe(struct mlx4_err_cqe *cqe,
+				     struct ib_wc *wc)
+{
+	if (cqe->syndrome == MLX4_CQE_SYNDROME_LOCAL_QP_OP_ERR) {
+		printk(KERN_DEBUG "local QP operation err "
+		       "(QPN %06x, WQE index %x, vendor syndrome %02x, "
+		       "opcode = %02x)\n",
+		       be32_to_cpu(cqe->my_qpn), be16_to_cpu(cqe->wqe_index),
+		       cqe->vendor_err_syndrome,
+		       cqe->owner_sr_opcode & ~MLX4_CQE_OWNER_MASK);
+		dump_cqe(cqe);
+	}
+
+	switch (cqe->syndrome) {
+	case MLX4_CQE_SYNDROME_LOCAL_LENGTH_ERR:
+		wc->status = IB_WC_LOC_LEN_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_LOCAL_QP_OP_ERR:
+		wc->status = IB_WC_LOC_QP_OP_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_LOCAL_PROT_ERR:
+		wc->status = IB_WC_LOC_PROT_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_WR_FLUSH_ERR:
+		wc->status = IB_WC_WR_FLUSH_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_MW_BIND_ERR:
+		wc->status = IB_WC_MW_BIND_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_BAD_RESP_ERR:
+		wc->status = IB_WC_BAD_RESP_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_LOCAL_ACCESS_ERR:
+		wc->status = IB_WC_LOC_ACCESS_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_REMOTE_INVAL_REQ_ERR:
+		wc->status = IB_WC_REM_INV_REQ_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_REMOTE_ACCESS_ERR:
+		wc->status = IB_WC_REM_ACCESS_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_REMOTE_OP_ERR:
+		wc->status = IB_WC_REM_OP_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_TRANSPORT_RETRY_EXC_ERR:
+		wc->status = IB_WC_RETRY_EXC_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_RNR_RETRY_EXC_ERR:
+		wc->status = IB_WC_RNR_RETRY_EXC_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_REMOTE_ABORTED_ERR:
+		wc->status = IB_WC_REM_ABORT_ERR;
+		break;
+	default:
+		wc->status = IB_WC_GENERAL_ERR;
+		break;
+	}
+
+	wc->vendor_err = cqe->vendor_err_syndrome;
+}
+
+static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
+			    struct mlx4_ib_qp **cur_qp,
+			    struct ib_wc *wc)
+{
+	struct mlx4_cqe *cqe;
+	struct mlx4_qp *mqp;
+	struct mlx4_ib_wq *wq;
+	struct mlx4_ib_srq *srq;
+	int is_send;
+	int is_error;
+	u16 wqe_ctr;
+
+	cqe = next_cqe_sw(cq);
+	if (!cqe)
+		return -EAGAIN;
+
+	++cq->mcq.cons_index;
+
+	/*
+	 * Make sure we read CQ entry contents after we've checked the
+	 * ownership bit.
+	 */
+	rmb();
+
+	is_send  = cqe->owner_sr_opcode & MLX4_CQE_IS_SEND_MASK;
+	is_error = (cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) ==
+		MLX4_CQE_OPCODE_ERROR;
+
+	if (!*cur_qp ||
+	    (be32_to_cpu(cqe->my_qpn) & 0xffffff) != (*cur_qp)->mqp.qpn) {
+		/*
+		 * We do not have to take the QP table lock here,
+		 * because CQs will be locked while QPs are removed
+		 * from the table.
+		 */
+		mqp = __mlx4_qp_lookup(to_mdev(cq->ibcq.device)->dev,
+				       be32_to_cpu(cqe->my_qpn));
+		if (unlikely(!mqp)) {
+			printk(KERN_WARNING "CQ %06x with entry for unknown QPN %06x\n",
+			       cq->mcq.cqn, be32_to_cpu(cqe->my_qpn) & 0xffffff);
+			return -EINVAL;
+		}
+
+		*cur_qp = to_mibqp(mqp);
+	}
+
+	wc->qp = &(*cur_qp)->ibqp;
+
+	if (is_send) {
+		wq = &(*cur_qp)->sq;
+		wqe_ctr = be16_to_cpu(cqe->wqe_index);
+		wq->tail += wqe_ctr - (u16) wq->tail;
+		wc->wr_id = wq->wrid[wq->tail & (wq->max - 1)];
+		++wq->tail;
+	} else if ((*cur_qp)->ibqp.srq) {
+		srq = to_msrq((*cur_qp)->ibqp.srq);
+		wqe_ctr = be16_to_cpu(cqe->wqe_index);
+		wc->wr_id = srq->wrid[wqe_ctr];
+		mlx4_ib_free_srq_wqe(srq, wqe_ctr);
+	} else {
+		wq	  = &(*cur_qp)->rq;
+		wc->wr_id = wq->wrid[wq->tail & (wq->max - 1)];
+		++wq->tail;
+	}
+
+	if (unlikely(is_error)) {
+		mlx4_ib_handle_error_cqe((struct mlx4_err_cqe *) cqe, wc);
+		return 0;
+	}
+
+	wc->status = IB_WC_SUCCESS;
+
+	if (is_send) {
+		wc->wc_flags = 0;
+		switch (cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) {
+		case MLX4_OPCODE_RDMA_WRITE_IMM:
+			wc->wc_flags |= IB_WC_WITH_IMM;
+		case MLX4_OPCODE_RDMA_WRITE:
+			wc->opcode    = IB_WC_RDMA_WRITE;
+			break;
+		case MLX4_OPCODE_SEND_IMM:
+			wc->wc_flags |= IB_WC_WITH_IMM;
+		case MLX4_OPCODE_SEND:
+			wc->opcode    = IB_WC_SEND;
+			break;
+		case MLX4_OPCODE_RDMA_READ:
+			wc->opcode    = IB_WC_SEND;
+			wc->byte_len  = be32_to_cpu(cqe->byte_cnt);
+			break;
+		case MLX4_OPCODE_ATOMIC_CS:
+			wc->opcode    = IB_WC_COMP_SWAP;
+			wc->byte_len  = 8;
+			break;
+		case MLX4_OPCODE_ATOMIC_FA:
+			wc->opcode    = IB_WC_FETCH_ADD;
+			wc->byte_len  = 8;
+			break;
+		case MLX4_OPCODE_BIND_MW:
+			wc->opcode    = IB_WC_BIND_MW;
+			break;
+		}
+	} else {
+		wc->byte_len = be32_to_cpu(cqe->byte_cnt);
+
+		switch (cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) {
+		case MLX4_RECV_OPCODE_RDMA_WRITE_IMM:
+			wc->opcode   = IB_WC_RECV_RDMA_WITH_IMM;
+			wc->wc_flags = IB_WC_WITH_IMM;
+			wc->imm_data = cqe->immed_rss_invalid;
+			break;
+		case MLX4_RECV_OPCODE_SEND:
+			wc->opcode   = IB_WC_RECV;
+			wc->wc_flags = 0;
+			break;
+		case MLX4_RECV_OPCODE_SEND_IMM:
+			wc->opcode   = IB_WC_RECV;
+			wc->wc_flags = IB_WC_WITH_IMM;
+			wc->imm_data = cqe->immed_rss_invalid;
+			break;
+		}
+
+		wc->slid	   = be16_to_cpu(cqe->rlid);
+		wc->sl		   = cqe->sl >> 4;
+		wc->src_qp	   = be32_to_cpu(cqe->g_mlpath_rqpn) & 0xffffff;
+		wc->dlid_path_bits = (be32_to_cpu(cqe->g_mlpath_rqpn) >> 24) & 0x7f;
+		wc->wc_flags      |= be32_to_cpu(cqe->g_mlpath_rqpn) & 0x80000000 ?
+			IB_WC_GRH : 0;
+		wc->pkey_index     = be32_to_cpu(cqe->immed_rss_invalid) >> 16;
+	}
+
+	return 0;
+}
+
+int mlx4_ib_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc)
+{
+	struct mlx4_ib_cq *cq = to_mcq(ibcq);
+	struct mlx4_ib_qp *cur_qp = NULL;
+	unsigned long flags;
+	int npolled;
+	int err = 0;
+
+	spin_lock_irqsave(&cq->lock, flags);
+
+	for (npolled = 0; npolled < num_entries; ++npolled) {
+		err = mlx4_ib_poll_one(cq, &cur_qp, wc + npolled);
+		if (err)
+			break;
+	}
+
+	if (npolled)
+		mlx4_cq_set_ci(&cq->mcq);
+
+	spin_unlock_irqrestore(&cq->lock, flags);
+
+	if (err == 0 || err == -EAGAIN)
+		return npolled;
+	else
+		return err;
+}
+
+int mlx4_ib_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify_flags flags)
+{
+	mlx4_cq_arm(&to_mcq(ibcq)->mcq,
+		    (flags & IB_CQ_SOLICITED_MASK) == IB_CQ_SOLICITED ?
+		    MLX4_CQ_DB_REQ_NOT_SOL : MLX4_CQ_DB_REQ_NOT,
+		    to_mdev(ibcq->device)->uar_map,
+		    MLX4_GET_DOORBELL_LOCK(&to_mdev(ibcq->device)->uar_lock));
+
+	return 0;
+}
+
+void __mlx4_ib_cq_clean(struct mlx4_ib_cq *cq, u32 qpn, struct mlx4_ib_srq *srq)
+{
+	u32 prod_index;
+	int nfreed = 0;
+	struct mlx4_cqe *cqe;
+
+	/*
+	 * First we need to find the current producer index, so we
+	 * know where to start cleaning from.  It doesn't matter if HW
+	 * adds new entries after this loop -- the QP we're worried
+	 * about is already in RESET, so the new entries won't come
+	 * from our QP and therefore don't need to be checked.
+	 */
+	for (prod_index = cq->mcq.cons_index; get_sw_cqe(cq, prod_index); ++prod_index)
+		if (prod_index == cq->mcq.cons_index + cq->ibcq.cqe)
+			break;
+
+	/*
+	 * Now sweep backwards through the CQ, removing CQ entries
+	 * that match our QP by copying older entries on top of them.
+	 */
+	while ((int) --prod_index - (int) cq->mcq.cons_index >= 0) {
+		cqe = get_cqe(cq, prod_index & cq->ibcq.cqe);
+		if ((be32_to_cpu(cqe->my_qpn) & 0xffffff) == qpn) {
+			if (srq && !(cqe->owner_sr_opcode & MLX4_CQE_IS_SEND_MASK))
+				mlx4_ib_free_srq_wqe(srq, be16_to_cpu(cqe->wqe_index));
+			++nfreed;
+		} else if (nfreed)
+			memcpy(get_cqe(cq, (prod_index + nfreed) & cq->ibcq.cqe),
+			       cqe, sizeof *cqe);
+	}
+
+	if (nfreed) {
+		cq->mcq.cons_index += nfreed;
+		/*
+		 * Make sure update of buffer contents is done before
+		 * updating consumer index.
+		 */
+		wmb();
+		mlx4_cq_set_ci(&cq->mcq);
+	}
+}
+
+void mlx4_ib_cq_clean(struct mlx4_ib_cq *cq, u32 qpn, struct mlx4_ib_srq *srq)
+{
+	spin_lock_irq(&cq->lock);
+	__mlx4_ib_cq_clean(cq, qpn, srq);
+	spin_unlock_irq(&cq->lock);
+}
diff --git a/drivers/infiniband/hw/mlx4/doorbell.c b/drivers/infiniband/hw/mlx4/doorbell.c
new file mode 100644
index 000000000000..1c36087aef14
--- /dev/null
+++ b/drivers/infiniband/hw/mlx4/doorbell.c
@@ -0,0 +1,216 @@
+/*
+ * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/slab.h>
+
+#include "mlx4_ib.h"
+
+struct mlx4_ib_db_pgdir {
+	struct list_head	list;
+	DECLARE_BITMAP(order0, MLX4_IB_DB_PER_PAGE);
+	DECLARE_BITMAP(order1, MLX4_IB_DB_PER_PAGE / 2);
+	unsigned long	       *bits[2];
+	__be32		       *db_page;
+	dma_addr_t		db_dma;
+};
+
+static struct mlx4_ib_db_pgdir *mlx4_ib_alloc_db_pgdir(struct mlx4_ib_dev *dev)
+{
+	struct mlx4_ib_db_pgdir *pgdir;
+
+	pgdir = kzalloc(sizeof *pgdir, GFP_KERNEL);
+	if (!pgdir)
+		return NULL;
+
+	bitmap_fill(pgdir->order1, MLX4_IB_DB_PER_PAGE / 2);
+	pgdir->bits[0] = pgdir->order0;
+	pgdir->bits[1] = pgdir->order1;
+	pgdir->db_page = dma_alloc_coherent(dev->ib_dev.dma_device,
+					    PAGE_SIZE, &pgdir->db_dma,
+					    GFP_KERNEL);
+	if (!pgdir->db_page) {
+		kfree(pgdir);
+		return NULL;
+	}
+
+	return pgdir;
+}
+
+static int mlx4_ib_alloc_db_from_pgdir(struct mlx4_ib_db_pgdir *pgdir,
+				       struct mlx4_ib_db *db, int order)
+{
+	int o;
+	int i;
+
+	for (o = order; o <= 1; ++o) {
+		i = find_first_bit(pgdir->bits[o], MLX4_IB_DB_PER_PAGE >> o);
+		if (i < MLX4_IB_DB_PER_PAGE >> o)
+			goto found;
+	}
+
+	return -ENOMEM;
+
+found:
+	clear_bit(i, pgdir->bits[o]);
+
+	i <<= o;
+
+	if (o > order)
+		set_bit(i ^ 1, pgdir->bits[order]);
+
+	db->u.pgdir = pgdir;
+	db->index   = i;
+	db->db      = pgdir->db_page + db->index;
+	db->dma     = pgdir->db_dma  + db->index * 4;
+	db->order   = order;
+
+	return 0;
+}
+
+int mlx4_ib_db_alloc(struct mlx4_ib_dev *dev, struct mlx4_ib_db *db, int order)
+{
+	struct mlx4_ib_db_pgdir *pgdir;
+	int ret = 0;
+
+	mutex_lock(&dev->pgdir_mutex);
+
+	list_for_each_entry(pgdir, &dev->pgdir_list, list)
+		if (!mlx4_ib_alloc_db_from_pgdir(pgdir, db, order))
+			goto out;
+
+	pgdir = mlx4_ib_alloc_db_pgdir(dev);
+	if (!pgdir) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	list_add(&pgdir->list, &dev->pgdir_list);
+
+	/* This should never fail -- we just allocated an empty page: */
+	WARN_ON(mlx4_ib_alloc_db_from_pgdir(pgdir, db, order));
+
+out:
+	mutex_unlock(&dev->pgdir_mutex);
+
+	return ret;
+}
+
+void mlx4_ib_db_free(struct mlx4_ib_dev *dev, struct mlx4_ib_db *db)
+{
+	int o;
+	int i;
+
+	mutex_lock(&dev->pgdir_mutex);
+
+	o = db->order;
+	i = db->index;
+
+	if (db->order == 0 && test_bit(i ^ 1, db->u.pgdir->order0)) {
+		clear_bit(i ^ 1, db->u.pgdir->order0);
+		++o;
+	}
+
+	i >>= o;
+	set_bit(i, db->u.pgdir->bits[o]);
+
+	if (bitmap_full(db->u.pgdir->order1, MLX4_IB_DB_PER_PAGE / 2)) {
+		dma_free_coherent(dev->ib_dev.dma_device, PAGE_SIZE,
+				  db->u.pgdir->db_page, db->u.pgdir->db_dma);
+		list_del(&db->u.pgdir->list);
+		kfree(db->u.pgdir);
+	}
+
+	mutex_unlock(&dev->pgdir_mutex);
+}
+
+struct mlx4_ib_user_db_page {
+	struct list_head	list;
+	struct ib_umem	       *umem;
+	unsigned long		user_virt;
+	int			refcnt;
+};
+
+int mlx4_ib_db_map_user(struct mlx4_ib_ucontext *context, unsigned long virt,
+			struct mlx4_ib_db *db)
+{
+	struct mlx4_ib_user_db_page *page;
+	struct ib_umem_chunk *chunk;
+	int err = 0;
+
+	mutex_lock(&context->db_page_mutex);
+
+	list_for_each_entry(page, &context->db_page_list, list)
+		if (page->user_virt == (virt & PAGE_MASK))
+			goto found;
+
+	page = kmalloc(sizeof *page, GFP_KERNEL);
+	if (!page) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	page->user_virt = (virt & PAGE_MASK);
+	page->refcnt    = 0;
+	page->umem      = ib_umem_get(&context->ibucontext, virt & PAGE_MASK,
+				      PAGE_SIZE, 0);
+	if (IS_ERR(page->umem)) {
+		err = PTR_ERR(page->umem);
+		kfree(page);
+		goto out;
+	}
+
+	list_add(&page->list, &context->db_page_list);
+
+found:
+	chunk = list_entry(page->umem->chunk_list.next, struct ib_umem_chunk, list);
+	db->dma		= sg_dma_address(chunk->page_list) + (virt & ~PAGE_MASK);
+	db->u.user_page = page;
+	++page->refcnt;
+
+out:
+	mutex_unlock(&context->db_page_mutex);
+
+	return err;
+}
+
+void mlx4_ib_db_unmap_user(struct mlx4_ib_ucontext *context, struct mlx4_ib_db *db)
+{
+	mutex_lock(&context->db_page_mutex);
+
+	if (!--db->u.user_page->refcnt) {
+		list_del(&db->u.user_page->list);
+		ib_umem_release(db->u.user_page->umem);
+		kfree(db->u.user_page);
+	}
+
+	mutex_unlock(&context->db_page_mutex);
+}
diff --git a/drivers/infiniband/hw/mlx4/mad.c b/drivers/infiniband/hw/mlx4/mad.c
new file mode 100644
index 000000000000..333091787c5f
--- /dev/null
+++ b/drivers/infiniband/hw/mlx4/mad.c
@@ -0,0 +1,339 @@
+/*
+ * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <rdma/ib_mad.h>
+#include <rdma/ib_smi.h>
+
+#include <linux/mlx4/cmd.h>
+
+#include "mlx4_ib.h"
+
+enum {
+	MLX4_IB_VENDOR_CLASS1 = 0x9,
+	MLX4_IB_VENDOR_CLASS2 = 0xa
+};
+
+int mlx4_MAD_IFC(struct mlx4_ib_dev *dev, int ignore_mkey, int ignore_bkey,
+		 int port, struct ib_wc *in_wc, struct ib_grh *in_grh,
+		 void *in_mad, void *response_mad)
+{
+	struct mlx4_cmd_mailbox *inmailbox, *outmailbox;
+	void *inbox;
+	int err;
+	u32 in_modifier = port;
+	u8 op_modifier = 0;
+
+	inmailbox = mlx4_alloc_cmd_mailbox(dev->dev);
+	if (IS_ERR(inmailbox))
+		return PTR_ERR(inmailbox);
+	inbox = inmailbox->buf;
+
+	outmailbox = mlx4_alloc_cmd_mailbox(dev->dev);
+	if (IS_ERR(outmailbox)) {
+		mlx4_free_cmd_mailbox(dev->dev, inmailbox);
+		return PTR_ERR(outmailbox);
+	}
+
+	memcpy(inbox, in_mad, 256);
+
+	/*
+	 * Key check traps can't be generated unless we have in_wc to
+	 * tell us where to send the trap.
+	 */
+	if (ignore_mkey || !in_wc)
+		op_modifier |= 0x1;
+	if (ignore_bkey || !in_wc)
+		op_modifier |= 0x2;
+
+	if (in_wc) {
+		struct {
+			__be32		my_qpn;
+			u32		reserved1;
+			__be32		rqpn;
+			u8		sl;
+			u8		g_path;
+			u16		reserved2[2];
+			__be16		pkey;
+			u32		reserved3[11];
+			u8		grh[40];
+		} *ext_info;
+
+		memset(inbox + 256, 0, 256);
+		ext_info = inbox + 256;
+
+		ext_info->my_qpn = cpu_to_be32(in_wc->qp->qp_num);
+		ext_info->rqpn   = cpu_to_be32(in_wc->src_qp);
+		ext_info->sl     = in_wc->sl << 4;
+		ext_info->g_path = in_wc->dlid_path_bits |
+			(in_wc->wc_flags & IB_WC_GRH ? 0x80 : 0);
+		ext_info->pkey   = cpu_to_be16(in_wc->pkey_index);
+
+		if (in_grh)
+			memcpy(ext_info->grh, in_grh, 40);
+
+		op_modifier |= 0x4;
+
+		in_modifier |= in_wc->slid << 16;
+	}
+
+	err = mlx4_cmd_box(dev->dev, inmailbox->dma, outmailbox->dma,
+			   in_modifier, op_modifier,
+			   MLX4_CMD_MAD_IFC, MLX4_CMD_TIME_CLASS_C);
+
+	if (!err);
+		memcpy(response_mad, outmailbox->buf, 256);
+
+	mlx4_free_cmd_mailbox(dev->dev, inmailbox);
+	mlx4_free_cmd_mailbox(dev->dev, outmailbox);
+
+	return err;
+}
+
+static void update_sm_ah(struct mlx4_ib_dev *dev, u8 port_num, u16 lid, u8 sl)
+{
+	struct ib_ah *new_ah;
+	struct ib_ah_attr ah_attr;
+
+	if (!dev->send_agent[port_num - 1][0])
+		return;
+
+	memset(&ah_attr, 0, sizeof ah_attr);
+	ah_attr.dlid     = lid;
+	ah_attr.sl       = sl;
+	ah_attr.port_num = port_num;
+
+	new_ah = ib_create_ah(dev->send_agent[port_num - 1][0]->qp->pd,
+			      &ah_attr);
+	if (IS_ERR(new_ah))
+		return;
+
+	spin_lock(&dev->sm_lock);
+	if (dev->sm_ah[port_num - 1])
+		ib_destroy_ah(dev->sm_ah[port_num - 1]);
+	dev->sm_ah[port_num - 1] = new_ah;
+	spin_unlock(&dev->sm_lock);
+}
+
+/*
+ * Snoop SM MADs for port info and P_Key table sets, so we can
+ * synthesize LID change and P_Key change events.
+ */
+static void smp_snoop(struct ib_device *ibdev, u8 port_num, struct ib_mad *mad)
+{
+	struct ib_event event;
+
+	if ((mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_LID_ROUTED ||
+	     mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) &&
+	    mad->mad_hdr.method == IB_MGMT_METHOD_SET) {
+		if (mad->mad_hdr.attr_id == IB_SMP_ATTR_PORT_INFO) {
+			struct ib_port_info *pinfo =
+				(struct ib_port_info *) ((struct ib_smp *) mad)->data;
+
+			update_sm_ah(to_mdev(ibdev), port_num,
+				     be16_to_cpu(pinfo->sm_lid),
+				     pinfo->neighbormtu_mastersmsl & 0xf);
+
+			event.device	       = ibdev;
+			event.element.port_num = port_num;
+
+			if(pinfo->clientrereg_resv_subnetto & 0x80)
+				event.event    = IB_EVENT_CLIENT_REREGISTER;
+			else
+				event.event    = IB_EVENT_LID_CHANGE;
+
+			ib_dispatch_event(&event);
+		}
+
+		if (mad->mad_hdr.attr_id == IB_SMP_ATTR_PKEY_TABLE) {
+			event.device	       = ibdev;
+			event.event	       = IB_EVENT_PKEY_CHANGE;
+			event.element.port_num = port_num;
+			ib_dispatch_event(&event);
+		}
+	}
+}
+
+static void node_desc_override(struct ib_device *dev,
+			       struct ib_mad *mad)
+{
+	if ((mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_LID_ROUTED ||
+	     mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) &&
+	    mad->mad_hdr.method == IB_MGMT_METHOD_GET_RESP &&
+	    mad->mad_hdr.attr_id == IB_SMP_ATTR_NODE_DESC) {
+		spin_lock(&to_mdev(dev)->sm_lock);
+		memcpy(((struct ib_smp *) mad)->data, dev->node_desc, 64);
+		spin_unlock(&to_mdev(dev)->sm_lock);
+	}
+}
+
+static void forward_trap(struct mlx4_ib_dev *dev, u8 port_num, struct ib_mad *mad)
+{
+	int qpn = mad->mad_hdr.mgmt_class != IB_MGMT_CLASS_SUBN_LID_ROUTED;
+	struct ib_mad_send_buf *send_buf;
+	struct ib_mad_agent *agent = dev->send_agent[port_num - 1][qpn];
+	int ret;
+
+	if (agent) {
+		send_buf = ib_create_send_mad(agent, qpn, 0, 0, IB_MGMT_MAD_HDR,
+					      IB_MGMT_MAD_DATA, GFP_ATOMIC);
+		/*
+		 * We rely here on the fact that MLX QPs don't use the
+		 * address handle after the send is posted (this is
+		 * wrong following the IB spec strictly, but we know
+		 * it's OK for our devices).
+		 */
+		spin_lock(&dev->sm_lock);
+		memcpy(send_buf->mad, mad, sizeof *mad);
+		if ((send_buf->ah = dev->sm_ah[port_num - 1]))
+			ret = ib_post_send_mad(send_buf, NULL);
+		else
+			ret = -EINVAL;
+		spin_unlock(&dev->sm_lock);
+
+		if (ret)
+			ib_free_send_mad(send_buf);
+	}
+}
+
+int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags,	u8 port_num,
+			struct ib_wc *in_wc, struct ib_grh *in_grh,
+			struct ib_mad *in_mad, struct ib_mad *out_mad)
+{
+	u16 slid;
+	int err;
+
+	slid = in_wc ? in_wc->slid : be16_to_cpu(IB_LID_PERMISSIVE);
+
+	if (in_mad->mad_hdr.method == IB_MGMT_METHOD_TRAP && slid == 0) {
+		forward_trap(to_mdev(ibdev), port_num, in_mad);
+		return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_CONSUMED;
+	}
+
+	if (in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_LID_ROUTED ||
+	    in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) {
+		if (in_mad->mad_hdr.method   != IB_MGMT_METHOD_GET &&
+		    in_mad->mad_hdr.method   != IB_MGMT_METHOD_SET &&
+		    in_mad->mad_hdr.method   != IB_MGMT_METHOD_TRAP_REPRESS)
+			return IB_MAD_RESULT_SUCCESS;
+
+		/*
+		 * Don't process SMInfo queries or vendor-specific
+		 * MADs -- the SMA can't handle them.
+		 */
+		if (in_mad->mad_hdr.attr_id == IB_SMP_ATTR_SM_INFO ||
+		    ((in_mad->mad_hdr.attr_id & IB_SMP_ATTR_VENDOR_MASK) ==
+		     IB_SMP_ATTR_VENDOR_MASK))
+			return IB_MAD_RESULT_SUCCESS;
+	} else if (in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_PERF_MGMT ||
+		   in_mad->mad_hdr.mgmt_class == MLX4_IB_VENDOR_CLASS1   ||
+		   in_mad->mad_hdr.mgmt_class == MLX4_IB_VENDOR_CLASS2) {
+		if (in_mad->mad_hdr.method  != IB_MGMT_METHOD_GET &&
+		    in_mad->mad_hdr.method  != IB_MGMT_METHOD_SET)
+			return IB_MAD_RESULT_SUCCESS;
+	} else
+		return IB_MAD_RESULT_SUCCESS;
+
+	err = mlx4_MAD_IFC(to_mdev(ibdev),
+			   mad_flags & IB_MAD_IGNORE_MKEY,
+			   mad_flags & IB_MAD_IGNORE_BKEY,
+			   port_num, in_wc, in_grh, in_mad, out_mad);
+	if (err)
+		return IB_MAD_RESULT_FAILURE;
+
+	if (!out_mad->mad_hdr.status) {
+		smp_snoop(ibdev, port_num, in_mad);
+		node_desc_override(ibdev, out_mad);
+	}
+
+	/* set return bit in status of directed route responses */
+	if (in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE)
+		out_mad->mad_hdr.status |= cpu_to_be16(1 << 15);
+
+	if (in_mad->mad_hdr.method == IB_MGMT_METHOD_TRAP_REPRESS)
+		/* no response for trap repress */
+		return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_CONSUMED;
+
+	return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY;
+}
+
+static void send_handler(struct ib_mad_agent *agent,
+			 struct ib_mad_send_wc *mad_send_wc)
+{
+	ib_free_send_mad(mad_send_wc->send_buf);
+}
+
+int mlx4_ib_mad_init(struct mlx4_ib_dev *dev)
+{
+	struct ib_mad_agent *agent;
+	int p, q;
+	int ret;
+
+	for (p = 0; p < dev->dev->caps.num_ports; ++p)
+		for (q = 0; q <= 1; ++q) {
+			agent = ib_register_mad_agent(&dev->ib_dev, p + 1,
+						      q ? IB_QPT_GSI : IB_QPT_SMI,
+						      NULL, 0, send_handler,
+						      NULL, NULL);
+			if (IS_ERR(agent)) {
+				ret = PTR_ERR(agent);
+				goto err;
+			}
+			dev->send_agent[p][q] = agent;
+		}
+
+	return 0;
+
+err:
+	for (p = 0; p < dev->dev->caps.num_ports; ++p)
+		for (q = 0; q <= 1; ++q)
+			if (dev->send_agent[p][q])
+				ib_unregister_mad_agent(dev->send_agent[p][q]);
+
+	return ret;
+}
+
+void mlx4_ib_mad_cleanup(struct mlx4_ib_dev *dev)
+{
+	struct ib_mad_agent *agent;
+	int p, q;
+
+	for (p = 0; p < dev->dev->caps.num_ports; ++p) {
+		for (q = 0; q <= 1; ++q) {
+			agent = dev->send_agent[p][q];
+			dev->send_agent[p][q] = NULL;
+			ib_unregister_mad_agent(agent);
+		}
+
+		if (dev->sm_ah[p])
+			ib_destroy_ah(dev->sm_ah[p]);
+	}
+}
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
new file mode 100644
index 000000000000..688ecb4c39f3
--- /dev/null
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -0,0 +1,651 @@
+/*
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/errno.h>
+
+#include <rdma/ib_smi.h>
+#include <rdma/ib_user_verbs.h>
+
+#include <linux/mlx4/driver.h>
+#include <linux/mlx4/cmd.h>
+
+#include "mlx4_ib.h"
+#include "user.h"
+
+#define DRV_NAME	"mlx4_ib"
+#define DRV_VERSION	"0.01"
+#define DRV_RELDATE	"May 1, 2006"
+
+MODULE_AUTHOR("Roland Dreier");
+MODULE_DESCRIPTION("Mellanox ConnectX HCA InfiniBand driver");
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_VERSION(DRV_VERSION);
+
+static const char mlx4_ib_version[] __devinitdata =
+	DRV_NAME ": Mellanox ConnectX InfiniBand driver v"
+	DRV_VERSION " (" DRV_RELDATE ")\n";
+
+static void init_query_mad(struct ib_smp *mad)
+{
+	mad->base_version  = 1;
+	mad->mgmt_class    = IB_MGMT_CLASS_SUBN_LID_ROUTED;
+	mad->class_version = 1;
+	mad->method	   = IB_MGMT_METHOD_GET;
+}
+
+static int mlx4_ib_query_device(struct ib_device *ibdev,
+				struct ib_device_attr *props)
+{
+	struct mlx4_ib_dev *dev = to_mdev(ibdev);
+	struct ib_smp *in_mad  = NULL;
+	struct ib_smp *out_mad = NULL;
+	int err = -ENOMEM;
+
+	in_mad  = kzalloc(sizeof *in_mad, GFP_KERNEL);
+	out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL);
+	if (!in_mad || !out_mad)
+		goto out;
+
+	init_query_mad(in_mad);
+	in_mad->attr_id = IB_SMP_ATTR_NODE_INFO;
+
+	err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, 1, NULL, NULL, in_mad, out_mad);
+	if (err)
+		goto out;
+
+	memset(props, 0, sizeof *props);
+
+	props->fw_ver = dev->dev->caps.fw_ver;
+	props->device_cap_flags    = IB_DEVICE_CHANGE_PHY_PORT |
+		IB_DEVICE_PORT_ACTIVE_EVENT		|
+		IB_DEVICE_SYS_IMAGE_GUID		|
+		IB_DEVICE_RC_RNR_NAK_GEN;
+	if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_BAD_PKEY_CNTR)
+		props->device_cap_flags |= IB_DEVICE_BAD_PKEY_CNTR;
+	if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_BAD_QKEY_CNTR)
+		props->device_cap_flags |= IB_DEVICE_BAD_QKEY_CNTR;
+	if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_APM)
+		props->device_cap_flags |= IB_DEVICE_AUTO_PATH_MIG;
+	if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_UD_AV_PORT)
+		props->device_cap_flags |= IB_DEVICE_UD_AV_PORT_ENFORCE;
+
+	props->vendor_id	   = be32_to_cpup((__be32 *) (out_mad->data + 36)) &
+		0xffffff;
+	props->vendor_part_id	   = be16_to_cpup((__be16 *) (out_mad->data + 30));
+	props->hw_ver		   = be32_to_cpup((__be32 *) (out_mad->data + 32));
+	memcpy(&props->sys_image_guid, out_mad->data +	4, 8);
+
+	props->max_mr_size	   = ~0ull;
+	props->page_size_cap	   = dev->dev->caps.page_size_cap;
+	props->max_qp		   = dev->dev->caps.num_qps - dev->dev->caps.reserved_qps;
+	props->max_qp_wr	   = dev->dev->caps.max_wqes;
+	props->max_sge		   = min(dev->dev->caps.max_sq_sg,
+					 dev->dev->caps.max_rq_sg);
+	props->max_cq		   = dev->dev->caps.num_cqs - dev->dev->caps.reserved_cqs;
+	props->max_cqe		   = dev->dev->caps.max_cqes;
+	props->max_mr		   = dev->dev->caps.num_mpts - dev->dev->caps.reserved_mrws;
+	props->max_pd		   = dev->dev->caps.num_pds - dev->dev->caps.reserved_pds;
+	props->max_qp_rd_atom	   = dev->dev->caps.max_qp_dest_rdma;
+	props->max_qp_init_rd_atom = dev->dev->caps.max_qp_init_rdma;
+	props->max_res_rd_atom	   = props->max_qp_rd_atom * props->max_qp;
+	props->max_srq		   = dev->dev->caps.num_srqs - dev->dev->caps.reserved_srqs;
+	props->max_srq_wr	   = dev->dev->caps.max_srq_wqes;
+	props->max_srq_sge	   = dev->dev->caps.max_srq_sge;
+	props->local_ca_ack_delay  = dev->dev->caps.local_ca_ack_delay;
+	props->atomic_cap	   = dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_ATOMIC ?
+		IB_ATOMIC_HCA : IB_ATOMIC_NONE;
+	props->max_pkeys	   = dev->dev->caps.pkey_table_len;
+	props->max_mcast_grp	   = dev->dev->caps.num_mgms + dev->dev->caps.num_amgms;
+	props->max_mcast_qp_attach = dev->dev->caps.num_qp_per_mgm;
+	props->max_total_mcast_qp_attach = props->max_mcast_qp_attach *
+					   props->max_mcast_grp;
+	props->max_map_per_fmr = (1 << (32 - ilog2(dev->dev->caps.num_mpts))) - 1;
+
+out:
+	kfree(in_mad);
+	kfree(out_mad);
+
+	return err;
+}
+
+static int mlx4_ib_query_port(struct ib_device *ibdev, u8 port,
+			      struct ib_port_attr *props)
+{
+	struct ib_smp *in_mad  = NULL;
+	struct ib_smp *out_mad = NULL;
+	int err = -ENOMEM;
+
+	in_mad  = kzalloc(sizeof *in_mad, GFP_KERNEL);
+	out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL);
+	if (!in_mad || !out_mad)
+		goto out;
+
+	memset(props, 0, sizeof *props);
+
+	init_query_mad(in_mad);
+	in_mad->attr_id  = IB_SMP_ATTR_PORT_INFO;
+	in_mad->attr_mod = cpu_to_be32(port);
+
+	err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL, in_mad, out_mad);
+	if (err)
+		goto out;
+
+	props->lid		= be16_to_cpup((__be16 *) (out_mad->data + 16));
+	props->lmc		= out_mad->data[34] & 0x7;
+	props->sm_lid		= be16_to_cpup((__be16 *) (out_mad->data + 18));
+	props->sm_sl		= out_mad->data[36] & 0xf;
+	props->state		= out_mad->data[32] & 0xf;
+	props->phys_state	= out_mad->data[33] >> 4;
+	props->port_cap_flags	= be32_to_cpup((__be32 *) (out_mad->data + 20));
+	props->gid_tbl_len	= to_mdev(ibdev)->dev->caps.gid_table_len;
+	props->max_msg_sz	= 0x80000000;
+	props->pkey_tbl_len	= to_mdev(ibdev)->dev->caps.pkey_table_len;
+	props->bad_pkey_cntr	= be16_to_cpup((__be16 *) (out_mad->data + 46));
+	props->qkey_viol_cntr	= be16_to_cpup((__be16 *) (out_mad->data + 48));
+	props->active_width	= out_mad->data[31] & 0xf;
+	props->active_speed	= out_mad->data[35] >> 4;
+	props->max_mtu		= out_mad->data[41] & 0xf;
+	props->active_mtu	= out_mad->data[36] >> 4;
+	props->subnet_timeout	= out_mad->data[51] & 0x1f;
+	props->max_vl_num	= out_mad->data[37] >> 4;
+	props->init_type_reply	= out_mad->data[41] >> 4;
+
+out:
+	kfree(in_mad);
+	kfree(out_mad);
+
+	return err;
+}
+
+static int mlx4_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
+			     union ib_gid *gid)
+{
+	struct ib_smp *in_mad  = NULL;
+	struct ib_smp *out_mad = NULL;
+	int err = -ENOMEM;
+
+	in_mad  = kzalloc(sizeof *in_mad, GFP_KERNEL);
+	out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL);
+	if (!in_mad || !out_mad)
+		goto out;
+
+	init_query_mad(in_mad);
+	in_mad->attr_id  = IB_SMP_ATTR_PORT_INFO;
+	in_mad->attr_mod = cpu_to_be32(port);
+
+	err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL, in_mad, out_mad);
+	if (err)
+		goto out;
+
+	memcpy(gid->raw, out_mad->data + 8, 8);
+
+	init_query_mad(in_mad);
+	in_mad->attr_id  = IB_SMP_ATTR_GUID_INFO;
+	in_mad->attr_mod = cpu_to_be32(index / 8);
+
+	err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL, in_mad, out_mad);
+	if (err)
+		goto out;
+
+	memcpy(gid->raw + 8, out_mad->data + (index % 8) * 8, 8);
+
+out:
+	kfree(in_mad);
+	kfree(out_mad);
+	return err;
+}
+
+static int mlx4_ib_query_pkey(struct ib_device *ibdev, u8 port, u16 index,
+			      u16 *pkey)
+{
+	struct ib_smp *in_mad  = NULL;
+	struct ib_smp *out_mad = NULL;
+	int err = -ENOMEM;
+
+	in_mad  = kzalloc(sizeof *in_mad, GFP_KERNEL);
+	out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL);
+	if (!in_mad || !out_mad)
+		goto out;
+
+	init_query_mad(in_mad);
+	in_mad->attr_id  = IB_SMP_ATTR_PKEY_TABLE;
+	in_mad->attr_mod = cpu_to_be32(index / 32);
+
+	err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL, in_mad, out_mad);
+	if (err)
+		goto out;
+
+	*pkey = be16_to_cpu(((__be16 *) out_mad->data)[index % 32]);
+
+out:
+	kfree(in_mad);
+	kfree(out_mad);
+	return err;
+}
+
+static int mlx4_ib_modify_device(struct ib_device *ibdev, int mask,
+				 struct ib_device_modify *props)
+{
+	if (mask & ~IB_DEVICE_MODIFY_NODE_DESC)
+		return -EOPNOTSUPP;
+
+	if (mask & IB_DEVICE_MODIFY_NODE_DESC) {
+		spin_lock(&to_mdev(ibdev)->sm_lock);
+		memcpy(ibdev->node_desc, props->node_desc, 64);
+		spin_unlock(&to_mdev(ibdev)->sm_lock);
+	}
+
+	return 0;
+}
+
+static int mlx4_SET_PORT(struct mlx4_ib_dev *dev, u8 port, int reset_qkey_viols,
+			 u32 cap_mask)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev->dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	memset(mailbox->buf, 0, 256);
+	*(u8 *) mailbox->buf	     = !!reset_qkey_viols << 6;
+	((__be32 *) mailbox->buf)[2] = cpu_to_be32(cap_mask);
+
+	err = mlx4_cmd(dev->dev, mailbox->dma, port, 0, MLX4_CMD_SET_PORT,
+		       MLX4_CMD_TIME_CLASS_B);
+
+	mlx4_free_cmd_mailbox(dev->dev, mailbox);
+	return err;
+}
+
+static int mlx4_ib_modify_port(struct ib_device *ibdev, u8 port, int mask,
+			       struct ib_port_modify *props)
+{
+	struct ib_port_attr attr;
+	u32 cap_mask;
+	int err;
+
+	mutex_lock(&to_mdev(ibdev)->cap_mask_mutex);
+
+	err = mlx4_ib_query_port(ibdev, port, &attr);
+	if (err)
+		goto out;
+
+	cap_mask = (attr.port_cap_flags | props->set_port_cap_mask) &
+		~props->clr_port_cap_mask;
+
+	err = mlx4_SET_PORT(to_mdev(ibdev), port,
+			    !!(mask & IB_PORT_RESET_QKEY_CNTR),
+			    cap_mask);
+
+out:
+	mutex_unlock(&to_mdev(ibdev)->cap_mask_mutex);
+	return err;
+}
+
+static struct ib_ucontext *mlx4_ib_alloc_ucontext(struct ib_device *ibdev,
+						  struct ib_udata *udata)
+{
+	struct mlx4_ib_dev *dev = to_mdev(ibdev);
+	struct mlx4_ib_ucontext *context;
+	struct mlx4_ib_alloc_ucontext_resp resp;
+	int err;
+
+	resp.qp_tab_size      = dev->dev->caps.num_qps;
+	resp.bf_reg_size      = dev->dev->caps.bf_reg_size;
+	resp.bf_regs_per_page = dev->dev->caps.bf_regs_per_page;
+
+	context = kmalloc(sizeof *context, GFP_KERNEL);
+	if (!context)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlx4_uar_alloc(to_mdev(ibdev)->dev, &context->uar);
+	if (err) {
+		kfree(context);
+		return ERR_PTR(err);
+	}
+
+	INIT_LIST_HEAD(&context->db_page_list);
+	mutex_init(&context->db_page_mutex);
+
+	err = ib_copy_to_udata(udata, &resp, sizeof resp);
+	if (err) {
+		mlx4_uar_free(to_mdev(ibdev)->dev, &context->uar);
+		kfree(context);
+		return ERR_PTR(-EFAULT);
+	}
+
+	return &context->ibucontext;
+}
+
+static int mlx4_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
+{
+	struct mlx4_ib_ucontext *context = to_mucontext(ibcontext);
+
+	mlx4_uar_free(to_mdev(ibcontext->device)->dev, &context->uar);
+	kfree(context);
+
+	return 0;
+}
+
+static int mlx4_ib_mmap(struct ib_ucontext *context, struct vm_area_struct *vma)
+{
+	struct mlx4_ib_dev *dev = to_mdev(context->device);
+
+	if (vma->vm_end - vma->vm_start != PAGE_SIZE)
+		return -EINVAL;
+
+	if (vma->vm_pgoff == 0) {
+		vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
+
+		if (io_remap_pfn_range(vma, vma->vm_start,
+				       to_mucontext(context)->uar.pfn,
+				       PAGE_SIZE, vma->vm_page_prot))
+			return -EAGAIN;
+	} else if (vma->vm_pgoff == 1 && dev->dev->caps.bf_reg_size != 0) {
+		/* FIXME want pgprot_writecombine() for BlueFlame pages */
+		vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
+
+		if (io_remap_pfn_range(vma, vma->vm_start,
+				       to_mucontext(context)->uar.pfn +
+				       dev->dev->caps.num_uars,
+				       PAGE_SIZE, vma->vm_page_prot))
+			return -EAGAIN;
+	} else
+		return -EINVAL;
+
+	return 0;
+}
+
+static struct ib_pd *mlx4_ib_alloc_pd(struct ib_device *ibdev,
+				      struct ib_ucontext *context,
+				      struct ib_udata *udata)
+{
+	struct mlx4_ib_pd *pd;
+	int err;
+
+	pd = kmalloc(sizeof *pd, GFP_KERNEL);
+	if (!pd)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlx4_pd_alloc(to_mdev(ibdev)->dev, &pd->pdn);
+	if (err) {
+		kfree(pd);
+		return ERR_PTR(err);
+	}
+
+	if (context)
+		if (ib_copy_to_udata(udata, &pd->pdn, sizeof (__u32))) {
+			mlx4_pd_free(to_mdev(ibdev)->dev, pd->pdn);
+			kfree(pd);
+			return ERR_PTR(-EFAULT);
+		}
+
+	return &pd->ibpd;
+}
+
+static int mlx4_ib_dealloc_pd(struct ib_pd *pd)
+{
+	mlx4_pd_free(to_mdev(pd->device)->dev, to_mpd(pd)->pdn);
+	kfree(pd);
+
+	return 0;
+}
+
+static int mlx4_ib_mcg_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
+{
+	return mlx4_multicast_attach(to_mdev(ibqp->device)->dev,
+				     &to_mqp(ibqp)->mqp, gid->raw);
+}
+
+static int mlx4_ib_mcg_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
+{
+	return mlx4_multicast_detach(to_mdev(ibqp->device)->dev,
+				     &to_mqp(ibqp)->mqp, gid->raw);
+}
+
+static int init_node_data(struct mlx4_ib_dev *dev)
+{
+	struct ib_smp *in_mad  = NULL;
+	struct ib_smp *out_mad = NULL;
+	int err = -ENOMEM;
+
+	in_mad  = kzalloc(sizeof *in_mad, GFP_KERNEL);
+	out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL);
+	if (!in_mad || !out_mad)
+		goto out;
+
+	init_query_mad(in_mad);
+	in_mad->attr_id = IB_SMP_ATTR_NODE_DESC;
+
+	err = mlx4_MAD_IFC(dev, 1, 1, 1, NULL, NULL, in_mad, out_mad);
+	if (err)
+		goto out;
+
+	memcpy(dev->ib_dev.node_desc, out_mad->data, 64);
+
+	in_mad->attr_id = IB_SMP_ATTR_NODE_INFO;
+
+	err = mlx4_MAD_IFC(dev, 1, 1, 1, NULL, NULL, in_mad, out_mad);
+	if (err)
+		goto out;
+
+	memcpy(&dev->ib_dev.node_guid, out_mad->data + 12, 8);
+
+out:
+	kfree(in_mad);
+	kfree(out_mad);
+	return err;
+}
+
+static void *mlx4_ib_add(struct mlx4_dev *dev)
+{
+	struct mlx4_ib_dev *ibdev;
+
+	ibdev = (struct mlx4_ib_dev *) ib_alloc_device(sizeof *ibdev);
+	if (!ibdev) {
+		dev_err(&dev->pdev->dev, "Device struct alloc failed\n");
+		return NULL;
+	}
+
+	if (mlx4_pd_alloc(dev, &ibdev->priv_pdn))
+		goto err_dealloc;
+
+	if (mlx4_uar_alloc(dev, &ibdev->priv_uar))
+		goto err_pd;
+
+	ibdev->uar_map = ioremap(ibdev->priv_uar.pfn << PAGE_SHIFT, PAGE_SIZE);
+	if (!ibdev->uar_map)
+		goto err_uar;
+
+	INIT_LIST_HEAD(&ibdev->pgdir_list);
+	mutex_init(&ibdev->pgdir_mutex);
+
+	ibdev->dev = dev;
+
+	strlcpy(ibdev->ib_dev.name, "mlx4_%d", IB_DEVICE_NAME_MAX);
+	ibdev->ib_dev.owner		= THIS_MODULE;
+	ibdev->ib_dev.node_type		= RDMA_NODE_IB_CA;
+	ibdev->ib_dev.phys_port_cnt	= dev->caps.num_ports;
+	ibdev->ib_dev.num_comp_vectors	= 1;
+	ibdev->ib_dev.dma_device	= &dev->pdev->dev;
+
+	ibdev->ib_dev.uverbs_abi_ver	= MLX4_IB_UVERBS_ABI_VERSION;
+	ibdev->ib_dev.uverbs_cmd_mask	=
+		(1ull << IB_USER_VERBS_CMD_GET_CONTEXT)		|
+		(1ull << IB_USER_VERBS_CMD_QUERY_DEVICE)	|
+		(1ull << IB_USER_VERBS_CMD_QUERY_PORT)		|
+		(1ull << IB_USER_VERBS_CMD_ALLOC_PD)		|
+		(1ull << IB_USER_VERBS_CMD_DEALLOC_PD)		|
+		(1ull << IB_USER_VERBS_CMD_REG_MR)		|
+		(1ull << IB_USER_VERBS_CMD_DEREG_MR)		|
+		(1ull << IB_USER_VERBS_CMD_CREATE_COMP_CHANNEL)	|
+		(1ull << IB_USER_VERBS_CMD_CREATE_CQ)		|
+		(1ull << IB_USER_VERBS_CMD_DESTROY_CQ)		|
+		(1ull << IB_USER_VERBS_CMD_CREATE_QP)		|
+		(1ull << IB_USER_VERBS_CMD_MODIFY_QP)		|
+		(1ull << IB_USER_VERBS_CMD_DESTROY_QP)		|
+		(1ull << IB_USER_VERBS_CMD_ATTACH_MCAST)	|
+		(1ull << IB_USER_VERBS_CMD_DETACH_MCAST)	|
+		(1ull << IB_USER_VERBS_CMD_CREATE_SRQ)		|
+		(1ull << IB_USER_VERBS_CMD_MODIFY_SRQ)		|
+		(1ull << IB_USER_VERBS_CMD_DESTROY_SRQ);
+
+	ibdev->ib_dev.query_device	= mlx4_ib_query_device;
+	ibdev->ib_dev.query_port	= mlx4_ib_query_port;
+	ibdev->ib_dev.query_gid		= mlx4_ib_query_gid;
+	ibdev->ib_dev.query_pkey	= mlx4_ib_query_pkey;
+	ibdev->ib_dev.modify_device	= mlx4_ib_modify_device;
+	ibdev->ib_dev.modify_port	= mlx4_ib_modify_port;
+	ibdev->ib_dev.alloc_ucontext	= mlx4_ib_alloc_ucontext;
+	ibdev->ib_dev.dealloc_ucontext	= mlx4_ib_dealloc_ucontext;
+	ibdev->ib_dev.mmap		= mlx4_ib_mmap;
+	ibdev->ib_dev.alloc_pd		= mlx4_ib_alloc_pd;
+	ibdev->ib_dev.dealloc_pd	= mlx4_ib_dealloc_pd;
+	ibdev->ib_dev.create_ah		= mlx4_ib_create_ah;
+	ibdev->ib_dev.query_ah		= mlx4_ib_query_ah;
+	ibdev->ib_dev.destroy_ah	= mlx4_ib_destroy_ah;
+	ibdev->ib_dev.create_srq	= mlx4_ib_create_srq;
+	ibdev->ib_dev.modify_srq	= mlx4_ib_modify_srq;
+	ibdev->ib_dev.destroy_srq	= mlx4_ib_destroy_srq;
+	ibdev->ib_dev.post_srq_recv	= mlx4_ib_post_srq_recv;
+	ibdev->ib_dev.create_qp		= mlx4_ib_create_qp;
+	ibdev->ib_dev.modify_qp		= mlx4_ib_modify_qp;
+	ibdev->ib_dev.destroy_qp	= mlx4_ib_destroy_qp;
+	ibdev->ib_dev.post_send		= mlx4_ib_post_send;
+	ibdev->ib_dev.post_recv		= mlx4_ib_post_recv;
+	ibdev->ib_dev.create_cq		= mlx4_ib_create_cq;
+	ibdev->ib_dev.destroy_cq	= mlx4_ib_destroy_cq;
+	ibdev->ib_dev.poll_cq		= mlx4_ib_poll_cq;
+	ibdev->ib_dev.req_notify_cq	= mlx4_ib_arm_cq;
+	ibdev->ib_dev.get_dma_mr	= mlx4_ib_get_dma_mr;
+	ibdev->ib_dev.reg_user_mr	= mlx4_ib_reg_user_mr;
+	ibdev->ib_dev.dereg_mr		= mlx4_ib_dereg_mr;
+	ibdev->ib_dev.attach_mcast	= mlx4_ib_mcg_attach;
+	ibdev->ib_dev.detach_mcast	= mlx4_ib_mcg_detach;
+	ibdev->ib_dev.process_mad	= mlx4_ib_process_mad;
+
+	if (init_node_data(ibdev))
+		goto err_map;
+
+	spin_lock_init(&ibdev->sm_lock);
+	mutex_init(&ibdev->cap_mask_mutex);
+
+	if (ib_register_device(&ibdev->ib_dev))
+		goto err_map;
+
+	if (mlx4_ib_mad_init(ibdev))
+		goto err_reg;
+
+	return ibdev;
+
+err_reg:
+	ib_unregister_device(&ibdev->ib_dev);
+
+err_map:
+	iounmap(ibdev->uar_map);
+
+err_uar:
+	mlx4_uar_free(dev, &ibdev->priv_uar);
+
+err_pd:
+	mlx4_pd_free(dev, ibdev->priv_pdn);
+
+err_dealloc:
+	ib_dealloc_device(&ibdev->ib_dev);
+
+	return NULL;
+}
+
+static void mlx4_ib_remove(struct mlx4_dev *dev, void *ibdev_ptr)
+{
+	struct mlx4_ib_dev *ibdev = ibdev_ptr;
+	int p;
+
+	for (p = 1; p <= dev->caps.num_ports; ++p)
+		mlx4_CLOSE_PORT(dev, p);
+
+	mlx4_ib_mad_cleanup(ibdev);
+	ib_unregister_device(&ibdev->ib_dev);
+	iounmap(ibdev->uar_map);
+	mlx4_uar_free(dev, &ibdev->priv_uar);
+	mlx4_pd_free(dev, ibdev->priv_pdn);
+	ib_dealloc_device(&ibdev->ib_dev);
+}
+
+static void mlx4_ib_event(struct mlx4_dev *dev, void *ibdev_ptr,
+			  enum mlx4_dev_event event, int subtype,
+			  int port)
+{
+	struct ib_event ibev;
+
+	switch (event) {
+	case MLX4_EVENT_TYPE_PORT_CHANGE:
+		ibev.event = subtype == MLX4_PORT_CHANGE_SUBTYPE_ACTIVE ?
+			IB_EVENT_PORT_ACTIVE : IB_EVENT_PORT_ERR;
+		break;
+
+	case MLX4_EVENT_TYPE_LOCAL_CATAS_ERROR:
+		ibev.event = IB_EVENT_DEVICE_FATAL;
+		break;
+
+	default:
+		return;
+	}
+
+	ibev.device	      = ibdev_ptr;
+	ibev.element.port_num = port;
+
+	ib_dispatch_event(&ibev);
+}
+
+static struct mlx4_interface mlx4_ib_interface = {
+	.add	= mlx4_ib_add,
+	.remove	= mlx4_ib_remove,
+	.event	= mlx4_ib_event
+};
+
+static int __init mlx4_ib_init(void)
+{
+	return mlx4_register_interface(&mlx4_ib_interface);
+}
+
+static void __exit mlx4_ib_cleanup(void)
+{
+	mlx4_unregister_interface(&mlx4_ib_interface);
+}
+
+module_init(mlx4_ib_init);
+module_exit(mlx4_ib_cleanup);
diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h
new file mode 100644
index 000000000000..93dac71f3230
--- /dev/null
+++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h
@@ -0,0 +1,285 @@
+/*
+ * Copyright (c) 2006, 2007 Cisco Systems.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef MLX4_IB_H
+#define MLX4_IB_H
+
+#include <linux/compiler.h>
+#include <linux/list.h>
+
+#include <rdma/ib_verbs.h>
+#include <rdma/ib_umem.h>
+
+#include <linux/mlx4/device.h>
+#include <linux/mlx4/doorbell.h>
+
+enum {
+	MLX4_IB_DB_PER_PAGE	= PAGE_SIZE / 4
+};
+
+struct mlx4_ib_db_pgdir;
+struct mlx4_ib_user_db_page;
+
+struct mlx4_ib_db {
+	__be32		       *db;
+	union {
+		struct mlx4_ib_db_pgdir	       *pgdir;
+		struct mlx4_ib_user_db_page    *user_page;
+	}			u;
+	dma_addr_t		dma;
+	int			index;
+	int			order;
+};
+
+struct mlx4_ib_ucontext {
+	struct ib_ucontext	ibucontext;
+	struct mlx4_uar		uar;
+	struct list_head	db_page_list;
+	struct mutex		db_page_mutex;
+};
+
+struct mlx4_ib_pd {
+	struct ib_pd		ibpd;
+	u32			pdn;
+};
+
+struct mlx4_ib_cq_buf {
+	struct mlx4_buf		buf;
+	struct mlx4_mtt		mtt;
+};
+
+struct mlx4_ib_cq {
+	struct ib_cq		ibcq;
+	struct mlx4_cq		mcq;
+	struct mlx4_ib_cq_buf	buf;
+	struct mlx4_ib_db	db;
+	spinlock_t		lock;
+	struct ib_umem	       *umem;
+};
+
+struct mlx4_ib_mr {
+	struct ib_mr		ibmr;
+	struct mlx4_mr		mmr;
+	struct ib_umem	       *umem;
+};
+
+struct mlx4_ib_wq {
+	u64		       *wrid;
+	spinlock_t		lock;
+	int			max;
+	int			max_gs;
+	int			offset;
+	int			wqe_shift;
+	unsigned		head;
+	unsigned		tail;
+};
+
+struct mlx4_ib_qp {
+	struct ib_qp		ibqp;
+	struct mlx4_qp		mqp;
+	struct mlx4_buf		buf;
+
+	struct mlx4_ib_db	db;
+	struct mlx4_ib_wq	rq;
+
+	u32			doorbell_qpn;
+	__be32			sq_signal_bits;
+	struct mlx4_ib_wq	sq;
+
+	struct ib_umem	       *umem;
+	struct mlx4_mtt		mtt;
+	int			buf_size;
+	struct mutex		mutex;
+	u8			port;
+	u8			alt_port;
+	u8			atomic_rd_en;
+	u8			resp_depth;
+	u8			state;
+};
+
+struct mlx4_ib_srq {
+	struct ib_srq		ibsrq;
+	struct mlx4_srq		msrq;
+	struct mlx4_buf		buf;
+	struct mlx4_ib_db	db;
+	u64		       *wrid;
+	spinlock_t		lock;
+	int			head;
+	int			tail;
+	u16			wqe_ctr;
+	struct ib_umem	       *umem;
+	struct mlx4_mtt		mtt;
+	struct mutex		mutex;
+};
+
+struct mlx4_ib_ah {
+	struct ib_ah		ibah;
+	struct mlx4_av		av;
+};
+
+struct mlx4_ib_dev {
+	struct ib_device	ib_dev;
+	struct mlx4_dev	       *dev;
+	void __iomem	       *uar_map;
+
+	struct list_head	pgdir_list;
+	struct mutex		pgdir_mutex;
+
+	struct mlx4_uar		priv_uar;
+	u32			priv_pdn;
+	MLX4_DECLARE_DOORBELL_LOCK(uar_lock);
+
+	struct ib_mad_agent    *send_agent[MLX4_MAX_PORTS][2];
+	struct ib_ah	       *sm_ah[MLX4_MAX_PORTS];
+	spinlock_t		sm_lock;
+
+	struct mutex		cap_mask_mutex;
+};
+
+static inline struct mlx4_ib_dev *to_mdev(struct ib_device *ibdev)
+{
+	return container_of(ibdev, struct mlx4_ib_dev, ib_dev);
+}
+
+static inline struct mlx4_ib_ucontext *to_mucontext(struct ib_ucontext *ibucontext)
+{
+	return container_of(ibucontext, struct mlx4_ib_ucontext, ibucontext);
+}
+
+static inline struct mlx4_ib_pd *to_mpd(struct ib_pd *ibpd)
+{
+	return container_of(ibpd, struct mlx4_ib_pd, ibpd);
+}
+
+static inline struct mlx4_ib_cq *to_mcq(struct ib_cq *ibcq)
+{
+	return container_of(ibcq, struct mlx4_ib_cq, ibcq);
+}
+
+static inline struct mlx4_ib_cq *to_mibcq(struct mlx4_cq *mcq)
+{
+	return container_of(mcq, struct mlx4_ib_cq, mcq);
+}
+
+static inline struct mlx4_ib_mr *to_mmr(struct ib_mr *ibmr)
+{
+	return container_of(ibmr, struct mlx4_ib_mr, ibmr);
+}
+
+static inline struct mlx4_ib_qp *to_mqp(struct ib_qp *ibqp)
+{
+	return container_of(ibqp, struct mlx4_ib_qp, ibqp);
+}
+
+static inline struct mlx4_ib_qp *to_mibqp(struct mlx4_qp *mqp)
+{
+	return container_of(mqp, struct mlx4_ib_qp, mqp);
+}
+
+static inline struct mlx4_ib_srq *to_msrq(struct ib_srq *ibsrq)
+{
+	return container_of(ibsrq, struct mlx4_ib_srq, ibsrq);
+}
+
+static inline struct mlx4_ib_srq *to_mibsrq(struct mlx4_srq *msrq)
+{
+	return container_of(msrq, struct mlx4_ib_srq, msrq);
+}
+
+static inline struct mlx4_ib_ah *to_mah(struct ib_ah *ibah)
+{
+	return container_of(ibah, struct mlx4_ib_ah, ibah);
+}
+
+int mlx4_ib_db_alloc(struct mlx4_ib_dev *dev, struct mlx4_ib_db *db, int order);
+void mlx4_ib_db_free(struct mlx4_ib_dev *dev, struct mlx4_ib_db *db);
+int mlx4_ib_db_map_user(struct mlx4_ib_ucontext *context, unsigned long virt,
+			struct mlx4_ib_db *db);
+void mlx4_ib_db_unmap_user(struct mlx4_ib_ucontext *context, struct mlx4_ib_db *db);
+
+struct ib_mr *mlx4_ib_get_dma_mr(struct ib_pd *pd, int acc);
+int mlx4_ib_umem_write_mtt(struct mlx4_ib_dev *dev, struct mlx4_mtt *mtt,
+			   struct ib_umem *umem);
+struct ib_mr *mlx4_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
+				  u64 virt_addr, int access_flags,
+				  struct ib_udata *udata);
+int mlx4_ib_dereg_mr(struct ib_mr *mr);
+
+struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector,
+				struct ib_ucontext *context,
+				struct ib_udata *udata);
+int mlx4_ib_destroy_cq(struct ib_cq *cq);
+int mlx4_ib_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc);
+int mlx4_ib_arm_cq(struct ib_cq *cq, enum ib_cq_notify_flags flags);
+void __mlx4_ib_cq_clean(struct mlx4_ib_cq *cq, u32 qpn, struct mlx4_ib_srq *srq);
+void mlx4_ib_cq_clean(struct mlx4_ib_cq *cq, u32 qpn, struct mlx4_ib_srq *srq);
+
+struct ib_ah *mlx4_ib_create_ah(struct ib_pd *pd, struct ib_ah_attr *ah_attr);
+int mlx4_ib_query_ah(struct ib_ah *ibah, struct ib_ah_attr *ah_attr);
+int mlx4_ib_destroy_ah(struct ib_ah *ah);
+
+struct ib_srq *mlx4_ib_create_srq(struct ib_pd *pd,
+				  struct ib_srq_init_attr *init_attr,
+				  struct ib_udata *udata);
+int mlx4_ib_modify_srq(struct ib_srq *ibsrq, struct ib_srq_attr *attr,
+		       enum ib_srq_attr_mask attr_mask, struct ib_udata *udata);
+int mlx4_ib_destroy_srq(struct ib_srq *srq);
+void mlx4_ib_free_srq_wqe(struct mlx4_ib_srq *srq, int wqe_index);
+int mlx4_ib_post_srq_recv(struct ib_srq *ibsrq, struct ib_recv_wr *wr,
+			  struct ib_recv_wr **bad_wr);
+
+struct ib_qp *mlx4_ib_create_qp(struct ib_pd *pd,
+				struct ib_qp_init_attr *init_attr,
+				struct ib_udata *udata);
+int mlx4_ib_destroy_qp(struct ib_qp *qp);
+int mlx4_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
+		      int attr_mask, struct ib_udata *udata);
+int mlx4_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
+		      struct ib_send_wr **bad_wr);
+int mlx4_ib_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *wr,
+		      struct ib_recv_wr **bad_wr);
+
+int mlx4_MAD_IFC(struct mlx4_ib_dev *dev, int ignore_mkey, int ignore_bkey,
+		 int port, struct ib_wc *in_wc, struct ib_grh *in_grh,
+		 void *in_mad, void *response_mad);
+int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags,	u8 port_num,
+			struct ib_wc *in_wc, struct ib_grh *in_grh,
+			struct ib_mad *in_mad, struct ib_mad *out_mad);
+int mlx4_ib_mad_init(struct mlx4_ib_dev *dev);
+void mlx4_ib_mad_cleanup(struct mlx4_ib_dev *dev);
+
+static inline int mlx4_ib_ah_grh_present(struct mlx4_ib_ah *ah)
+{
+	return !!(ah->av.g_slid & 0x80);
+}
+
+#endif /* MLX4_IB_H */
diff --git a/drivers/infiniband/hw/mlx4/mr.c b/drivers/infiniband/hw/mlx4/mr.c
new file mode 100644
index 000000000000..85ae906f1d12
--- /dev/null
+++ b/drivers/infiniband/hw/mlx4/mr.c
@@ -0,0 +1,184 @@
+/*
+ * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "mlx4_ib.h"
+
+static u32 convert_access(int acc)
+{
+	return (acc & IB_ACCESS_REMOTE_ATOMIC ? MLX4_PERM_ATOMIC       : 0) |
+	       (acc & IB_ACCESS_REMOTE_WRITE  ? MLX4_PERM_REMOTE_WRITE : 0) |
+	       (acc & IB_ACCESS_REMOTE_READ   ? MLX4_PERM_REMOTE_READ  : 0) |
+	       (acc & IB_ACCESS_LOCAL_WRITE   ? MLX4_PERM_LOCAL_WRITE  : 0) |
+	       MLX4_PERM_LOCAL_READ;
+}
+
+struct ib_mr *mlx4_ib_get_dma_mr(struct ib_pd *pd, int acc)
+{
+	struct mlx4_ib_mr *mr;
+	int err;
+
+	mr = kmalloc(sizeof *mr, GFP_KERNEL);
+	if (!mr)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlx4_mr_alloc(to_mdev(pd->device)->dev, to_mpd(pd)->pdn, 0,
+			    ~0ull, convert_access(acc), 0, 0, &mr->mmr);
+	if (err)
+		goto err_free;
+
+	err = mlx4_mr_enable(to_mdev(pd->device)->dev, &mr->mmr);
+	if (err)
+		goto err_mr;
+
+	mr->ibmr.rkey = mr->ibmr.lkey = mr->mmr.key;
+	mr->umem = NULL;
+
+	return &mr->ibmr;
+
+err_mr:
+	mlx4_mr_free(to_mdev(pd->device)->dev, &mr->mmr);
+
+err_free:
+	kfree(mr);
+
+	return ERR_PTR(err);
+}
+
+int mlx4_ib_umem_write_mtt(struct mlx4_ib_dev *dev, struct mlx4_mtt *mtt,
+			   struct ib_umem *umem)
+{
+	u64 *pages;
+	struct ib_umem_chunk *chunk;
+	int i, j, k;
+	int n;
+	int len;
+	int err = 0;
+
+	pages = (u64 *) __get_free_page(GFP_KERNEL);
+	if (!pages)
+		return -ENOMEM;
+
+	i = n = 0;
+
+	list_for_each_entry(chunk, &umem->chunk_list, list)
+		for (j = 0; j < chunk->nmap; ++j) {
+			len = sg_dma_len(&chunk->page_list[j]) >> mtt->page_shift;
+			for (k = 0; k < len; ++k) {
+				pages[i++] = sg_dma_address(&chunk->page_list[j]) +
+					umem->page_size * k;
+				/*
+				 * Be friendly to WRITE_MTT firmware
+				 * command, and pass it chunks of
+				 * appropriate size.
+				 */
+				if (i == PAGE_SIZE / sizeof (u64) - 2) {
+					err = mlx4_write_mtt(dev->dev, mtt, n,
+							     i, pages);
+					if (err)
+						goto out;
+					n += i;
+					i = 0;
+				}
+			}
+		}
+
+	if (i)
+		err = mlx4_write_mtt(dev->dev, mtt, n, i, pages);
+
+out:
+	free_page((unsigned long) pages);
+	return err;
+}
+
+struct ib_mr *mlx4_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
+				  u64 virt_addr, int access_flags,
+				  struct ib_udata *udata)
+{
+	struct mlx4_ib_dev *dev = to_mdev(pd->device);
+	struct mlx4_ib_mr *mr;
+	int shift;
+	int err;
+	int n;
+
+	mr = kmalloc(sizeof *mr, GFP_KERNEL);
+	if (!mr)
+		return ERR_PTR(-ENOMEM);
+
+	mr->umem = ib_umem_get(pd->uobject->context, start, length, access_flags);
+	if (IS_ERR(mr->umem)) {
+		err = PTR_ERR(mr->umem);
+		goto err_free;
+	}
+
+	n = ib_umem_page_count(mr->umem);
+	shift = ilog2(mr->umem->page_size);
+
+	err = mlx4_mr_alloc(dev->dev, to_mpd(pd)->pdn, virt_addr, length,
+			    convert_access(access_flags), n, shift, &mr->mmr);
+	if (err)
+		goto err_umem;
+
+	err = mlx4_ib_umem_write_mtt(dev, &mr->mmr.mtt, mr->umem);
+	if (err)
+		goto err_mr;
+
+	err = mlx4_mr_enable(dev->dev, &mr->mmr);
+	if (err)
+		goto err_mr;
+
+	mr->ibmr.rkey = mr->ibmr.lkey = mr->mmr.key;
+
+	return &mr->ibmr;
+
+err_mr:
+	mlx4_mr_free(to_mdev(pd->device)->dev, &mr->mmr);
+
+err_umem:
+	ib_umem_release(mr->umem);
+
+err_free:
+	kfree(mr);
+
+	return ERR_PTR(err);
+}
+
+int mlx4_ib_dereg_mr(struct ib_mr *ibmr)
+{
+	struct mlx4_ib_mr *mr = to_mmr(ibmr);
+
+	mlx4_mr_free(to_mdev(ibmr->device)->dev, &mr->mmr);
+	if (mr->umem)
+		ib_umem_release(mr->umem);
+	kfree(mr);
+
+	return 0;
+}
diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c
new file mode 100644
index 000000000000..5cd706908450
--- /dev/null
+++ b/drivers/infiniband/hw/mlx4/qp.c
@@ -0,0 +1,1294 @@
+/*
+ * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <rdma/ib_cache.h>
+#include <rdma/ib_pack.h>
+
+#include <linux/mlx4/qp.h>
+
+#include "mlx4_ib.h"
+#include "user.h"
+
+enum {
+	MLX4_IB_ACK_REQ_FREQ	= 8,
+};
+
+enum {
+	MLX4_IB_DEFAULT_SCHED_QUEUE	= 0x83,
+	MLX4_IB_DEFAULT_QP0_SCHED_QUEUE	= 0x3f
+};
+
+enum {
+	/*
+	 * Largest possible UD header: send with GRH and immediate data.
+	 */
+	MLX4_IB_UD_HEADER_SIZE		= 72
+};
+
+struct mlx4_ib_sqp {
+	struct mlx4_ib_qp	qp;
+	int			pkey_index;
+	u32			qkey;
+	u32			send_psn;
+	struct ib_ud_header	ud_header;
+	u8			header_buf[MLX4_IB_UD_HEADER_SIZE];
+};
+
+static const __be32 mlx4_ib_opcode[] = {
+	[IB_WR_SEND]			= __constant_cpu_to_be32(MLX4_OPCODE_SEND),
+	[IB_WR_SEND_WITH_IMM]		= __constant_cpu_to_be32(MLX4_OPCODE_SEND_IMM),
+	[IB_WR_RDMA_WRITE]		= __constant_cpu_to_be32(MLX4_OPCODE_RDMA_WRITE),
+	[IB_WR_RDMA_WRITE_WITH_IMM]	= __constant_cpu_to_be32(MLX4_OPCODE_RDMA_WRITE_IMM),
+	[IB_WR_RDMA_READ]		= __constant_cpu_to_be32(MLX4_OPCODE_RDMA_READ),
+	[IB_WR_ATOMIC_CMP_AND_SWP]	= __constant_cpu_to_be32(MLX4_OPCODE_ATOMIC_CS),
+	[IB_WR_ATOMIC_FETCH_AND_ADD]	= __constant_cpu_to_be32(MLX4_OPCODE_ATOMIC_FA),
+};
+
+static struct mlx4_ib_sqp *to_msqp(struct mlx4_ib_qp *mqp)
+{
+	return container_of(mqp, struct mlx4_ib_sqp, qp);
+}
+
+static int is_sqp(struct mlx4_ib_dev *dev, struct mlx4_ib_qp *qp)
+{
+	return qp->mqp.qpn >= dev->dev->caps.sqp_start &&
+		qp->mqp.qpn <= dev->dev->caps.sqp_start + 3;
+}
+
+static int is_qp0(struct mlx4_ib_dev *dev, struct mlx4_ib_qp *qp)
+{
+	return qp->mqp.qpn >= dev->dev->caps.sqp_start &&
+		qp->mqp.qpn <= dev->dev->caps.sqp_start + 1;
+}
+
+static void *get_wqe(struct mlx4_ib_qp *qp, int offset)
+{
+	if (qp->buf.nbufs == 1)
+		return qp->buf.u.direct.buf + offset;
+	else
+		return qp->buf.u.page_list[offset >> PAGE_SHIFT].buf +
+			(offset & (PAGE_SIZE - 1));
+}
+
+static void *get_recv_wqe(struct mlx4_ib_qp *qp, int n)
+{
+	return get_wqe(qp, qp->rq.offset + (n << qp->rq.wqe_shift));
+}
+
+static void *get_send_wqe(struct mlx4_ib_qp *qp, int n)
+{
+	return get_wqe(qp, qp->sq.offset + (n << qp->sq.wqe_shift));
+}
+
+static void mlx4_ib_qp_event(struct mlx4_qp *qp, enum mlx4_event type)
+{
+	struct ib_event event;
+	struct ib_qp *ibqp = &to_mibqp(qp)->ibqp;
+
+	if (type == MLX4_EVENT_TYPE_PATH_MIG)
+		to_mibqp(qp)->port = to_mibqp(qp)->alt_port;
+
+	if (ibqp->event_handler) {
+		event.device     = ibqp->device;
+		event.element.qp = ibqp;
+		switch (type) {
+		case MLX4_EVENT_TYPE_PATH_MIG:
+			event.event = IB_EVENT_PATH_MIG;
+			break;
+		case MLX4_EVENT_TYPE_COMM_EST:
+			event.event = IB_EVENT_COMM_EST;
+			break;
+		case MLX4_EVENT_TYPE_SQ_DRAINED:
+			event.event = IB_EVENT_SQ_DRAINED;
+			break;
+		case MLX4_EVENT_TYPE_SRQ_QP_LAST_WQE:
+			event.event = IB_EVENT_QP_LAST_WQE_REACHED;
+			break;
+		case MLX4_EVENT_TYPE_WQ_CATAS_ERROR:
+			event.event = IB_EVENT_QP_FATAL;
+			break;
+		case MLX4_EVENT_TYPE_PATH_MIG_FAILED:
+			event.event = IB_EVENT_PATH_MIG_ERR;
+			break;
+		case MLX4_EVENT_TYPE_WQ_INVAL_REQ_ERROR:
+			event.event = IB_EVENT_QP_REQ_ERR;
+			break;
+		case MLX4_EVENT_TYPE_WQ_ACCESS_ERROR:
+			event.event = IB_EVENT_QP_ACCESS_ERR;
+			break;
+		default:
+			printk(KERN_WARNING "mlx4_ib: Unexpected event type %d "
+			       "on QP %06x\n", type, qp->qpn);
+			return;
+		}
+
+		ibqp->event_handler(&event, ibqp->qp_context);
+	}
+}
+
+static int send_wqe_overhead(enum ib_qp_type type)
+{
+	/*
+	 * UD WQEs must have a datagram segment.
+	 * RC and UC WQEs might have a remote address segment.
+	 * MLX WQEs need two extra inline data segments (for the UD
+	 * header and space for the ICRC).
+	 */
+	switch (type) {
+	case IB_QPT_UD:
+		return sizeof (struct mlx4_wqe_ctrl_seg) +
+			sizeof (struct mlx4_wqe_datagram_seg);
+	case IB_QPT_UC:
+		return sizeof (struct mlx4_wqe_ctrl_seg) +
+			sizeof (struct mlx4_wqe_raddr_seg);
+	case IB_QPT_RC:
+		return sizeof (struct mlx4_wqe_ctrl_seg) +
+			sizeof (struct mlx4_wqe_atomic_seg) +
+			sizeof (struct mlx4_wqe_raddr_seg);
+	case IB_QPT_SMI:
+	case IB_QPT_GSI:
+		return sizeof (struct mlx4_wqe_ctrl_seg) +
+			ALIGN(MLX4_IB_UD_HEADER_SIZE +
+			      sizeof (struct mlx4_wqe_inline_seg),
+			      sizeof (struct mlx4_wqe_data_seg)) +
+			ALIGN(4 +
+			      sizeof (struct mlx4_wqe_inline_seg),
+			      sizeof (struct mlx4_wqe_data_seg));
+	default:
+		return sizeof (struct mlx4_wqe_ctrl_seg);
+	}
+}
+
+static int set_qp_size(struct mlx4_ib_dev *dev, struct ib_qp_cap *cap,
+		       enum ib_qp_type type, struct mlx4_ib_qp *qp)
+{
+	/* Sanity check QP size before proceeding */
+	if (cap->max_send_wr	 > dev->dev->caps.max_wqes  ||
+	    cap->max_recv_wr	 > dev->dev->caps.max_wqes  ||
+	    cap->max_send_sge	 > dev->dev->caps.max_sq_sg ||
+	    cap->max_recv_sge	 > dev->dev->caps.max_rq_sg ||
+	    cap->max_inline_data + send_wqe_overhead(type) +
+	    sizeof (struct mlx4_wqe_inline_seg) > dev->dev->caps.max_sq_desc_sz)
+		return -EINVAL;
+
+	/*
+	 * For MLX transport we need 2 extra S/G entries:
+	 * one for the header and one for the checksum at the end
+	 */
+	if ((type == IB_QPT_SMI || type == IB_QPT_GSI) &&
+	    cap->max_send_sge + 2 > dev->dev->caps.max_sq_sg)
+		return -EINVAL;
+
+	qp->rq.max = cap->max_recv_wr ? roundup_pow_of_two(cap->max_recv_wr) : 0;
+	qp->sq.max = cap->max_send_wr ? roundup_pow_of_two(cap->max_send_wr) : 0;
+
+	qp->rq.wqe_shift = ilog2(roundup_pow_of_two(cap->max_recv_sge *
+						    sizeof (struct mlx4_wqe_data_seg)));
+	qp->rq.max_gs    = (1 << qp->rq.wqe_shift) / sizeof (struct mlx4_wqe_data_seg);
+
+	qp->sq.wqe_shift = ilog2(roundup_pow_of_two(max(cap->max_send_sge *
+							sizeof (struct mlx4_wqe_data_seg),
+							cap->max_inline_data +
+							sizeof (struct mlx4_wqe_inline_seg)) +
+						    send_wqe_overhead(type)));
+	qp->sq.max_gs    = ((1 << qp->sq.wqe_shift) - send_wqe_overhead(type)) /
+		sizeof (struct mlx4_wqe_data_seg);
+
+	qp->buf_size = (qp->rq.max << qp->rq.wqe_shift) +
+		(qp->sq.max << qp->sq.wqe_shift);
+	if (qp->rq.wqe_shift > qp->sq.wqe_shift) {
+		qp->rq.offset = 0;
+		qp->sq.offset = qp->rq.max << qp->rq.wqe_shift;
+	} else {
+		qp->rq.offset = qp->sq.max << qp->sq.wqe_shift;
+		qp->sq.offset = 0;
+	}
+
+	cap->max_send_wr  = qp->sq.max;
+	cap->max_recv_wr  = qp->rq.max;
+	cap->max_send_sge = qp->sq.max_gs;
+	cap->max_recv_sge = qp->rq.max_gs;
+	cap->max_inline_data = (1 << qp->sq.wqe_shift) - send_wqe_overhead(type) -
+		sizeof (struct mlx4_wqe_inline_seg);
+
+	return 0;
+}
+
+static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd,
+			    struct ib_qp_init_attr *init_attr,
+			    struct ib_udata *udata, int sqpn, struct mlx4_ib_qp *qp)
+{
+	struct mlx4_wqe_ctrl_seg *ctrl;
+	int err;
+	int i;
+
+	mutex_init(&qp->mutex);
+	spin_lock_init(&qp->sq.lock);
+	spin_lock_init(&qp->rq.lock);
+
+	qp->state	 = IB_QPS_RESET;
+	qp->atomic_rd_en = 0;
+	qp->resp_depth   = 0;
+
+	qp->rq.head	    = 0;
+	qp->rq.tail	    = 0;
+	qp->sq.head	    = 0;
+	qp->sq.tail	    = 0;
+
+	err = set_qp_size(dev, &init_attr->cap, init_attr->qp_type, qp);
+	if (err)
+		goto err;
+
+	if (pd->uobject) {
+		struct mlx4_ib_create_qp ucmd;
+
+		if (ib_copy_from_udata(&ucmd, udata, sizeof ucmd)) {
+			err = -EFAULT;
+			goto err;
+		}
+
+		qp->umem = ib_umem_get(pd->uobject->context, ucmd.buf_addr,
+				       qp->buf_size, 0);
+		if (IS_ERR(qp->umem)) {
+			err = PTR_ERR(qp->umem);
+			goto err;
+		}
+
+		err = mlx4_mtt_init(dev->dev, ib_umem_page_count(qp->umem),
+				    ilog2(qp->umem->page_size), &qp->mtt);
+		if (err)
+			goto err_buf;
+
+		err = mlx4_ib_umem_write_mtt(dev, &qp->mtt, qp->umem);
+		if (err)
+			goto err_mtt;
+
+		err = mlx4_ib_db_map_user(to_mucontext(pd->uobject->context),
+					  ucmd.db_addr, &qp->db);
+		if (err)
+			goto err_mtt;
+	} else {
+		err = mlx4_ib_db_alloc(dev, &qp->db, 0);
+		if (err)
+			goto err;
+
+		*qp->db.db = 0;
+
+		if (mlx4_buf_alloc(dev->dev, qp->buf_size, PAGE_SIZE * 2, &qp->buf)) {
+			err = -ENOMEM;
+			goto err_db;
+		}
+
+		err = mlx4_mtt_init(dev->dev, qp->buf.npages, qp->buf.page_shift,
+				    &qp->mtt);
+		if (err)
+			goto err_buf;
+
+		err = mlx4_buf_write_mtt(dev->dev, &qp->mtt, &qp->buf);
+		if (err)
+			goto err_mtt;
+
+		for (i = 0; i < qp->sq.max; ++i) {
+			ctrl = get_send_wqe(qp, i);
+			ctrl->owner_opcode = cpu_to_be32(1 << 31);
+		}
+
+		qp->sq.wrid  = kmalloc(qp->sq.max * sizeof (u64), GFP_KERNEL);
+		qp->rq.wrid  = kmalloc(qp->rq.max * sizeof (u64), GFP_KERNEL);
+
+		if (!qp->sq.wrid || !qp->rq.wrid) {
+			err = -ENOMEM;
+			goto err_wrid;
+		}
+
+		/* We don't support inline sends for kernel QPs (yet) */
+		init_attr->cap.max_inline_data = 0;
+	}
+
+	err = mlx4_qp_alloc(dev->dev, sqpn, &qp->mqp);
+	if (err)
+		goto err_wrid;
+
+	/*
+	 * Hardware wants QPN written in big-endian order (after
+	 * shifting) for send doorbell.  Precompute this value to save
+	 * a little bit when posting sends.
+	 */
+	qp->doorbell_qpn = swab32(qp->mqp.qpn << 8);
+
+	if (init_attr->sq_sig_type == IB_SIGNAL_ALL_WR)
+		qp->sq_signal_bits = cpu_to_be32(MLX4_WQE_CTRL_CQ_UPDATE);
+	else
+		qp->sq_signal_bits = 0;
+
+	qp->mqp.event = mlx4_ib_qp_event;
+
+	return 0;
+
+err_wrid:
+	if (pd->uobject)
+		mlx4_ib_db_unmap_user(to_mucontext(pd->uobject->context), &qp->db);
+	else {
+		kfree(qp->sq.wrid);
+		kfree(qp->rq.wrid);
+	}
+
+err_mtt:
+	mlx4_mtt_cleanup(dev->dev, &qp->mtt);
+
+err_buf:
+	if (pd->uobject)
+		ib_umem_release(qp->umem);
+	else
+		mlx4_buf_free(dev->dev, qp->buf_size, &qp->buf);
+
+err_db:
+	if (!pd->uobject)
+		mlx4_ib_db_free(dev, &qp->db);
+
+err:
+	return err;
+}
+
+static enum mlx4_qp_state to_mlx4_state(enum ib_qp_state state)
+{
+	switch (state) {
+	case IB_QPS_RESET:	return MLX4_QP_STATE_RST;
+	case IB_QPS_INIT:	return MLX4_QP_STATE_INIT;
+	case IB_QPS_RTR:	return MLX4_QP_STATE_RTR;
+	case IB_QPS_RTS:	return MLX4_QP_STATE_RTS;
+	case IB_QPS_SQD:	return MLX4_QP_STATE_SQD;
+	case IB_QPS_SQE:	return MLX4_QP_STATE_SQER;
+	case IB_QPS_ERR:	return MLX4_QP_STATE_ERR;
+	default:		return -1;
+	}
+}
+
+static void mlx4_ib_lock_cqs(struct mlx4_ib_cq *send_cq, struct mlx4_ib_cq *recv_cq)
+{
+	if (send_cq == recv_cq)
+		spin_lock_irq(&send_cq->lock);
+	else if (send_cq->mcq.cqn < recv_cq->mcq.cqn) {
+		spin_lock_irq(&send_cq->lock);
+		spin_lock_nested(&recv_cq->lock, SINGLE_DEPTH_NESTING);
+	} else {
+		spin_lock_irq(&recv_cq->lock);
+		spin_lock_nested(&send_cq->lock, SINGLE_DEPTH_NESTING);
+	}
+}
+
+static void mlx4_ib_unlock_cqs(struct mlx4_ib_cq *send_cq, struct mlx4_ib_cq *recv_cq)
+{
+	if (send_cq == recv_cq)
+		spin_unlock_irq(&send_cq->lock);
+	else if (send_cq->mcq.cqn < recv_cq->mcq.cqn) {
+		spin_unlock(&recv_cq->lock);
+		spin_unlock_irq(&send_cq->lock);
+	} else {
+		spin_unlock(&send_cq->lock);
+		spin_unlock_irq(&recv_cq->lock);
+	}
+}
+
+static void destroy_qp_common(struct mlx4_ib_dev *dev, struct mlx4_ib_qp *qp,
+			      int is_user)
+{
+	struct mlx4_ib_cq *send_cq, *recv_cq;
+
+	if (qp->state != IB_QPS_RESET)
+		if (mlx4_qp_modify(dev->dev, NULL, to_mlx4_state(qp->state),
+				   MLX4_QP_STATE_RST, NULL, 0, 0, &qp->mqp))
+			printk(KERN_WARNING "mlx4_ib: modify QP %06x to RESET failed.\n",
+			       qp->mqp.qpn);
+
+	send_cq = to_mcq(qp->ibqp.send_cq);
+	recv_cq = to_mcq(qp->ibqp.recv_cq);
+
+	mlx4_ib_lock_cqs(send_cq, recv_cq);
+
+	if (!is_user) {
+		__mlx4_ib_cq_clean(recv_cq, qp->mqp.qpn,
+				 qp->ibqp.srq ? to_msrq(qp->ibqp.srq): NULL);
+		if (send_cq != recv_cq)
+			__mlx4_ib_cq_clean(send_cq, qp->mqp.qpn, NULL);
+	}
+
+	mlx4_qp_remove(dev->dev, &qp->mqp);
+
+	mlx4_ib_unlock_cqs(send_cq, recv_cq);
+
+	mlx4_qp_free(dev->dev, &qp->mqp);
+	mlx4_mtt_cleanup(dev->dev, &qp->mtt);
+
+	if (is_user) {
+		mlx4_ib_db_unmap_user(to_mucontext(qp->ibqp.uobject->context),
+				      &qp->db);
+		ib_umem_release(qp->umem);
+	} else {
+		kfree(qp->sq.wrid);
+		kfree(qp->rq.wrid);
+		mlx4_buf_free(dev->dev, qp->buf_size, &qp->buf);
+		mlx4_ib_db_free(dev, &qp->db);
+	}
+}
+
+struct ib_qp *mlx4_ib_create_qp(struct ib_pd *pd,
+				struct ib_qp_init_attr *init_attr,
+				struct ib_udata *udata)
+{
+	struct mlx4_ib_dev *dev = to_mdev(pd->device);
+	struct mlx4_ib_sqp *sqp;
+	struct mlx4_ib_qp *qp;
+	int err;
+
+	switch (init_attr->qp_type) {
+	case IB_QPT_RC:
+	case IB_QPT_UC:
+	case IB_QPT_UD:
+	{
+		qp = kmalloc(sizeof *qp, GFP_KERNEL);
+		if (!qp)
+			return ERR_PTR(-ENOMEM);
+
+		err = create_qp_common(dev, pd, init_attr, udata, 0, qp);
+		if (err) {
+			kfree(qp);
+			return ERR_PTR(err);
+		}
+
+		qp->ibqp.qp_num = qp->mqp.qpn;
+
+		break;
+	}
+	case IB_QPT_SMI:
+	case IB_QPT_GSI:
+	{
+		/* Userspace is not allowed to create special QPs: */
+		if (pd->uobject)
+			return ERR_PTR(-EINVAL);
+
+		sqp = kmalloc(sizeof *sqp, GFP_KERNEL);
+		if (!sqp)
+			return ERR_PTR(-ENOMEM);
+
+		qp = &sqp->qp;
+
+		err = create_qp_common(dev, pd, init_attr, udata,
+				       dev->dev->caps.sqp_start +
+				       (init_attr->qp_type == IB_QPT_SMI ? 0 : 2) +
+				       init_attr->port_num - 1,
+				       qp);
+		if (err) {
+			kfree(sqp);
+			return ERR_PTR(err);
+		}
+
+		qp->port	= init_attr->port_num;
+		qp->ibqp.qp_num = init_attr->qp_type == IB_QPT_SMI ? 0 : 1;
+
+		break;
+	}
+	default:
+		/* Don't support raw QPs */
+		return ERR_PTR(-EINVAL);
+	}
+
+	return &qp->ibqp;
+}
+
+int mlx4_ib_destroy_qp(struct ib_qp *qp)
+{
+	struct mlx4_ib_dev *dev = to_mdev(qp->device);
+	struct mlx4_ib_qp *mqp = to_mqp(qp);
+
+	if (is_qp0(dev, mqp))
+		mlx4_CLOSE_PORT(dev->dev, mqp->port);
+
+	destroy_qp_common(dev, mqp, !!qp->pd->uobject);
+
+	if (is_sqp(dev, mqp))
+		kfree(to_msqp(mqp));
+	else
+		kfree(mqp);
+
+	return 0;
+}
+
+static void init_port(struct mlx4_ib_dev *dev, int port)
+{
+	struct mlx4_init_port_param param;
+	int err;
+
+	memset(&param, 0, sizeof param);
+
+	param.port_width_cap = dev->dev->caps.port_width_cap;
+	param.vl_cap	     = dev->dev->caps.vl_cap;
+	param.mtu	     = ib_mtu_enum_to_int(dev->dev->caps.mtu_cap);
+	param.max_gid	     = dev->dev->caps.gid_table_len;
+	param.max_pkey	     = dev->dev->caps.pkey_table_len;
+
+	err = mlx4_INIT_PORT(dev->dev, &param, port);
+	if (err)
+		printk(KERN_WARNING "INIT_PORT failed, return code %d.\n", err);
+}
+
+static int to_mlx4_st(enum ib_qp_type type)
+{
+	switch (type) {
+	case IB_QPT_RC:		return MLX4_QP_ST_RC;
+	case IB_QPT_UC:		return MLX4_QP_ST_UC;
+	case IB_QPT_UD:		return MLX4_QP_ST_UD;
+	case IB_QPT_SMI:
+	case IB_QPT_GSI:	return MLX4_QP_ST_MLX;
+	default:		return -1;
+	}
+}
+
+static __be32 to_mlx4_access_flags(struct mlx4_ib_qp *qp, struct ib_qp_attr *attr,
+				   int attr_mask)
+{
+	u8 dest_rd_atomic;
+	u32 access_flags;
+	u32 hw_access_flags = 0;
+
+	if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC)
+		dest_rd_atomic = attr->max_dest_rd_atomic;
+	else
+		dest_rd_atomic = qp->resp_depth;
+
+	if (attr_mask & IB_QP_ACCESS_FLAGS)
+		access_flags = attr->qp_access_flags;
+	else
+		access_flags = qp->atomic_rd_en;
+
+	if (!dest_rd_atomic)
+		access_flags &= IB_ACCESS_REMOTE_WRITE;
+
+	if (access_flags & IB_ACCESS_REMOTE_READ)
+		hw_access_flags |= MLX4_QP_BIT_RRE;
+	if (access_flags & IB_ACCESS_REMOTE_ATOMIC)
+		hw_access_flags |= MLX4_QP_BIT_RAE;
+	if (access_flags & IB_ACCESS_REMOTE_WRITE)
+		hw_access_flags |= MLX4_QP_BIT_RWE;
+
+	return cpu_to_be32(hw_access_flags);
+}
+
+static void store_sqp_attrs(struct mlx4_ib_sqp *sqp, struct ib_qp_attr *attr,
+			    int attr_mask)
+{
+	if (attr_mask & IB_QP_PKEY_INDEX)
+		sqp->pkey_index = attr->pkey_index;
+	if (attr_mask & IB_QP_QKEY)
+		sqp->qkey = attr->qkey;
+	if (attr_mask & IB_QP_SQ_PSN)
+		sqp->send_psn = attr->sq_psn;
+}
+
+static void mlx4_set_sched(struct mlx4_qp_path *path, u8 port)
+{
+	path->sched_queue = (path->sched_queue & 0xbf) | ((port - 1) << 6);
+}
+
+static int mlx4_set_path(struct mlx4_ib_dev *dev, struct ib_ah_attr *ah,
+			 struct mlx4_qp_path *path, u8 port)
+{
+	path->grh_mylmc     = ah->src_path_bits & 0x7f;
+	path->rlid	    = cpu_to_be16(ah->dlid);
+	if (ah->static_rate) {
+		path->static_rate = ah->static_rate + MLX4_STAT_RATE_OFFSET;
+		while (path->static_rate > IB_RATE_2_5_GBPS + MLX4_STAT_RATE_OFFSET &&
+		       !(1 << path->static_rate & dev->dev->caps.stat_rate_support))
+			--path->static_rate;
+	} else
+		path->static_rate = 0;
+	path->counter_index = 0xff;
+
+	if (ah->ah_flags & IB_AH_GRH) {
+		if (ah->grh.sgid_index >= dev->dev->caps.gid_table_len) {
+			printk(KERN_ERR "sgid_index (%u) too large. max is %d\n",
+			       ah->grh.sgid_index, dev->dev->caps.gid_table_len - 1);
+			return -1;
+		}
+
+		path->grh_mylmc |= 1 << 7;
+		path->mgid_index = ah->grh.sgid_index;
+		path->hop_limit  = ah->grh.hop_limit;
+		path->tclass_flowlabel =
+			cpu_to_be32((ah->grh.traffic_class << 20) |
+				    (ah->grh.flow_label));
+		memcpy(path->rgid, ah->grh.dgid.raw, 16);
+	}
+
+	path->sched_queue = MLX4_IB_DEFAULT_SCHED_QUEUE |
+		((port - 1) << 6) | ((ah->sl & 0xf) << 2);
+
+	return 0;
+}
+
+int mlx4_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
+		      int attr_mask, struct ib_udata *udata)
+{
+	struct mlx4_ib_dev *dev = to_mdev(ibqp->device);
+	struct mlx4_ib_qp *qp = to_mqp(ibqp);
+	struct mlx4_qp_context *context;
+	enum mlx4_qp_optpar optpar = 0;
+	enum ib_qp_state cur_state, new_state;
+	int sqd_event;
+	int err = -EINVAL;
+
+	context = kzalloc(sizeof *context, GFP_KERNEL);
+	if (!context)
+		return -ENOMEM;
+
+	mutex_lock(&qp->mutex);
+
+	cur_state = attr_mask & IB_QP_CUR_STATE ? attr->cur_qp_state : qp->state;
+	new_state = attr_mask & IB_QP_STATE ? attr->qp_state : cur_state;
+
+	if (!ib_modify_qp_is_ok(cur_state, new_state, ibqp->qp_type, attr_mask))
+		goto out;
+
+	if ((attr_mask & IB_QP_PKEY_INDEX) &&
+	     attr->pkey_index >= dev->dev->caps.pkey_table_len) {
+		goto out;
+	}
+
+	if ((attr_mask & IB_QP_PORT) &&
+	    (attr->port_num == 0 || attr->port_num > dev->dev->caps.num_ports)) {
+		goto out;
+	}
+
+	if (attr_mask & IB_QP_MAX_QP_RD_ATOMIC &&
+	    attr->max_rd_atomic > dev->dev->caps.max_qp_init_rdma) {
+		goto out;
+	}
+
+	if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC &&
+	    attr->max_dest_rd_atomic > 1 << dev->dev->caps.max_qp_dest_rdma) {
+		goto out;
+	}
+
+	context->flags = cpu_to_be32((to_mlx4_state(new_state) << 28) |
+				     (to_mlx4_st(ibqp->qp_type) << 16));
+	context->flags     |= cpu_to_be32(1 << 8); /* DE? */
+
+	if (!(attr_mask & IB_QP_PATH_MIG_STATE))
+		context->flags |= cpu_to_be32(MLX4_QP_PM_MIGRATED << 11);
+	else {
+		optpar |= MLX4_QP_OPTPAR_PM_STATE;
+		switch (attr->path_mig_state) {
+		case IB_MIG_MIGRATED:
+			context->flags |= cpu_to_be32(MLX4_QP_PM_MIGRATED << 11);
+			break;
+		case IB_MIG_REARM:
+			context->flags |= cpu_to_be32(MLX4_QP_PM_REARM << 11);
+			break;
+		case IB_MIG_ARMED:
+			context->flags |= cpu_to_be32(MLX4_QP_PM_ARMED << 11);
+			break;
+		}
+	}
+
+	if (ibqp->qp_type == IB_QPT_GSI || ibqp->qp_type == IB_QPT_SMI ||
+	    ibqp->qp_type == IB_QPT_UD)
+		context->mtu_msgmax = (IB_MTU_4096 << 5) | 11;
+	else if (attr_mask & IB_QP_PATH_MTU) {
+		if (attr->path_mtu < IB_MTU_256 || attr->path_mtu > IB_MTU_4096) {
+			printk(KERN_ERR "path MTU (%u) is invalid\n",
+			       attr->path_mtu);
+			return -EINVAL;
+		}
+		context->mtu_msgmax = (attr->path_mtu << 5) | 31;
+	}
+
+	if (qp->rq.max)
+		context->rq_size_stride = ilog2(qp->rq.max) << 3;
+	context->rq_size_stride |= qp->rq.wqe_shift - 4;
+
+	if (qp->sq.max)
+		context->sq_size_stride = ilog2(qp->sq.max) << 3;
+	context->sq_size_stride |= qp->sq.wqe_shift - 4;
+
+	if (qp->ibqp.uobject)
+		context->usr_page = cpu_to_be32(to_mucontext(ibqp->uobject->context)->uar.index);
+	else
+		context->usr_page = cpu_to_be32(dev->priv_uar.index);
+
+	if (attr_mask & IB_QP_DEST_QPN)
+		context->remote_qpn = cpu_to_be32(attr->dest_qp_num);
+
+	if (attr_mask & IB_QP_PORT) {
+		if (cur_state == IB_QPS_SQD && new_state == IB_QPS_SQD &&
+		    !(attr_mask & IB_QP_AV)) {
+			mlx4_set_sched(&context->pri_path, attr->port_num);
+			optpar |= MLX4_QP_OPTPAR_SCHED_QUEUE;
+		}
+	}
+
+	if (attr_mask & IB_QP_PKEY_INDEX) {
+		context->pri_path.pkey_index = attr->pkey_index;
+		optpar |= MLX4_QP_OPTPAR_PKEY_INDEX;
+	}
+
+	if (attr_mask & IB_QP_RNR_RETRY) {
+		context->params1 |= cpu_to_be32(attr->rnr_retry << 13);
+		optpar |= MLX4_QP_OPTPAR_RNR_RETRY;
+	}
+
+	if (attr_mask & IB_QP_AV) {
+		if (mlx4_set_path(dev, &attr->ah_attr, &context->pri_path,
+				  attr_mask & IB_QP_PORT ? attr->port_num : qp->port)) {
+			err = -EINVAL;
+			goto out;
+		}
+
+		optpar |= (MLX4_QP_OPTPAR_PRIMARY_ADDR_PATH |
+			   MLX4_QP_OPTPAR_SCHED_QUEUE);
+	}
+
+	if (attr_mask & IB_QP_TIMEOUT) {
+		context->pri_path.ackto = attr->timeout << 3;
+		optpar |= MLX4_QP_OPTPAR_ACK_TIMEOUT;
+	}
+
+	if (attr_mask & IB_QP_ALT_PATH) {
+		if (attr->alt_pkey_index >= dev->dev->caps.pkey_table_len)
+			return -EINVAL;
+
+		if (attr->alt_port_num == 0 ||
+		    attr->alt_port_num > dev->dev->caps.num_ports)
+			return -EINVAL;
+
+		if (mlx4_set_path(dev, &attr->alt_ah_attr, &context->alt_path,
+				  attr->alt_port_num))
+			return -EINVAL;
+
+		context->alt_path.pkey_index = attr->alt_pkey_index;
+		context->alt_path.ackto = attr->alt_timeout << 3;
+		optpar |= MLX4_QP_OPTPAR_ALT_ADDR_PATH;
+	}
+
+	context->pd	    = cpu_to_be32(to_mpd(ibqp->pd)->pdn);
+	context->params1    = cpu_to_be32(MLX4_IB_ACK_REQ_FREQ << 28);
+	if (attr_mask & IB_QP_RETRY_CNT) {
+		context->params1 |= cpu_to_be32(attr->retry_cnt << 16);
+		optpar |= MLX4_QP_OPTPAR_RETRY_COUNT;
+	}
+
+	if (attr_mask & IB_QP_MAX_QP_RD_ATOMIC) {
+		if (attr->max_rd_atomic)
+			context->params1 |=
+				cpu_to_be32(fls(attr->max_rd_atomic - 1) << 21);
+		optpar |= MLX4_QP_OPTPAR_SRA_MAX;
+	}
+
+	if (attr_mask & IB_QP_SQ_PSN)
+		context->next_send_psn = cpu_to_be32(attr->sq_psn);
+
+	context->cqn_send = cpu_to_be32(to_mcq(ibqp->send_cq)->mcq.cqn);
+
+	if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC) {
+		if (attr->max_dest_rd_atomic)
+			context->params2 |=
+				cpu_to_be32(fls(attr->max_dest_rd_atomic - 1) << 21);
+		optpar |= MLX4_QP_OPTPAR_RRA_MAX;
+	}
+
+	if (attr_mask & (IB_QP_ACCESS_FLAGS | IB_QP_MAX_DEST_RD_ATOMIC)) {
+		context->params2 |= to_mlx4_access_flags(qp, attr, attr_mask);
+		optpar |= MLX4_QP_OPTPAR_RWE | MLX4_QP_OPTPAR_RRE | MLX4_QP_OPTPAR_RAE;
+	}
+
+	if (ibqp->srq)
+		context->params2 |= cpu_to_be32(MLX4_QP_BIT_RIC);
+
+	if (attr_mask & IB_QP_MIN_RNR_TIMER) {
+		context->rnr_nextrecvpsn |= cpu_to_be32(attr->min_rnr_timer << 24);
+		optpar |= MLX4_QP_OPTPAR_RNR_TIMEOUT;
+	}
+	if (attr_mask & IB_QP_RQ_PSN)
+		context->rnr_nextrecvpsn |= cpu_to_be32(attr->rq_psn);
+
+	context->cqn_recv = cpu_to_be32(to_mcq(ibqp->recv_cq)->mcq.cqn);
+
+	if (attr_mask & IB_QP_QKEY) {
+		context->qkey = cpu_to_be32(attr->qkey);
+		optpar |= MLX4_QP_OPTPAR_Q_KEY;
+	}
+
+	if (ibqp->srq)
+		context->srqn = cpu_to_be32(1 << 24 | to_msrq(ibqp->srq)->msrq.srqn);
+
+	if (cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT)
+		context->db_rec_addr = cpu_to_be64(qp->db.dma);
+
+	if (cur_state == IB_QPS_INIT &&
+	    new_state == IB_QPS_RTR  &&
+	    (ibqp->qp_type == IB_QPT_GSI || ibqp->qp_type == IB_QPT_SMI ||
+	     ibqp->qp_type == IB_QPT_UD)) {
+		context->pri_path.sched_queue = (qp->port - 1) << 6;
+		if (is_qp0(dev, qp))
+			context->pri_path.sched_queue |= MLX4_IB_DEFAULT_QP0_SCHED_QUEUE;
+		else
+			context->pri_path.sched_queue |= MLX4_IB_DEFAULT_SCHED_QUEUE;
+	}
+
+	if (cur_state == IB_QPS_RTS && new_state == IB_QPS_SQD	&&
+	    attr_mask & IB_QP_EN_SQD_ASYNC_NOTIFY && attr->en_sqd_async_notify)
+		sqd_event = 1;
+	else
+		sqd_event = 0;
+
+	err = mlx4_qp_modify(dev->dev, &qp->mtt, to_mlx4_state(cur_state),
+			     to_mlx4_state(new_state), context, optpar,
+			     sqd_event, &qp->mqp);
+	if (err)
+		goto out;
+
+	qp->state = new_state;
+
+	if (attr_mask & IB_QP_ACCESS_FLAGS)
+		qp->atomic_rd_en = attr->qp_access_flags;
+	if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC)
+		qp->resp_depth = attr->max_dest_rd_atomic;
+	if (attr_mask & IB_QP_PORT)
+		qp->port = attr->port_num;
+	if (attr_mask & IB_QP_ALT_PATH)
+		qp->alt_port = attr->alt_port_num;
+
+	if (is_sqp(dev, qp))
+		store_sqp_attrs(to_msqp(qp), attr, attr_mask);
+
+	/*
+	 * If we moved QP0 to RTR, bring the IB link up; if we moved
+	 * QP0 to RESET or ERROR, bring the link back down.
+	 */
+	if (is_qp0(dev, qp)) {
+		if (cur_state != IB_QPS_RTR && new_state == IB_QPS_RTR)
+			init_port(dev, qp->port);
+
+		if (cur_state != IB_QPS_RESET && cur_state != IB_QPS_ERR &&
+		    (new_state == IB_QPS_RESET || new_state == IB_QPS_ERR))
+			mlx4_CLOSE_PORT(dev->dev, qp->port);
+	}
+
+	/*
+	 * If we moved a kernel QP to RESET, clean up all old CQ
+	 * entries and reinitialize the QP.
+	 */
+	if (new_state == IB_QPS_RESET && !ibqp->uobject) {
+		mlx4_ib_cq_clean(to_mcq(ibqp->recv_cq), qp->mqp.qpn,
+				 ibqp->srq ? to_msrq(ibqp->srq): NULL);
+		if (ibqp->send_cq != ibqp->recv_cq)
+			mlx4_ib_cq_clean(to_mcq(ibqp->send_cq), qp->mqp.qpn, NULL);
+
+		qp->rq.head = 0;
+		qp->rq.tail = 0;
+		qp->sq.head = 0;
+		qp->sq.tail = 0;
+		*qp->db.db  = 0;
+	}
+
+out:
+	mutex_unlock(&qp->mutex);
+	kfree(context);
+	return err;
+}
+
+static int build_mlx_header(struct mlx4_ib_sqp *sqp, struct ib_send_wr *wr,
+			    void *wqe)
+{
+	struct ib_device *ib_dev = &to_mdev(sqp->qp.ibqp.device)->ib_dev;
+	struct mlx4_wqe_mlx_seg *mlx = wqe;
+	struct mlx4_wqe_inline_seg *inl = wqe + sizeof *mlx;
+	struct mlx4_ib_ah *ah = to_mah(wr->wr.ud.ah);
+	u16 pkey;
+	int send_size;
+	int header_size;
+	int i;
+
+	send_size = 0;
+	for (i = 0; i < wr->num_sge; ++i)
+		send_size += wr->sg_list[i].length;
+
+	ib_ud_header_init(send_size, mlx4_ib_ah_grh_present(ah), &sqp->ud_header);
+
+	sqp->ud_header.lrh.service_level   =
+		be32_to_cpu(ah->av.sl_tclass_flowlabel) >> 28;
+	sqp->ud_header.lrh.destination_lid = ah->av.dlid;
+	sqp->ud_header.lrh.source_lid      = cpu_to_be16(ah->av.g_slid & 0x7f);
+	if (mlx4_ib_ah_grh_present(ah)) {
+		sqp->ud_header.grh.traffic_class =
+			(be32_to_cpu(ah->av.sl_tclass_flowlabel) >> 20) & 0xff;
+		sqp->ud_header.grh.flow_label    =
+			ah->av.sl_tclass_flowlabel & cpu_to_be32(0xfffff);
+		ib_get_cached_gid(ib_dev, be32_to_cpu(ah->av.port_pd) >> 24,
+				  ah->av.gid_index, &sqp->ud_header.grh.source_gid);
+		memcpy(sqp->ud_header.grh.destination_gid.raw,
+		       ah->av.dgid, 16);
+	}
+
+	mlx->flags &= cpu_to_be32(MLX4_WQE_CTRL_CQ_UPDATE);
+	mlx->flags |= cpu_to_be32((!sqp->qp.ibqp.qp_num ? MLX4_WQE_MLX_VL15 : 0) |
+				  (sqp->ud_header.lrh.destination_lid ==
+				   IB_LID_PERMISSIVE ? MLX4_WQE_MLX_SLR : 0) |
+				  (sqp->ud_header.lrh.service_level << 8));
+	mlx->rlid   = sqp->ud_header.lrh.destination_lid;
+
+	switch (wr->opcode) {
+	case IB_WR_SEND:
+		sqp->ud_header.bth.opcode	 = IB_OPCODE_UD_SEND_ONLY;
+		sqp->ud_header.immediate_present = 0;
+		break;
+	case IB_WR_SEND_WITH_IMM:
+		sqp->ud_header.bth.opcode	 = IB_OPCODE_UD_SEND_ONLY_WITH_IMMEDIATE;
+		sqp->ud_header.immediate_present = 1;
+		sqp->ud_header.immediate_data    = wr->imm_data;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	sqp->ud_header.lrh.virtual_lane    = !sqp->qp.ibqp.qp_num ? 15 : 0;
+	if (sqp->ud_header.lrh.destination_lid == IB_LID_PERMISSIVE)
+		sqp->ud_header.lrh.source_lid = IB_LID_PERMISSIVE;
+	sqp->ud_header.bth.solicited_event = !!(wr->send_flags & IB_SEND_SOLICITED);
+	if (!sqp->qp.ibqp.qp_num)
+		ib_get_cached_pkey(ib_dev, sqp->qp.port, sqp->pkey_index, &pkey);
+	else
+		ib_get_cached_pkey(ib_dev, sqp->qp.port, wr->wr.ud.pkey_index, &pkey);
+	sqp->ud_header.bth.pkey = cpu_to_be16(pkey);
+	sqp->ud_header.bth.destination_qpn = cpu_to_be32(wr->wr.ud.remote_qpn);
+	sqp->ud_header.bth.psn = cpu_to_be32((sqp->send_psn++) & ((1 << 24) - 1));
+	sqp->ud_header.deth.qkey = cpu_to_be32(wr->wr.ud.remote_qkey & 0x80000000 ?
+					       sqp->qkey : wr->wr.ud.remote_qkey);
+	sqp->ud_header.deth.source_qpn = cpu_to_be32(sqp->qp.ibqp.qp_num);
+
+	header_size = ib_ud_header_pack(&sqp->ud_header, sqp->header_buf);
+
+	if (0) {
+		printk(KERN_ERR "built UD header of size %d:\n", header_size);
+		for (i = 0; i < header_size / 4; ++i) {
+			if (i % 8 == 0)
+				printk("  [%02x] ", i * 4);
+			printk(" %08x",
+			       be32_to_cpu(((__be32 *) sqp->header_buf)[i]));
+			if ((i + 1) % 8 == 0)
+				printk("\n");
+		}
+		printk("\n");
+	}
+
+	inl->byte_count = cpu_to_be32(1 << 31 | header_size);
+	memcpy(inl + 1, sqp->header_buf, header_size);
+
+	return ALIGN(sizeof (struct mlx4_wqe_inline_seg) + header_size, 16);
+}
+
+static int mlx4_wq_overflow(struct mlx4_ib_wq *wq, int nreq, struct ib_cq *ib_cq)
+{
+	unsigned cur;
+	struct mlx4_ib_cq *cq;
+
+	cur = wq->head - wq->tail;
+	if (likely(cur + nreq < wq->max))
+		return 0;
+
+	cq = to_mcq(ib_cq);
+	spin_lock(&cq->lock);
+	cur = wq->head - wq->tail;
+	spin_unlock(&cq->lock);
+
+	return cur + nreq >= wq->max;
+}
+
+int mlx4_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
+		      struct ib_send_wr **bad_wr)
+{
+	struct mlx4_ib_qp *qp = to_mqp(ibqp);
+	void *wqe;
+	struct mlx4_wqe_ctrl_seg *ctrl;
+	unsigned long flags;
+	int nreq;
+	int err = 0;
+	int ind;
+	int size;
+	int i;
+
+	spin_lock_irqsave(&qp->rq.lock, flags);
+
+	ind = qp->sq.head;
+
+	for (nreq = 0; wr; ++nreq, wr = wr->next) {
+		if (mlx4_wq_overflow(&qp->sq, nreq, qp->ibqp.send_cq)) {
+			err = -ENOMEM;
+			*bad_wr = wr;
+			goto out;
+		}
+
+		if (unlikely(wr->num_sge > qp->sq.max_gs)) {
+			err = -EINVAL;
+			*bad_wr = wr;
+			goto out;
+		}
+
+		ctrl = wqe = get_send_wqe(qp, ind & (qp->sq.max - 1));
+		qp->sq.wrid[ind & (qp->sq.max - 1)] = wr->wr_id;
+
+		ctrl->srcrb_flags =
+			(wr->send_flags & IB_SEND_SIGNALED ?
+			 cpu_to_be32(MLX4_WQE_CTRL_CQ_UPDATE) : 0) |
+			(wr->send_flags & IB_SEND_SOLICITED ?
+			 cpu_to_be32(MLX4_WQE_CTRL_SOLICITED) : 0) |
+			qp->sq_signal_bits;
+
+		if (wr->opcode == IB_WR_SEND_WITH_IMM ||
+		    wr->opcode == IB_WR_RDMA_WRITE_WITH_IMM)
+			ctrl->imm = wr->imm_data;
+		else
+			ctrl->imm = 0;
+
+		wqe += sizeof *ctrl;
+		size = sizeof *ctrl / 16;
+
+		switch (ibqp->qp_type) {
+		case IB_QPT_RC:
+		case IB_QPT_UC:
+			switch (wr->opcode) {
+			case IB_WR_ATOMIC_CMP_AND_SWP:
+			case IB_WR_ATOMIC_FETCH_AND_ADD:
+				((struct mlx4_wqe_raddr_seg *) wqe)->raddr =
+					cpu_to_be64(wr->wr.atomic.remote_addr);
+				((struct mlx4_wqe_raddr_seg *) wqe)->rkey =
+					cpu_to_be32(wr->wr.atomic.rkey);
+				((struct mlx4_wqe_raddr_seg *) wqe)->reserved = 0;
+
+				wqe  += sizeof (struct mlx4_wqe_raddr_seg);
+
+				if (wr->opcode == IB_WR_ATOMIC_CMP_AND_SWP) {
+					((struct mlx4_wqe_atomic_seg *) wqe)->swap_add =
+						cpu_to_be64(wr->wr.atomic.swap);
+					((struct mlx4_wqe_atomic_seg *) wqe)->compare =
+						cpu_to_be64(wr->wr.atomic.compare_add);
+				} else {
+					((struct mlx4_wqe_atomic_seg *) wqe)->swap_add =
+						cpu_to_be64(wr->wr.atomic.compare_add);
+					((struct mlx4_wqe_atomic_seg *) wqe)->compare = 0;
+				}
+
+				wqe  += sizeof (struct mlx4_wqe_atomic_seg);
+				size += (sizeof (struct mlx4_wqe_raddr_seg) +
+					 sizeof (struct mlx4_wqe_atomic_seg)) / 16;
+
+				break;
+
+			case IB_WR_RDMA_READ:
+			case IB_WR_RDMA_WRITE:
+			case IB_WR_RDMA_WRITE_WITH_IMM:
+				((struct mlx4_wqe_raddr_seg *) wqe)->raddr =
+					cpu_to_be64(wr->wr.rdma.remote_addr);
+				((struct mlx4_wqe_raddr_seg *) wqe)->rkey =
+					cpu_to_be32(wr->wr.rdma.rkey);
+				((struct mlx4_wqe_raddr_seg *) wqe)->reserved = 0;
+
+				wqe  += sizeof (struct mlx4_wqe_raddr_seg);
+				size += sizeof (struct mlx4_wqe_raddr_seg) / 16;
+
+				break;
+
+			default:
+				/* No extra segments required for sends */
+				break;
+			}
+			break;
+
+		case IB_QPT_UD:
+			memcpy(((struct mlx4_wqe_datagram_seg *) wqe)->av,
+			       &to_mah(wr->wr.ud.ah)->av, sizeof (struct mlx4_av));
+			((struct mlx4_wqe_datagram_seg *) wqe)->dqpn =
+				cpu_to_be32(wr->wr.ud.remote_qpn);
+			((struct mlx4_wqe_datagram_seg *) wqe)->qkey =
+				cpu_to_be32(wr->wr.ud.remote_qkey);
+
+			wqe  += sizeof (struct mlx4_wqe_datagram_seg);
+			size += sizeof (struct mlx4_wqe_datagram_seg) / 16;
+			break;
+
+		case IB_QPT_SMI:
+		case IB_QPT_GSI:
+			err = build_mlx_header(to_msqp(qp), wr, ctrl);
+			if (err < 0) {
+				*bad_wr = wr;
+				goto out;
+			}
+			wqe  += err;
+			size += err / 16;
+
+			err = 0;
+			break;
+
+		default:
+			break;
+		}
+
+		for (i = 0; i < wr->num_sge; ++i) {
+			((struct mlx4_wqe_data_seg *) wqe)->byte_count =
+				cpu_to_be32(wr->sg_list[i].length);
+			((struct mlx4_wqe_data_seg *) wqe)->lkey =
+				cpu_to_be32(wr->sg_list[i].lkey);
+			((struct mlx4_wqe_data_seg *) wqe)->addr =
+				cpu_to_be64(wr->sg_list[i].addr);
+
+			wqe  += sizeof (struct mlx4_wqe_data_seg);
+			size += sizeof (struct mlx4_wqe_data_seg) / 16;
+		}
+
+		/* Add one more inline data segment for ICRC for MLX sends */
+		if (qp->ibqp.qp_type == IB_QPT_SMI || qp->ibqp.qp_type == IB_QPT_GSI) {
+			((struct mlx4_wqe_inline_seg *) wqe)->byte_count =
+				cpu_to_be32((1 << 31) | 4);
+			((u32 *) wqe)[1] = 0;
+			wqe  += sizeof (struct mlx4_wqe_data_seg);
+			size += sizeof (struct mlx4_wqe_data_seg) / 16;
+		}
+
+		ctrl->fence_size = (wr->send_flags & IB_SEND_FENCE ?
+				    MLX4_WQE_CTRL_FENCE : 0) | size;
+
+		/*
+		 * Make sure descriptor is fully written before
+		 * setting ownership bit (because HW can start
+		 * executing as soon as we do).
+		 */
+		wmb();
+
+		if (wr->opcode < 0 || wr->opcode > ARRAY_SIZE(mlx4_ib_opcode)) {
+			err = -EINVAL;
+			goto out;
+		}
+
+		ctrl->owner_opcode = mlx4_ib_opcode[wr->opcode] |
+			(ind & qp->sq.max ? cpu_to_be32(1 << 31) : 0);
+
+		++ind;
+	}
+
+out:
+	if (likely(nreq)) {
+		qp->sq.head += nreq;
+
+		/*
+		 * Make sure that descriptors are written before
+		 * doorbell record.
+		 */
+		wmb();
+
+		writel(qp->doorbell_qpn,
+		       to_mdev(ibqp->device)->uar_map + MLX4_SEND_DOORBELL);
+
+		/*
+		 * Make sure doorbells don't leak out of SQ spinlock
+		 * and reach the HCA out of order.
+		 */
+		mmiowb();
+	}
+
+	spin_unlock_irqrestore(&qp->rq.lock, flags);
+
+	return err;
+}
+
+int mlx4_ib_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *wr,
+		      struct ib_recv_wr **bad_wr)
+{
+	struct mlx4_ib_qp *qp = to_mqp(ibqp);
+	struct mlx4_wqe_data_seg *scat;
+	unsigned long flags;
+	int err = 0;
+	int nreq;
+	int ind;
+	int i;
+
+	spin_lock_irqsave(&qp->rq.lock, flags);
+
+	ind = qp->rq.head & (qp->rq.max - 1);
+
+	for (nreq = 0; wr; ++nreq, wr = wr->next) {
+		if (mlx4_wq_overflow(&qp->rq, nreq, qp->ibqp.send_cq)) {
+			err = -ENOMEM;
+			*bad_wr = wr;
+			goto out;
+		}
+
+		if (unlikely(wr->num_sge > qp->rq.max_gs)) {
+			err = -EINVAL;
+			*bad_wr = wr;
+			goto out;
+		}
+
+		scat = get_recv_wqe(qp, ind);
+
+		for (i = 0; i < wr->num_sge; ++i) {
+			scat[i].byte_count = cpu_to_be32(wr->sg_list[i].length);
+			scat[i].lkey       = cpu_to_be32(wr->sg_list[i].lkey);
+			scat[i].addr       = cpu_to_be64(wr->sg_list[i].addr);
+		}
+
+		if (i < qp->rq.max_gs) {
+			scat[i].byte_count = 0;
+			scat[i].lkey       = cpu_to_be32(MLX4_INVALID_LKEY);
+			scat[i].addr       = 0;
+		}
+
+		qp->rq.wrid[ind] = wr->wr_id;
+
+		ind = (ind + 1) & (qp->rq.max - 1);
+	}
+
+out:
+	if (likely(nreq)) {
+		qp->rq.head += nreq;
+
+		/*
+		 * Make sure that descriptors are written before
+		 * doorbell record.
+		 */
+		wmb();
+
+		*qp->db.db = cpu_to_be32(qp->rq.head & 0xffff);
+	}
+
+	spin_unlock_irqrestore(&qp->rq.lock, flags);
+
+	return err;
+}
diff --git a/drivers/infiniband/hw/mlx4/srq.c b/drivers/infiniband/hw/mlx4/srq.c
new file mode 100644
index 000000000000..42ab4a801d6a
--- /dev/null
+++ b/drivers/infiniband/hw/mlx4/srq.c
@@ -0,0 +1,334 @@
+/*
+ * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx4/qp.h>
+#include <linux/mlx4/srq.h>
+
+#include "mlx4_ib.h"
+#include "user.h"
+
+static void *get_wqe(struct mlx4_ib_srq *srq, int n)
+{
+	int offset = n << srq->msrq.wqe_shift;
+
+	if (srq->buf.nbufs == 1)
+		return srq->buf.u.direct.buf + offset;
+	else
+		return srq->buf.u.page_list[offset >> PAGE_SHIFT].buf +
+			(offset & (PAGE_SIZE - 1));
+}
+
+static void mlx4_ib_srq_event(struct mlx4_srq *srq, enum mlx4_event type)
+{
+	struct ib_event event;
+	struct ib_srq *ibsrq = &to_mibsrq(srq)->ibsrq;
+
+	if (ibsrq->event_handler) {
+		event.device      = ibsrq->device;
+		event.element.srq = ibsrq;
+		switch (type) {
+		case MLX4_EVENT_TYPE_SRQ_LIMIT:
+			event.event = IB_EVENT_SRQ_LIMIT_REACHED;
+			break;
+		case MLX4_EVENT_TYPE_SRQ_CATAS_ERROR:
+			event.event = IB_EVENT_SRQ_ERR;
+			break;
+		default:
+			printk(KERN_WARNING "mlx4_ib: Unexpected event type %d "
+			       "on SRQ %06x\n", type, srq->srqn);
+			return;
+		}
+
+		ibsrq->event_handler(&event, ibsrq->srq_context);
+	}
+}
+
+struct ib_srq *mlx4_ib_create_srq(struct ib_pd *pd,
+				  struct ib_srq_init_attr *init_attr,
+				  struct ib_udata *udata)
+{
+	struct mlx4_ib_dev *dev = to_mdev(pd->device);
+	struct mlx4_ib_srq *srq;
+	struct mlx4_wqe_srq_next_seg *next;
+	int desc_size;
+	int buf_size;
+	int err;
+	int i;
+
+	/* Sanity check SRQ size before proceeding */
+	if (init_attr->attr.max_wr  >= dev->dev->caps.max_srq_wqes ||
+	    init_attr->attr.max_sge >  dev->dev->caps.max_srq_sge)
+		return ERR_PTR(-EINVAL);
+
+	srq = kmalloc(sizeof *srq, GFP_KERNEL);
+	if (!srq)
+		return ERR_PTR(-ENOMEM);
+
+	mutex_init(&srq->mutex);
+	spin_lock_init(&srq->lock);
+	srq->msrq.max    = roundup_pow_of_two(init_attr->attr.max_wr + 1);
+	srq->msrq.max_gs = init_attr->attr.max_sge;
+
+	desc_size = max(32UL,
+			roundup_pow_of_two(sizeof (struct mlx4_wqe_srq_next_seg) +
+					   srq->msrq.max_gs *
+					   sizeof (struct mlx4_wqe_data_seg)));
+	srq->msrq.wqe_shift = ilog2(desc_size);
+
+	buf_size = srq->msrq.max * desc_size;
+
+	if (pd->uobject) {
+		struct mlx4_ib_create_srq ucmd;
+
+		if (ib_copy_from_udata(&ucmd, udata, sizeof ucmd)) {
+			err = -EFAULT;
+			goto err_srq;
+		}
+
+		srq->umem = ib_umem_get(pd->uobject->context, ucmd.buf_addr,
+					buf_size, 0);
+		if (IS_ERR(srq->umem)) {
+			err = PTR_ERR(srq->umem);
+			goto err_srq;
+		}
+
+		err = mlx4_mtt_init(dev->dev, ib_umem_page_count(srq->umem),
+				    ilog2(srq->umem->page_size), &srq->mtt);
+		if (err)
+			goto err_buf;
+
+		err = mlx4_ib_umem_write_mtt(dev, &srq->mtt, srq->umem);
+		if (err)
+			goto err_mtt;
+
+		err = mlx4_ib_db_map_user(to_mucontext(pd->uobject->context),
+					  ucmd.db_addr, &srq->db);
+		if (err)
+			goto err_mtt;
+	} else {
+		err = mlx4_ib_db_alloc(dev, &srq->db, 0);
+		if (err)
+			goto err_srq;
+
+		*srq->db.db = 0;
+
+		if (mlx4_buf_alloc(dev->dev, buf_size, PAGE_SIZE * 2, &srq->buf)) {
+			err = -ENOMEM;
+			goto err_db;
+		}
+
+		srq->head    = 0;
+		srq->tail    = srq->msrq.max - 1;
+		srq->wqe_ctr = 0;
+
+		for (i = 0; i < srq->msrq.max; ++i) {
+			next = get_wqe(srq, i);
+			next->next_wqe_index =
+				cpu_to_be16((i + 1) & (srq->msrq.max - 1));
+		}
+
+		err = mlx4_mtt_init(dev->dev, srq->buf.npages, srq->buf.page_shift,
+				    &srq->mtt);
+		if (err)
+			goto err_buf;
+
+		err = mlx4_buf_write_mtt(dev->dev, &srq->mtt, &srq->buf);
+		if (err)
+			goto err_mtt;
+
+		srq->wrid = kmalloc(srq->msrq.max * sizeof (u64), GFP_KERNEL);
+		if (!srq->wrid) {
+			err = -ENOMEM;
+			goto err_mtt;
+		}
+	}
+
+	err = mlx4_srq_alloc(dev->dev, to_mpd(pd)->pdn, &srq->mtt,
+			     srq->db.dma, &srq->msrq);
+	if (err)
+		goto err_wrid;
+
+	srq->msrq.event = mlx4_ib_srq_event;
+
+	if (pd->uobject)
+		if (ib_copy_to_udata(udata, &srq->msrq.srqn, sizeof (__u32))) {
+			err = -EFAULT;
+			goto err_wrid;
+		}
+
+	init_attr->attr.max_wr = srq->msrq.max - 1;
+
+	return &srq->ibsrq;
+
+err_wrid:
+	if (pd->uobject)
+		mlx4_ib_db_unmap_user(to_mucontext(pd->uobject->context), &srq->db);
+	else
+		kfree(srq->wrid);
+
+err_mtt:
+	mlx4_mtt_cleanup(dev->dev, &srq->mtt);
+
+err_buf:
+	if (pd->uobject)
+		ib_umem_release(srq->umem);
+	else
+		mlx4_buf_free(dev->dev, buf_size, &srq->buf);
+
+err_db:
+	if (!pd->uobject)
+		mlx4_ib_db_free(dev, &srq->db);
+
+err_srq:
+	kfree(srq);
+
+	return ERR_PTR(err);
+}
+
+int mlx4_ib_modify_srq(struct ib_srq *ibsrq, struct ib_srq_attr *attr,
+		       enum ib_srq_attr_mask attr_mask, struct ib_udata *udata)
+{
+	struct mlx4_ib_dev *dev = to_mdev(ibsrq->device);
+	struct mlx4_ib_srq *srq = to_msrq(ibsrq);
+	int ret;
+
+	/* We don't support resizing SRQs (yet?) */
+	if (attr_mask & IB_SRQ_MAX_WR)
+		return -EINVAL;
+
+	if (attr_mask & IB_SRQ_LIMIT) {
+		if (attr->srq_limit >= srq->msrq.max)
+			return -EINVAL;
+
+		mutex_lock(&srq->mutex);
+		ret = mlx4_srq_arm(dev->dev, &srq->msrq, attr->srq_limit);
+		mutex_unlock(&srq->mutex);
+
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+int mlx4_ib_destroy_srq(struct ib_srq *srq)
+{
+	struct mlx4_ib_dev *dev = to_mdev(srq->device);
+	struct mlx4_ib_srq *msrq = to_msrq(srq);
+
+	mlx4_srq_free(dev->dev, &msrq->msrq);
+	mlx4_mtt_cleanup(dev->dev, &msrq->mtt);
+
+	if (srq->uobject) {
+		mlx4_ib_db_unmap_user(to_mucontext(srq->uobject->context), &msrq->db);
+		ib_umem_release(msrq->umem);
+	} else {
+		kfree(msrq->wrid);
+		mlx4_buf_free(dev->dev, msrq->msrq.max << msrq->msrq.wqe_shift,
+			      &msrq->buf);
+		mlx4_ib_db_free(dev, &msrq->db);
+	}
+
+	kfree(msrq);
+
+	return 0;
+}
+
+void mlx4_ib_free_srq_wqe(struct mlx4_ib_srq *srq, int wqe_index)
+{
+	struct mlx4_wqe_srq_next_seg *next;
+
+	/* always called with interrupts disabled. */
+	spin_lock(&srq->lock);
+
+	next = get_wqe(srq, srq->tail);
+	next->next_wqe_index = cpu_to_be16(wqe_index);
+	srq->tail = wqe_index;
+
+	spin_unlock(&srq->lock);
+}
+
+int mlx4_ib_post_srq_recv(struct ib_srq *ibsrq, struct ib_recv_wr *wr,
+			  struct ib_recv_wr **bad_wr)
+{
+	struct mlx4_ib_srq *srq = to_msrq(ibsrq);
+	struct mlx4_wqe_srq_next_seg *next;
+	struct mlx4_wqe_data_seg *scat;
+	unsigned long flags;
+	int err = 0;
+	int nreq;
+	int i;
+
+	spin_lock_irqsave(&srq->lock, flags);
+
+	for (nreq = 0; wr; ++nreq, wr = wr->next) {
+		if (unlikely(wr->num_sge > srq->msrq.max_gs)) {
+			err = -EINVAL;
+			*bad_wr = wr;
+			break;
+		}
+
+		srq->wrid[srq->head] = wr->wr_id;
+
+		next      = get_wqe(srq, srq->head);
+		srq->head = be16_to_cpu(next->next_wqe_index);
+		scat      = (struct mlx4_wqe_data_seg *) (next + 1);
+
+		for (i = 0; i < wr->num_sge; ++i) {
+			scat[i].byte_count = cpu_to_be32(wr->sg_list[i].length);
+			scat[i].lkey       = cpu_to_be32(wr->sg_list[i].lkey);
+			scat[i].addr       = cpu_to_be64(wr->sg_list[i].addr);
+		}
+
+		if (i < srq->msrq.max_gs) {
+			scat[i].byte_count = 0;
+			scat[i].lkey       = cpu_to_be32(MLX4_INVALID_LKEY);
+			scat[i].addr       = 0;
+		}
+	}
+
+	if (likely(nreq)) {
+		srq->wqe_ctr += nreq;
+
+		/*
+		 * Make sure that descriptors are written before
+		 * doorbell record.
+		 */
+		wmb();
+
+		*srq->db.db = cpu_to_be32(srq->wqe_ctr);
+	}
+
+	spin_unlock_irqrestore(&srq->lock, flags);
+
+	return err;
+}
diff --git a/drivers/infiniband/hw/mlx4/user.h b/drivers/infiniband/hw/mlx4/user.h
new file mode 100644
index 000000000000..5b8eddc9fa83
--- /dev/null
+++ b/drivers/infiniband/hw/mlx4/user.h
@@ -0,0 +1,92 @@
+/*
+ * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef MLX4_IB_USER_H
+#define MLX4_IB_USER_H
+
+#include <linux/types.h>
+
+/*
+ * Increment this value if any changes that break userspace ABI
+ * compatibility are made.
+ */
+#define MLX4_IB_UVERBS_ABI_VERSION	1
+
+/*
+ * Make sure that all structs defined in this file remain laid out so
+ * that they pack the same way on 32-bit and 64-bit architectures (to
+ * avoid incompatibility between 32-bit userspace and 64-bit kernels).
+ * In particular do not use pointer types -- pass pointers in __u64
+ * instead.
+ */
+
+struct mlx4_ib_alloc_ucontext_resp {
+	__u32	qp_tab_size;
+	__u16	bf_reg_size;
+	__u16	bf_regs_per_page;
+};
+
+struct mlx4_ib_alloc_pd_resp {
+	__u32	pdn;
+	__u32	reserved;
+};
+
+struct mlx4_ib_create_cq {
+	__u64	buf_addr;
+	__u64	db_addr;
+};
+
+struct mlx4_ib_create_cq_resp {
+	__u32	cqn;
+	__u32	reserved;
+};
+
+struct mlx4_ib_resize_cq {
+	__u64	buf_addr;
+};
+
+struct mlx4_ib_create_srq {
+	__u64	buf_addr;
+	__u64	db_addr;
+};
+
+struct mlx4_ib_create_srq_resp {
+	__u32	srqn;
+	__u32	reserved;
+};
+
+struct mlx4_ib_create_qp {
+	__u64	buf_addr;
+	__u64	db_addr;
+};
+
+#endif /* MLX4_IB_USER_H */
diff --git a/drivers/infiniband/hw/mthca/mthca_provider.c b/drivers/infiniband/hw/mthca/mthca_provider.c
index 1c05486c3c68..6bcde1cb9688 100644
--- a/drivers/infiniband/hw/mthca/mthca_provider.c
+++ b/drivers/infiniband/hw/mthca/mthca_provider.c
@@ -37,6 +37,7 @@
  */
 
 #include <rdma/ib_smi.h>
+#include <rdma/ib_umem.h>
 #include <rdma/ib_user_verbs.h>
 #include <linux/mm.h>
 
@@ -908,6 +909,8 @@ static struct ib_mr *mthca_get_dma_mr(struct ib_pd *pd, int acc)
 		return ERR_PTR(err);
 	}
 
+	mr->umem = NULL;
+
 	return &mr->ibmr;
 }
 
@@ -1003,11 +1006,13 @@ static struct ib_mr *mthca_reg_phys_mr(struct ib_pd       *pd,
 	}
 
 	kfree(page_list);
+	mr->umem = NULL;
+
 	return &mr->ibmr;
 }
 
-static struct ib_mr *mthca_reg_user_mr(struct ib_pd *pd, struct ib_umem *region,
-				       int acc, struct ib_udata *udata)
+static struct ib_mr *mthca_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
+				       u64 virt, int acc, struct ib_udata *udata)
 {
 	struct mthca_dev *dev = to_mdev(pd->device);
 	struct ib_umem_chunk *chunk;
@@ -1018,20 +1023,26 @@ static struct ib_mr *mthca_reg_user_mr(struct ib_pd *pd, struct ib_umem *region,
 	int err = 0;
 	int write_mtt_size;
 
-	shift = ffs(region->page_size) - 1;
-
 	mr = kmalloc(sizeof *mr, GFP_KERNEL);
 	if (!mr)
 		return ERR_PTR(-ENOMEM);
 
+	mr->umem = ib_umem_get(pd->uobject->context, start, length, acc);
+	if (IS_ERR(mr->umem)) {
+		err = PTR_ERR(mr->umem);
+		goto err;
+	}
+
+	shift = ffs(mr->umem->page_size) - 1;
+
 	n = 0;
-	list_for_each_entry(chunk, &region->chunk_list, list)
+	list_for_each_entry(chunk, &mr->umem->chunk_list, list)
 		n += chunk->nents;
 
 	mr->mtt = mthca_alloc_mtt(dev, n);
 	if (IS_ERR(mr->mtt)) {
 		err = PTR_ERR(mr->mtt);
-		goto err;
+		goto err_umem;
 	}
 
 	pages = (u64 *) __get_free_page(GFP_KERNEL);
@@ -1044,12 +1055,12 @@ static struct ib_mr *mthca_reg_user_mr(struct ib_pd *pd, struct ib_umem *region,
 
 	write_mtt_size = min(mthca_write_mtt_size(dev), (int) (PAGE_SIZE / sizeof *pages));
 
-	list_for_each_entry(chunk, &region->chunk_list, list)
+	list_for_each_entry(chunk, &mr->umem->chunk_list, list)
 		for (j = 0; j < chunk->nmap; ++j) {
 			len = sg_dma_len(&chunk->page_list[j]) >> shift;
 			for (k = 0; k < len; ++k) {
 				pages[i++] = sg_dma_address(&chunk->page_list[j]) +
-					region->page_size * k;
+					mr->umem->page_size * k;
 				/*
 				 * Be friendly to write_mtt and pass it chunks
 				 * of appropriate size.
@@ -1071,8 +1082,8 @@ mtt_done:
 	if (err)
 		goto err_mtt;
 
-	err = mthca_mr_alloc(dev, to_mpd(pd)->pd_num, shift, region->virt_base,
-			     region->length, convert_access(acc), mr);
+	err = mthca_mr_alloc(dev, to_mpd(pd)->pd_num, shift, virt, length,
+			     convert_access(acc), mr);
 
 	if (err)
 		goto err_mtt;
@@ -1082,6 +1093,9 @@ mtt_done:
 err_mtt:
 	mthca_free_mtt(dev, mr->mtt);
 
+err_umem:
+	ib_umem_release(mr->umem);
+
 err:
 	kfree(mr);
 	return ERR_PTR(err);
@@ -1090,8 +1104,12 @@ err:
 static int mthca_dereg_mr(struct ib_mr *mr)
 {
 	struct mthca_mr *mmr = to_mmr(mr);
+
 	mthca_free_mr(to_mdev(mr->device), mmr);
+	if (mmr->umem)
+		ib_umem_release(mmr->umem);
 	kfree(mmr);
+
 	return 0;
 }
 
diff --git a/drivers/infiniband/hw/mthca/mthca_provider.h b/drivers/infiniband/hw/mthca/mthca_provider.h
index 1d266ac2e094..262616c8ebb6 100644
--- a/drivers/infiniband/hw/mthca/mthca_provider.h
+++ b/drivers/infiniband/hw/mthca/mthca_provider.h
@@ -73,6 +73,7 @@ struct mthca_mtt;
 
 struct mthca_mr {
 	struct ib_mr      ibmr;
+	struct ib_umem   *umem;
 	struct mthca_mtt *mtt;
 };
 
diff --git a/drivers/isdn/capi/Kconfig b/drivers/isdn/capi/Kconfig
index c921d6c522f5..c92f9d764fce 100644
--- a/drivers/isdn/capi/Kconfig
+++ b/drivers/isdn/capi/Kconfig
@@ -17,7 +17,7 @@ config CAPI_TRACE
 	help
 	  If you say Y here, the kernelcapi driver can make verbose traces
 	  of CAPI messages. This feature can be enabled/disabled via IOCTL for
-	  every controler (default disabled).
+	  every controller (default disabled).
 	  This will increase the size of the kernelcapi module by 20 KB.
 	  If unsure, say Y.
 
diff --git a/drivers/isdn/hardware/eicon/divasync.h b/drivers/isdn/hardware/eicon/divasync.h
index af3eb9e795b5..85784a7ffb25 100644
--- a/drivers/isdn/hardware/eicon/divasync.h
+++ b/drivers/isdn/hardware/eicon/divasync.h
@@ -216,7 +216,7 @@ typedef struct
 #define SERIAL_HOOK_RING 0x85
 #define SERIAL_HOOK_DETACH 0x8f
  unsigned char Flags;           /* function refinements   */
- /* parameters passed by the the ATTACH request      */
+ /* parameters passed by the ATTACH request      */
  SERIAL_INT_CB InterruptHandler; /* called on each interrupt  */
  SERIAL_DPC_CB DeferredHandler; /* called on hook state changes */
  void   *HandlerContext; /* context for both handlers */
diff --git a/drivers/isdn/hisax/hfc_usb.c b/drivers/isdn/hisax/hfc_usb.c
index 99e70d4103b6..1f18f1993387 100644
--- a/drivers/isdn/hisax/hfc_usb.c
+++ b/drivers/isdn/hisax/hfc_usb.c
@@ -1217,11 +1217,11 @@ usb_init(hfcusb_data * hfc)
 	/* aux = output, reset off */
 	write_usb(hfc, HFCUSB_CIRM, 0x10);
 
-	/* set USB_SIZE to match the the wMaxPacketSize for INT or BULK transfers */
+	/* set USB_SIZE to match the wMaxPacketSize for INT or BULK transfers */
 	write_usb(hfc, HFCUSB_USB_SIZE,
 		  (hfc->packet_size / 8) | ((hfc->packet_size / 8) << 4));
 
-	/* set USB_SIZE_I to match the the wMaxPacketSize for ISO transfers */
+	/* set USB_SIZE_I to match the wMaxPacketSize for ISO transfers */
 	write_usb(hfc, HFCUSB_USB_SIZE_I, hfc->iso_packet_size);
 
 	/* enable PCM/GCI master mode */
diff --git a/drivers/kvm/kvm_main.c b/drivers/kvm/kvm_main.c
index c8b8cfa332bb..0d892600ff00 100644
--- a/drivers/kvm/kvm_main.c
+++ b/drivers/kvm/kvm_main.c
@@ -2889,7 +2889,9 @@ static int kvm_cpu_hotplug(struct notifier_block *notifier, unsigned long val,
 
 	switch (val) {
 	case CPU_DOWN_PREPARE:
+	case CPU_DOWN_PREPARE_FROZEN:
 	case CPU_UP_CANCELED:
+	case CPU_UP_CANCELED_FROZEN:
 		printk(KERN_INFO "kvm: disabling virtualization on CPU%d\n",
 		       cpu);
 		decache_vcpus_on_cpu(cpu);
@@ -2897,6 +2899,7 @@ static int kvm_cpu_hotplug(struct notifier_block *notifier, unsigned long val,
 					 NULL, 0, 1);
 		break;
 	case CPU_ONLINE:
+	case CPU_ONLINE_FROZEN:
 		printk(KERN_INFO "kvm: enabling virtualization on CPU%d\n",
 		       cpu);
 		smp_call_function_single(cpu, kvm_arch_ops->hardware_enable,
diff --git a/drivers/leds/leds-h1940.c b/drivers/leds/leds-h1940.c
index 1d49d2ade557..677c99325be5 100644
--- a/drivers/leds/leds-h1940.c
+++ b/drivers/leds/leds-h1940.c
@@ -1,5 +1,5 @@
 /*
- * drivers/leds/h1940-leds.c
+ * drivers/leds/leds-h1940.c
  * Copyright (c) Arnaud Patard <arnaud.patard@rtp-net.org>
  *
  * This file is subject to the terms and conditions of the GNU General Public
diff --git a/drivers/macintosh/Kconfig b/drivers/macintosh/Kconfig
index a32c91e27b3c..58926da0ae18 100644
--- a/drivers/macintosh/Kconfig
+++ b/drivers/macintosh/Kconfig
@@ -237,7 +237,7 @@ config PMAC_RACKMETER
 	tristate "Support for Apple XServe front panel LEDs"
 	depends on PPC_PMAC
 	help
-	  This driver procides some support to control the front panel
+	  This driver provides some support to control the front panel
           blue LEDs "vu-meter" of the XServer macs.
 
 endif # MACINTOSH_DRIVERS
diff --git a/drivers/macintosh/mediabay.c b/drivers/macintosh/mediabay.c
index 0acf2f7fd9d7..c803d2bba65d 100644
--- a/drivers/macintosh/mediabay.c
+++ b/drivers/macintosh/mediabay.c
@@ -563,7 +563,7 @@ static void media_bay_step(int i)
 				ide_init_hwif_ports(&hw, (unsigned long) bay->cd_base, (unsigned long) 0, NULL);
 				hw.irq = bay->cd_irq;
 				hw.chipset = ide_pmac;
-				bay->cd_index = ide_register_hw(&hw, NULL);
+				bay->cd_index = ide_register_hw(&hw, 0, NULL);
 				pmu_resume();
 			}
 			if (bay->cd_index == -1) {
diff --git a/drivers/mca/mca-bus.c b/drivers/mca/mca-bus.c
index da862e4632dd..67b8e9453b19 100644
--- a/drivers/mca/mca-bus.c
+++ b/drivers/mca/mca-bus.c
@@ -47,19 +47,25 @@ static int mca_bus_match (struct device *dev, struct device_driver *drv)
 {
 	struct mca_device *mca_dev = to_mca_device (dev);
 	struct mca_driver *mca_drv = to_mca_driver (drv);
-	const short *mca_ids = mca_drv->id_table;
-	int i;
-
-	if (!mca_ids)
-		return 0;
-
-	for(i = 0; mca_ids[i]; i++) {
-		if (mca_ids[i] == mca_dev->pos_id) {
-			mca_dev->index = i;
-			return 1;
+	const unsigned short *mca_ids = mca_drv->id_table;
+	int i = 0;
+
+	if (mca_ids) {
+		for(i = 0; mca_ids[i]; i++) {
+			if (mca_ids[i] == mca_dev->pos_id) {
+				mca_dev->index = i;
+				return 1;
+			}
 		}
 	}
-
+	/* If the integrated id is present, treat it as though it were an
+	 * additional id in the id_table (it can't be because by definition,
+	 * integrated id's overflow a short */
+	if (mca_drv->integrated_id && mca_dev->pos_id ==
+	    mca_drv->integrated_id) {
+		mca_dev->index = i;
+		return 1;
+	}
 	return 0;
 }
 
diff --git a/drivers/mca/mca-driver.c b/drivers/mca/mca-driver.c
index 2223466b3d8a..32cd39bcc715 100644
--- a/drivers/mca/mca-driver.c
+++ b/drivers/mca/mca-driver.c
@@ -36,12 +36,25 @@ int mca_register_driver(struct mca_driver *mca_drv)
 		mca_drv->driver.bus = &mca_bus_type;
 		if ((r = driver_register(&mca_drv->driver)) < 0)
 			return r;
+		mca_drv->integrated_id = 0;
 	}
 
 	return 0;
 }
 EXPORT_SYMBOL(mca_register_driver);
 
+int mca_register_driver_integrated(struct mca_driver *mca_driver,
+				   int integrated_id)
+{
+	int r = mca_register_driver(mca_driver);
+
+	if (!r)
+		mca_driver->integrated_id = integrated_id;
+
+	return r;
+}
+EXPORT_SYMBOL(mca_register_driver_integrated);
+
 void mca_unregister_driver(struct mca_driver *mca_drv)
 {
 	if (MCA_bus)
diff --git a/drivers/md/Kconfig b/drivers/md/Kconfig
index 4540ade6b6b5..7df934d69134 100644
--- a/drivers/md/Kconfig
+++ b/drivers/md/Kconfig
@@ -262,6 +262,15 @@ config DM_MULTIPATH_EMC
 	---help---
 	  Multipath support for EMC CX/AX series hardware.
 
+config DM_DELAY
+	tristate "I/O delaying target (EXPERIMENTAL)"
+	depends on BLK_DEV_DM && EXPERIMENTAL
+	---help---
+	A target that delays reads and/or writes and can send
+	them to different devices.  Useful for testing.
+
+	If unsure, say N.
+
 endmenu
 
 endif
diff --git a/drivers/md/Makefile b/drivers/md/Makefile
index 34957a68d921..38754084eac7 100644
--- a/drivers/md/Makefile
+++ b/drivers/md/Makefile
@@ -31,6 +31,7 @@ obj-$(CONFIG_MD_FAULTY)		+= faulty.o
 obj-$(CONFIG_BLK_DEV_MD)	+= md-mod.o
 obj-$(CONFIG_BLK_DEV_DM)	+= dm-mod.o
 obj-$(CONFIG_DM_CRYPT)		+= dm-crypt.o
+obj-$(CONFIG_DM_DELAY)		+= dm-delay.o
 obj-$(CONFIG_DM_MULTIPATH)	+= dm-multipath.o dm-round-robin.o
 obj-$(CONFIG_DM_MULTIPATH_EMC)	+= dm-emc.o
 obj-$(CONFIG_DM_SNAPSHOT)	+= dm-snapshot.o
diff --git a/drivers/md/dm-bio-list.h b/drivers/md/dm-bio-list.h
index da4349649f7f..c6be88826fae 100644
--- a/drivers/md/dm-bio-list.h
+++ b/drivers/md/dm-bio-list.h
@@ -8,17 +8,43 @@
 #define DM_BIO_LIST_H
 
 #include <linux/bio.h>
+#include <linux/prefetch.h>
 
 struct bio_list {
 	struct bio *head;
 	struct bio *tail;
 };
 
+static inline int bio_list_empty(const struct bio_list *bl)
+{
+	return bl->head == NULL;
+}
+
+#define BIO_LIST_INIT { .head = NULL, .tail = NULL }
+
+#define BIO_LIST(bl) \
+	struct bio_list bl = BIO_LIST_INIT
+
 static inline void bio_list_init(struct bio_list *bl)
 {
 	bl->head = bl->tail = NULL;
 }
 
+#define bio_list_for_each(bio, bl) \
+	for (bio = (bl)->head; bio && ({ prefetch(bio->bi_next); 1; }); \
+	     bio = bio->bi_next)
+
+static inline unsigned bio_list_size(const struct bio_list *bl)
+{
+	unsigned sz = 0;
+	struct bio *bio;
+
+	bio_list_for_each(bio, bl)
+		sz++;
+
+	return sz;
+}
+
 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
 {
 	bio->bi_next = NULL;
diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c
index d8121234c347..7b0fcfc9eaa5 100644
--- a/drivers/md/dm-crypt.c
+++ b/drivers/md/dm-crypt.c
@@ -33,7 +33,6 @@
 struct crypt_io {
 	struct dm_target *target;
 	struct bio *base_bio;
-	struct bio *first_clone;
 	struct work_struct work;
 	atomic_t pending;
 	int error;
@@ -107,6 +106,8 @@ struct crypt_config {
 
 static struct kmem_cache *_crypt_io_pool;
 
+static void clone_init(struct crypt_io *, struct bio *);
+
 /*
  * Different IV generation algorithms:
  *
@@ -120,6 +121,9 @@ static struct kmem_cache *_crypt_io_pool;
  * benbi: the 64-bit "big-endian 'narrow block'-count", starting at 1
  *        (needed for LRW-32-AES and possible other narrow block modes)
  *
+ * null: the initial vector is always zero.  Provides compatibility with
+ *       obsolete loop_fish2 devices.  Do not use for new devices.
+ *
  * plumb: unimplemented, see:
  * http://article.gmane.org/gmane.linux.kernel.device-mapper.dm-crypt/454
  */
@@ -256,6 +260,13 @@ static int crypt_iv_benbi_gen(struct crypt_config *cc, u8 *iv, sector_t sector)
 	return 0;
 }
 
+static int crypt_iv_null_gen(struct crypt_config *cc, u8 *iv, sector_t sector)
+{
+	memset(iv, 0, cc->iv_size);
+
+	return 0;
+}
+
 static struct crypt_iv_operations crypt_iv_plain_ops = {
 	.generator = crypt_iv_plain_gen
 };
@@ -272,6 +283,10 @@ static struct crypt_iv_operations crypt_iv_benbi_ops = {
 	.generator = crypt_iv_benbi_gen
 };
 
+static struct crypt_iv_operations crypt_iv_null_ops = {
+	.generator = crypt_iv_null_gen
+};
+
 static int
 crypt_convert_scatterlist(struct crypt_config *cc, struct scatterlist *out,
                           struct scatterlist *in, unsigned int length,
@@ -378,36 +393,21 @@ static int crypt_convert(struct crypt_config *cc,
  * This should never violate the device limitations
  * May return a smaller bio when running out of pages
  */
-static struct bio *
-crypt_alloc_buffer(struct crypt_config *cc, unsigned int size,
-                   struct bio *base_bio, unsigned int *bio_vec_idx)
+static struct bio *crypt_alloc_buffer(struct crypt_io *io, unsigned int size)
 {
+	struct crypt_config *cc = io->target->private;
 	struct bio *clone;
 	unsigned int nr_iovecs = (size + PAGE_SIZE - 1) >> PAGE_SHIFT;
 	gfp_t gfp_mask = GFP_NOIO | __GFP_HIGHMEM;
 	unsigned int i;
 
-	if (base_bio) {
-		clone = bio_alloc_bioset(GFP_NOIO, base_bio->bi_max_vecs, cc->bs);
-		__bio_clone(clone, base_bio);
-	} else
-		clone = bio_alloc_bioset(GFP_NOIO, nr_iovecs, cc->bs);
-
+	clone = bio_alloc_bioset(GFP_NOIO, nr_iovecs, cc->bs);
 	if (!clone)
 		return NULL;
 
-	clone->bi_destructor = dm_crypt_bio_destructor;
-
-	/* if the last bio was not complete, continue where that one ended */
-	clone->bi_idx = *bio_vec_idx;
-	clone->bi_vcnt = *bio_vec_idx;
-	clone->bi_size = 0;
-	clone->bi_flags &= ~(1 << BIO_SEG_VALID);
-
-	/* clone->bi_idx pages have already been allocated */
-	size -= clone->bi_idx * PAGE_SIZE;
+	clone_init(io, clone);
 
-	for (i = clone->bi_idx; i < nr_iovecs; i++) {
+	for (i = 0; i < nr_iovecs; i++) {
 		struct bio_vec *bv = bio_iovec_idx(clone, i);
 
 		bv->bv_page = mempool_alloc(cc->page_pool, gfp_mask);
@@ -419,7 +419,7 @@ crypt_alloc_buffer(struct crypt_config *cc, unsigned int size,
 		 * return a partially allocated bio, the caller will then try
 		 * to allocate additional bios while submitting this partial bio
 		 */
-		if ((i - clone->bi_idx) == (MIN_BIO_PAGES - 1))
+		if (i == (MIN_BIO_PAGES - 1))
 			gfp_mask = (gfp_mask | __GFP_NOWARN) & ~__GFP_WAIT;
 
 		bv->bv_offset = 0;
@@ -438,12 +438,6 @@ crypt_alloc_buffer(struct crypt_config *cc, unsigned int size,
 		return NULL;
 	}
 
-	/*
-	 * Remember the last bio_vec allocated to be able
-	 * to correctly continue after the splitting.
-	 */
-	*bio_vec_idx = clone->bi_vcnt;
-
 	return clone;
 }
 
@@ -495,9 +489,6 @@ static void dec_pending(struct crypt_io *io, int error)
 	if (!atomic_dec_and_test(&io->pending))
 		return;
 
-	if (io->first_clone)
-		bio_put(io->first_clone);
-
 	bio_endio(io->base_bio, io->base_bio->bi_size, io->error);
 
 	mempool_free(io, cc->io_pool);
@@ -562,6 +553,7 @@ static void clone_init(struct crypt_io *io, struct bio *clone)
 	clone->bi_end_io  = crypt_endio;
 	clone->bi_bdev    = cc->dev->bdev;
 	clone->bi_rw      = io->base_bio->bi_rw;
+	clone->bi_destructor = dm_crypt_bio_destructor;
 }
 
 static void process_read(struct crypt_io *io)
@@ -585,7 +577,6 @@ static void process_read(struct crypt_io *io)
 	}
 
 	clone_init(io, clone);
-	clone->bi_destructor = dm_crypt_bio_destructor;
 	clone->bi_idx = 0;
 	clone->bi_vcnt = bio_segments(base_bio);
 	clone->bi_size = base_bio->bi_size;
@@ -604,7 +595,6 @@ static void process_write(struct crypt_io *io)
 	struct convert_context ctx;
 	unsigned remaining = base_bio->bi_size;
 	sector_t sector = base_bio->bi_sector - io->target->begin;
-	unsigned bvec_idx = 0;
 
 	atomic_inc(&io->pending);
 
@@ -615,14 +605,14 @@ static void process_write(struct crypt_io *io)
 	 * so repeat the whole process until all the data can be handled.
 	 */
 	while (remaining) {
-		clone = crypt_alloc_buffer(cc, base_bio->bi_size,
-					   io->first_clone, &bvec_idx);
+		clone = crypt_alloc_buffer(io, remaining);
 		if (unlikely(!clone)) {
 			dec_pending(io, -ENOMEM);
 			return;
 		}
 
 		ctx.bio_out = clone;
+		ctx.idx_out = 0;
 
 		if (unlikely(crypt_convert(cc, &ctx) < 0)) {
 			crypt_free_buffer_pages(cc, clone, clone->bi_size);
@@ -631,31 +621,26 @@ static void process_write(struct crypt_io *io)
 			return;
 		}
 
-		clone_init(io, clone);
-		clone->bi_sector = cc->start + sector;
-
-		if (!io->first_clone) {
-			/*
-			 * hold a reference to the first clone, because it
-			 * holds the bio_vec array and that can't be freed
-			 * before all other clones are released
-			 */
-			bio_get(clone);
-			io->first_clone = clone;
-		}
+		/* crypt_convert should have filled the clone bio */
+		BUG_ON(ctx.idx_out < clone->bi_vcnt);
 
+		clone->bi_sector = cc->start + sector;
 		remaining -= clone->bi_size;
 		sector += bio_sectors(clone);
 
-		/* prevent bio_put of first_clone */
+		/* Grab another reference to the io struct
+		 * before we kick off the request */
 		if (remaining)
 			atomic_inc(&io->pending);
 
 		generic_make_request(clone);
 
+		/* Do not reference clone after this - it
+		 * may be gone already. */
+
 		/* out of memory -> run queues */
 		if (remaining)
-			congestion_wait(bio_data_dir(clone), HZ/100);
+			congestion_wait(WRITE, HZ/100);
 	}
 }
 
@@ -832,6 +817,8 @@ static int crypt_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 		cc->iv_gen_ops = &crypt_iv_essiv_ops;
 	else if (strcmp(ivmode, "benbi") == 0)
 		cc->iv_gen_ops = &crypt_iv_benbi_ops;
+	else if (strcmp(ivmode, "null") == 0)
+		cc->iv_gen_ops = &crypt_iv_null_ops;
 	else {
 		ti->error = "Invalid IV mode";
 		goto bad2;
@@ -954,10 +941,12 @@ static int crypt_map(struct dm_target *ti, struct bio *bio,
 	struct crypt_config *cc = ti->private;
 	struct crypt_io *io;
 
+	if (bio_barrier(bio))
+		return -EOPNOTSUPP;
+
 	io = mempool_alloc(cc->io_pool, GFP_NOIO);
 	io->target = ti;
 	io->base_bio = bio;
-	io->first_clone = NULL;
 	io->error = io->post_process = 0;
 	atomic_set(&io->pending, 0);
 	kcryptd_queue_io(io);
@@ -1057,7 +1046,7 @@ error:
 
 static struct target_type crypt_target = {
 	.name   = "crypt",
-	.version= {1, 3, 0},
+	.version= {1, 5, 0},
 	.module = THIS_MODULE,
 	.ctr    = crypt_ctr,
 	.dtr    = crypt_dtr,
diff --git a/drivers/md/dm-delay.c b/drivers/md/dm-delay.c
new file mode 100644
index 000000000000..52c7cf9e5803
--- /dev/null
+++ b/drivers/md/dm-delay.c
@@ -0,0 +1,383 @@
+/*
+ * Copyright (C) 2005-2007 Red Hat GmbH
+ *
+ * A target that delays reads and/or writes and can send
+ * them to different devices.
+ *
+ * This file is released under the GPL.
+ */
+
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/blkdev.h>
+#include <linux/bio.h>
+#include <linux/slab.h>
+
+#include "dm.h"
+#include "dm-bio-list.h"
+
+#define DM_MSG_PREFIX "delay"
+
+struct delay_c {
+	struct timer_list delay_timer;
+	struct semaphore timer_lock;
+	struct work_struct flush_expired_bios;
+	struct list_head delayed_bios;
+	atomic_t may_delay;
+	mempool_t *delayed_pool;
+
+	struct dm_dev *dev_read;
+	sector_t start_read;
+	unsigned read_delay;
+	unsigned reads;
+
+	struct dm_dev *dev_write;
+	sector_t start_write;
+	unsigned write_delay;
+	unsigned writes;
+};
+
+struct delay_info {
+	struct delay_c *context;
+	struct list_head list;
+	struct bio *bio;
+	unsigned long expires;
+};
+
+static DEFINE_MUTEX(delayed_bios_lock);
+
+static struct workqueue_struct *kdelayd_wq;
+static struct kmem_cache *delayed_cache;
+
+static void handle_delayed_timer(unsigned long data)
+{
+	struct delay_c *dc = (struct delay_c *)data;
+
+	queue_work(kdelayd_wq, &dc->flush_expired_bios);
+}
+
+static void queue_timeout(struct delay_c *dc, unsigned long expires)
+{
+	down(&dc->timer_lock);
+
+	if (!timer_pending(&dc->delay_timer) || expires < dc->delay_timer.expires)
+		mod_timer(&dc->delay_timer, expires);
+
+	up(&dc->timer_lock);
+}
+
+static void flush_bios(struct bio *bio)
+{
+	struct bio *n;
+
+	while (bio) {
+		n = bio->bi_next;
+		bio->bi_next = NULL;
+		generic_make_request(bio);
+		bio = n;
+	}
+}
+
+static struct bio *flush_delayed_bios(struct delay_c *dc, int flush_all)
+{
+	struct delay_info *delayed, *next;
+	unsigned long next_expires = 0;
+	int start_timer = 0;
+	BIO_LIST(flush_bios);
+
+	mutex_lock(&delayed_bios_lock);
+	list_for_each_entry_safe(delayed, next, &dc->delayed_bios, list) {
+		if (flush_all || time_after_eq(jiffies, delayed->expires)) {
+			list_del(&delayed->list);
+			bio_list_add(&flush_bios, delayed->bio);
+			if ((bio_data_dir(delayed->bio) == WRITE))
+				delayed->context->writes--;
+			else
+				delayed->context->reads--;
+			mempool_free(delayed, dc->delayed_pool);
+			continue;
+		}
+
+		if (!start_timer) {
+			start_timer = 1;
+			next_expires = delayed->expires;
+		} else
+			next_expires = min(next_expires, delayed->expires);
+	}
+
+	mutex_unlock(&delayed_bios_lock);
+
+	if (start_timer)
+		queue_timeout(dc, next_expires);
+
+	return bio_list_get(&flush_bios);
+}
+
+static void flush_expired_bios(struct work_struct *work)
+{
+	struct delay_c *dc;
+
+	dc = container_of(work, struct delay_c, flush_expired_bios);
+	flush_bios(flush_delayed_bios(dc, 0));
+}
+
+/*
+ * Mapping parameters:
+ *    <device> <offset> <delay> [<write_device> <write_offset> <write_delay>]
+ *
+ * With separate write parameters, the first set is only used for reads.
+ * Delays are specified in milliseconds.
+ */
+static int delay_ctr(struct dm_target *ti, unsigned int argc, char **argv)
+{
+	struct delay_c *dc;
+	unsigned long long tmpll;
+
+	if (argc != 3 && argc != 6) {
+		ti->error = "requires exactly 3 or 6 arguments";
+		return -EINVAL;
+	}
+
+	dc = kmalloc(sizeof(*dc), GFP_KERNEL);
+	if (!dc) {
+		ti->error = "Cannot allocate context";
+		return -ENOMEM;
+	}
+
+	dc->reads = dc->writes = 0;
+
+	if (sscanf(argv[1], "%llu", &tmpll) != 1) {
+		ti->error = "Invalid device sector";
+		goto bad;
+	}
+	dc->start_read = tmpll;
+
+	if (sscanf(argv[2], "%u", &dc->read_delay) != 1) {
+		ti->error = "Invalid delay";
+		goto bad;
+	}
+
+	if (dm_get_device(ti, argv[0], dc->start_read, ti->len,
+			  dm_table_get_mode(ti->table), &dc->dev_read)) {
+		ti->error = "Device lookup failed";
+		goto bad;
+	}
+
+	if (argc == 3) {
+		dc->dev_write = NULL;
+		goto out;
+	}
+
+	if (sscanf(argv[4], "%llu", &tmpll) != 1) {
+		ti->error = "Invalid write device sector";
+		goto bad;
+	}
+	dc->start_write = tmpll;
+
+	if (sscanf(argv[5], "%u", &dc->write_delay) != 1) {
+		ti->error = "Invalid write delay";
+		goto bad;
+	}
+
+	if (dm_get_device(ti, argv[3], dc->start_write, ti->len,
+			  dm_table_get_mode(ti->table), &dc->dev_write)) {
+		ti->error = "Write device lookup failed";
+		dm_put_device(ti, dc->dev_read);
+		goto bad;
+	}
+
+out:
+	dc->delayed_pool = mempool_create_slab_pool(128, delayed_cache);
+	if (!dc->delayed_pool) {
+		DMERR("Couldn't create delayed bio pool.");
+		goto bad;
+	}
+
+	init_timer(&dc->delay_timer);
+	dc->delay_timer.function = handle_delayed_timer;
+	dc->delay_timer.data = (unsigned long)dc;
+
+	INIT_WORK(&dc->flush_expired_bios, flush_expired_bios);
+	INIT_LIST_HEAD(&dc->delayed_bios);
+	init_MUTEX(&dc->timer_lock);
+	atomic_set(&dc->may_delay, 1);
+
+	ti->private = dc;
+	return 0;
+
+bad:
+	kfree(dc);
+	return -EINVAL;
+}
+
+static void delay_dtr(struct dm_target *ti)
+{
+	struct delay_c *dc = ti->private;
+
+	flush_workqueue(kdelayd_wq);
+
+	dm_put_device(ti, dc->dev_read);
+
+	if (dc->dev_write)
+		dm_put_device(ti, dc->dev_write);
+
+	mempool_destroy(dc->delayed_pool);
+	kfree(dc);
+}
+
+static int delay_bio(struct delay_c *dc, int delay, struct bio *bio)
+{
+	struct delay_info *delayed;
+	unsigned long expires = 0;
+
+	if (!delay || !atomic_read(&dc->may_delay))
+		return 1;
+
+	delayed = mempool_alloc(dc->delayed_pool, GFP_NOIO);
+
+	delayed->context = dc;
+	delayed->bio = bio;
+	delayed->expires = expires = jiffies + (delay * HZ / 1000);
+
+	mutex_lock(&delayed_bios_lock);
+
+	if (bio_data_dir(bio) == WRITE)
+		dc->writes++;
+	else
+		dc->reads++;
+
+	list_add_tail(&delayed->list, &dc->delayed_bios);
+
+	mutex_unlock(&delayed_bios_lock);
+
+	queue_timeout(dc, expires);
+
+	return 0;
+}
+
+static void delay_presuspend(struct dm_target *ti)
+{
+	struct delay_c *dc = ti->private;
+
+	atomic_set(&dc->may_delay, 0);
+	del_timer_sync(&dc->delay_timer);
+	flush_bios(flush_delayed_bios(dc, 1));
+}
+
+static void delay_resume(struct dm_target *ti)
+{
+	struct delay_c *dc = ti->private;
+
+	atomic_set(&dc->may_delay, 1);
+}
+
+static int delay_map(struct dm_target *ti, struct bio *bio,
+		     union map_info *map_context)
+{
+	struct delay_c *dc = ti->private;
+
+	if ((bio_data_dir(bio) == WRITE) && (dc->dev_write)) {
+		bio->bi_bdev = dc->dev_write->bdev;
+		bio->bi_sector = dc->start_write +
+				 (bio->bi_sector - ti->begin);
+
+		return delay_bio(dc, dc->write_delay, bio);
+	}
+
+	bio->bi_bdev = dc->dev_read->bdev;
+	bio->bi_sector = dc->start_read +
+			 (bio->bi_sector - ti->begin);
+
+	return delay_bio(dc, dc->read_delay, bio);
+}
+
+static int delay_status(struct dm_target *ti, status_type_t type,
+			char *result, unsigned maxlen)
+{
+	struct delay_c *dc = ti->private;
+	int sz = 0;
+
+	switch (type) {
+	case STATUSTYPE_INFO:
+		DMEMIT("%u %u", dc->reads, dc->writes);
+		break;
+
+	case STATUSTYPE_TABLE:
+		DMEMIT("%s %llu %u", dc->dev_read->name,
+		       (unsigned long long) dc->start_read,
+		       dc->read_delay);
+		if (dc->dev_write)
+			DMEMIT("%s %llu %u", dc->dev_write->name,
+			       (unsigned long long) dc->start_write,
+			       dc->write_delay);
+		break;
+	}
+
+	return 0;
+}
+
+static struct target_type delay_target = {
+	.name	     = "delay",
+	.version     = {1, 0, 2},
+	.module      = THIS_MODULE,
+	.ctr	     = delay_ctr,
+	.dtr	     = delay_dtr,
+	.map	     = delay_map,
+	.presuspend  = delay_presuspend,
+	.resume	     = delay_resume,
+	.status	     = delay_status,
+};
+
+static int __init dm_delay_init(void)
+{
+	int r = -ENOMEM;
+
+	kdelayd_wq = create_workqueue("kdelayd");
+	if (!kdelayd_wq) {
+		DMERR("Couldn't start kdelayd");
+		goto bad_queue;
+	}
+
+	delayed_cache = kmem_cache_create("dm-delay",
+					  sizeof(struct delay_info),
+					  __alignof__(struct delay_info),
+					  0, NULL, NULL);
+	if (!delayed_cache) {
+		DMERR("Couldn't create delayed bio cache.");
+		goto bad_memcache;
+	}
+
+	r = dm_register_target(&delay_target);
+	if (r < 0) {
+		DMERR("register failed %d", r);
+		goto bad_register;
+	}
+
+	return 0;
+
+bad_register:
+	kmem_cache_destroy(delayed_cache);
+bad_memcache:
+	destroy_workqueue(kdelayd_wq);
+bad_queue:
+	return r;
+}
+
+static void __exit dm_delay_exit(void)
+{
+	int r = dm_unregister_target(&delay_target);
+
+	if (r < 0)
+		DMERR("unregister failed %d", r);
+
+	kmem_cache_destroy(delayed_cache);
+	destroy_workqueue(kdelayd_wq);
+}
+
+/* Module hooks */
+module_init(dm_delay_init);
+module_exit(dm_delay_exit);
+
+MODULE_DESCRIPTION(DM_NAME " delay target");
+MODULE_AUTHOR("Heinz Mauelshagen <mauelshagen@redhat.com>");
+MODULE_LICENSE("GPL");
diff --git a/drivers/md/dm-exception-store.c b/drivers/md/dm-exception-store.c
index 99cdffa7fbfe..07e0a0c84f6e 100644
--- a/drivers/md/dm-exception-store.c
+++ b/drivers/md/dm-exception-store.c
@@ -1,7 +1,8 @@
 /*
- * dm-snapshot.c
+ * dm-exception-store.c
  *
  * Copyright (C) 2001-2002 Sistina Software (UK) Limited.
+ * Copyright (C) 2006 Red Hat GmbH
  *
  * This file is released under the GPL.
  */
@@ -123,6 +124,7 @@ struct pstore {
 	atomic_t pending_count;
 	uint32_t callback_count;
 	struct commit_callback *callbacks;
+	struct dm_io_client *io_client;
 };
 
 static inline unsigned int sectors_to_pages(unsigned int sectors)
@@ -159,14 +161,20 @@ static void free_area(struct pstore *ps)
  */
 static int chunk_io(struct pstore *ps, uint32_t chunk, int rw)
 {
-	struct io_region where;
-	unsigned long bits;
-
-	where.bdev = ps->snap->cow->bdev;
-	where.sector = ps->snap->chunk_size * chunk;
-	where.count = ps->snap->chunk_size;
-
-	return dm_io_sync_vm(1, &where, rw, ps->area, &bits);
+	struct io_region where = {
+		.bdev = ps->snap->cow->bdev,
+		.sector = ps->snap->chunk_size * chunk,
+		.count = ps->snap->chunk_size,
+	};
+	struct dm_io_request io_req = {
+		.bi_rw = rw,
+		.mem.type = DM_IO_VMA,
+		.mem.ptr.vma = ps->area,
+		.client = ps->io_client,
+		.notify.fn = NULL,
+	};
+
+	return dm_io(&io_req, 1, &where, NULL);
 }
 
 /*
@@ -213,17 +221,18 @@ static int read_header(struct pstore *ps, int *new_snapshot)
 		chunk_size_supplied = 0;
 	}
 
-	r = dm_io_get(sectors_to_pages(ps->snap->chunk_size));
-	if (r)
-		return r;
+	ps->io_client = dm_io_client_create(sectors_to_pages(ps->snap->
+							     chunk_size));
+	if (IS_ERR(ps->io_client))
+		return PTR_ERR(ps->io_client);
 
 	r = alloc_area(ps);
 	if (r)
-		goto bad1;
+		return r;
 
 	r = chunk_io(ps, 0, READ);
 	if (r)
-		goto bad2;
+		goto bad;
 
 	dh = (struct disk_header *) ps->area;
 
@@ -235,7 +244,7 @@ static int read_header(struct pstore *ps, int *new_snapshot)
 	if (le32_to_cpu(dh->magic) != SNAP_MAGIC) {
 		DMWARN("Invalid or corrupt snapshot");
 		r = -ENXIO;
-		goto bad2;
+		goto bad;
 	}
 
 	*new_snapshot = 0;
@@ -252,27 +261,22 @@ static int read_header(struct pstore *ps, int *new_snapshot)
 	       (unsigned long long)ps->snap->chunk_size);
 
 	/* We had a bogus chunk_size. Fix stuff up. */
-	dm_io_put(sectors_to_pages(ps->snap->chunk_size));
 	free_area(ps);
 
 	ps->snap->chunk_size = chunk_size;
 	ps->snap->chunk_mask = chunk_size - 1;
 	ps->snap->chunk_shift = ffs(chunk_size) - 1;
 
-	r = dm_io_get(sectors_to_pages(chunk_size));
+	r = dm_io_client_resize(sectors_to_pages(ps->snap->chunk_size),
+				ps->io_client);
 	if (r)
 		return r;
 
 	r = alloc_area(ps);
-	if (r)
-		goto bad1;
-
-	return 0;
+	return r;
 
-bad2:
+bad:
 	free_area(ps);
-bad1:
-	dm_io_put(sectors_to_pages(ps->snap->chunk_size));
 	return r;
 }
 
@@ -405,7 +409,7 @@ static void persistent_destroy(struct exception_store *store)
 {
 	struct pstore *ps = get_info(store);
 
-	dm_io_put(sectors_to_pages(ps->snap->chunk_size));
+	dm_io_client_destroy(ps->io_client);
 	vfree(ps->callbacks);
 	free_area(ps);
 	kfree(ps);
diff --git a/drivers/md/dm-hw-handler.h b/drivers/md/dm-hw-handler.h
index 32eff28e4adc..e0832e6fcf36 100644
--- a/drivers/md/dm-hw-handler.h
+++ b/drivers/md/dm-hw-handler.h
@@ -16,6 +16,7 @@
 struct hw_handler_type;
 struct hw_handler {
 	struct hw_handler_type *type;
+	struct mapped_device *md;
 	void *context;
 };
 
diff --git a/drivers/md/dm-io.c b/drivers/md/dm-io.c
index 8bdc8a87b249..352c6fbeac53 100644
--- a/drivers/md/dm-io.c
+++ b/drivers/md/dm-io.c
@@ -1,5 +1,6 @@
 /*
  * Copyright (C) 2003 Sistina Software
+ * Copyright (C) 2006 Red Hat GmbH
  *
  * This file is released under the GPL.
  */
@@ -12,13 +13,17 @@
 #include <linux/sched.h>
 #include <linux/slab.h>
 
-static struct bio_set *_bios;
+struct dm_io_client {
+	mempool_t *pool;
+	struct bio_set *bios;
+};
 
 /* FIXME: can we shrink this ? */
 struct io {
 	unsigned long error;
 	atomic_t count;
 	struct task_struct *sleeper;
+	struct dm_io_client *client;
 	io_notify_fn callback;
 	void *context;
 };
@@ -26,63 +31,58 @@ struct io {
 /*
  * io contexts are only dynamically allocated for asynchronous
  * io.  Since async io is likely to be the majority of io we'll
- * have the same number of io contexts as buffer heads ! (FIXME:
- * must reduce this).
+ * have the same number of io contexts as bios! (FIXME: must reduce this).
  */
-static unsigned _num_ios;
-static mempool_t *_io_pool;
 
 static unsigned int pages_to_ios(unsigned int pages)
 {
 	return 4 * pages;	/* too many ? */
 }
 
-static int resize_pool(unsigned int new_ios)
+/*
+ * Create a client with mempool and bioset.
+ */
+struct dm_io_client *dm_io_client_create(unsigned num_pages)
 {
-	int r = 0;
-
-	if (_io_pool) {
-		if (new_ios == 0) {
-			/* free off the pool */
-			mempool_destroy(_io_pool);
-			_io_pool = NULL;
-			bioset_free(_bios);
-
-		} else {
-			/* resize the pool */
-			r = mempool_resize(_io_pool, new_ios, GFP_KERNEL);
-		}
+	unsigned ios = pages_to_ios(num_pages);
+	struct dm_io_client *client;
 
-	} else {
-		/* create new pool */
-		_io_pool = mempool_create_kmalloc_pool(new_ios,
-						       sizeof(struct io));
-		if (!_io_pool)
-			return -ENOMEM;
-
-		_bios = bioset_create(16, 16);
-		if (!_bios) {
-			mempool_destroy(_io_pool);
-			_io_pool = NULL;
-			return -ENOMEM;
-		}
-	}
+	client = kmalloc(sizeof(*client), GFP_KERNEL);
+	if (!client)
+		return ERR_PTR(-ENOMEM);
+
+	client->pool = mempool_create_kmalloc_pool(ios, sizeof(struct io));
+	if (!client->pool)
+		goto bad;
 
-	if (!r)
-		_num_ios = new_ios;
+	client->bios = bioset_create(16, 16);
+	if (!client->bios)
+		goto bad;
 
-	return r;
+	return client;
+
+   bad:
+	if (client->pool)
+		mempool_destroy(client->pool);
+	kfree(client);
+	return ERR_PTR(-ENOMEM);
 }
+EXPORT_SYMBOL(dm_io_client_create);
 
-int dm_io_get(unsigned int num_pages)
+int dm_io_client_resize(unsigned num_pages, struct dm_io_client *client)
 {
-	return resize_pool(_num_ios + pages_to_ios(num_pages));
+	return mempool_resize(client->pool, pages_to_ios(num_pages),
+			      GFP_KERNEL);
 }
+EXPORT_SYMBOL(dm_io_client_resize);
 
-void dm_io_put(unsigned int num_pages)
+void dm_io_client_destroy(struct dm_io_client *client)
 {
-	resize_pool(_num_ios - pages_to_ios(num_pages));
+	mempool_destroy(client->pool);
+	bioset_free(client->bios);
+	kfree(client);
 }
+EXPORT_SYMBOL(dm_io_client_destroy);
 
 /*-----------------------------------------------------------------
  * We need to keep track of which region a bio is doing io for.
@@ -118,7 +118,7 @@ static void dec_count(struct io *io, unsigned int region, int error)
 			io_notify_fn fn = io->callback;
 			void *context = io->context;
 
-			mempool_free(io, _io_pool);
+			mempool_free(io, io->client->pool);
 			fn(r, context);
 		}
 	}
@@ -126,7 +126,8 @@ static void dec_count(struct io *io, unsigned int region, int error)
 
 static int endio(struct bio *bio, unsigned int done, int error)
 {
-	struct io *io = (struct io *) bio->bi_private;
+	struct io *io;
+	unsigned region;
 
 	/* keep going until we've finished */
 	if (bio->bi_size)
@@ -135,10 +136,17 @@ static int endio(struct bio *bio, unsigned int done, int error)
 	if (error && bio_data_dir(bio) == READ)
 		zero_fill_bio(bio);
 
-	dec_count(io, bio_get_region(bio), error);
+	/*
+	 * The bio destructor in bio_put() may use the io object.
+	 */
+	io = bio->bi_private;
+	region = bio_get_region(bio);
+
 	bio->bi_max_vecs++;
 	bio_put(bio);
 
+	dec_count(io, region, error);
+
 	return 0;
 }
 
@@ -209,6 +217,9 @@ static void bvec_dp_init(struct dpages *dp, struct bio_vec *bvec)
 	dp->context_ptr = bvec;
 }
 
+/*
+ * Functions for getting the pages from a VMA.
+ */
 static void vm_get_page(struct dpages *dp,
 		 struct page **p, unsigned long *len, unsigned *offset)
 {
@@ -233,7 +244,34 @@ static void vm_dp_init(struct dpages *dp, void *data)
 
 static void dm_bio_destructor(struct bio *bio)
 {
-	bio_free(bio, _bios);
+	struct io *io = bio->bi_private;
+
+	bio_free(bio, io->client->bios);
+}
+
+/*
+ * Functions for getting the pages from kernel memory.
+ */
+static void km_get_page(struct dpages *dp, struct page **p, unsigned long *len,
+			unsigned *offset)
+{
+	*p = virt_to_page(dp->context_ptr);
+	*offset = dp->context_u;
+	*len = PAGE_SIZE - dp->context_u;
+}
+
+static void km_next_page(struct dpages *dp)
+{
+	dp->context_ptr += PAGE_SIZE - dp->context_u;
+	dp->context_u = 0;
+}
+
+static void km_dp_init(struct dpages *dp, void *data)
+{
+	dp->get_page = km_get_page;
+	dp->next_page = km_next_page;
+	dp->context_u = ((unsigned long) data) & (PAGE_SIZE - 1);
+	dp->context_ptr = data;
 }
 
 /*-----------------------------------------------------------------
@@ -256,7 +294,7 @@ static void do_region(int rw, unsigned int region, struct io_region *where,
 		 * to hide it from bio_add_page().
 		 */
 		num_bvecs = (remaining / (PAGE_SIZE >> SECTOR_SHIFT)) + 2;
-		bio = bio_alloc_bioset(GFP_NOIO, num_bvecs, _bios);
+		bio = bio_alloc_bioset(GFP_NOIO, num_bvecs, io->client->bios);
 		bio->bi_sector = where->sector + (where->count - remaining);
 		bio->bi_bdev = where->bdev;
 		bio->bi_end_io = endio;
@@ -311,8 +349,9 @@ static void dispatch_io(int rw, unsigned int num_regions,
 	dec_count(io, 0, 0);
 }
 
-static int sync_io(unsigned int num_regions, struct io_region *where,
-	    int rw, struct dpages *dp, unsigned long *error_bits)
+static int sync_io(struct dm_io_client *client, unsigned int num_regions,
+		   struct io_region *where, int rw, struct dpages *dp,
+		   unsigned long *error_bits)
 {
 	struct io io;
 
@@ -324,6 +363,7 @@ static int sync_io(unsigned int num_regions, struct io_region *where,
 	io.error = 0;
 	atomic_set(&io.count, 1); /* see dispatch_io() */
 	io.sleeper = current;
+	io.client = client;
 
 	dispatch_io(rw, num_regions, where, dp, &io, 1);
 
@@ -340,12 +380,15 @@ static int sync_io(unsigned int num_regions, struct io_region *where,
 	if (atomic_read(&io.count))
 		return -EINTR;
 
-	*error_bits = io.error;
+	if (error_bits)
+		*error_bits = io.error;
+
 	return io.error ? -EIO : 0;
 }
 
-static int async_io(unsigned int num_regions, struct io_region *where, int rw,
-	     struct dpages *dp, io_notify_fn fn, void *context)
+static int async_io(struct dm_io_client *client, unsigned int num_regions,
+		    struct io_region *where, int rw, struct dpages *dp,
+		    io_notify_fn fn, void *context)
 {
 	struct io *io;
 
@@ -355,10 +398,11 @@ static int async_io(unsigned int num_regions, struct io_region *where, int rw,
 		return -EIO;
 	}
 
-	io = mempool_alloc(_io_pool, GFP_NOIO);
+	io = mempool_alloc(client->pool, GFP_NOIO);
 	io->error = 0;
 	atomic_set(&io->count, 1); /* see dispatch_io() */
 	io->sleeper = NULL;
+	io->client = client;
 	io->callback = fn;
 	io->context = context;
 
@@ -366,61 +410,51 @@ static int async_io(unsigned int num_regions, struct io_region *where, int rw,
 	return 0;
 }
 
-int dm_io_sync(unsigned int num_regions, struct io_region *where, int rw,
-	       struct page_list *pl, unsigned int offset,
-	       unsigned long *error_bits)
+static int dp_init(struct dm_io_request *io_req, struct dpages *dp)
 {
-	struct dpages dp;
-	list_dp_init(&dp, pl, offset);
-	return sync_io(num_regions, where, rw, &dp, error_bits);
-}
+	/* Set up dpages based on memory type */
+	switch (io_req->mem.type) {
+	case DM_IO_PAGE_LIST:
+		list_dp_init(dp, io_req->mem.ptr.pl, io_req->mem.offset);
+		break;
+
+	case DM_IO_BVEC:
+		bvec_dp_init(dp, io_req->mem.ptr.bvec);
+		break;
+
+	case DM_IO_VMA:
+		vm_dp_init(dp, io_req->mem.ptr.vma);
+		break;
+
+	case DM_IO_KMEM:
+		km_dp_init(dp, io_req->mem.ptr.addr);
+		break;
+
+	default:
+		return -EINVAL;
+	}
 
-int dm_io_sync_bvec(unsigned int num_regions, struct io_region *where, int rw,
-		    struct bio_vec *bvec, unsigned long *error_bits)
-{
-	struct dpages dp;
-	bvec_dp_init(&dp, bvec);
-	return sync_io(num_regions, where, rw, &dp, error_bits);
+	return 0;
 }
 
-int dm_io_sync_vm(unsigned int num_regions, struct io_region *where, int rw,
-		  void *data, unsigned long *error_bits)
+/*
+ * New collapsed (a)synchronous interface
+ */
+int dm_io(struct dm_io_request *io_req, unsigned num_regions,
+	  struct io_region *where, unsigned long *sync_error_bits)
 {
+	int r;
 	struct dpages dp;
-	vm_dp_init(&dp, data);
-	return sync_io(num_regions, where, rw, &dp, error_bits);
-}
 
-int dm_io_async(unsigned int num_regions, struct io_region *where, int rw,
-		struct page_list *pl, unsigned int offset,
-		io_notify_fn fn, void *context)
-{
-	struct dpages dp;
-	list_dp_init(&dp, pl, offset);
-	return async_io(num_regions, where, rw, &dp, fn, context);
-}
+	r = dp_init(io_req, &dp);
+	if (r)
+		return r;
 
-int dm_io_async_bvec(unsigned int num_regions, struct io_region *where, int rw,
-		     struct bio_vec *bvec, io_notify_fn fn, void *context)
-{
-	struct dpages dp;
-	bvec_dp_init(&dp, bvec);
-	return async_io(num_regions, where, rw, &dp, fn, context);
-}
+	if (!io_req->notify.fn)
+		return sync_io(io_req->client, num_regions, where,
+			       io_req->bi_rw, &dp, sync_error_bits);
 
-int dm_io_async_vm(unsigned int num_regions, struct io_region *where, int rw,
-		   void *data, io_notify_fn fn, void *context)
-{
-	struct dpages dp;
-	vm_dp_init(&dp, data);
-	return async_io(num_regions, where, rw, &dp, fn, context);
+	return async_io(io_req->client, num_regions, where, io_req->bi_rw,
+			&dp, io_req->notify.fn, io_req->notify.context);
 }
-
-EXPORT_SYMBOL(dm_io_get);
-EXPORT_SYMBOL(dm_io_put);
-EXPORT_SYMBOL(dm_io_sync);
-EXPORT_SYMBOL(dm_io_async);
-EXPORT_SYMBOL(dm_io_sync_bvec);
-EXPORT_SYMBOL(dm_io_async_bvec);
-EXPORT_SYMBOL(dm_io_sync_vm);
-EXPORT_SYMBOL(dm_io_async_vm);
+EXPORT_SYMBOL(dm_io);
diff --git a/drivers/md/dm-io.h b/drivers/md/dm-io.h
index f9035bfd1a9f..f647e2cceaa6 100644
--- a/drivers/md/dm-io.h
+++ b/drivers/md/dm-io.h
@@ -12,7 +12,7 @@
 struct io_region {
 	struct block_device *bdev;
 	sector_t sector;
-	sector_t count;
+	sector_t count;		/* If this is zero the region is ignored. */
 };
 
 struct page_list {
@@ -20,55 +20,60 @@ struct page_list {
 	struct page *page;
 };
 
-
-/*
- * 'error' is a bitset, with each bit indicating whether an error
- * occurred doing io to the corresponding region.
- */
 typedef void (*io_notify_fn)(unsigned long error, void *context);
 
+enum dm_io_mem_type {
+	DM_IO_PAGE_LIST,/* Page list */
+	DM_IO_BVEC,	/* Bio vector */
+	DM_IO_VMA,	/* Virtual memory area */
+	DM_IO_KMEM,	/* Kernel memory */
+};
+
+struct dm_io_memory {
+	enum dm_io_mem_type type;
+
+	union {
+		struct page_list *pl;
+		struct bio_vec *bvec;
+		void *vma;
+		void *addr;
+	} ptr;
+
+	unsigned offset;
+};
+
+struct dm_io_notify {
+	io_notify_fn fn;	/* Callback for asynchronous requests */
+	void *context;		/* Passed to callback */
+};
 
 /*
- * Before anyone uses the IO interface they should call
- * dm_io_get(), specifying roughly how many pages they are
- * expecting to perform io on concurrently.
- *
- * This function may block.
+ * IO request structure
  */
-int dm_io_get(unsigned int num_pages);
-void dm_io_put(unsigned int num_pages);
+struct dm_io_client;
+struct dm_io_request {
+	int bi_rw;			/* READ|WRITE - not READA */
+	struct dm_io_memory mem;	/* Memory to use for io */
+	struct dm_io_notify notify;	/* Synchronous if notify.fn is NULL */
+	struct dm_io_client *client;	/* Client memory handler */
+};
 
 /*
- * Synchronous IO.
+ * For async io calls, users can alternatively use the dm_io() function below
+ * and dm_io_client_create() to create private mempools for the client.
  *
- * Please ensure that the rw flag in the next two functions is
- * either READ or WRITE, ie. we don't take READA.  Any
- * regions with a zero count field will be ignored.
+ * Create/destroy may block.
  */
-int dm_io_sync(unsigned int num_regions, struct io_region *where, int rw,
-	       struct page_list *pl, unsigned int offset,
-	       unsigned long *error_bits);
-
-int dm_io_sync_bvec(unsigned int num_regions, struct io_region *where, int rw,
-		    struct bio_vec *bvec, unsigned long *error_bits);
-
-int dm_io_sync_vm(unsigned int num_regions, struct io_region *where, int rw,
-		  void *data, unsigned long *error_bits);
+struct dm_io_client *dm_io_client_create(unsigned num_pages);
+int dm_io_client_resize(unsigned num_pages, struct dm_io_client *client);
+void dm_io_client_destroy(struct dm_io_client *client);
 
 /*
- * Aynchronous IO.
- *
- * The 'where' array may be safely allocated on the stack since
- * the function takes a copy.
+ * IO interface using private per-client pools.
+ * Each bit in the optional 'sync_error_bits' bitset indicates whether an
+ * error occurred doing io to the corresponding region.
  */
-int dm_io_async(unsigned int num_regions, struct io_region *where, int rw,
-		struct page_list *pl, unsigned int offset,
-		io_notify_fn fn, void *context);
-
-int dm_io_async_bvec(unsigned int num_regions, struct io_region *where, int rw,
-		     struct bio_vec *bvec, io_notify_fn fn, void *context);
-
-int dm_io_async_vm(unsigned int num_regions, struct io_region *where, int rw,
-		   void *data, io_notify_fn fn, void *context);
+int dm_io(struct dm_io_request *io_req, unsigned num_regions,
+	  struct io_region *region, unsigned long *sync_error_bits);
 
 #endif
diff --git a/drivers/md/dm-log.c b/drivers/md/dm-log.c
index 6a9261351848..a66428d860fe 100644
--- a/drivers/md/dm-log.c
+++ b/drivers/md/dm-log.c
@@ -149,9 +149,12 @@ struct log_c {
 		FORCESYNC,	/* Force a sync to happen */
 	} sync;
 
+	struct dm_io_request io_req;
+
 	/*
 	 * Disk log fields
 	 */
+	int log_dev_failed;
 	struct dm_dev *log_dev;
 	struct log_header header;
 
@@ -199,13 +202,20 @@ static void header_from_disk(struct log_header *core, struct log_header *disk)
 	core->nr_regions = le64_to_cpu(disk->nr_regions);
 }
 
+static int rw_header(struct log_c *lc, int rw)
+{
+	lc->io_req.bi_rw = rw;
+	lc->io_req.mem.ptr.vma = lc->disk_header;
+	lc->io_req.notify.fn = NULL;
+
+	return dm_io(&lc->io_req, 1, &lc->header_location, NULL);
+}
+
 static int read_header(struct log_c *log)
 {
 	int r;
-	unsigned long ebits;
 
-	r = dm_io_sync_vm(1, &log->header_location, READ,
-			  log->disk_header, &ebits);
+	r = rw_header(log, READ);
 	if (r)
 		return r;
 
@@ -233,11 +243,8 @@ static int read_header(struct log_c *log)
 
 static inline int write_header(struct log_c *log)
 {
-	unsigned long ebits;
-
 	header_to_disk(&log->header, log->disk_header);
-	return dm_io_sync_vm(1, &log->header_location, WRITE,
-			     log->disk_header, &ebits);
+	return rw_header(log, WRITE);
 }
 
 /*----------------------------------------------------------------
@@ -256,6 +263,7 @@ static int create_log_context(struct dirty_log *log, struct dm_target *ti,
 	uint32_t region_size;
 	unsigned int region_count;
 	size_t bitset_size, buf_size;
+	int r;
 
 	if (argc < 1 || argc > 2) {
 		DMWARN("wrong number of arguments to mirror log");
@@ -315,6 +323,7 @@ static int create_log_context(struct dirty_log *log, struct dm_target *ti,
 		lc->disk_header = NULL;
 	} else {
 		lc->log_dev = dev;
+		lc->log_dev_failed = 0;
 		lc->header_location.bdev = lc->log_dev->bdev;
 		lc->header_location.sector = 0;
 
@@ -324,6 +333,15 @@ static int create_log_context(struct dirty_log *log, struct dm_target *ti,
 		buf_size = dm_round_up((LOG_OFFSET << SECTOR_SHIFT) +
 				       bitset_size, ti->limits.hardsect_size);
 		lc->header_location.count = buf_size >> SECTOR_SHIFT;
+		lc->io_req.mem.type = DM_IO_VMA;
+		lc->io_req.client = dm_io_client_create(dm_div_up(buf_size,
+								   PAGE_SIZE));
+		if (IS_ERR(lc->io_req.client)) {
+			r = PTR_ERR(lc->io_req.client);
+			DMWARN("couldn't allocate disk io client");
+			kfree(lc);
+			return -ENOMEM;
+		}
 
 		lc->disk_header = vmalloc(buf_size);
 		if (!lc->disk_header) {
@@ -424,6 +442,7 @@ static void disk_dtr(struct dirty_log *log)
 
 	dm_put_device(lc->ti, lc->log_dev);
 	vfree(lc->disk_header);
+	dm_io_client_destroy(lc->io_req.client);
 	destroy_log_context(lc);
 }
 
@@ -437,6 +456,15 @@ static int count_bits32(uint32_t *addr, unsigned size)
 	return count;
 }
 
+static void fail_log_device(struct log_c *lc)
+{
+	if (lc->log_dev_failed)
+		return;
+
+	lc->log_dev_failed = 1;
+	dm_table_event(lc->ti->table);
+}
+
 static int disk_resume(struct dirty_log *log)
 {
 	int r;
@@ -446,8 +474,19 @@ static int disk_resume(struct dirty_log *log)
 
 	/* read the disk header */
 	r = read_header(lc);
-	if (r)
-		return r;
+	if (r) {
+		DMWARN("%s: Failed to read header on mirror log device",
+		       lc->log_dev->name);
+		fail_log_device(lc);
+		/*
+		 * If the log device cannot be read, we must assume
+		 * all regions are out-of-sync.  If we simply return
+		 * here, the state will be uninitialized and could
+		 * lead us to return 'in-sync' status for regions
+		 * that are actually 'out-of-sync'.
+		 */
+		lc->header.nr_regions = 0;
+	}
 
 	/* set or clear any new bits -- device has grown */
 	if (lc->sync == NOSYNC)
@@ -472,7 +511,14 @@ static int disk_resume(struct dirty_log *log)
 	lc->header.nr_regions = lc->region_count;
 
 	/* write the new header */
-	return write_header(lc);
+	r = write_header(lc);
+	if (r) {
+		DMWARN("%s: Failed to write header on mirror log device",
+		       lc->log_dev->name);
+		fail_log_device(lc);
+	}
+
+	return r;
 }
 
 static uint32_t core_get_region_size(struct dirty_log *log)
@@ -516,7 +562,9 @@ static int disk_flush(struct dirty_log *log)
 		return 0;
 
 	r = write_header(lc);
-	if (!r)
+	if (r)
+		fail_log_device(lc);
+	else
 		lc->touched = 0;
 
 	return r;
@@ -591,6 +639,7 @@ static int core_status(struct dirty_log *log, status_type_t status,
 
 	switch(status) {
 	case STATUSTYPE_INFO:
+		DMEMIT("1 %s", log->type->name);
 		break;
 
 	case STATUSTYPE_TABLE:
@@ -606,17 +655,17 @@ static int disk_status(struct dirty_log *log, status_type_t status,
 		       char *result, unsigned int maxlen)
 {
 	int sz = 0;
-	char buffer[16];
 	struct log_c *lc = log->context;
 
 	switch(status) {
 	case STATUSTYPE_INFO:
+		DMEMIT("3 %s %s %c", log->type->name, lc->log_dev->name,
+		       lc->log_dev_failed ? 'D' : 'A');
 		break;
 
 	case STATUSTYPE_TABLE:
-		format_dev_t(buffer, lc->log_dev->bdev->bd_dev);
 		DMEMIT("%s %u %s %u ", log->type->name,
-		       lc->sync == DEFAULTSYNC ? 2 : 3, buffer,
+		       lc->sync == DEFAULTSYNC ? 2 : 3, lc->log_dev->name,
 		       lc->region_size);
 		DMEMIT_SYNC;
 	}
diff --git a/drivers/md/dm-mpath.c b/drivers/md/dm-mpath.c
index 3aa013506967..de54b39e6ffe 100644
--- a/drivers/md/dm-mpath.c
+++ b/drivers/md/dm-mpath.c
@@ -668,6 +668,9 @@ static int parse_hw_handler(struct arg_set *as, struct multipath *m)
 		return -EINVAL;
 	}
 
+	m->hw_handler.md = dm_table_get_md(ti->table);
+	dm_put(m->hw_handler.md);
+
 	r = hwht->create(&m->hw_handler, hw_argc - 1, as->argv);
 	if (r) {
 		dm_put_hw_handler(hwht);
diff --git a/drivers/md/dm-raid1.c b/drivers/md/dm-raid1.c
index 23a642619bed..ef124b71ccc8 100644
--- a/drivers/md/dm-raid1.c
+++ b/drivers/md/dm-raid1.c
@@ -21,15 +21,11 @@
 #include <linux/workqueue.h>
 
 #define DM_MSG_PREFIX "raid1"
+#define DM_IO_PAGES 64
 
-static struct workqueue_struct *_kmirrord_wq;
-static struct work_struct _kmirrord_work;
-static DECLARE_WAIT_QUEUE_HEAD(_kmirrord_recovery_stopped);
+#define DM_RAID1_HANDLE_ERRORS 0x01
 
-static inline void wake(void)
-{
-	queue_work(_kmirrord_wq, &_kmirrord_work);
-}
+static DECLARE_WAIT_QUEUE_HEAD(_kmirrord_recovery_stopped);
 
 /*-----------------------------------------------------------------
  * Region hash
@@ -125,17 +121,23 @@ struct mirror_set {
 	struct list_head list;
 	struct region_hash rh;
 	struct kcopyd_client *kcopyd_client;
+	uint64_t features;
 
 	spinlock_t lock;	/* protects the next two lists */
 	struct bio_list reads;
 	struct bio_list writes;
 
+	struct dm_io_client *io_client;
+
 	/* recovery */
 	region_t nr_regions;
 	int in_sync;
 
 	struct mirror *default_mirror;	/* Default mirror */
 
+	struct workqueue_struct *kmirrord_wq;
+	struct work_struct kmirrord_work;
+
 	unsigned int nr_mirrors;
 	struct mirror mirror[0];
 };
@@ -153,6 +155,11 @@ static inline sector_t region_to_sector(struct region_hash *rh, region_t region)
 	return region << rh->region_shift;
 }
 
+static void wake(struct mirror_set *ms)
+{
+	queue_work(ms->kmirrord_wq, &ms->kmirrord_work);
+}
+
 /* FIXME move this */
 static void queue_bio(struct mirror_set *ms, struct bio *bio, int rw);
 
@@ -398,8 +405,7 @@ static void rh_update_states(struct region_hash *rh)
 		mempool_free(reg, rh->region_pool);
 	}
 
-	if (!list_empty(&recovered))
-		rh->log->type->flush(rh->log);
+	rh->log->type->flush(rh->log);
 
 	list_for_each_entry_safe (reg, next, &clean, list)
 		mempool_free(reg, rh->region_pool);
@@ -471,7 +477,7 @@ static void rh_dec(struct region_hash *rh, region_t region)
 	spin_unlock_irqrestore(&rh->region_lock, flags);
 
 	if (should_wake)
-		wake();
+		wake(rh->ms);
 }
 
 /*
@@ -558,7 +564,7 @@ static void rh_recovery_end(struct region *reg, int success)
 	list_add(&reg->list, &reg->rh->recovered_regions);
 	spin_unlock_irq(&rh->region_lock);
 
-	wake();
+	wake(rh->ms);
 }
 
 static void rh_flush(struct region_hash *rh)
@@ -592,7 +598,7 @@ static void rh_start_recovery(struct region_hash *rh)
 	for (i = 0; i < MAX_RECOVERY; i++)
 		up(&rh->recovery_count);
 
-	wake();
+	wake(rh->ms);
 }
 
 /*
@@ -735,7 +741,7 @@ static void do_reads(struct mirror_set *ms, struct bio_list *reads)
 		/*
 		 * We can only read balance if the region is in sync.
 		 */
-		if (rh_in_sync(&ms->rh, region, 0))
+		if (rh_in_sync(&ms->rh, region, 1))
 			m = choose_mirror(ms, bio->bi_sector);
 		else
 			m = ms->default_mirror;
@@ -792,6 +798,14 @@ static void do_write(struct mirror_set *ms, struct bio *bio)
 	unsigned int i;
 	struct io_region io[KCOPYD_MAX_REGIONS+1];
 	struct mirror *m;
+	struct dm_io_request io_req = {
+		.bi_rw = WRITE,
+		.mem.type = DM_IO_BVEC,
+		.mem.ptr.bvec = bio->bi_io_vec + bio->bi_idx,
+		.notify.fn = write_callback,
+		.notify.context = bio,
+		.client = ms->io_client,
+	};
 
 	for (i = 0; i < ms->nr_mirrors; i++) {
 		m = ms->mirror + i;
@@ -802,9 +816,8 @@ static void do_write(struct mirror_set *ms, struct bio *bio)
 	}
 
 	bio_set_ms(bio, ms);
-	dm_io_async_bvec(ms->nr_mirrors, io, WRITE,
-			 bio->bi_io_vec + bio->bi_idx,
-			 write_callback, bio);
+
+	(void) dm_io(&io_req, ms->nr_mirrors, io, NULL);
 }
 
 static void do_writes(struct mirror_set *ms, struct bio_list *writes)
@@ -870,11 +883,10 @@ static void do_writes(struct mirror_set *ms, struct bio_list *writes)
 /*-----------------------------------------------------------------
  * kmirrord
  *---------------------------------------------------------------*/
-static LIST_HEAD(_mirror_sets);
-static DECLARE_RWSEM(_mirror_sets_lock);
-
-static void do_mirror(struct mirror_set *ms)
+static void do_mirror(struct work_struct *work)
 {
+	struct mirror_set *ms =container_of(work, struct mirror_set,
+					    kmirrord_work);
 	struct bio_list reads, writes;
 
 	spin_lock(&ms->lock);
@@ -890,16 +902,6 @@ static void do_mirror(struct mirror_set *ms)
 	do_writes(ms, &writes);
 }
 
-static void do_work(struct work_struct *ignored)
-{
-	struct mirror_set *ms;
-
-	down_read(&_mirror_sets_lock);
-	list_for_each_entry (ms, &_mirror_sets, list)
-		do_mirror(ms);
-	up_read(&_mirror_sets_lock);
-}
-
 /*-----------------------------------------------------------------
  * Target functions
  *---------------------------------------------------------------*/
@@ -931,6 +933,13 @@ static struct mirror_set *alloc_context(unsigned int nr_mirrors,
 	ms->in_sync = 0;
 	ms->default_mirror = &ms->mirror[DEFAULT_MIRROR];
 
+	ms->io_client = dm_io_client_create(DM_IO_PAGES);
+	if (IS_ERR(ms->io_client)) {
+		ti->error = "Error creating dm_io client";
+		kfree(ms);
+ 		return NULL;
+	}
+
 	if (rh_init(&ms->rh, ms, dl, region_size, ms->nr_regions)) {
 		ti->error = "Error creating dirty region hash";
 		kfree(ms);
@@ -946,6 +955,7 @@ static void free_context(struct mirror_set *ms, struct dm_target *ti,
 	while (m--)
 		dm_put_device(ti, ms->mirror[m].dev);
 
+	dm_io_client_destroy(ms->io_client);
 	rh_exit(&ms->rh);
 	kfree(ms);
 }
@@ -978,23 +988,6 @@ static int get_mirror(struct mirror_set *ms, struct dm_target *ti,
 	return 0;
 }
 
-static int add_mirror_set(struct mirror_set *ms)
-{
-	down_write(&_mirror_sets_lock);
-	list_add_tail(&ms->list, &_mirror_sets);
-	up_write(&_mirror_sets_lock);
-	wake();
-
-	return 0;
-}
-
-static void del_mirror_set(struct mirror_set *ms)
-{
-	down_write(&_mirror_sets_lock);
-	list_del(&ms->list);
-	up_write(&_mirror_sets_lock);
-}
-
 /*
  * Create dirty log: log_type #log_params <log_params>
  */
@@ -1037,16 +1030,55 @@ static struct dirty_log *create_dirty_log(struct dm_target *ti,
 	return dl;
 }
 
+static int parse_features(struct mirror_set *ms, unsigned argc, char **argv,
+			  unsigned *args_used)
+{
+	unsigned num_features;
+	struct dm_target *ti = ms->ti;
+
+	*args_used = 0;
+
+	if (!argc)
+		return 0;
+
+	if (sscanf(argv[0], "%u", &num_features) != 1) {
+		ti->error = "Invalid number of features";
+		return -EINVAL;
+	}
+
+	argc--;
+	argv++;
+	(*args_used)++;
+
+	if (num_features > argc) {
+		ti->error = "Not enough arguments to support feature count";
+		return -EINVAL;
+	}
+
+	if (!strcmp("handle_errors", argv[0]))
+		ms->features |= DM_RAID1_HANDLE_ERRORS;
+	else {
+		ti->error = "Unrecognised feature requested";
+		return -EINVAL;
+	}
+
+	(*args_used)++;
+
+	return 0;
+}
+
 /*
  * Construct a mirror mapping:
  *
  * log_type #log_params <log_params>
  * #mirrors [mirror_path offset]{2,}
+ * [#features <features>]
  *
  * log_type is "core" or "disk"
  * #log_params is between 1 and 3
+ *
+ * If present, features must be "handle_errors".
  */
-#define DM_IO_PAGES 64
 static int mirror_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 {
 	int r;
@@ -1070,8 +1102,8 @@ static int mirror_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 
 	argv++, argc--;
 
-	if (argc != nr_mirrors * 2) {
-		ti->error = "Wrong number of mirror arguments";
+	if (argc < nr_mirrors * 2) {
+		ti->error = "Too few mirror arguments";
 		dm_destroy_dirty_log(dl);
 		return -EINVAL;
 	}
@@ -1096,13 +1128,37 @@ static int mirror_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 	ti->private = ms;
  	ti->split_io = ms->rh.region_size;
 
+	ms->kmirrord_wq = create_singlethread_workqueue("kmirrord");
+	if (!ms->kmirrord_wq) {
+		DMERR("couldn't start kmirrord");
+		free_context(ms, ti, m);
+		return -ENOMEM;
+	}
+	INIT_WORK(&ms->kmirrord_work, do_mirror);
+
+	r = parse_features(ms, argc, argv, &args_used);
+	if (r) {
+		free_context(ms, ti, ms->nr_mirrors);
+		return r;
+	}
+
+	argv += args_used;
+	argc -= args_used;
+
+	if (argc) {
+		ti->error = "Too many mirror arguments";
+		free_context(ms, ti, ms->nr_mirrors);
+		return -EINVAL;
+	}
+
 	r = kcopyd_client_create(DM_IO_PAGES, &ms->kcopyd_client);
 	if (r) {
+		destroy_workqueue(ms->kmirrord_wq);
 		free_context(ms, ti, ms->nr_mirrors);
 		return r;
 	}
 
-	add_mirror_set(ms);
+	wake(ms);
 	return 0;
 }
 
@@ -1110,8 +1166,9 @@ static void mirror_dtr(struct dm_target *ti)
 {
 	struct mirror_set *ms = (struct mirror_set *) ti->private;
 
-	del_mirror_set(ms);
+	flush_workqueue(ms->kmirrord_wq);
 	kcopyd_client_destroy(ms->kcopyd_client);
+	destroy_workqueue(ms->kmirrord_wq);
 	free_context(ms, ti, ms->nr_mirrors);
 }
 
@@ -1127,7 +1184,7 @@ static void queue_bio(struct mirror_set *ms, struct bio *bio, int rw)
 	spin_unlock(&ms->lock);
 
 	if (should_wake)
-		wake();
+		wake(ms);
 }
 
 /*
@@ -1222,11 +1279,9 @@ static void mirror_resume(struct dm_target *ti)
 static int mirror_status(struct dm_target *ti, status_type_t type,
 			 char *result, unsigned int maxlen)
 {
-	unsigned int m, sz;
+	unsigned int m, sz = 0;
 	struct mirror_set *ms = (struct mirror_set *) ti->private;
 
-	sz = ms->rh.log->type->status(ms->rh.log, type, result, maxlen);
-
 	switch (type) {
 	case STATUSTYPE_INFO:
 		DMEMIT("%d ", ms->nr_mirrors);
@@ -1237,13 +1292,21 @@ static int mirror_status(struct dm_target *ti, status_type_t type,
 			(unsigned long long)ms->rh.log->type->
 				get_sync_count(ms->rh.log),
 			(unsigned long long)ms->nr_regions);
+
+		sz = ms->rh.log->type->status(ms->rh.log, type, result, maxlen);
+
 		break;
 
 	case STATUSTYPE_TABLE:
+		sz = ms->rh.log->type->status(ms->rh.log, type, result, maxlen);
+
 		DMEMIT("%d", ms->nr_mirrors);
 		for (m = 0; m < ms->nr_mirrors; m++)
 			DMEMIT(" %s %llu", ms->mirror[m].dev->name,
 				(unsigned long long)ms->mirror[m].offset);
+
+		if (ms->features & DM_RAID1_HANDLE_ERRORS)
+			DMEMIT(" 1 handle_errors");
 	}
 
 	return 0;
@@ -1251,7 +1314,7 @@ static int mirror_status(struct dm_target *ti, status_type_t type,
 
 static struct target_type mirror_target = {
 	.name	 = "mirror",
-	.version = {1, 0, 2},
+	.version = {1, 0, 3},
 	.module	 = THIS_MODULE,
 	.ctr	 = mirror_ctr,
 	.dtr	 = mirror_dtr,
@@ -1270,20 +1333,11 @@ static int __init dm_mirror_init(void)
 	if (r)
 		return r;
 
-	_kmirrord_wq = create_singlethread_workqueue("kmirrord");
-	if (!_kmirrord_wq) {
-		DMERR("couldn't start kmirrord");
-		dm_dirty_log_exit();
-		return r;
-	}
-	INIT_WORK(&_kmirrord_work, do_work);
-
 	r = dm_register_target(&mirror_target);
 	if (r < 0) {
 		DMERR("%s: Failed to register mirror target",
 		      mirror_target.name);
 		dm_dirty_log_exit();
-		destroy_workqueue(_kmirrord_wq);
 	}
 
 	return r;
@@ -1297,7 +1351,6 @@ static void __exit dm_mirror_exit(void)
 	if (r < 0)
 		DMERR("%s: unregister failed %d", mirror_target.name, r);
 
-	destroy_workqueue(_kmirrord_wq);
 	dm_dirty_log_exit();
 }
 
diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index 05befa91807a..2fc199b0016b 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -425,13 +425,15 @@ static void close_dev(struct dm_dev *d, struct mapped_device *md)
 }
 
 /*
- * If possible (ie. blk_size[major] is set), this checks an area
- * of a destination device is valid.
+ * If possible, this checks an area of a destination device is valid.
  */
 static int check_device_area(struct dm_dev *dd, sector_t start, sector_t len)
 {
-	sector_t dev_size;
-	dev_size = dd->bdev->bd_inode->i_size >> SECTOR_SHIFT;
+	sector_t dev_size = dd->bdev->bd_inode->i_size >> SECTOR_SHIFT;
+
+	if (!dev_size)
+		return 1;
+
 	return ((start < dev_size) && (len <= (dev_size - start)));
 }
 
diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index 11a98df298ec..2717a355dc5b 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -1236,6 +1236,7 @@ void dm_put(struct mapped_device *md)
 		free_dev(md);
 	}
 }
+EXPORT_SYMBOL_GPL(dm_put);
 
 /*
  * Process the deferred bios
diff --git a/drivers/md/kcopyd.c b/drivers/md/kcopyd.c
index b46f6c575f7e..dbc234e3c69f 100644
--- a/drivers/md/kcopyd.c
+++ b/drivers/md/kcopyd.c
@@ -1,5 +1,6 @@
 /*
  * Copyright (C) 2002 Sistina Software (UK) Limited.
+ * Copyright (C) 2006 Red Hat GmbH
  *
  * This file is released under the GPL.
  *
@@ -45,6 +46,8 @@ struct kcopyd_client {
 	unsigned int nr_pages;
 	unsigned int nr_free_pages;
 
+	struct dm_io_client *io_client;
+
 	wait_queue_head_t destroyq;
 	atomic_t nr_jobs;
 };
@@ -342,16 +345,20 @@ static void complete_io(unsigned long error, void *context)
 static int run_io_job(struct kcopyd_job *job)
 {
 	int r;
+	struct dm_io_request io_req = {
+		.bi_rw = job->rw,
+		.mem.type = DM_IO_PAGE_LIST,
+		.mem.ptr.pl = job->pages,
+		.mem.offset = job->offset,
+		.notify.fn = complete_io,
+		.notify.context = job,
+		.client = job->kc->io_client,
+	};
 
 	if (job->rw == READ)
-		r = dm_io_async(1, &job->source, job->rw,
-				job->pages,
-				job->offset, complete_io, job);
-
+		r = dm_io(&io_req, 1, &job->source, NULL);
 	else
-		r = dm_io_async(job->num_dests, job->dests, job->rw,
-				job->pages,
-				job->offset, complete_io, job);
+		r = dm_io(&io_req, job->num_dests, job->dests, NULL);
 
 	return r;
 }
@@ -670,8 +677,9 @@ int kcopyd_client_create(unsigned int nr_pages, struct kcopyd_client **result)
 		return r;
 	}
 
-	r = dm_io_get(nr_pages);
-	if (r) {
+	kc->io_client = dm_io_client_create(nr_pages);
+	if (IS_ERR(kc->io_client)) {
+		r = PTR_ERR(kc->io_client);
 		client_free_pages(kc);
 		kfree(kc);
 		kcopyd_exit();
@@ -691,7 +699,7 @@ void kcopyd_client_destroy(struct kcopyd_client *kc)
 	/* Wait for completion of all jobs submitted by this client. */
 	wait_event(kc->destroyq, !atomic_read(&kc->nr_jobs));
 
-	dm_io_put(kc->nr_pages);
+	dm_io_client_destroy(kc->io_client);
 	client_free_pages(kc);
 	client_del(kc);
 	kfree(kc);
diff --git a/drivers/md/md.c b/drivers/md/md.c
index 2b4315d7e5d6..65814b0340cb 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -33,6 +33,7 @@
 */
 
 #include <linux/module.h>
+#include <linux/kernel.h>
 #include <linux/kthread.h>
 #include <linux/linkage.h>
 #include <linux/raid/md.h>
@@ -273,6 +274,7 @@ static mddev_t * mddev_find(dev_t unit)
 	atomic_set(&new->active, 1);
 	spin_lock_init(&new->write_lock);
 	init_waitqueue_head(&new->sb_wait);
+	new->reshape_position = MaxSector;
 
 	new->queue = blk_alloc_queue(GFP_KERNEL);
 	if (!new->queue) {
@@ -589,14 +591,41 @@ abort:
 	return ret;
 }
 
+
+static u32 md_csum_fold(u32 csum)
+{
+	csum = (csum & 0xffff) + (csum >> 16);
+	return (csum & 0xffff) + (csum >> 16);
+}
+
 static unsigned int calc_sb_csum(mdp_super_t * sb)
 {
+	u64 newcsum = 0;
+	u32 *sb32 = (u32*)sb;
+	int i;
 	unsigned int disk_csum, csum;
 
 	disk_csum = sb->sb_csum;
 	sb->sb_csum = 0;
-	csum = csum_partial((void *)sb, MD_SB_BYTES, 0);
+
+	for (i = 0; i < MD_SB_BYTES/4 ; i++)
+		newcsum += sb32[i];
+	csum = (newcsum & 0xffffffff) + (newcsum>>32);
+
+
+#ifdef CONFIG_ALPHA
+	/* This used to use csum_partial, which was wrong for several
+	 * reasons including that different results are returned on
+	 * different architectures.  It isn't critical that we get exactly
+	 * the same return value as before (we always csum_fold before
+	 * testing, and that removes any differences).  However as we
+	 * know that csum_partial always returned a 16bit value on
+	 * alphas, do a fold to maximise conformity to previous behaviour.
+	 */
+	sb->sb_csum = md_csum_fold(disk_csum);
+#else
 	sb->sb_csum = disk_csum;
+#endif
 	return csum;
 }
 
@@ -684,7 +713,7 @@ static int super_90_load(mdk_rdev_t *rdev, mdk_rdev_t *refdev, int minor_version
 	if (sb->raid_disks <= 0)
 		goto abort;
 
-	if (csum_fold(calc_sb_csum(sb)) != csum_fold(sb->sb_csum)) {
+	if (md_csum_fold(calc_sb_csum(sb)) != md_csum_fold(sb->sb_csum)) {
 		printk(KERN_WARNING "md: invalid superblock checksum on %s\n",
 			b);
 		goto abort;
@@ -694,6 +723,17 @@ static int super_90_load(mdk_rdev_t *rdev, mdk_rdev_t *refdev, int minor_version
 	rdev->data_offset = 0;
 	rdev->sb_size = MD_SB_BYTES;
 
+	if (sb->state & (1<<MD_SB_BITMAP_PRESENT)) {
+		if (sb->level != 1 && sb->level != 4
+		    && sb->level != 5 && sb->level != 6
+		    && sb->level != 10) {
+			/* FIXME use a better test */
+			printk(KERN_WARNING
+			       "md: bitmaps not supported for this level.\n");
+			goto abort;
+		}
+	}
+
 	if (sb->level == LEVEL_MULTIPATH)
 		rdev->desc_nr = -1;
 	else
@@ -792,16 +832,8 @@ static int super_90_validate(mddev_t *mddev, mdk_rdev_t *rdev)
 		mddev->max_disks = MD_SB_DISKS;
 
 		if (sb->state & (1<<MD_SB_BITMAP_PRESENT) &&
-		    mddev->bitmap_file == NULL) {
-			if (mddev->level != 1 && mddev->level != 4
-			    && mddev->level != 5 && mddev->level != 6
-			    && mddev->level != 10) {
-				/* FIXME use a better test */
-				printk(KERN_WARNING "md: bitmaps not supported for this level.\n");
-				return -EINVAL;
-			}
+		    mddev->bitmap_file == NULL)
 			mddev->bitmap_offset = mddev->default_bitmap_offset;
-		}
 
 	} else if (mddev->pers == NULL) {
 		/* Insist on good event counter while assembling */
@@ -1058,6 +1090,18 @@ static int super_1_load(mdk_rdev_t *rdev, mdk_rdev_t *refdev, int minor_version)
 		       bdevname(rdev->bdev,b));
 		return -EINVAL;
 	}
+	if ((le32_to_cpu(sb->feature_map) & MD_FEATURE_BITMAP_OFFSET)) {
+		if (sb->level != cpu_to_le32(1) &&
+		    sb->level != cpu_to_le32(4) &&
+		    sb->level != cpu_to_le32(5) &&
+		    sb->level != cpu_to_le32(6) &&
+		    sb->level != cpu_to_le32(10)) {
+			printk(KERN_WARNING
+			       "md: bitmaps not supported for this level.\n");
+			return -EINVAL;
+		}
+	}
+
 	rdev->preferred_minor = 0xffff;
 	rdev->data_offset = le64_to_cpu(sb->data_offset);
 	atomic_set(&rdev->corrected_errors, le32_to_cpu(sb->cnt_corrected_read));
@@ -1141,14 +1185,9 @@ static int super_1_validate(mddev_t *mddev, mdk_rdev_t *rdev)
 		mddev->max_disks =  (4096-256)/2;
 
 		if ((le32_to_cpu(sb->feature_map) & MD_FEATURE_BITMAP_OFFSET) &&
-		    mddev->bitmap_file == NULL ) {
-			if (mddev->level != 1 && mddev->level != 5 && mddev->level != 6
-			    && mddev->level != 10) {
-				printk(KERN_WARNING "md: bitmaps not supported for this level.\n");
-				return -EINVAL;
-			}
+		    mddev->bitmap_file == NULL )
 			mddev->bitmap_offset = (__s32)le32_to_cpu(sb->bitmap_offset);
-		}
+
 		if ((le32_to_cpu(sb->feature_map) & MD_FEATURE_RESHAPE_ACTIVE)) {
 			mddev->reshape_position = le64_to_cpu(sb->reshape_position);
 			mddev->delta_disks = le32_to_cpu(sb->delta_disks);
@@ -2204,6 +2243,10 @@ static ssize_t
 layout_show(mddev_t *mddev, char *page)
 {
 	/* just a number, not meaningful for all levels */
+	if (mddev->reshape_position != MaxSector &&
+	    mddev->layout != mddev->new_layout)
+		return sprintf(page, "%d (%d)\n",
+			       mddev->new_layout, mddev->layout);
 	return sprintf(page, "%d\n", mddev->layout);
 }
 
@@ -2212,13 +2255,16 @@ layout_store(mddev_t *mddev, const char *buf, size_t len)
 {
 	char *e;
 	unsigned long n = simple_strtoul(buf, &e, 10);
-	if (mddev->pers)
-		return -EBUSY;
 
 	if (!*buf || (*e && *e != '\n'))
 		return -EINVAL;
 
-	mddev->layout = n;
+	if (mddev->pers)
+		return -EBUSY;
+	if (mddev->reshape_position != MaxSector)
+		mddev->new_layout = n;
+	else
+		mddev->layout = n;
 	return len;
 }
 static struct md_sysfs_entry md_layout =
@@ -2230,6 +2276,10 @@ raid_disks_show(mddev_t *mddev, char *page)
 {
 	if (mddev->raid_disks == 0)
 		return 0;
+	if (mddev->reshape_position != MaxSector &&
+	    mddev->delta_disks != 0)
+		return sprintf(page, "%d (%d)\n", mddev->raid_disks,
+			       mddev->raid_disks - mddev->delta_disks);
 	return sprintf(page, "%d\n", mddev->raid_disks);
 }
 
@@ -2247,7 +2297,11 @@ raid_disks_store(mddev_t *mddev, const char *buf, size_t len)
 
 	if (mddev->pers)
 		rv = update_raid_disks(mddev, n);
-	else
+	else if (mddev->reshape_position != MaxSector) {
+		int olddisks = mddev->raid_disks - mddev->delta_disks;
+		mddev->delta_disks = n - olddisks;
+		mddev->raid_disks = n;
+	} else
 		mddev->raid_disks = n;
 	return rv ? rv : len;
 }
@@ -2257,6 +2311,10 @@ __ATTR(raid_disks, S_IRUGO|S_IWUSR, raid_disks_show, raid_disks_store);
 static ssize_t
 chunk_size_show(mddev_t *mddev, char *page)
 {
+	if (mddev->reshape_position != MaxSector &&
+	    mddev->chunk_size != mddev->new_chunk)
+		return sprintf(page, "%d (%d)\n", mddev->new_chunk,
+			       mddev->chunk_size);
 	return sprintf(page, "%d\n", mddev->chunk_size);
 }
 
@@ -2267,12 +2325,15 @@ chunk_size_store(mddev_t *mddev, const char *buf, size_t len)
 	char *e;
 	unsigned long n = simple_strtoul(buf, &e, 10);
 
-	if (mddev->pers)
-		return -EBUSY;
 	if (!*buf || (*e && *e != '\n'))
 		return -EINVAL;
 
-	mddev->chunk_size = n;
+	if (mddev->pers)
+		return -EBUSY;
+	else if (mddev->reshape_position != MaxSector)
+		mddev->new_chunk = n;
+	else
+		mddev->chunk_size = n;
 	return len;
 }
 static struct md_sysfs_entry md_chunk_size =
@@ -2637,8 +2698,7 @@ metadata_store(mddev_t *mddev, const char *buf, size_t len)
 	minor = simple_strtoul(buf, &e, 10);
 	if (e==buf || (*e && *e != '\n') )
 		return -EINVAL;
-	if (major >= sizeof(super_types)/sizeof(super_types[0]) ||
-	    super_types[major].name == NULL)
+	if (major >= ARRAY_SIZE(super_types) || super_types[major].name == NULL)
 		return -ENOENT;
 	mddev->major_version = major;
 	mddev->minor_version = minor;
@@ -2859,6 +2919,37 @@ suspend_hi_store(mddev_t *mddev, const char *buf, size_t len)
 static struct md_sysfs_entry md_suspend_hi =
 __ATTR(suspend_hi, S_IRUGO|S_IWUSR, suspend_hi_show, suspend_hi_store);
 
+static ssize_t
+reshape_position_show(mddev_t *mddev, char *page)
+{
+	if (mddev->reshape_position != MaxSector)
+		return sprintf(page, "%llu\n",
+			       (unsigned long long)mddev->reshape_position);
+	strcpy(page, "none\n");
+	return 5;
+}
+
+static ssize_t
+reshape_position_store(mddev_t *mddev, const char *buf, size_t len)
+{
+	char *e;
+	unsigned long long new = simple_strtoull(buf, &e, 10);
+	if (mddev->pers)
+		return -EBUSY;
+	if (buf == e || (*e && *e != '\n'))
+		return -EINVAL;
+	mddev->reshape_position = new;
+	mddev->delta_disks = 0;
+	mddev->new_level = mddev->level;
+	mddev->new_layout = mddev->layout;
+	mddev->new_chunk = mddev->chunk_size;
+	return len;
+}
+
+static struct md_sysfs_entry md_reshape_position =
+__ATTR(reshape_position, S_IRUGO|S_IWUSR, reshape_position_show,
+       reshape_position_store);
+
 
 static struct attribute *md_default_attrs[] = {
 	&md_level.attr,
@@ -2871,6 +2962,7 @@ static struct attribute *md_default_attrs[] = {
 	&md_new_device.attr,
 	&md_safe_delay.attr,
 	&md_array_state.attr,
+	&md_reshape_position.attr,
 	NULL,
 };
 
@@ -3409,6 +3501,7 @@ static int do_md_stop(mddev_t * mddev, int mode)
 		mddev->size = 0;
 		mddev->raid_disks = 0;
 		mddev->recovery_cp = 0;
+		mddev->reshape_position = MaxSector;
 
 	} else if (mddev->pers)
 		printk(KERN_INFO "md: %s switched to read-only mode.\n",
@@ -4019,7 +4112,7 @@ static int set_array_info(mddev_t * mddev, mdu_array_info_t *info)
 	if (info->raid_disks == 0) {
 		/* just setting version number for superblock loading */
 		if (info->major_version < 0 ||
-		    info->major_version >= sizeof(super_types)/sizeof(super_types[0]) ||
+		    info->major_version >= ARRAY_SIZE(super_types) ||
 		    super_types[info->major_version].name == NULL) {
 			/* maybe try to auto-load a module? */
 			printk(KERN_INFO 
@@ -4941,15 +5034,6 @@ static int md_seq_open(struct inode *inode, struct file *file)
 	return error;
 }
 
-static int md_seq_release(struct inode *inode, struct file *file)
-{
-	struct seq_file *m = file->private_data;
-	struct mdstat_info *mi = m->private;
-	m->private = NULL;
-	kfree(mi);
-	return seq_release(inode, file);
-}
-
 static unsigned int mdstat_poll(struct file *filp, poll_table *wait)
 {
 	struct seq_file *m = filp->private_data;
@@ -4971,7 +5055,7 @@ static const struct file_operations md_seq_fops = {
 	.open           = md_seq_open,
 	.read           = seq_read,
 	.llseek         = seq_lseek,
-	.release	= md_seq_release,
+	.release	= seq_release_private,
 	.poll		= mdstat_poll,
 };
 
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 8d59914f2057..061375ee6592 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -353,8 +353,8 @@ static int grow_stripes(raid5_conf_t *conf, int num)
 	struct kmem_cache *sc;
 	int devs = conf->raid_disks;
 
-	sprintf(conf->cache_name[0], "raid5/%s", mdname(conf->mddev));
-	sprintf(conf->cache_name[1], "raid5/%s-alt", mdname(conf->mddev));
+	sprintf(conf->cache_name[0], "raid5-%s", mdname(conf->mddev));
+	sprintf(conf->cache_name[1], "raid5-%s-alt", mdname(conf->mddev));
 	conf->active_name = 0;
 	sc = kmem_cache_create(conf->cache_name[conf->active_name],
 			       sizeof(struct stripe_head)+(devs-1)*sizeof(struct r5dev),
diff --git a/drivers/media/dvb/dvb-usb/dvb-usb-remote.c b/drivers/media/dvb/dvb-usb/dvb-usb-remote.c
index 68ed3a788083..9200a30dd1b9 100644
--- a/drivers/media/dvb/dvb-usb/dvb-usb-remote.c
+++ b/drivers/media/dvb/dvb-usb/dvb-usb-remote.c
@@ -3,7 +3,7 @@
  * Copyright (C) 2004-6 Patrick Boettcher (patrick.boettcher@desy.de)
  * see dvb-usb-init.c for copyright information.
  *
- * This file contains functions for initializing the the input-device and for handling remote-control-queries.
+ * This file contains functions for initializing the input-device and for handling remote-control-queries.
  */
 #include "dvb-usb-common.h"
 #include <linux/usb/input.h>
diff --git a/drivers/media/dvb/frontends/dib7000m.c b/drivers/media/dvb/frontends/dib7000m.c
index f5d40aa3d27f..f64546c6aeb5 100644
--- a/drivers/media/dvb/frontends/dib7000m.c
+++ b/drivers/media/dvb/frontends/dib7000m.c
@@ -266,7 +266,7 @@ static int dib7000m_sad_calib(struct dib7000m_state *state)
 {
 
 /* internal */
-//	dib7000m_write_word(state, 928, (3 << 14) | (1 << 12) | (524 << 0)); // sampling clock of the SAD is writting in set_bandwidth
+//	dib7000m_write_word(state, 928, (3 << 14) | (1 << 12) | (524 << 0)); // sampling clock of the SAD is written in set_bandwidth
 	dib7000m_write_word(state, 929, (0 << 1) | (0 << 0));
 	dib7000m_write_word(state, 930, 776); // 0.625*3.3 / 4096
 
diff --git a/drivers/media/dvb/frontends/dib7000p.c b/drivers/media/dvb/frontends/dib7000p.c
index 0349a4b5da3f..aece458cfe12 100644
--- a/drivers/media/dvb/frontends/dib7000p.c
+++ b/drivers/media/dvb/frontends/dib7000p.c
@@ -223,7 +223,7 @@ static int dib7000p_set_bandwidth(struct dvb_frontend *demod, u8 BW_Idx)
 static int dib7000p_sad_calib(struct dib7000p_state *state)
 {
 /* internal */
-//	dib7000p_write_word(state, 72, (3 << 14) | (1 << 12) | (524 << 0)); // sampling clock of the SAD is writting in set_bandwidth
+//	dib7000p_write_word(state, 72, (3 << 14) | (1 << 12) | (524 << 0)); // sampling clock of the SAD is written in set_bandwidth
 	dib7000p_write_word(state, 73, (0 << 1) | (0 << 0));
 	dib7000p_write_word(state, 74, 776); // 0.625*3.3 / 4096
 
diff --git a/drivers/media/dvb/frontends/tda10021.c b/drivers/media/dvb/frontends/tda10021.c
index 110536843e8e..e725f612a6b7 100644
--- a/drivers/media/dvb/frontends/tda10021.c
+++ b/drivers/media/dvb/frontends/tda10021.c
@@ -1,6 +1,6 @@
 /*
     TDA10021  - Single Chip Cable Channel Receiver driver module
-	       used on the the Siemens DVB-C cards
+	       used on the Siemens DVB-C cards
 
     Copyright (C) 1999 Convergence Integrated Media GmbH <ralph@convergence.de>
     Copyright (C) 2004 Markus Schulz <msc@antzsystem.de>
diff --git a/drivers/media/dvb/frontends/ves1x93.c b/drivers/media/dvb/frontends/ves1x93.c
index 54d7b07571b8..23fd0303c91b 100644
--- a/drivers/media/dvb/frontends/ves1x93.c
+++ b/drivers/media/dvb/frontends/ves1x93.c
@@ -306,7 +306,7 @@ static int ves1x93_read_status(struct dvb_frontend* fe, fe_status_t* status)
 	 * The ves1893 sometimes returns sync values that make no sense,
 	 * because, e.g., the SIGNAL bit is 0, while some of the higher
 	 * bits are 1 (and how can there be a CARRIER w/o a SIGNAL?).
-	 * Tests showed that the the VITERBI and SYNC bits are returned
+	 * Tests showed that the VITERBI and SYNC bits are returned
 	 * reliably, while the SIGNAL and CARRIER bits ar sometimes wrong.
 	 * If such a case occurs, we read the value again, until we get a
 	 * valid value.
diff --git a/drivers/media/video/em28xx/em28xx-i2c.c b/drivers/media/video/em28xx/em28xx-i2c.c
index 563a8319e608..54ccc6e1f92e 100644
--- a/drivers/media/video/em28xx/em28xx-i2c.c
+++ b/drivers/media/video/em28xx/em28xx-i2c.c
@@ -70,7 +70,7 @@ static int em2800_i2c_send_max4(struct em28xx *dev, unsigned char addr,
 
 	ret = dev->em28xx_write_regs(dev, 4 - len, &b2[4 - len], 2 + len);
 	if (ret != 2 + len) {
-		em28xx_warn("writting to i2c device failed (error=%i)\n", ret);
+		em28xx_warn("writing to i2c device failed (error=%i)\n", ret);
 		return -EIO;
 	}
 	for (write_timeout = EM2800_I2C_WRITE_TIMEOUT; write_timeout > 0;
diff --git a/drivers/media/video/em28xx/em28xx-video.c b/drivers/media/video/em28xx/em28xx-video.c
index bec67609500f..2c7b158ce7e1 100644
--- a/drivers/media/video/em28xx/em28xx-video.c
+++ b/drivers/media/video/em28xx/em28xx-video.c
@@ -1729,7 +1729,7 @@ static int em28xx_usb_probe(struct usb_interface *interface,
 
 	endpoint = &interface->cur_altsetting->endpoint[1].desc;
 
-	/* check if the the device has the iso in endpoint at the correct place */
+	/* check if the device has the iso in endpoint at the correct place */
 	if ((endpoint->bmAttributes & USB_ENDPOINT_XFERTYPE_MASK) !=
 	    USB_ENDPOINT_XFER_ISOC) {
 		em28xx_err(DRIVER_NAME " probing error: endpoint is non-ISO endpoint!\n");
diff --git a/drivers/media/video/pwc/philips.txt b/drivers/media/video/pwc/philips.txt
index f5e848410311..f9f3584281d8 100644
--- a/drivers/media/video/pwc/philips.txt
+++ b/drivers/media/video/pwc/philips.txt
@@ -54,9 +54,9 @@ fps
    Specifies the desired framerate. Is an integer in the range of 4-30.
 
 fbufs
-   This paramter specifies the number of internal buffers to use for storing
+   This parameter specifies the number of internal buffers to use for storing
    frames from the cam. This will help if the process that reads images from
-   the cam is a bit slow or momentarely busy. However, on slow machines it
+   the cam is a bit slow or momentarily busy. However, on slow machines it
    only introduces lag, so choose carefully. The default is 3, which is
    reasonable. You can set it between 2 and 5.
 
@@ -209,7 +209,7 @@ trace
 
      128   0x80   PWCX debugging                                      Off
 
-   For example, to trace the open() & read() fuctions, sum 8 + 4 = 12,
+   For example, to trace the open() & read() functions, sum 8 + 4 = 12,
    so you would supply trace=12 during insmod or modprobe. If
    you want to turn the initialization and probing tracing off, set trace=0.
    The default value for trace is 35 (0x23).
diff --git a/drivers/media/video/usbvideo/vicam.c b/drivers/media/video/usbvideo/vicam.c
index 876fd2768242..982b115193f8 100644
--- a/drivers/media/video/usbvideo/vicam.c
+++ b/drivers/media/video/usbvideo/vicam.c
@@ -28,7 +28,7 @@
  *
  * Portions of this code were also copied from usbvideo.c
  *
- * Special thanks to the the whole team at Sourceforge for help making
+ * Special thanks to the whole team at Sourceforge for help making
  * this driver become a reality.  Notably:
  * Andy Armstrong who reverse engineered the color encoding and
  * Pavel Machek and Chris Cheney who worked on reverse engineering the
diff --git a/drivers/message/fusion/lsi/mpi_history.txt b/drivers/message/fusion/lsi/mpi_history.txt
index d6b4c607453b..ddc7ae029dd3 100644
--- a/drivers/message/fusion/lsi/mpi_history.txt
+++ b/drivers/message/fusion/lsi/mpi_history.txt
@@ -571,7 +571,7 @@ mpi_fc.h
  *  11-02-00  01.01.01  Original release for post 1.0 work
  *  12-04-00  01.01.02  Added messages for Common Transport Send and
  *                      Primitive Send.
- *  01-09-01  01.01.03  Modifed some of the new flags to have an MPI prefix
+ *  01-09-01  01.01.03  Modified some of the new flags to have an MPI prefix
  *                      and modified the FcPrimitiveSend flags.
  *  01-25-01  01.01.04  Move InitiatorIndex in LinkServiceRsp reply to a larger
  *                      field.
diff --git a/drivers/message/fusion/mptbase.c b/drivers/message/fusion/mptbase.c
index 97471af4309c..5021d1a2a1d4 100644
--- a/drivers/message/fusion/mptbase.c
+++ b/drivers/message/fusion/mptbase.c
@@ -3585,7 +3585,7 @@ initChainBuffers(MPT_ADAPTER *ioc)
 	 * index = chain_idx
 	 *
 	 * Calculate the number of chain buffers needed(plus 1) per I/O
-	 * then multiply the the maximum number of simultaneous cmds
+	 * then multiply the maximum number of simultaneous cmds
 	 *
 	 * num_sge = num sge in request frame + last chain buffer
 	 * scale = num sge per chain buffer if no chain element
diff --git a/drivers/mmc/core/core.c b/drivers/mmc/core/core.c
index b6c16704aaab..7385acfa1dd9 100644
--- a/drivers/mmc/core/core.c
+++ b/drivers/mmc/core/core.c
@@ -501,9 +501,9 @@ void mmc_detect_change(struct mmc_host *host, unsigned long delay)
 {
 #ifdef CONFIG_MMC_DEBUG
 	unsigned long flags;
-	spin_lock_irqsave(host->lock, flags);
+	spin_lock_irqsave(&host->lock, flags);
 	BUG_ON(host->removed);
-	spin_unlock_irqrestore(host->lock, flags);
+	spin_unlock_irqrestore(&host->lock, flags);
 #endif
 
 	mmc_schedule_delayed_work(&host->detect, delay);
diff --git a/drivers/mtd/chips/Kconfig b/drivers/mtd/chips/Kconfig
index d28e0fc85e12..479d32b57a1e 100644
--- a/drivers/mtd/chips/Kconfig
+++ b/drivers/mtd/chips/Kconfig
@@ -1,5 +1,4 @@
 # drivers/mtd/chips/Kconfig
-# $Id: Kconfig,v 1.18 2005/11/07 11:14:22 gleixner Exp $
 
 menu "RAM/ROM/Flash chip drivers"
 	depends on MTD!=n
@@ -231,45 +230,6 @@ config MTD_ABSENT
 	  the system regardless of media presence.  Device nodes created
 	  with this driver will return -ENODEV upon access.
 
-config MTD_OBSOLETE_CHIPS
-	bool "Older (theoretically obsoleted now) drivers for non-CFI chips"
-	help
-	  This option does not enable any code directly, but will allow you to
-	  select some other chip drivers which are now considered obsolete,
-	  because the generic CONFIG_JEDECPROBE code above should now detect
-	  the chips which are supported by these drivers, and allow the generic
-	  CFI-compatible drivers to drive the chips. Say 'N' here unless you have
-	  already tried the CONFIG_JEDECPROBE method and reported its failure
-	  to the MTD mailing list at <linux-mtd@lists.infradead.org>
-
-config MTD_AMDSTD
-	tristate "AMD compatible flash chip support (non-CFI)"
-	depends on MTD_OBSOLETE_CHIPS && BROKEN
-	help
-	  This option enables support for flash chips using AMD-compatible
-	  commands, including some which are not CFI-compatible and hence
-	  cannot be used with the CONFIG_MTD_CFI_AMDSTD option.
-
-	  It also works on AMD compatible chips that do conform to CFI.
-
-config MTD_SHARP
-	tristate "pre-CFI Sharp chip support"
-	depends on MTD_OBSOLETE_CHIPS
-	help
-	  This option enables support for flash chips using Sharp-compatible
-	  commands, including some which are not CFI-compatible and hence
-	  cannot be used with the CONFIG_MTD_CFI_INTELxxx options.
-
-config MTD_JEDEC
-	tristate "JEDEC device support"
-	depends on MTD_OBSOLETE_CHIPS && BROKEN
-	help
-	  Enable older JEDEC flash interface devices for self
-	  programming flash.  It is commonly used in older AMD chips.  It is
-	  only called JEDEC because the JEDEC association
-	  <http://www.jedec.org/> distributes the identification codes for the
-	  chips.
-
 config MTD_XIP
 	bool "XIP aware MTD support"
 	depends on !SMP && (MTD_CFI_INTELEXT || MTD_CFI_AMDSTD) && EXPERIMENTAL && ARCH_MTD_XIP
diff --git a/drivers/mtd/chips/Makefile b/drivers/mtd/chips/Makefile
index 75bc1c2a0f43..36582412ccda 100644
--- a/drivers/mtd/chips/Makefile
+++ b/drivers/mtd/chips/Makefile
@@ -1,19 +1,15 @@
 #
 # linux/drivers/chips/Makefile
 #
-# $Id: Makefile.common,v 1.5 2005/11/07 11:14:22 gleixner Exp $
 
 obj-$(CONFIG_MTD)		+= chipreg.o
-obj-$(CONFIG_MTD_AMDSTD)	+= amd_flash.o
 obj-$(CONFIG_MTD_CFI)		+= cfi_probe.o
 obj-$(CONFIG_MTD_CFI_UTIL)	+= cfi_util.o
 obj-$(CONFIG_MTD_CFI_STAA)	+= cfi_cmdset_0020.o
 obj-$(CONFIG_MTD_CFI_AMDSTD)	+= cfi_cmdset_0002.o
 obj-$(CONFIG_MTD_CFI_INTELEXT)	+= cfi_cmdset_0001.o
 obj-$(CONFIG_MTD_GEN_PROBE)	+= gen_probe.o
-obj-$(CONFIG_MTD_JEDEC)		+= jedec.o
 obj-$(CONFIG_MTD_JEDECPROBE)	+= jedec_probe.o
 obj-$(CONFIG_MTD_RAM)		+= map_ram.o
 obj-$(CONFIG_MTD_ROM)		+= map_rom.o
-obj-$(CONFIG_MTD_SHARP)		+= sharp.o
 obj-$(CONFIG_MTD_ABSENT)	+= map_absent.o
diff --git a/drivers/mtd/chips/amd_flash.c b/drivers/mtd/chips/amd_flash.c
deleted file mode 100644
index e7999f15d85a..000000000000
--- a/drivers/mtd/chips/amd_flash.c
+++ /dev/null
@@ -1,1396 +0,0 @@
-/*
- * MTD map driver for AMD compatible flash chips (non-CFI)
- *
- * Author: Jonas Holmberg <jonas.holmberg@axis.com>
- *
- * $Id: amd_flash.c,v 1.28 2005/11/07 11:14:22 gleixner Exp $
- *
- * Copyright (c) 2001 Axis Communications AB
- *
- * This file is under GPL.
- *
- */
-
-#include <linux/module.h>
-#include <linux/types.h>
-#include <linux/kernel.h>
-#include <linux/sched.h>
-#include <linux/errno.h>
-#include <linux/slab.h>
-#include <linux/delay.h>
-#include <linux/interrupt.h>
-#include <linux/init.h>
-#include <linux/mtd/map.h>
-#include <linux/mtd/mtd.h>
-#include <linux/mtd/flashchip.h>
-
-/* There's no limit. It exists only to avoid realloc. */
-#define MAX_AMD_CHIPS 8
-
-#define DEVICE_TYPE_X8	(8 / 8)
-#define DEVICE_TYPE_X16	(16 / 8)
-#define DEVICE_TYPE_X32	(32 / 8)
-
-/* Addresses */
-#define ADDR_MANUFACTURER		0x0000
-#define ADDR_DEVICE_ID			0x0001
-#define ADDR_SECTOR_LOCK		0x0002
-#define ADDR_HANDSHAKE			0x0003
-#define ADDR_UNLOCK_1			0x0555
-#define ADDR_UNLOCK_2			0x02AA
-
-/* Commands */
-#define CMD_UNLOCK_DATA_1		0x00AA
-#define CMD_UNLOCK_DATA_2		0x0055
-#define CMD_MANUFACTURER_UNLOCK_DATA	0x0090
-#define CMD_UNLOCK_BYPASS_MODE		0x0020
-#define CMD_PROGRAM_UNLOCK_DATA		0x00A0
-#define CMD_RESET_DATA			0x00F0
-#define CMD_SECTOR_ERASE_UNLOCK_DATA	0x0080
-#define CMD_SECTOR_ERASE_UNLOCK_DATA_2	0x0030
-
-#define CMD_UNLOCK_SECTOR		0x0060
-
-/* Manufacturers */
-#define MANUFACTURER_AMD	0x0001
-#define MANUFACTURER_ATMEL	0x001F
-#define MANUFACTURER_FUJITSU	0x0004
-#define MANUFACTURER_ST		0x0020
-#define MANUFACTURER_SST	0x00BF
-#define MANUFACTURER_TOSHIBA	0x0098
-
-/* AMD */
-#define AM29F800BB	0x2258
-#define AM29F800BT	0x22D6
-#define AM29LV800BB	0x225B
-#define AM29LV800BT	0x22DA
-#define AM29LV160DT	0x22C4
-#define AM29LV160DB	0x2249
-#define AM29BDS323D     0x22D1
-
-/* Atmel */
-#define AT49xV16x	0x00C0
-#define AT49xV16xT	0x00C2
-
-/* Fujitsu */
-#define MBM29LV160TE	0x22C4
-#define MBM29LV160BE	0x2249
-#define MBM29LV800BB	0x225B
-
-/* ST - www.st.com */
-#define M29W800T	0x00D7
-#define M29W160DT	0x22C4
-#define M29W160DB	0x2249
-
-/* SST */
-#define SST39LF800	0x2781
-#define SST39LF160	0x2782
-
-/* Toshiba */
-#define TC58FVT160	0x00C2
-#define TC58FVB160	0x0043
-
-#define D6_MASK	0x40
-
-struct amd_flash_private {
-	int device_type;
-	int interleave;
-	int numchips;
-	unsigned long chipshift;
-	struct flchip chips[0];
-};
-
-struct amd_flash_info {
-	const __u16 mfr_id;
-	const __u16 dev_id;
-	const char *name;
-	const u_long size;
-	const int numeraseregions;
-	const struct mtd_erase_region_info regions[4];
-};
-
-
-
-static int amd_flash_read(struct mtd_info *, loff_t, size_t, size_t *,
-			  u_char *);
-static int amd_flash_write(struct mtd_info *, loff_t, size_t, size_t *,
-			   const u_char *);
-static int amd_flash_erase(struct mtd_info *, struct erase_info *);
-static void amd_flash_sync(struct mtd_info *);
-static int amd_flash_suspend(struct mtd_info *);
-static void amd_flash_resume(struct mtd_info *);
-static void amd_flash_destroy(struct mtd_info *);
-static struct mtd_info *amd_flash_probe(struct map_info *map);
-
-
-static struct mtd_chip_driver amd_flash_chipdrv = {
-	.probe = amd_flash_probe,
-	.destroy = amd_flash_destroy,
-	.name = "amd_flash",
-	.module = THIS_MODULE
-};
-
-static inline __u32 wide_read(struct map_info *map, __u32 addr)
-{
-	if (map->buswidth == 1) {
-		return map_read8(map, addr);
-	} else if (map->buswidth == 2) {
-		return map_read16(map, addr);
-	} else if (map->buswidth == 4) {
-		return map_read32(map, addr);
-        }
-
-	return 0;
-}
-
-static inline void wide_write(struct map_info *map, __u32 val, __u32 addr)
-{
-	if (map->buswidth == 1) {
-		map_write8(map, val, addr);
-	} else if (map->buswidth == 2) {
-		map_write16(map, val, addr);
-	} else if (map->buswidth == 4) {
-		map_write32(map, val, addr);
-	}
-}
-
-static inline __u32 make_cmd(struct map_info *map, __u32 cmd)
-{
-	const struct amd_flash_private *private = map->fldrv_priv;
-	if ((private->interleave == 2) &&
-	    (private->device_type == DEVICE_TYPE_X16)) {
-		cmd |= (cmd << 16);
-	}
-
-	return cmd;
-}
-
-static inline void send_unlock(struct map_info *map, unsigned long base)
-{
-	wide_write(map, (CMD_UNLOCK_DATA_1 << 16) | CMD_UNLOCK_DATA_1,
-		   base + (map->buswidth * ADDR_UNLOCK_1));
-	wide_write(map, (CMD_UNLOCK_DATA_2 << 16) | CMD_UNLOCK_DATA_2,
-		   base + (map->buswidth * ADDR_UNLOCK_2));
-}
-
-static inline void send_cmd(struct map_info *map, unsigned long base, __u32 cmd)
-{
-	send_unlock(map, base);
-	wide_write(map, make_cmd(map, cmd),
-		   base + (map->buswidth * ADDR_UNLOCK_1));
-}
-
-static inline void send_cmd_to_addr(struct map_info *map, unsigned long base,
-				    __u32 cmd, unsigned long addr)
-{
-	send_unlock(map, base);
-	wide_write(map, make_cmd(map, cmd), addr);
-}
-
-static inline int flash_is_busy(struct map_info *map, unsigned long addr,
-				int interleave)
-{
-
-	if ((interleave == 2) && (map->buswidth == 4)) {
-		__u32 read1, read2;
-
-		read1 = wide_read(map, addr);
-		read2 = wide_read(map, addr);
-
-		return (((read1 >> 16) & D6_MASK) !=
-			((read2 >> 16) & D6_MASK)) ||
-		       (((read1 & 0xffff) & D6_MASK) !=
-			((read2 & 0xffff) & D6_MASK));
-	}
-
-	return ((wide_read(map, addr) & D6_MASK) !=
-		(wide_read(map, addr) & D6_MASK));
-}
-
-static inline void unlock_sector(struct map_info *map, unsigned long sect_addr,
-				 int unlock)
-{
-	/* Sector lock address. A6 = 1 for unlock, A6 = 0 for lock */
-	int SLA = unlock ?
-		(sect_addr |  (0x40 * map->buswidth)) :
-		(sect_addr & ~(0x40 * map->buswidth)) ;
-
-	__u32 cmd = make_cmd(map, CMD_UNLOCK_SECTOR);
-
-	wide_write(map, make_cmd(map, CMD_RESET_DATA), 0);
-	wide_write(map, cmd, SLA); /* 1st cycle: write cmd to any address */
-	wide_write(map, cmd, SLA); /* 2nd cycle: write cmd to any address */
-	wide_write(map, cmd, SLA); /* 3rd cycle: write cmd to SLA */
-}
-
-static inline int is_sector_locked(struct map_info *map,
-				   unsigned long sect_addr)
-{
-	int status;
-
-	wide_write(map, CMD_RESET_DATA, 0);
-	send_cmd(map, sect_addr, CMD_MANUFACTURER_UNLOCK_DATA);
-
-	/* status is 0x0000 for unlocked and 0x0001 for locked */
-	status = wide_read(map, sect_addr + (map->buswidth * ADDR_SECTOR_LOCK));
-	wide_write(map, CMD_RESET_DATA, 0);
-	return status;
-}
-
-static int amd_flash_do_unlock(struct mtd_info *mtd, loff_t ofs, size_t len,
-			       int is_unlock)
-{
-	struct map_info *map;
-	struct mtd_erase_region_info *merip;
-	int eraseoffset, erasesize, eraseblocks;
-	int i;
-	int retval = 0;
-	int lock_status;
-
-	map = mtd->priv;
-
-	/* Pass the whole chip through sector by sector and check for each
-	   sector if the sector and the given interval overlap */
-	for(i = 0; i < mtd->numeraseregions; i++) {
-		merip = &mtd->eraseregions[i];
-
-		eraseoffset = merip->offset;
-		erasesize = merip->erasesize;
-		eraseblocks = merip->numblocks;
-
-		if (ofs > eraseoffset + erasesize)
-			continue;
-
-		while (eraseblocks > 0) {
-			if (ofs < eraseoffset + erasesize && ofs + len > eraseoffset) {
-				unlock_sector(map, eraseoffset, is_unlock);
-
-				lock_status = is_sector_locked(map, eraseoffset);
-
-				if (is_unlock && lock_status) {
-					printk("Cannot unlock sector at address %x length %xx\n",
-					       eraseoffset, merip->erasesize);
-					retval = -1;
-				} else if (!is_unlock && !lock_status) {
-					printk("Cannot lock sector at address %x length %x\n",
-					       eraseoffset, merip->erasesize);
-					retval = -1;
-				}
-			}
-			eraseoffset += erasesize;
-			eraseblocks --;
-		}
-	}
-	return retval;
-}
-
-static int amd_flash_unlock(struct mtd_info *mtd, loff_t ofs, size_t len)
-{
-	return amd_flash_do_unlock(mtd, ofs, len, 1);
-}
-
-static int amd_flash_lock(struct mtd_info *mtd, loff_t ofs, size_t len)
-{
-	return amd_flash_do_unlock(mtd, ofs, len, 0);
-}
-
-
-/*
- * Reads JEDEC manufacturer ID and device ID and returns the index of the first
- * matching table entry (-1 if not found or alias for already found chip).
- */
-static int probe_new_chip(struct mtd_info *mtd, __u32 base,
-			  struct flchip *chips,
-			  struct amd_flash_private *private,
-			  const struct amd_flash_info *table, int table_size)
-{
-	__u32 mfr_id;
-	__u32 dev_id;
-	struct map_info *map = mtd->priv;
-	struct amd_flash_private temp;
-	int i;
-
-	temp.device_type = DEVICE_TYPE_X16;	// Assume X16 (FIXME)
-	temp.interleave = 2;
-	map->fldrv_priv = &temp;
-
-	/* Enter autoselect mode. */
-	send_cmd(map, base, CMD_RESET_DATA);
-	send_cmd(map, base, CMD_MANUFACTURER_UNLOCK_DATA);
-
-	mfr_id = wide_read(map, base + (map->buswidth * ADDR_MANUFACTURER));
-	dev_id = wide_read(map, base + (map->buswidth * ADDR_DEVICE_ID));
-
-	if ((map->buswidth == 4) && ((mfr_id >> 16) == (mfr_id & 0xffff)) &&
-	    ((dev_id >> 16) == (dev_id & 0xffff))) {
-		mfr_id &= 0xffff;
-		dev_id &= 0xffff;
-	} else {
-		temp.interleave = 1;
-	}
-
-	for (i = 0; i < table_size; i++) {
-		if ((mfr_id == table[i].mfr_id) &&
-		    (dev_id == table[i].dev_id)) {
-			if (chips) {
-				int j;
-
-				/* Is this an alias for an already found chip?
-				 * In that case that chip should be in
-				 * autoselect mode now.
-				 */
-				for (j = 0; j < private->numchips; j++) {
-					__u32 mfr_id_other;
-					__u32 dev_id_other;
-
-					mfr_id_other =
-						wide_read(map, chips[j].start +
-							       (map->buswidth *
-								ADDR_MANUFACTURER
-							       ));
-					dev_id_other =
-						wide_read(map, chips[j].start +
-					    		       (map->buswidth *
-							        ADDR_DEVICE_ID));
-					if (temp.interleave == 2) {
-						mfr_id_other &= 0xffff;
-						dev_id_other &= 0xffff;
-					}
-					if ((mfr_id_other == mfr_id) &&
-					    (dev_id_other == dev_id)) {
-
-						/* Exit autoselect mode. */
-						send_cmd(map, base,
-							 CMD_RESET_DATA);
-
-						return -1;
-					}
-				}
-
-				if (private->numchips == MAX_AMD_CHIPS) {
-					printk(KERN_WARNING
-					       "%s: Too many flash chips "
-					       "detected. Increase "
-					       "MAX_AMD_CHIPS from %d.\n",
-					       map->name, MAX_AMD_CHIPS);
-
-					return -1;
-				}
-
-				chips[private->numchips].start = base;
-				chips[private->numchips].state = FL_READY;
-				chips[private->numchips].mutex =
-					&chips[private->numchips]._spinlock;
-				private->numchips++;
-			}
-
-			printk("%s: Found %d x %ldMiB %s at 0x%x\n", map->name,
-			       temp.interleave, (table[i].size)/(1024*1024),
-			       table[i].name, base);
-
-			mtd->size += table[i].size * temp.interleave;
-			mtd->numeraseregions += table[i].numeraseregions;
-
-			break;
-		}
-	}
-
-	/* Exit autoselect mode. */
-	send_cmd(map, base, CMD_RESET_DATA);
-
-	if (i == table_size) {
-		printk(KERN_DEBUG "%s: unknown flash device at 0x%x, "
-		       "mfr id 0x%x, dev id 0x%x\n", map->name,
-		       base, mfr_id, dev_id);
-		map->fldrv_priv = NULL;
-
-		return -1;
-	}
-
-	private->device_type = temp.device_type;
-	private->interleave = temp.interleave;
-
-	return i;
-}
-
-
-
-static struct mtd_info *amd_flash_probe(struct map_info *map)
-{
-	static const struct amd_flash_info table[] = {
-	{
-		.mfr_id = MANUFACTURER_AMD,
-		.dev_id = AM29LV160DT,
-		.name = "AMD AM29LV160DT",
-		.size = 0x00200000,
-		.numeraseregions = 4,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x10000, .numblocks = 31 },
-			{ .offset = 0x1F0000, .erasesize = 0x08000, .numblocks =  1 },
-			{ .offset = 0x1F8000, .erasesize = 0x02000, .numblocks =  2 },
-			{ .offset = 0x1FC000, .erasesize = 0x04000, .numblocks =  1 }
-		}
-	}, {
-		.mfr_id = MANUFACTURER_AMD,
-		.dev_id = AM29LV160DB,
-		.name = "AMD AM29LV160DB",
-		.size = 0x00200000,
-		.numeraseregions = 4,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x04000, .numblocks =  1 },
-			{ .offset = 0x004000, .erasesize = 0x02000, .numblocks =  2 },
-			{ .offset = 0x008000, .erasesize = 0x08000, .numblocks =  1 },
-			{ .offset = 0x010000, .erasesize = 0x10000, .numblocks = 31 }
-		}
-	}, {
-		.mfr_id = MANUFACTURER_TOSHIBA,
-		.dev_id = TC58FVT160,
-		.name = "Toshiba TC58FVT160",
-		.size = 0x00200000,
-		.numeraseregions = 4,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x10000, .numblocks = 31 },
-			{ .offset = 0x1F0000, .erasesize = 0x08000, .numblocks =  1 },
-			{ .offset = 0x1F8000, .erasesize = 0x02000, .numblocks =  2 },
-			{ .offset = 0x1FC000, .erasesize = 0x04000, .numblocks =  1 }
-		}
-	}, {
-		.mfr_id = MANUFACTURER_FUJITSU,
-		.dev_id = MBM29LV160TE,
-		.name = "Fujitsu MBM29LV160TE",
-		.size = 0x00200000,
-		.numeraseregions = 4,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x10000, .numblocks = 31 },
-			{ .offset = 0x1F0000, .erasesize = 0x08000, .numblocks =  1 },
-			{ .offset = 0x1F8000, .erasesize = 0x02000, .numblocks =  2 },
-			{ .offset = 0x1FC000, .erasesize = 0x04000, .numblocks =  1 }
-		}
-	}, {
-		.mfr_id = MANUFACTURER_TOSHIBA,
-		.dev_id = TC58FVB160,
-		.name = "Toshiba TC58FVB160",
-		.size = 0x00200000,
-		.numeraseregions = 4,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x04000, .numblocks =  1 },
-			{ .offset = 0x004000, .erasesize = 0x02000, .numblocks =  2 },
-			{ .offset = 0x008000, .erasesize = 0x08000, .numblocks =  1 },
-			{ .offset = 0x010000, .erasesize = 0x10000, .numblocks = 31 }
-		}
-	}, {
-		.mfr_id = MANUFACTURER_FUJITSU,
-		.dev_id = MBM29LV160BE,
-		.name = "Fujitsu MBM29LV160BE",
-		.size = 0x00200000,
-		.numeraseregions = 4,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x04000, .numblocks =  1 },
-			{ .offset = 0x004000, .erasesize = 0x02000, .numblocks =  2 },
-			{ .offset = 0x008000, .erasesize = 0x08000, .numblocks =  1 },
-			{ .offset = 0x010000, .erasesize = 0x10000, .numblocks = 31 }
-		}
-	}, {
-		.mfr_id = MANUFACTURER_AMD,
-		.dev_id = AM29LV800BB,
-		.name = "AMD AM29LV800BB",
-		.size = 0x00100000,
-		.numeraseregions = 4,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x04000, .numblocks =  1 },
-			{ .offset = 0x004000, .erasesize = 0x02000, .numblocks =  2 },
-			{ .offset = 0x008000, .erasesize = 0x08000, .numblocks =  1 },
-			{ .offset = 0x010000, .erasesize = 0x10000, .numblocks = 15 }
-		}
-	}, {
-		.mfr_id = MANUFACTURER_AMD,
-		.dev_id = AM29F800BB,
-		.name = "AMD AM29F800BB",
-		.size = 0x00100000,
-		.numeraseregions = 4,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x04000, .numblocks =  1 },
-			{ .offset = 0x004000, .erasesize = 0x02000, .numblocks =  2 },
-			{ .offset = 0x008000, .erasesize = 0x08000, .numblocks =  1 },
-			{ .offset = 0x010000, .erasesize = 0x10000, .numblocks = 15 }
-		}
-	}, {
-		.mfr_id = MANUFACTURER_AMD,
-		.dev_id = AM29LV800BT,
-		.name = "AMD AM29LV800BT",
-		.size = 0x00100000,
-		.numeraseregions = 4,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x10000, .numblocks = 15 },
-			{ .offset = 0x0F0000, .erasesize = 0x08000, .numblocks =  1 },
-			{ .offset = 0x0F8000, .erasesize = 0x02000, .numblocks =  2 },
-			{ .offset = 0x0FC000, .erasesize = 0x04000, .numblocks =  1 }
-		}
-	}, {
-		.mfr_id = MANUFACTURER_AMD,
-		.dev_id = AM29F800BT,
-		.name = "AMD AM29F800BT",
-		.size = 0x00100000,
-		.numeraseregions = 4,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x10000, .numblocks = 15 },
-			{ .offset = 0x0F0000, .erasesize = 0x08000, .numblocks =  1 },
-			{ .offset = 0x0F8000, .erasesize = 0x02000, .numblocks =  2 },
-			{ .offset = 0x0FC000, .erasesize = 0x04000, .numblocks =  1 }
-		}
-	}, {
-		.mfr_id = MANUFACTURER_AMD,
-		.dev_id = AM29LV800BB,
-		.name = "AMD AM29LV800BB",
-		.size = 0x00100000,
-		.numeraseregions = 4,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x10000, .numblocks = 15 },
-			{ .offset = 0x0F0000, .erasesize = 0x08000, .numblocks =  1 },
-			{ .offset = 0x0F8000, .erasesize = 0x02000, .numblocks =  2 },
-			{ .offset = 0x0FC000, .erasesize = 0x04000, .numblocks =  1 }
-		}
-	}, {
-		.mfr_id = MANUFACTURER_FUJITSU,
-		.dev_id = MBM29LV800BB,
-		.name = "Fujitsu MBM29LV800BB",
-		.size = 0x00100000,
-		.numeraseregions = 4,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x04000, .numblocks =  1 },
-			{ .offset = 0x004000, .erasesize = 0x02000, .numblocks =  2 },
-			{ .offset = 0x008000, .erasesize = 0x08000, .numblocks =  1 },
-			{ .offset = 0x010000, .erasesize = 0x10000, .numblocks = 15 }
-		}
-	}, {
-		.mfr_id = MANUFACTURER_ST,
-		.dev_id = M29W800T,
-		.name = "ST M29W800T",
-		.size = 0x00100000,
-		.numeraseregions = 4,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x10000, .numblocks = 15 },
-			{ .offset = 0x0F0000, .erasesize = 0x08000, .numblocks =  1 },
-			{ .offset = 0x0F8000, .erasesize = 0x02000, .numblocks =  2 },
-			{ .offset = 0x0FC000, .erasesize = 0x04000, .numblocks =  1 }
-		}
-	}, {
-		.mfr_id = MANUFACTURER_ST,
-		.dev_id = M29W160DT,
-		.name = "ST M29W160DT",
-		.size = 0x00200000,
-		.numeraseregions = 4,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x10000, .numblocks = 31 },
-			{ .offset = 0x1F0000, .erasesize = 0x08000, .numblocks =  1 },
-			{ .offset = 0x1F8000, .erasesize = 0x02000, .numblocks =  2 },
-			{ .offset = 0x1FC000, .erasesize = 0x04000, .numblocks =  1 }
-		}
-	}, {
-		.mfr_id = MANUFACTURER_ST,
-		.dev_id = M29W160DB,
-		.name = "ST M29W160DB",
-		.size = 0x00200000,
-		.numeraseregions = 4,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x04000, .numblocks =  1 },
-			{ .offset = 0x004000, .erasesize = 0x02000, .numblocks =  2 },
-			{ .offset = 0x008000, .erasesize = 0x08000, .numblocks =  1 },
-			{ .offset = 0x010000, .erasesize = 0x10000, .numblocks = 31 }
-		}
-	}, {
-		.mfr_id = MANUFACTURER_AMD,
-		.dev_id = AM29BDS323D,
-		.name = "AMD AM29BDS323D",
-		.size = 0x00400000,
-		.numeraseregions = 3,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x10000, .numblocks = 48 },
-			{ .offset = 0x300000, .erasesize = 0x10000, .numblocks = 15 },
-			{ .offset = 0x3f0000, .erasesize = 0x02000, .numblocks =  8 },
-		}
-	}, {
-		.mfr_id = MANUFACTURER_ATMEL,
-		.dev_id = AT49xV16x,
-		.name = "Atmel AT49xV16x",
-		.size = 0x00200000,
-		.numeraseregions = 2,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x02000, .numblocks =  8 },
-			{ .offset = 0x010000, .erasesize = 0x10000, .numblocks = 31 }
-		}
-	}, {
-		.mfr_id = MANUFACTURER_ATMEL,
-		.dev_id = AT49xV16xT,
-		.name = "Atmel AT49xV16xT",
-		.size = 0x00200000,
-		.numeraseregions = 2,
-		.regions = {
-			{ .offset = 0x000000, .erasesize = 0x10000, .numblocks = 31 },
-			{ .offset = 0x1F0000, .erasesize = 0x02000, .numblocks =  8 }
-		}
-	}
-	};
-
-	struct mtd_info *mtd;
-	struct flchip chips[MAX_AMD_CHIPS];
-	int table_pos[MAX_AMD_CHIPS];
-	struct amd_flash_private temp;
-	struct amd_flash_private *private;
-	u_long size;
-	unsigned long base;
-	int i;
-	int reg_idx;
-	int offset;
-
-	mtd = kzalloc(sizeof(*mtd), GFP_KERNEL);
-	if (!mtd) {
-		printk(KERN_WARNING
-		       "%s: kmalloc failed for info structure\n", map->name);
-		return NULL;
-	}
-	mtd->priv = map;
-
-	memset(&temp, 0, sizeof(temp));
-
-	printk("%s: Probing for AMD compatible flash...\n", map->name);
-
-	if ((table_pos[0] = probe_new_chip(mtd, 0, NULL, &temp, table,
-					   ARRAY_SIZE(table)))
-	    == -1) {
-		printk(KERN_WARNING
-		       "%s: Found no AMD compatible device at location zero\n",
-		       map->name);
-		kfree(mtd);
-
-		return NULL;
-	}
-
-	chips[0].start = 0;
-	chips[0].state = FL_READY;
-	chips[0].mutex = &chips[0]._spinlock;
-	temp.numchips = 1;
-	for (size = mtd->size; size > 1; size >>= 1) {
-		temp.chipshift++;
-	}
-	switch (temp.interleave) {
-		case 2:
-			temp.chipshift += 1;
-			break;
-		case 4:
-			temp.chipshift += 2;
-			break;
-	}
-
-	/* Find out if there are any more chips in the map. */
-	for (base = (1 << temp.chipshift);
-	     base < map->size;
-	     base += (1 << temp.chipshift)) {
-	     	int numchips = temp.numchips;
-		table_pos[numchips] = probe_new_chip(mtd, base, chips,
-			&temp, table, ARRAY_SIZE(table));
-	}
-
-	mtd->eraseregions = kmalloc(sizeof(struct mtd_erase_region_info) *
-				    mtd->numeraseregions, GFP_KERNEL);
-	if (!mtd->eraseregions) {
-		printk(KERN_WARNING "%s: Failed to allocate "
-		       "memory for MTD erase region info\n", map->name);
-		kfree(mtd);
-		map->fldrv_priv = NULL;
-		return NULL;
-	}
-
-	reg_idx = 0;
-	offset = 0;
-	for (i = 0; i < temp.numchips; i++) {
-		int dev_size;
-		int j;
-
-		dev_size = 0;
-		for (j = 0; j < table[table_pos[i]].numeraseregions; j++) {
-			mtd->eraseregions[reg_idx].offset = offset +
-				(table[table_pos[i]].regions[j].offset *
-				 temp.interleave);
-			mtd->eraseregions[reg_idx].erasesize =
-				table[table_pos[i]].regions[j].erasesize *
-				temp.interleave;
-			mtd->eraseregions[reg_idx].numblocks =
-				table[table_pos[i]].regions[j].numblocks;
-			if (mtd->erasesize <
-			    mtd->eraseregions[reg_idx].erasesize) {
-				mtd->erasesize =
-					mtd->eraseregions[reg_idx].erasesize;
-			}
-			dev_size += mtd->eraseregions[reg_idx].erasesize *
-				    mtd->eraseregions[reg_idx].numblocks;
-			reg_idx++;
-		}
-		offset += dev_size;
-	}
-	mtd->type = MTD_NORFLASH;
-	mtd->writesize = 1;
-	mtd->flags = MTD_CAP_NORFLASH;
-	mtd->name = map->name;
-	mtd->erase = amd_flash_erase;
-	mtd->read = amd_flash_read;
-	mtd->write = amd_flash_write;
-	mtd->sync = amd_flash_sync;
-	mtd->suspend = amd_flash_suspend;
-	mtd->resume = amd_flash_resume;
-	mtd->lock = amd_flash_lock;
-	mtd->unlock = amd_flash_unlock;
-
-	private = kmalloc(sizeof(*private) + (sizeof(struct flchip) *
-					      temp.numchips), GFP_KERNEL);
-	if (!private) {
-		printk(KERN_WARNING
-		       "%s: kmalloc failed for private structure\n", map->name);
-		kfree(mtd);
-		map->fldrv_priv = NULL;
-		return NULL;
-	}
-	memcpy(private, &temp, sizeof(temp));
-	memcpy(private->chips, chips,
-	       sizeof(struct flchip) * private->numchips);
-	for (i = 0; i < private->numchips; i++) {
-		init_waitqueue_head(&private->chips[i].wq);
-		spin_lock_init(&private->chips[i]._spinlock);
-	}
-
-	map->fldrv_priv = private;
-
-	map->fldrv = &amd_flash_chipdrv;
-
-	__module_get(THIS_MODULE);
-	return mtd;
-}
-
-
-
-static inline int read_one_chip(struct map_info *map, struct flchip *chip,
-			       loff_t adr, size_t len, u_char *buf)
-{
-	DECLARE_WAITQUEUE(wait, current);
-	unsigned long timeo = jiffies + HZ;
-
-retry:
-	spin_lock_bh(chip->mutex);
-
-	if (chip->state != FL_READY){
-		printk(KERN_INFO "%s: waiting for chip to read, state = %d\n",
-		       map->name, chip->state);
-		set_current_state(TASK_UNINTERRUPTIBLE);
-		add_wait_queue(&chip->wq, &wait);
-
-		spin_unlock_bh(chip->mutex);
-
-		schedule();
-		remove_wait_queue(&chip->wq, &wait);
-
-		if(signal_pending(current)) {
-			return -EINTR;
-		}
-
-		timeo = jiffies + HZ;
-
-		goto retry;
-	}
-
-	adr += chip->start;
-
-	chip->state = FL_READY;
-
-	map_copy_from(map, buf, adr, len);
-
-	wake_up(&chip->wq);
-	spin_unlock_bh(chip->mutex);
-
-	return 0;
-}
-
-
-
-static int amd_flash_read(struct mtd_info *mtd, loff_t from, size_t len,
-			  size_t *retlen, u_char *buf)
-{
-	struct map_info *map = mtd->priv;
-	struct amd_flash_private *private = map->fldrv_priv;
-	unsigned long ofs;
-	int chipnum;
-	int ret = 0;
-
-	if ((from + len) > mtd->size) {
-		printk(KERN_WARNING "%s: read request past end of device "
-		       "(0x%lx)\n", map->name, (unsigned long)from + len);
-
-		return -EINVAL;
-	}
-
-	/* Offset within the first chip that the first read should start. */
-	chipnum = (from >> private->chipshift);
-	ofs = from - (chipnum <<  private->chipshift);
-
-	*retlen = 0;
-
-	while (len) {
-		unsigned long this_len;
-
-		if (chipnum >= private->numchips) {
-			break;
-		}
-
-		if ((len + ofs - 1) >> private->chipshift) {
-			this_len = (1 << private->chipshift) - ofs;
-		} else {
-			this_len = len;
-		}
-
-		ret = read_one_chip(map, &private->chips[chipnum], ofs,
-				    this_len, buf);
-		if (ret) {
-			break;
-		}
-
-		*retlen += this_len;
-		len -= this_len;
-		buf += this_len;
-
-		ofs = 0;
-		chipnum++;
-	}
-
-	return ret;
-}
-
-
-
-static int write_one_word(struct map_info *map, struct flchip *chip,
-			  unsigned long adr, __u32 datum)
-{
-	unsigned long timeo = jiffies + HZ;
-	struct amd_flash_private *private = map->fldrv_priv;
-	DECLARE_WAITQUEUE(wait, current);
-	int ret = 0;
-	int times_left;
-
-retry:
-	spin_lock_bh(chip->mutex);
-
-	if (chip->state != FL_READY){
-		printk("%s: waiting for chip to write, state = %d\n",
-		       map->name, chip->state);
-		set_current_state(TASK_UNINTERRUPTIBLE);
-		add_wait_queue(&chip->wq, &wait);
-
-		spin_unlock_bh(chip->mutex);
-
-		schedule();
-		remove_wait_queue(&chip->wq, &wait);
-		printk(KERN_INFO "%s: woke up to write\n", map->name);
-		if(signal_pending(current))
-			return -EINTR;
-
-		timeo = jiffies + HZ;
-
-		goto retry;
-	}
-
-	chip->state = FL_WRITING;
-
-	adr += chip->start;
-	ENABLE_VPP(map);
-	send_cmd(map, chip->start, CMD_PROGRAM_UNLOCK_DATA);
-	wide_write(map, datum, adr);
-
-	times_left = 500000;
-	while (times_left-- && flash_is_busy(map, adr, private->interleave)) {
-		if (need_resched()) {
-			spin_unlock_bh(chip->mutex);
-			schedule();
-			spin_lock_bh(chip->mutex);
-		}
-	}
-
-	if (!times_left) {
-		printk(KERN_WARNING "%s: write to 0x%lx timed out!\n",
-		       map->name, adr);
-		ret = -EIO;
-	} else {
-		__u32 verify;
-		if ((verify = wide_read(map, adr)) != datum) {
-			printk(KERN_WARNING "%s: write to 0x%lx failed. "
-			       "datum = %x, verify = %x\n",
-			       map->name, adr, datum, verify);
-			ret = -EIO;
-		}
-	}
-
-	DISABLE_VPP(map);
-	chip->state = FL_READY;
-	wake_up(&chip->wq);
-	spin_unlock_bh(chip->mutex);
-
-	return ret;
-}
-
-
-
-static int amd_flash_write(struct mtd_info *mtd, loff_t to , size_t len,
-			   size_t *retlen, const u_char *buf)
-{
-	struct map_info *map = mtd->priv;
-	struct amd_flash_private *private = map->fldrv_priv;
-	int ret = 0;
-	int chipnum;
-	unsigned long ofs;
-	unsigned long chipstart;
-
-	*retlen = 0;
-	if (!len) {
-		return 0;
-	}
-
-	chipnum = to >> private->chipshift;
-	ofs = to  - (chipnum << private->chipshift);
-	chipstart = private->chips[chipnum].start;
-
-	/* If it's not bus-aligned, do the first byte write. */
-	if (ofs & (map->buswidth - 1)) {
-		unsigned long bus_ofs = ofs & ~(map->buswidth - 1);
-		int i = ofs - bus_ofs;
-		int n = 0;
-		u_char tmp_buf[4];
-		__u32 datum;
-
-		map_copy_from(map, tmp_buf,
-			       bus_ofs + private->chips[chipnum].start,
-			       map->buswidth);
-		while (len && i < map->buswidth)
-			tmp_buf[i++] = buf[n++], len--;
-
-		if (map->buswidth == 2) {
-			datum = *(__u16*)tmp_buf;
-		} else if (map->buswidth == 4) {
-			datum = *(__u32*)tmp_buf;
-		} else {
-			return -EINVAL;  /* should never happen, but be safe */
-		}
-
-		ret = write_one_word(map, &private->chips[chipnum], bus_ofs,
-				     datum);
-		if (ret) {
-			return ret;
-		}
-
-		ofs += n;
-		buf += n;
-		(*retlen) += n;
-
-		if (ofs >> private->chipshift) {
-			chipnum++;
-			ofs = 0;
-			if (chipnum == private->numchips) {
-				return 0;
-			}
-		}
-	}
-
-	/* We are now aligned, write as much as possible. */
-	while(len >= map->buswidth) {
-		__u32 datum;
-
-		if (map->buswidth == 1) {
-			datum = *(__u8*)buf;
-		} else if (map->buswidth == 2) {
-			datum = *(__u16*)buf;
-		} else if (map->buswidth == 4) {
-			datum = *(__u32*)buf;
-		} else {
-			return -EINVAL;
-		}
-
-		ret = write_one_word(map, &private->chips[chipnum], ofs, datum);
-
-		if (ret) {
-			return ret;
-		}
-
-		ofs += map->buswidth;
-		buf += map->buswidth;
-		(*retlen) += map->buswidth;
-		len -= map->buswidth;
-
-		if (ofs >> private->chipshift) {
-			chipnum++;
-			ofs = 0;
-			if (chipnum == private->numchips) {
-				return 0;
-			}
-			chipstart = private->chips[chipnum].start;
-		}
-	}
-
-	if (len & (map->buswidth - 1)) {
-		int i = 0, n = 0;
-		u_char tmp_buf[2];
-		__u32 datum;
-
-		map_copy_from(map, tmp_buf,
-			       ofs + private->chips[chipnum].start,
-			       map->buswidth);
-		while (len--) {
-			tmp_buf[i++] = buf[n++];
-		}
-
-		if (map->buswidth == 2) {
-			datum = *(__u16*)tmp_buf;
-		} else if (map->buswidth == 4) {
-			datum = *(__u32*)tmp_buf;
-		} else {
-			return -EINVAL;  /* should never happen, but be safe */
-		}
-
-		ret = write_one_word(map, &private->chips[chipnum], ofs, datum);
-
-		if (ret) {
-			return ret;
-		}
-
-		(*retlen) += n;
-	}
-
-	return 0;
-}
-
-
-
-static inline int erase_one_block(struct map_info *map, struct flchip *chip,
-				  unsigned long adr, u_long size)
-{
-	unsigned long timeo = jiffies + HZ;
-	struct amd_flash_private *private = map->fldrv_priv;
-	DECLARE_WAITQUEUE(wait, current);
-
-retry:
-	spin_lock_bh(chip->mutex);
-
-	if (chip->state != FL_READY){
-		set_current_state(TASK_UNINTERRUPTIBLE);
-		add_wait_queue(&chip->wq, &wait);
-
-		spin_unlock_bh(chip->mutex);
-
-		schedule();
-		remove_wait_queue(&chip->wq, &wait);
-
-		if (signal_pending(current)) {
-			return -EINTR;
-		}
-
-		timeo = jiffies + HZ;
-
-		goto retry;
-	}
-
-	chip->state = FL_ERASING;
-
-	adr += chip->start;
-	ENABLE_VPP(map);
-	send_cmd(map, chip->start, CMD_SECTOR_ERASE_UNLOCK_DATA);
-	send_cmd_to_addr(map, chip->start, CMD_SECTOR_ERASE_UNLOCK_DATA_2, adr);
-
-	timeo = jiffies + (HZ * 20);
-
-	spin_unlock_bh(chip->mutex);
-	msleep(1000);
-	spin_lock_bh(chip->mutex);
-
-	while (flash_is_busy(map, adr, private->interleave)) {
-
-		if (chip->state != FL_ERASING) {
-			/* Someone's suspended the erase. Sleep */
-			set_current_state(TASK_UNINTERRUPTIBLE);
-			add_wait_queue(&chip->wq, &wait);
-
-			spin_unlock_bh(chip->mutex);
-			printk(KERN_INFO "%s: erase suspended. Sleeping\n",
-			       map->name);
-			schedule();
-			remove_wait_queue(&chip->wq, &wait);
-
-			if (signal_pending(current)) {
-				return -EINTR;
-			}
-
-			timeo = jiffies + (HZ*2); /* FIXME */
-			spin_lock_bh(chip->mutex);
-			continue;
-		}
-
-		/* OK Still waiting */
-		if (time_after(jiffies, timeo)) {
-			chip->state = FL_READY;
-			spin_unlock_bh(chip->mutex);
-			printk(KERN_WARNING "%s: waiting for erase to complete "
-			       "timed out.\n", map->name);
-			DISABLE_VPP(map);
-
-			return -EIO;
-		}
-
-		/* Latency issues. Drop the lock, wait a while and retry */
-		spin_unlock_bh(chip->mutex);
-
-		if (need_resched())
-			schedule();
-		else
-			udelay(1);
-
-		spin_lock_bh(chip->mutex);
-	}
-
-	/* Verify every single word */
-	{
-		int address;
-		int error = 0;
-		__u8 verify;
-
-		for (address = adr; address < (adr + size); address++) {
-			if ((verify = map_read8(map, address)) != 0xFF) {
-				error = 1;
-				break;
-			}
-		}
-		if (error) {
-			chip->state = FL_READY;
-			spin_unlock_bh(chip->mutex);
-			printk(KERN_WARNING
-			       "%s: verify error at 0x%x, size %ld.\n",
-			       map->name, address, size);
-			DISABLE_VPP(map);
-
-			return -EIO;
-		}
-	}
-
-	DISABLE_VPP(map);
-	chip->state = FL_READY;
-	wake_up(&chip->wq);
-	spin_unlock_bh(chip->mutex);
-
-	return 0;
-}
-
-
-
-static int amd_flash_erase(struct mtd_info *mtd, struct erase_info *instr)
-{
-	struct map_info *map = mtd->priv;
-	struct amd_flash_private *private = map->fldrv_priv;
-	unsigned long adr, len;
-	int chipnum;
-	int ret = 0;
-	int i;
-	int first;
-	struct mtd_erase_region_info *regions = mtd->eraseregions;
-
-	if (instr->addr > mtd->size) {
-		return -EINVAL;
-	}
-
-	if ((instr->len + instr->addr) > mtd->size) {
-		return -EINVAL;
-	}
-
-	/* Check that both start and end of the requested erase are
-	 * aligned with the erasesize at the appropriate addresses.
-	 */
-
-	i = 0;
-
-        /* Skip all erase regions which are ended before the start of
-           the requested erase. Actually, to save on the calculations,
-           we skip to the first erase region which starts after the
-           start of the requested erase, and then go back one.
-        */
-
-        while ((i < mtd->numeraseregions) &&
-	       (instr->addr >= regions[i].offset)) {
-               i++;
-	}
-        i--;
-
-	/* OK, now i is pointing at the erase region in which this
-	 * erase request starts. Check the start of the requested
-	 * erase range is aligned with the erase size which is in
-	 * effect here.
-	 */
-
-	if (instr->addr & (regions[i].erasesize-1)) {
-		return -EINVAL;
-	}
-
-	/* Remember the erase region we start on. */
-
-	first = i;
-
-	/* Next, check that the end of the requested erase is aligned
-	 * with the erase region at that address.
-	 */
-
-	while ((i < mtd->numeraseregions) &&
-	       ((instr->addr + instr->len) >= regions[i].offset)) {
-                i++;
-	}
-
-	/* As before, drop back one to point at the region in which
-	 * the address actually falls.
-	 */
-
-	i--;
-
-	if ((instr->addr + instr->len) & (regions[i].erasesize-1)) {
-                return -EINVAL;
-	}
-
-	chipnum = instr->addr >> private->chipshift;
-	adr = instr->addr - (chipnum << private->chipshift);
-	len = instr->len;
-
-	i = first;
-
-	while (len) {
-		ret = erase_one_block(map, &private->chips[chipnum], adr,
-				      regions[i].erasesize);
-
-		if (ret) {
-			return ret;
-		}
-
-		adr += regions[i].erasesize;
-		len -= regions[i].erasesize;
-
-		if ((adr % (1 << private->chipshift)) ==
-		    ((regions[i].offset + (regions[i].erasesize *
-		    			   regions[i].numblocks))
-		     % (1 << private->chipshift))) {
-			i++;
-		}
-
-		if (adr >> private->chipshift) {
-			adr = 0;
-			chipnum++;
-			if (chipnum >= private->numchips) {
-				break;
-			}
-		}
-	}
-
-	instr->state = MTD_ERASE_DONE;
-	mtd_erase_callback(instr);
-
-	return 0;
-}
-
-
-
-static void amd_flash_sync(struct mtd_info *mtd)
-{
-	struct map_info *map = mtd->priv;
-	struct amd_flash_private *private = map->fldrv_priv;
-	int i;
-	struct flchip *chip;
-	int ret = 0;
-	DECLARE_WAITQUEUE(wait, current);
-
-	for (i = 0; !ret && (i < private->numchips); i++) {
-		chip = &private->chips[i];
-
-	retry:
-		spin_lock_bh(chip->mutex);
-
-		switch(chip->state) {
-		case FL_READY:
-		case FL_STATUS:
-		case FL_CFI_QUERY:
-		case FL_JEDEC_QUERY:
-			chip->oldstate = chip->state;
-			chip->state = FL_SYNCING;
-			/* No need to wake_up() on this state change -
-			 * as the whole point is that nobody can do anything
-			 * with the chip now anyway.
-			 */
-		case FL_SYNCING:
-			spin_unlock_bh(chip->mutex);
-			break;
-
-		default:
-			/* Not an idle state */
-			add_wait_queue(&chip->wq, &wait);
-
-			spin_unlock_bh(chip->mutex);
-
-			schedule();
-
-		        remove_wait_queue(&chip->wq, &wait);
-
-			goto retry;
-		}
-	}
-
-	/* Unlock the chips again */
-	for (i--; i >= 0; i--) {
-		chip = &private->chips[i];
-
-		spin_lock_bh(chip->mutex);
-
-		if (chip->state == FL_SYNCING) {
-			chip->state = chip->oldstate;
-			wake_up(&chip->wq);
-		}
-		spin_unlock_bh(chip->mutex);
-	}
-}
-
-
-
-static int amd_flash_suspend(struct mtd_info *mtd)
-{
-printk("amd_flash_suspend(): not implemented!\n");
-	return -EINVAL;
-}
-
-
-
-static void amd_flash_resume(struct mtd_info *mtd)
-{
-printk("amd_flash_resume(): not implemented!\n");
-}
-
-
-
-static void amd_flash_destroy(struct mtd_info *mtd)
-{
-	struct map_info *map = mtd->priv;
-	struct amd_flash_private *private = map->fldrv_priv;
-	kfree(private);
-}
-
-int __init amd_flash_init(void)
-{
-	register_mtd_chip_driver(&amd_flash_chipdrv);
-	return 0;
-}
-
-void __exit amd_flash_exit(void)
-{
-	unregister_mtd_chip_driver(&amd_flash_chipdrv);
-}
-
-module_init(amd_flash_init);
-module_exit(amd_flash_exit);
-
-MODULE_LICENSE("GPL");
-MODULE_AUTHOR("Jonas Holmberg <jonas.holmberg@axis.com>");
-MODULE_DESCRIPTION("Old MTD chip driver for AMD flash chips");
diff --git a/drivers/mtd/chips/jedec.c b/drivers/mtd/chips/jedec.c
deleted file mode 100644
index 14e57b2bf842..000000000000
--- a/drivers/mtd/chips/jedec.c
+++ /dev/null
@@ -1,935 +0,0 @@
-
-/* JEDEC Flash Interface.
- * This is an older type of interface for self programming flash. It is
- * commonly use in older AMD chips and is obsolete compared with CFI.
- * It is called JEDEC because the JEDEC association distributes the ID codes
- * for the chips.
- *
- * See the AMD flash databook for information on how to operate the interface.
- *
- * This code does not support anything wider than 8 bit flash chips, I am
- * not going to guess how to send commands to them, plus I expect they will
- * all speak CFI..
- *
- * $Id: jedec.c,v 1.22 2005/01/05 18:05:11 dwmw2 Exp $
- */
-
-#include <linux/init.h>
-#include <linux/module.h>
-#include <linux/kernel.h>
-#include <linux/slab.h>
-#include <linux/mtd/jedec.h>
-#include <linux/mtd/map.h>
-#include <linux/mtd/mtd.h>
-#include <linux/mtd/compatmac.h>
-
-static struct mtd_info *jedec_probe(struct map_info *);
-static int jedec_probe8(struct map_info *map,unsigned long base,
-		  struct jedec_private *priv);
-static int jedec_probe16(struct map_info *map,unsigned long base,
-		  struct jedec_private *priv);
-static int jedec_probe32(struct map_info *map,unsigned long base,
-		  struct jedec_private *priv);
-static void jedec_flash_chip_scan(struct jedec_private *priv,unsigned long start,
-			    unsigned long len);
-static int flash_erase(struct mtd_info *mtd, struct erase_info *instr);
-static int flash_write(struct mtd_info *mtd, loff_t start, size_t len,
-		       size_t *retlen, const u_char *buf);
-
-static unsigned long my_bank_size;
-
-/* Listing of parts and sizes. We need this table to learn the sector
-   size of the chip and the total length */
-static const struct JEDECTable JEDEC_table[] = {
-	{
-		.jedec		= 0x013D,
-		.name		= "AMD Am29F017D",
-		.size		= 2*1024*1024,
-		.sectorsize	= 64*1024,
-		.capabilities	= MTD_CAP_NORFLASH
-	},
-	{
-		.jedec		= 0x01AD,
-		.name		= "AMD Am29F016",
-		.size		= 2*1024*1024,
-		.sectorsize	= 64*1024,
-		.capabilities	= MTD_CAP_NORFLASH
-	},
-	{
-		.jedec		= 0x01D5,
-		.name		= "AMD Am29F080",
-		.size		= 1*1024*1024,
-		.sectorsize	= 64*1024,
-		.capabilities	= MTD_CAP_NORFLASH
-	},
-	{
-		.jedec		= 0x01A4,
-		.name		= "AMD Am29F040",
-		.size		= 512*1024,
-		.sectorsize	= 64*1024,
-		.capabilities	= MTD_CAP_NORFLASH
-	},
-	{
-		.jedec		= 0x20E3,
-		.name		= "AMD Am29W040B",
-		.size		= 512*1024,
-		.sectorsize	= 64*1024,
-		.capabilities	= MTD_CAP_NORFLASH
-	},
-	{
-		.jedec		= 0xC2AD,
-		.name		= "Macronix MX29F016",
-		.size		= 2*1024*1024,
-		.sectorsize	= 64*1024,
-		.capabilities	= MTD_CAP_NORFLASH
-	},
-	{ .jedec = 0x0 }
-};
-
-static const struct JEDECTable *jedec_idtoinf(__u8 mfr,__u8 id);
-static void jedec_sync(struct mtd_info *mtd) {};
-static int jedec_read(struct mtd_info *mtd, loff_t from, size_t len,
-		      size_t *retlen, u_char *buf);
-static int jedec_read_banked(struct mtd_info *mtd, loff_t from, size_t len,
-			     size_t *retlen, u_char *buf);
-
-static struct mtd_info *jedec_probe(struct map_info *map);
-
-
-
-static struct mtd_chip_driver jedec_chipdrv = {
-	.probe	= jedec_probe,
-	.name	= "jedec",
-	.module	= THIS_MODULE
-};
-
-/* Probe entry point */
-
-static struct mtd_info *jedec_probe(struct map_info *map)
-{
-   struct mtd_info *MTD;
-   struct jedec_private *priv;
-   unsigned long Base;
-   unsigned long SectorSize;
-   unsigned count;
-   unsigned I,Uniq;
-   char Part[200];
-   memset(&priv,0,sizeof(priv));
-
-   MTD = kzalloc(sizeof(struct mtd_info) + sizeof(struct jedec_private), GFP_KERNEL);
-   if (!MTD)
-	   return NULL;
-
-   priv = (struct jedec_private *)&MTD[1];
-
-   my_bank_size = map->size;
-
-   if (map->size/my_bank_size > MAX_JEDEC_CHIPS)
-   {
-      printk("mtd: Increase MAX_JEDEC_CHIPS, too many banks.\n");
-      kfree(MTD);
-      return NULL;
-   }
-
-   for (Base = 0; Base < map->size; Base += my_bank_size)
-   {
-      // Perhaps zero could designate all tests?
-      if (map->buswidth == 0)
-	 map->buswidth = 1;
-
-      if (map->buswidth == 1){
-	 if (jedec_probe8(map,Base,priv) == 0) {
-		 printk("did recognize jedec chip\n");
-		 kfree(MTD);
-	         return NULL;
-	 }
-      }
-      if (map->buswidth == 2)
-	 jedec_probe16(map,Base,priv);
-      if (map->buswidth == 4)
-	 jedec_probe32(map,Base,priv);
-   }
-
-   // Get the biggest sector size
-   SectorSize = 0;
-   for (I = 0; priv->chips[I].jedec != 0 && I < MAX_JEDEC_CHIPS; I++)
-   {
-	   //	   printk("priv->chips[%d].jedec is %x\n",I,priv->chips[I].jedec);
-	   //	   printk("priv->chips[%d].sectorsize is %lx\n",I,priv->chips[I].sectorsize);
-      if (priv->chips[I].sectorsize > SectorSize)
-	 SectorSize = priv->chips[I].sectorsize;
-   }
-
-   // Quickly ensure that the other sector sizes are factors of the largest
-   for (I = 0; priv->chips[I].jedec != 0 && I < MAX_JEDEC_CHIPS; I++)
-   {
-      if ((SectorSize/priv->chips[I].sectorsize)*priv->chips[I].sectorsize != SectorSize)
-      {
-	 printk("mtd: Failed. Device has incompatible mixed sector sizes\n");
-	 kfree(MTD);
-	 return NULL;
-      }
-   }
-
-   /* Generate a part name that includes the number of different chips and
-      other configuration information */
-   count = 1;
-   strlcpy(Part,map->name,sizeof(Part)-10);
-   strcat(Part," ");
-   Uniq = 0;
-   for (I = 0; priv->chips[I].jedec != 0 && I < MAX_JEDEC_CHIPS; I++)
-   {
-      const struct JEDECTable *JEDEC;
-
-      if (priv->chips[I+1].jedec == priv->chips[I].jedec)
-      {
-	 count++;
-	 continue;
-      }
-
-      // Locate the chip in the jedec table
-      JEDEC = jedec_idtoinf(priv->chips[I].jedec >> 8,priv->chips[I].jedec);
-      if (JEDEC == 0)
-      {
-	 printk("mtd: Internal Error, JEDEC not set\n");
-	 kfree(MTD);
-	 return NULL;
-      }
-
-      if (Uniq != 0)
-	 strcat(Part,",");
-      Uniq++;
-
-      if (count != 1)
-	 sprintf(Part+strlen(Part),"%x*[%s]",count,JEDEC->name);
-      else
-	 sprintf(Part+strlen(Part),"%s",JEDEC->name);
-      if (strlen(Part) > sizeof(Part)*2/3)
-	 break;
-      count = 1;
-   }
-
-   /* Determine if the chips are organized in a linear fashion, or if there
-      are empty banks. Note, the last bank does not count here, only the
-      first banks are important. Holes on non-bank boundaries can not exist
-      due to the way the detection algorithm works. */
-   if (priv->size < my_bank_size)
-      my_bank_size = priv->size;
-   priv->is_banked = 0;
-   //printk("priv->size is %x, my_bank_size is %x\n",priv->size,my_bank_size);
-   //printk("priv->bank_fill[0] is %x\n",priv->bank_fill[0]);
-   if (!priv->size) {
-	   printk("priv->size is zero\n");
-	   kfree(MTD);
-	   return NULL;
-   }
-   if (priv->size/my_bank_size) {
-	   if (priv->size/my_bank_size == 1) {
-		   priv->size = my_bank_size;
-	   }
-	   else {
-		   for (I = 0; I != priv->size/my_bank_size - 1; I++)
-		   {
-		      if (priv->bank_fill[I] != my_bank_size)
-			 priv->is_banked = 1;
-
-		      /* This even could be eliminated, but new de-optimized read/write
-			 functions have to be written */
-		      printk("priv->bank_fill[%d] is %lx, priv->bank_fill[0] is %lx\n",I,priv->bank_fill[I],priv->bank_fill[0]);
-		      if (priv->bank_fill[I] != priv->bank_fill[0])
-		      {
-			 printk("mtd: Failed. Cannot handle unsymmetric banking\n");
-			 kfree(MTD);
-			 return NULL;
-		      }
-		   }
-	   }
-   }
-   if (priv->is_banked == 1)
-      strcat(Part,", banked");
-
-   //   printk("Part: '%s'\n",Part);
-
-   memset(MTD,0,sizeof(*MTD));
-  // strlcpy(MTD->name,Part,sizeof(MTD->name));
-   MTD->name = map->name;
-   MTD->type = MTD_NORFLASH;
-   MTD->flags = MTD_CAP_NORFLASH;
-   MTD->writesize = 1;
-   MTD->erasesize = SectorSize*(map->buswidth);
-   //   printk("MTD->erasesize is %x\n",(unsigned int)MTD->erasesize);
-   MTD->size = priv->size;
-   //   printk("MTD->size is %x\n",(unsigned int)MTD->size);
-   //MTD->module = THIS_MODULE; // ? Maybe this should be the low level module?
-   MTD->erase = flash_erase;
-   if (priv->is_banked == 1)
-      MTD->read = jedec_read_banked;
-   else
-      MTD->read = jedec_read;
-   MTD->write = flash_write;
-   MTD->sync = jedec_sync;
-   MTD->priv = map;
-   map->fldrv_priv = priv;
-   map->fldrv = &jedec_chipdrv;
-   __module_get(THIS_MODULE);
-   return MTD;
-}
-
-/* Helper for the JEDEC function, JEDEC numbers all have odd parity */
-static int checkparity(u_char C)
-{
-   u_char parity = 0;
-   while (C != 0)
-   {
-      parity ^= C & 1;
-      C >>= 1;
-   }
-
-   return parity == 1;
-}
-
-
-/* Take an array of JEDEC numbers that represent interleved flash chips
-   and process them. Check to make sure they are good JEDEC numbers, look
-   them up and then add them to the chip list */
-static int handle_jedecs(struct map_info *map,__u8 *Mfg,__u8 *Id,unsigned Count,
-		  unsigned long base,struct jedec_private *priv)
-{
-   unsigned I,J;
-   unsigned long Size;
-   unsigned long SectorSize;
-   const struct JEDECTable *JEDEC;
-
-   // Test #2 JEDEC numbers exhibit odd parity
-   for (I = 0; I != Count; I++)
-   {
-      if (checkparity(Mfg[I]) == 0 || checkparity(Id[I]) == 0)
-	 return 0;
-   }
-
-   // Finally, just make sure all the chip sizes are the same
-   JEDEC = jedec_idtoinf(Mfg[0],Id[0]);
-
-   if (JEDEC == 0)
-   {
-      printk("mtd: Found JEDEC flash chip, but do not have a table entry for %x:%x\n",Mfg[0],Mfg[1]);
-      return 0;
-   }
-
-   Size = JEDEC->size;
-   SectorSize = JEDEC->sectorsize;
-   for (I = 0; I != Count; I++)
-   {
-      JEDEC = jedec_idtoinf(Mfg[0],Id[0]);
-      if (JEDEC == 0)
-      {
-	 printk("mtd: Found JEDEC flash chip, but do not have a table entry for %x:%x\n",Mfg[0],Mfg[1]);
-	 return 0;
-      }
-
-      if (Size != JEDEC->size || SectorSize != JEDEC->sectorsize)
-      {
-	 printk("mtd: Failed. Interleved flash does not have matching characteristics\n");
-	 return 0;
-      }
-   }
-
-   // Load the Chips
-   for (I = 0; I != MAX_JEDEC_CHIPS; I++)
-   {
-      if (priv->chips[I].jedec == 0)
-	 break;
-   }
-
-   if (I + Count > MAX_JEDEC_CHIPS)
-   {
-      printk("mtd: Device has too many chips. Increase MAX_JEDEC_CHIPS\n");
-      return 0;
-   }
-
-   // Add them to the table
-   for (J = 0; J != Count; J++)
-   {
-      unsigned long Bank;
-
-      JEDEC = jedec_idtoinf(Mfg[J],Id[J]);
-      priv->chips[I].jedec = (Mfg[J] << 8) | Id[J];
-      priv->chips[I].size = JEDEC->size;
-      priv->chips[I].sectorsize = JEDEC->sectorsize;
-      priv->chips[I].base = base + J;
-      priv->chips[I].datashift = J*8;
-      priv->chips[I].capabilities = JEDEC->capabilities;
-      priv->chips[I].offset = priv->size + J;
-
-      // log2 n :|
-      priv->chips[I].addrshift = 0;
-      for (Bank = Count; Bank != 1; Bank >>= 1, priv->chips[I].addrshift++);
-
-      // Determine how filled this bank is.
-      Bank = base & (~(my_bank_size-1));
-      if (priv->bank_fill[Bank/my_bank_size] < base +
-	  (JEDEC->size << priv->chips[I].addrshift) - Bank)
-	 priv->bank_fill[Bank/my_bank_size] =  base + (JEDEC->size << priv->chips[I].addrshift) - Bank;
-      I++;
-   }
-
-   priv->size += priv->chips[I-1].size*Count;
-
-   return priv->chips[I-1].size;
-}
-
-/* Lookup the chip information from the JEDEC ID table. */
-static const struct JEDECTable *jedec_idtoinf(__u8 mfr,__u8 id)
-{
-   __u16 Id = (mfr << 8) | id;
-   unsigned long I = 0;
-   for (I = 0; JEDEC_table[I].jedec != 0; I++)
-      if (JEDEC_table[I].jedec == Id)
-	 return JEDEC_table + I;
-   return NULL;
-}
-
-// Look for flash using an 8 bit bus interface
-static int jedec_probe8(struct map_info *map,unsigned long base,
-		  struct jedec_private *priv)
-{
-   #define flread(x) map_read8(map,base+x)
-   #define flwrite(v,x) map_write8(map,v,base+x)
-
-   const unsigned long AutoSel1 = 0xAA;
-   const unsigned long AutoSel2 = 0x55;
-   const unsigned long AutoSel3 = 0x90;
-   const unsigned long Reset = 0xF0;
-   __u32 OldVal;
-   __u8 Mfg[1];
-   __u8 Id[1];
-   unsigned I;
-   unsigned long Size;
-
-   // Wait for any write/erase operation to settle
-   OldVal = flread(base);
-   for (I = 0; OldVal != flread(base) && I < 10000; I++)
-      OldVal = flread(base);
-
-   // Reset the chip
-   flwrite(Reset,0x555);
-
-   // Send the sequence
-   flwrite(AutoSel1,0x555);
-   flwrite(AutoSel2,0x2AA);
-   flwrite(AutoSel3,0x555);
-
-   //  Get the JEDEC numbers
-   Mfg[0] = flread(0);
-   Id[0] = flread(1);
-   //   printk("Mfg is %x, Id is %x\n",Mfg[0],Id[0]);
-
-   Size = handle_jedecs(map,Mfg,Id,1,base,priv);
-   //   printk("handle_jedecs Size is %x\n",(unsigned int)Size);
-   if (Size == 0)
-   {
-      flwrite(Reset,0x555);
-      return 0;
-   }
-
-
-   // Reset.
-   flwrite(Reset,0x555);
-
-   return 1;
-
-   #undef flread
-   #undef flwrite
-}
-
-// Look for flash using a 16 bit bus interface (ie 2 8-bit chips)
-static int jedec_probe16(struct map_info *map,unsigned long base,
-		  struct jedec_private *priv)
-{
-   return 0;
-}
-
-// Look for flash using a 32 bit bus interface (ie 4 8-bit chips)
-static int jedec_probe32(struct map_info *map,unsigned long base,
-		  struct jedec_private *priv)
-{
-   #define flread(x) map_read32(map,base+((x)<<2))
-   #define flwrite(v,x) map_write32(map,v,base+((x)<<2))
-
-   const unsigned long AutoSel1 = 0xAAAAAAAA;
-   const unsigned long AutoSel2 = 0x55555555;
-   const unsigned long AutoSel3 = 0x90909090;
-   const unsigned long Reset = 0xF0F0F0F0;
-   __u32 OldVal;
-   __u8 Mfg[4];
-   __u8 Id[4];
-   unsigned I;
-   unsigned long Size;
-
-   // Wait for any write/erase operation to settle
-   OldVal = flread(base);
-   for (I = 0; OldVal != flread(base) && I < 10000; I++)
-      OldVal = flread(base);
-
-   // Reset the chip
-   flwrite(Reset,0x555);
-
-   // Send the sequence
-   flwrite(AutoSel1,0x555);
-   flwrite(AutoSel2,0x2AA);
-   flwrite(AutoSel3,0x555);
-
-   // Test #1, JEDEC numbers are readable from 0x??00/0x??01
-   if (flread(0) != flread(0x100) ||
-       flread(1) != flread(0x101))
-   {
-      flwrite(Reset,0x555);
-      return 0;
-   }
-
-   // Split up the JEDEC numbers
-   OldVal = flread(0);
-   for (I = 0; I != 4; I++)
-      Mfg[I] = (OldVal >> (I*8));
-   OldVal = flread(1);
-   for (I = 0; I != 4; I++)
-      Id[I] = (OldVal >> (I*8));
-
-   Size = handle_jedecs(map,Mfg,Id,4,base,priv);
-   if (Size == 0)
-   {
-      flwrite(Reset,0x555);
-      return 0;
-   }
-
-   /* Check if there is address wrap around within a single bank, if this
-      returns JEDEC numbers then we assume that it is wrap around. Notice
-      we call this routine with the JEDEC return still enabled, if two or
-      more flashes have a truncated address space the probe test will still
-      work */
-   if (base + (Size<<2)+0x555 < map->size &&
-       base + (Size<<2)+0x555 < (base & (~(my_bank_size-1))) + my_bank_size)
-   {
-      if (flread(base+Size) != flread(base+Size + 0x100) ||
-	  flread(base+Size + 1) != flread(base+Size + 0x101))
-      {
-	 jedec_probe32(map,base+Size,priv);
-      }
-   }
-
-   // Reset.
-   flwrite(0xF0F0F0F0,0x555);
-
-   return 1;
-
-   #undef flread
-   #undef flwrite
-}
-
-/* Linear read. */
-static int jedec_read(struct mtd_info *mtd, loff_t from, size_t len,
-		      size_t *retlen, u_char *buf)
-{
-   struct map_info *map = mtd->priv;
-
-   map_copy_from(map, buf, from, len);
-   *retlen = len;
-   return 0;
-}
-
-/* Banked read. Take special care to jump past the holes in the bank
-   mapping. This version assumes symetry in the holes.. */
-static int jedec_read_banked(struct mtd_info *mtd, loff_t from, size_t len,
-			     size_t *retlen, u_char *buf)
-{
-   struct map_info *map = mtd->priv;
-   struct jedec_private *priv = map->fldrv_priv;
-
-   *retlen = 0;
-   while (len > 0)
-   {
-      // Determine what bank and offset into that bank the first byte is
-      unsigned long bank = from & (~(priv->bank_fill[0]-1));
-      unsigned long offset = from & (priv->bank_fill[0]-1);
-      unsigned long get = len;
-      if (priv->bank_fill[0] - offset < len)
-	 get = priv->bank_fill[0] - offset;
-
-      bank /= priv->bank_fill[0];
-      map_copy_from(map,buf + *retlen,bank*my_bank_size + offset,get);
-
-      len -= get;
-      *retlen += get;
-      from += get;
-   }
-   return 0;
-}
-
-/* Pass the flags value that the flash return before it re-entered read
-   mode. */
-static void jedec_flash_failed(unsigned char code)
-{
-   /* Bit 5 being high indicates that there was an internal device
-      failure, erasure time limits exceeded or something */
-   if ((code & (1 << 5)) != 0)
-   {
-      printk("mtd: Internal Flash failure\n");
-      return;
-   }
-   printk("mtd: Programming didn't take\n");
-}
-
-/* This uses the erasure function described in the AMD Flash Handbook,
-   it will work for flashes with a fixed sector size only. Flashes with
-   a selection of sector sizes (ie the AMD Am29F800B) will need a different
-   routine. This routine tries to parallize erasing multiple chips/sectors
-   where possible */
-static int flash_erase(struct mtd_info *mtd, struct erase_info *instr)
-{
-   // Does IO to the currently selected chip
-   #define flread(x) map_read8(map,chip->base+((x)<<chip->addrshift))
-   #define flwrite(v,x) map_write8(map,v,chip->base+((x)<<chip->addrshift))
-
-   unsigned long Time = 0;
-   unsigned long NoTime = 0;
-   unsigned long start = instr->addr, len = instr->len;
-   unsigned int I;
-   struct map_info *map = mtd->priv;
-   struct jedec_private *priv = map->fldrv_priv;
-
-   // Verify the arguments..
-   if (start + len > mtd->size ||
-       (start % mtd->erasesize) != 0 ||
-       (len % mtd->erasesize) != 0 ||
-       (len/mtd->erasesize) == 0)
-      return -EINVAL;
-
-   jedec_flash_chip_scan(priv,start,len);
-
-   // Start the erase sequence on each chip
-   for (I = 0; priv->chips[I].jedec != 0 && I < MAX_JEDEC_CHIPS; I++)
-   {
-      unsigned long off;
-      struct jedec_flash_chip *chip = priv->chips + I;
-
-      if (chip->length == 0)
-	 continue;
-
-      if (chip->start + chip->length > chip->size)
-      {
-	 printk("DIE\n");
-	 return -EIO;
-      }
-
-      flwrite(0xF0,chip->start + 0x555);
-      flwrite(0xAA,chip->start + 0x555);
-      flwrite(0x55,chip->start + 0x2AA);
-      flwrite(0x80,chip->start + 0x555);
-      flwrite(0xAA,chip->start + 0x555);
-      flwrite(0x55,chip->start + 0x2AA);
-
-      /* Once we start selecting the erase sectors the delay between each
-         command must not exceed 50us or it will immediately start erasing
-         and ignore the other sectors */
-      for (off = 0; off < len; off += chip->sectorsize)
-      {
-	 // Check to make sure we didn't timeout
-	 flwrite(0x30,chip->start + off);
-	 if (off == 0)
-	    continue;
-	 if ((flread(chip->start + off) & (1 << 3)) != 0)
-	 {
-	    printk("mtd: Ack! We timed out the erase timer!\n");
-	    return -EIO;
-	 }
-      }
-   }
-
-   /* We could split this into a timer routine and return early, performing
-      background erasure.. Maybe later if the need warrents */
-
-   /* Poll the flash for erasure completion, specs say this can take as long
-      as 480 seconds to do all the sectors (for a 2 meg flash).
-      Erasure time is dependent on chip age, temp and wear.. */
-
-   /* This being a generic routine assumes a 32 bit bus. It does read32s
-      and bundles interleved chips into the same grouping. This will work
-      for all bus widths */
-   Time = 0;
-   NoTime = 0;
-   for (I = 0; priv->chips[I].jedec != 0 && I < MAX_JEDEC_CHIPS; I++)
-   {
-      struct jedec_flash_chip *chip = priv->chips + I;
-      unsigned long off = 0;
-      unsigned todo[4] = {0,0,0,0};
-      unsigned todo_left = 0;
-      unsigned J;
-
-      if (chip->length == 0)
-	 continue;
-
-      /* Find all chips in this data line, realistically this is all
-         or nothing up to the interleve count */
-      for (J = 0; priv->chips[J].jedec != 0 && J < MAX_JEDEC_CHIPS; J++)
-      {
-	 if ((priv->chips[J].base & (~((1<<chip->addrshift)-1))) ==
-	     (chip->base & (~((1<<chip->addrshift)-1))))
-	 {
-	    todo_left++;
-	    todo[priv->chips[J].base & ((1<<chip->addrshift)-1)] = 1;
-	 }
-      }
-
-      /*      printk("todo: %x %x %x %x\n",(short)todo[0],(short)todo[1],
-	      (short)todo[2],(short)todo[3]);
-      */
-      while (1)
-      {
-	 __u32 Last[4];
-	 unsigned long Count = 0;
-
-	 /* During erase bit 7 is held low and bit 6 toggles, we watch this,
-	    should it stop toggling or go high then the erase is completed,
-  	    or this is not really flash ;> */
-	 switch (map->buswidth) {
-	 case 1:
-	    Last[0] = map_read8(map,(chip->base >> chip->addrshift) + chip->start + off);
-	    Last[1] = map_read8(map,(chip->base >> chip->addrshift) + chip->start + off);
-	    Last[2] = map_read8(map,(chip->base >> chip->addrshift) + chip->start + off);
-	    break;
-	 case 2:
-	    Last[0] = map_read16(map,(chip->base >> chip->addrshift) + chip->start + off);
-	    Last[1] = map_read16(map,(chip->base >> chip->addrshift) + chip->start + off);
-	    Last[2] = map_read16(map,(chip->base >> chip->addrshift) + chip->start + off);
-	    break;
-	 case 3:
-	    Last[0] = map_read32(map,(chip->base >> chip->addrshift) + chip->start + off);
-	    Last[1] = map_read32(map,(chip->base >> chip->addrshift) + chip->start + off);
-	    Last[2] = map_read32(map,(chip->base >> chip->addrshift) + chip->start + off);
-	    break;
-	 }
-	 Count = 3;
-	 while (todo_left != 0)
-	 {
-	    for (J = 0; J != 4; J++)
-	    {
-	       __u8 Byte1 = (Last[(Count-1)%4] >> (J*8)) & 0xFF;
-	       __u8 Byte2 = (Last[(Count-2)%4] >> (J*8)) & 0xFF;
-	       __u8 Byte3 = (Last[(Count-3)%4] >> (J*8)) & 0xFF;
-	       if (todo[J] == 0)
-		  continue;
-
-	       if ((Byte1 & (1 << 7)) == 0 && Byte1 != Byte2)
-	       {
-//		  printk("Check %x %x %x\n",(short)J,(short)Byte1,(short)Byte2);
-		  continue;
-	       }
-
-	       if (Byte1 == Byte2)
-	       {
-		  jedec_flash_failed(Byte3);
-		  return -EIO;
-	       }
-
-	       todo[J] = 0;
-	       todo_left--;
-	    }
-
-/*	    if (NoTime == 0)
-	       Time += HZ/10 - schedule_timeout(HZ/10);*/
-	    NoTime = 0;
-
-	    switch (map->buswidth) {
-	    case 1:
-	       Last[Count % 4] = map_read8(map,(chip->base >> chip->addrshift) + chip->start + off);
-	      break;
-	    case 2:
-	       Last[Count % 4] = map_read16(map,(chip->base >> chip->addrshift) + chip->start + off);
-	      break;
-	    case 4:
-	       Last[Count % 4] = map_read32(map,(chip->base >> chip->addrshift) + chip->start + off);
-	      break;
-	    }
-	    Count++;
-
-/*	    // Count time, max of 15s per sector (according to AMD)
-	    if (Time > 15*len/mtd->erasesize*HZ)
-	    {
-	       printk("mtd: Flash Erase Timed out\n");
-	       return -EIO;
-	    }	    */
-	 }
-
-	 // Skip to the next chip if we used chip erase
-	 if (chip->length == chip->size)
-	    off = chip->size;
-	 else
-	    off += chip->sectorsize;
-
-	 if (off >= chip->length)
-	    break;
-	 NoTime = 1;
-      }
-
-      for (J = 0; priv->chips[J].jedec != 0 && J < MAX_JEDEC_CHIPS; J++)
-      {
-	 if ((priv->chips[J].base & (~((1<<chip->addrshift)-1))) ==
-	     (chip->base & (~((1<<chip->addrshift)-1))))
-	    priv->chips[J].length = 0;
-      }
-   }
-
-   //printk("done\n");
-   instr->state = MTD_ERASE_DONE;
-   mtd_erase_callback(instr);
-   return 0;
-
-   #undef flread
-   #undef flwrite
-}
-
-/* This is the simple flash writing function. It writes to every byte, in
-   sequence. It takes care of how to properly address the flash if
-   the flash is interleved. It can only be used if all the chips in the
-   array are identical!*/
-static int flash_write(struct mtd_info *mtd, loff_t start, size_t len,
-		       size_t *retlen, const u_char *buf)
-{
-   /* Does IO to the currently selected chip. It takes the bank addressing
-      base (which is divisible by the chip size) adds the necessary lower bits
-      of addrshift (interleave index) and then adds the control register index. */
-   #define flread(x) map_read8(map,base+(off&((1<<chip->addrshift)-1))+((x)<<chip->addrshift))
-   #define flwrite(v,x) map_write8(map,v,base+(off&((1<<chip->addrshift)-1))+((x)<<chip->addrshift))
-
-   struct map_info *map = mtd->priv;
-   struct jedec_private *priv = map->fldrv_priv;
-   unsigned long base;
-   unsigned long off;
-   size_t save_len = len;
-
-   if (start + len > mtd->size)
-      return -EIO;
-
-   //printk("Here");
-
-   //printk("flash_write: start is %x, len is %x\n",start,(unsigned long)len);
-   while (len != 0)
-   {
-      struct jedec_flash_chip *chip = priv->chips;
-      unsigned long bank;
-      unsigned long boffset;
-
-      // Compute the base of the flash.
-      off = ((unsigned long)start) % (chip->size << chip->addrshift);
-      base = start - off;
-
-      // Perform banked addressing translation.
-      bank = base & (~(priv->bank_fill[0]-1));
-      boffset = base & (priv->bank_fill[0]-1);
-      bank = (bank/priv->bank_fill[0])*my_bank_size;
-      base = bank + boffset;
-
-    //  printk("Flasing %X %X %X\n",base,chip->size,len);
-     // printk("off is %x, compare with %x\n",off,chip->size << chip->addrshift);
-
-      // Loop over this page
-      for (; off != (chip->size << chip->addrshift) && len != 0; start++, len--, off++,buf++)
-      {
-	 unsigned char oldbyte = map_read8(map,base+off);
-	 unsigned char Last[4];
-	 unsigned long Count = 0;
-
-	 if (oldbyte == *buf) {
-	//	 printk("oldbyte and *buf is %x,len is %x\n",oldbyte,len);
-	    continue;
-	 }
-	 if (((~oldbyte) & *buf) != 0)
-	    printk("mtd: warn: Trying to set a 0 to a 1\n");
-
-	 // Write
-	 flwrite(0xAA,0x555);
-	 flwrite(0x55,0x2AA);
-	 flwrite(0xA0,0x555);
-	 map_write8(map,*buf,base + off);
-	 Last[0] = map_read8(map,base + off);
-	 Last[1] = map_read8(map,base + off);
-	 Last[2] = map_read8(map,base + off);
-
-	 /* Wait for the flash to finish the operation. We store the last 4
-	    status bytes that have been retrieved so we can determine why
-	    it failed. The toggle bits keep toggling when there is a
-	    failure */
-	 for (Count = 3; Last[(Count - 1) % 4] != Last[(Count - 2) % 4] &&
-	      Count < 10000; Count++)
-	    Last[Count % 4] = map_read8(map,base + off);
-	 if (Last[(Count - 1) % 4] != *buf)
-	 {
-	    jedec_flash_failed(Last[(Count - 3) % 4]);
-	    return -EIO;
-	 }
-      }
-   }
-   *retlen = save_len;
-   return 0;
-}
-
-/* This is used to enhance the speed of the erase routine,
-   when things are being done to multiple chips it is possible to
-   parallize the operations, particularly full memory erases of multi
-   chip memories benifit */
-static void jedec_flash_chip_scan(struct jedec_private *priv,unsigned long start,
-		     unsigned long len)
-{
-   unsigned int I;
-
-   // Zero the records
-   for (I = 0; priv->chips[I].jedec != 0 && I < MAX_JEDEC_CHIPS; I++)
-      priv->chips[I].start = priv->chips[I].length = 0;
-
-   // Intersect the region with each chip
-   for (I = 0; priv->chips[I].jedec != 0 && I < MAX_JEDEC_CHIPS; I++)
-   {
-      struct jedec_flash_chip *chip = priv->chips + I;
-      unsigned long ByteStart;
-      unsigned long ChipEndByte = chip->offset + (chip->size << chip->addrshift);
-
-      // End is before this chip or the start is after it
-      if (start+len < chip->offset ||
-	  ChipEndByte - (1 << chip->addrshift) < start)
-	 continue;
-
-      if (start < chip->offset)
-      {
-	 ByteStart = chip->offset;
-	 chip->start = 0;
-      }
-      else
-      {
-	 chip->start = (start - chip->offset + (1 << chip->addrshift)-1) >> chip->addrshift;
-	 ByteStart = start;
-      }
-
-      if (start + len >= ChipEndByte)
-	 chip->length = (ChipEndByte - ByteStart) >> chip->addrshift;
-      else
-	 chip->length = (start + len - ByteStart + (1 << chip->addrshift)-1) >> chip->addrshift;
-   }
-}
-
-int __init jedec_init(void)
-{
-	register_mtd_chip_driver(&jedec_chipdrv);
-	return 0;
-}
-
-static void __exit jedec_exit(void)
-{
-	unregister_mtd_chip_driver(&jedec_chipdrv);
-}
-
-module_init(jedec_init);
-module_exit(jedec_exit);
-
-MODULE_LICENSE("GPL");
-MODULE_AUTHOR("Jason Gunthorpe <jgg@deltatee.com> et al.");
-MODULE_DESCRIPTION("Old MTD chip driver for JEDEC-compliant flash chips");
diff --git a/drivers/mtd/chips/sharp.c b/drivers/mtd/chips/sharp.c
deleted file mode 100644
index c9cd3d21ccfa..000000000000
--- a/drivers/mtd/chips/sharp.c
+++ /dev/null
@@ -1,601 +0,0 @@
-/*
- * MTD chip driver for pre-CFI Sharp flash chips
- *
- * Copyright 2000,2001 David A. Schleef <ds@schleef.org>
- *           2000,2001 Lineo, Inc.
- *
- * $Id: sharp.c,v 1.17 2005/11/29 14:28:28 gleixner Exp $
- *
- * Devices supported:
- *   LH28F016SCT Symmetrical block flash memory, 2Mx8
- *   LH28F008SCT Symmetrical block flash memory, 1Mx8
- *
- * Documentation:
- *   http://www.sharpmeg.com/datasheets/memic/flashcmp/
- *   http://www.sharpmeg.com/datasheets/memic/flashcmp/01symf/16m/016sctl9.pdf
- *   016sctl9.pdf
- *
- * Limitations:
- *   This driver only supports 4x1 arrangement of chips.
- *   Not tested on anything but PowerPC.
- */
-
-#include <linux/kernel.h>
-#include <linux/module.h>
-#include <linux/types.h>
-#include <linux/sched.h>
-#include <linux/errno.h>
-#include <linux/interrupt.h>
-#include <linux/mtd/map.h>
-#include <linux/mtd/mtd.h>
-#include <linux/mtd/cfi.h>
-#include <linux/delay.h>
-#include <linux/init.h>
-#include <linux/slab.h>
-
-#define CMD_RESET		0xffffffff
-#define CMD_READ_ID		0x90909090
-#define CMD_READ_STATUS		0x70707070
-#define CMD_CLEAR_STATUS	0x50505050
-#define CMD_BLOCK_ERASE_1	0x20202020
-#define CMD_BLOCK_ERASE_2	0xd0d0d0d0
-#define CMD_BYTE_WRITE		0x40404040
-#define CMD_SUSPEND		0xb0b0b0b0
-#define CMD_RESUME		0xd0d0d0d0
-#define CMD_SET_BLOCK_LOCK_1	0x60606060
-#define CMD_SET_BLOCK_LOCK_2	0x01010101
-#define CMD_SET_MASTER_LOCK_1	0x60606060
-#define CMD_SET_MASTER_LOCK_2	0xf1f1f1f1
-#define CMD_CLEAR_BLOCK_LOCKS_1	0x60606060
-#define CMD_CLEAR_BLOCK_LOCKS_2	0xd0d0d0d0
-
-#define SR_READY		0x80808080 // 1 = ready
-#define SR_ERASE_SUSPEND	0x40404040 // 1 = block erase suspended
-#define SR_ERROR_ERASE		0x20202020 // 1 = error in block erase or clear lock bits
-#define SR_ERROR_WRITE		0x10101010 // 1 = error in byte write or set lock bit
-#define	SR_VPP			0x08080808 // 1 = Vpp is low
-#define SR_WRITE_SUSPEND	0x04040404 // 1 = byte write suspended
-#define SR_PROTECT		0x02020202 // 1 = lock bit set
-#define SR_RESERVED		0x01010101
-
-#define SR_ERRORS (SR_ERROR_ERASE|SR_ERROR_WRITE|SR_VPP|SR_PROTECT)
-
-/* Configuration options */
-
-#undef AUTOUNLOCK  /* automatically unlocks blocks before erasing */
-
-static struct mtd_info *sharp_probe(struct map_info *);
-
-static int sharp_probe_map(struct map_info *map,struct mtd_info *mtd);
-
-static int sharp_read(struct mtd_info *mtd, loff_t from, size_t len,
-	size_t *retlen, u_char *buf);
-static int sharp_write(struct mtd_info *mtd, loff_t from, size_t len,
-	size_t *retlen, const u_char *buf);
-static int sharp_erase(struct mtd_info *mtd, struct erase_info *instr);
-static void sharp_sync(struct mtd_info *mtd);
-static int sharp_suspend(struct mtd_info *mtd);
-static void sharp_resume(struct mtd_info *mtd);
-static void sharp_destroy(struct mtd_info *mtd);
-
-static int sharp_write_oneword(struct map_info *map, struct flchip *chip,
-	unsigned long adr, __u32 datum);
-static int sharp_erase_oneblock(struct map_info *map, struct flchip *chip,
-	unsigned long adr);
-#ifdef AUTOUNLOCK
-static void sharp_unlock_oneblock(struct map_info *map, struct flchip *chip,
-	unsigned long adr);
-#endif
-
-
-struct sharp_info{
-	struct flchip *chip;
-	int bogus;
-	int chipshift;
-	int numchips;
-	struct flchip chips[1];
-};
-
-static void sharp_destroy(struct mtd_info *mtd);
-
-static struct mtd_chip_driver sharp_chipdrv = {
-	.probe		= sharp_probe,
-	.destroy	= sharp_destroy,
-	.name		= "sharp",
-	.module		= THIS_MODULE
-};
-
-
-static struct mtd_info *sharp_probe(struct map_info *map)
-{
-	struct mtd_info *mtd = NULL;
-	struct sharp_info *sharp = NULL;
-	int width;
-
-	mtd = kzalloc(sizeof(*mtd), GFP_KERNEL);
-	if(!mtd)
-		return NULL;
-
-	sharp = kzalloc(sizeof(*sharp), GFP_KERNEL);
-	if(!sharp) {
-		kfree(mtd);
-		return NULL;
-	}
-
-	width = sharp_probe_map(map,mtd);
-	if(!width){
-		kfree(mtd);
-		kfree(sharp);
-		return NULL;
-	}
-
-	mtd->priv = map;
-	mtd->type = MTD_NORFLASH;
-	mtd->erase = sharp_erase;
-	mtd->read = sharp_read;
-	mtd->write = sharp_write;
-	mtd->sync = sharp_sync;
-	mtd->suspend = sharp_suspend;
-	mtd->resume = sharp_resume;
-	mtd->flags = MTD_CAP_NORFLASH;
-	mtd->writesize = 1;
-	mtd->name = map->name;
-
-	sharp->chipshift = 23;
-	sharp->numchips = 1;
-	sharp->chips[0].start = 0;
-	sharp->chips[0].state = FL_READY;
-	sharp->chips[0].mutex = &sharp->chips[0]._spinlock;
-	sharp->chips[0].word_write_time = 0;
-	init_waitqueue_head(&sharp->chips[0].wq);
-	spin_lock_init(&sharp->chips[0]._spinlock);
-
-	map->fldrv = &sharp_chipdrv;
-	map->fldrv_priv = sharp;
-
-	__module_get(THIS_MODULE);
-	return mtd;
-}
-
-static inline void sharp_send_cmd(struct map_info *map, unsigned long cmd, unsigned long adr)
-{
-	map_word map_cmd;
-	map_cmd.x[0] = cmd;
-	map_write(map, map_cmd, adr);
-}
-
-static int sharp_probe_map(struct map_info *map,struct mtd_info *mtd)
-{
-	map_word tmp, read0, read4;
-	unsigned long base = 0;
-	int width = 4;
-
-	tmp = map_read(map, base+0);
-
-	sharp_send_cmd(map, CMD_READ_ID, base+0);
-
-	read0 = map_read(map, base+0);
-	read4 = map_read(map, base+4);
-	if(read0.x[0] == 0x89898989){
-		printk("Looks like sharp flash\n");
-		switch(read4.x[0]){
-		case 0xaaaaaaaa:
-		case 0xa0a0a0a0:
-			/* aa - LH28F016SCT-L95 2Mx8, 32 64k blocks*/
-			/* a0 - LH28F016SCT-Z4  2Mx8, 32 64k blocks*/
-			mtd->erasesize = 0x10000 * width;
-			mtd->size = 0x200000 * width;
-			return width;
-		case 0xa6a6a6a6:
-			/* a6 - LH28F008SCT-L12 1Mx8, 16 64k blocks*/
-			/* a6 - LH28F008SCR-L85 1Mx8, 16 64k blocks*/
-			mtd->erasesize = 0x10000 * width;
-			mtd->size = 0x100000 * width;
-			return width;
-#if 0
-		case 0x00000000: /* unknown */
-			/* XX - LH28F004SCT 512kx8, 8 64k blocks*/
-			mtd->erasesize = 0x10000 * width;
-			mtd->size = 0x80000 * width;
-			return width;
-#endif
-		default:
-			printk("Sort-of looks like sharp flash, 0x%08lx 0x%08lx\n",
-				read0.x[0], read4.x[0]);
-		}
-	}else if((map_read(map, base+0).x[0] == CMD_READ_ID)){
-		/* RAM, probably */
-		printk("Looks like RAM\n");
-		map_write(map, tmp, base+0);
-	}else{
-		printk("Doesn't look like sharp flash, 0x%08lx 0x%08lx\n",
-			read0.x[0], read4.x[0]);
-	}
-
-	return 0;
-}
-
-/* This function returns with the chip->mutex lock held. */
-static int sharp_wait(struct map_info *map, struct flchip *chip)
-{
-	int i;
-	map_word status;
-	unsigned long timeo = jiffies + HZ;
-	DECLARE_WAITQUEUE(wait, current);
-	int adr = 0;
-
-retry:
-	spin_lock_bh(chip->mutex);
-
-	switch(chip->state){
-	case FL_READY:
-		sharp_send_cmd(map, CMD_READ_STATUS, adr);
-		chip->state = FL_STATUS;
-	case FL_STATUS:
-		for(i=0;i<100;i++){
-			status = map_read(map, adr);
-			if((status.x[0] & SR_READY)==SR_READY)
-				break;
-			udelay(1);
-		}
-		break;
-	default:
-		printk("Waiting for chip\n");
-
-		set_current_state(TASK_INTERRUPTIBLE);
-		add_wait_queue(&chip->wq, &wait);
-
-		spin_unlock_bh(chip->mutex);
-
-		schedule();
-		remove_wait_queue(&chip->wq, &wait);
-
-		if(signal_pending(current))
-			return -EINTR;
-
-		timeo = jiffies + HZ;
-
-		goto retry;
-	}
-
-	sharp_send_cmd(map, CMD_RESET, adr);
-
-	chip->state = FL_READY;
-
-	return 0;
-}
-
-static void sharp_release(struct flchip *chip)
-{
-	wake_up(&chip->wq);
-	spin_unlock_bh(chip->mutex);
-}
-
-static int sharp_read(struct mtd_info *mtd, loff_t from, size_t len,
-	size_t *retlen, u_char *buf)
-{
-	struct map_info *map = mtd->priv;
-	struct sharp_info *sharp = map->fldrv_priv;
-	int chipnum;
-	int ret = 0;
-	int ofs = 0;
-
-	chipnum = (from >> sharp->chipshift);
-	ofs = from & ((1 << sharp->chipshift)-1);
-
-	*retlen = 0;
-
-	while(len){
-		unsigned long thislen;
-
-		if(chipnum>=sharp->numchips)
-			break;
-
-		thislen = len;
-		if(ofs+thislen >= (1<<sharp->chipshift))
-			thislen = (1<<sharp->chipshift) - ofs;
-
-		ret = sharp_wait(map,&sharp->chips[chipnum]);
-		if(ret<0)
-			break;
-
-		map_copy_from(map,buf,ofs,thislen);
-
-		sharp_release(&sharp->chips[chipnum]);
-
-		*retlen += thislen;
-		len -= thislen;
-		buf += thislen;
-
-		ofs = 0;
-		chipnum++;
-	}
-	return ret;
-}
-
-static int sharp_write(struct mtd_info *mtd, loff_t to, size_t len,
-	size_t *retlen, const u_char *buf)
-{
-	struct map_info *map = mtd->priv;
-	struct sharp_info *sharp = map->fldrv_priv;
-	int ret = 0;
-	int i,j;
-	int chipnum;
-	unsigned long ofs;
-	union { u32 l; unsigned char uc[4]; } tbuf;
-
-	*retlen = 0;
-
-	while(len){
-		tbuf.l = 0xffffffff;
-		chipnum = to >> sharp->chipshift;
-		ofs = to & ((1<<sharp->chipshift)-1);
-
-		j=0;
-		for(i=ofs&3;i<4 && len;i++){
-			tbuf.uc[i] = *buf;
-			buf++;
-			to++;
-			len--;
-			j++;
-		}
-		sharp_write_oneword(map, &sharp->chips[chipnum], ofs&~3, tbuf.l);
-		if(ret<0)
-			return ret;
-		(*retlen)+=j;
-	}
-
-	return 0;
-}
-
-static int sharp_write_oneword(struct map_info *map, struct flchip *chip,
-	unsigned long adr, __u32 datum)
-{
-	int ret;
-	int timeo;
-	int try;
-	int i;
-	map_word data, status;
-
-	status.x[0] = 0;
-	ret = sharp_wait(map,chip);
-
-	for(try=0;try<10;try++){
-		sharp_send_cmd(map, CMD_BYTE_WRITE, adr);
-		/* cpu_to_le32 -> hack to fix the writel be->le conversion */
-		data.x[0] = cpu_to_le32(datum);
-		map_write(map, data, adr);
-
-		chip->state = FL_WRITING;
-
-		timeo = jiffies + (HZ/2);
-
-		sharp_send_cmd(map, CMD_READ_STATUS, adr);
-		for(i=0;i<100;i++){
-			status = map_read(map, adr);
-			if((status.x[0] & SR_READY) == SR_READY)
-				break;
-		}
-		if(i==100){
-			printk("sharp: timed out writing\n");
-		}
-
-		if(!(status.x[0] & SR_ERRORS))
-			break;
-
-		printk("sharp: error writing byte at addr=%08lx status=%08lx\n", adr, status.x[0]);
-
-		sharp_send_cmd(map, CMD_CLEAR_STATUS, adr);
-	}
-	sharp_send_cmd(map, CMD_RESET, adr);
-	chip->state = FL_READY;
-
-	wake_up(&chip->wq);
-	spin_unlock_bh(chip->mutex);
-
-	return 0;
-}
-
-static int sharp_erase(struct mtd_info *mtd, struct erase_info *instr)
-{
-	struct map_info *map = mtd->priv;
-	struct sharp_info *sharp = map->fldrv_priv;
-	unsigned long adr,len;
-	int chipnum, ret=0;
-
-//printk("sharp_erase()\n");
-	if(instr->addr & (mtd->erasesize - 1))
-		return -EINVAL;
-	if(instr->len & (mtd->erasesize - 1))
-		return -EINVAL;
-	if(instr->len + instr->addr > mtd->size)
-		return -EINVAL;
-
-	chipnum = instr->addr >> sharp->chipshift;
-	adr = instr->addr & ((1<<sharp->chipshift)-1);
-	len = instr->len;
-
-	while(len){
-		ret = sharp_erase_oneblock(map, &sharp->chips[chipnum], adr);
-		if(ret)return ret;
-
-		adr += mtd->erasesize;
-		len -= mtd->erasesize;
-		if(adr >> sharp->chipshift){
-			adr = 0;
-			chipnum++;
-			if(chipnum>=sharp->numchips)
-				break;
-		}
-	}
-
-	instr->state = MTD_ERASE_DONE;
-	mtd_erase_callback(instr);
-
-	return 0;
-}
-
-static int sharp_do_wait_for_ready(struct map_info *map, struct flchip *chip,
-	unsigned long adr)
-{
-	int ret;
-	unsigned long timeo;
-	map_word status;
-	DECLARE_WAITQUEUE(wait, current);
-
-	sharp_send_cmd(map, CMD_READ_STATUS, adr);
-	status = map_read(map, adr);
-
-	timeo = jiffies + HZ;
-
-	while(time_before(jiffies, timeo)){
-		sharp_send_cmd(map, CMD_READ_STATUS, adr);
-		status = map_read(map, adr);
-		if((status.x[0] & SR_READY)==SR_READY){
-			ret = 0;
-			goto out;
-		}
-		set_current_state(TASK_INTERRUPTIBLE);
-		add_wait_queue(&chip->wq, &wait);
-
-		//spin_unlock_bh(chip->mutex);
-
-		schedule_timeout(1);
-		schedule();
-		remove_wait_queue(&chip->wq, &wait);
-
-		//spin_lock_bh(chip->mutex);
-
-		if (signal_pending(current)){
-			ret = -EINTR;
-			goto out;
-		}
-
-	}
-	ret = -ETIME;
-out:
-	return ret;
-}
-
-static int sharp_erase_oneblock(struct map_info *map, struct flchip *chip,
-	unsigned long adr)
-{
-	int ret;
-	//int timeo;
-	map_word status;
-	//int i;
-
-//printk("sharp_erase_oneblock()\n");
-
-#ifdef AUTOUNLOCK
-	/* This seems like a good place to do an unlock */
-	sharp_unlock_oneblock(map,chip,adr);
-#endif
-
-	sharp_send_cmd(map, CMD_BLOCK_ERASE_1, adr);
-	sharp_send_cmd(map, CMD_BLOCK_ERASE_2, adr);
-
-	chip->state = FL_ERASING;
-
-	ret = sharp_do_wait_for_ready(map,chip,adr);
-	if(ret<0)return ret;
-
-	sharp_send_cmd(map, CMD_READ_STATUS, adr);
-	status = map_read(map, adr);
-
-	if(!(status.x[0] & SR_ERRORS)){
-		sharp_send_cmd(map, CMD_RESET, adr);
-		chip->state = FL_READY;
-		//spin_unlock_bh(chip->mutex);
-		return 0;
-	}
-
-	printk("sharp: error erasing block at addr=%08lx status=%08lx\n", adr, status.x[0]);
-	sharp_send_cmd(map, CMD_CLEAR_STATUS, adr);
-
-	//spin_unlock_bh(chip->mutex);
-
-	return -EIO;
-}
-
-#ifdef AUTOUNLOCK
-static void sharp_unlock_oneblock(struct map_info *map, struct flchip *chip,
-	unsigned long adr)
-{
-	int i;
-	map_word status;
-
-	sharp_send_cmd(map, CMD_CLEAR_BLOCK_LOCKS_1, adr);
-	sharp_send_cmd(map, CMD_CLEAR_BLOCK_LOCKS_2, adr);
-
-	udelay(100);
-
-	status = map_read(map, adr);
-	printk("status=%08lx\n", status.x[0]);
-
-	for(i=0;i<1000;i++){
-		//sharp_send_cmd(map, CMD_READ_STATUS, adr);
-		status = map_read(map, adr);
-		if((status.x[0] & SR_READY) == SR_READY)
-			break;
-		udelay(100);
-	}
-	if(i==1000){
-		printk("sharp: timed out unlocking block\n");
-	}
-
-	if(!(status.x[0] & SR_ERRORS)){
-		sharp_send_cmd(map, CMD_RESET, adr);
-		chip->state = FL_READY;
-		return;
-	}
-
-	printk("sharp: error unlocking block at addr=%08lx status=%08lx\n", adr, status.x[0]);
-	sharp_send_cmd(map, CMD_CLEAR_STATUS, adr);
-}
-#endif
-
-static void sharp_sync(struct mtd_info *mtd)
-{
-	//printk("sharp_sync()\n");
-}
-
-static int sharp_suspend(struct mtd_info *mtd)
-{
-	printk("sharp_suspend()\n");
-	return -EINVAL;
-}
-
-static void sharp_resume(struct mtd_info *mtd)
-{
-	printk("sharp_resume()\n");
-
-}
-
-static void sharp_destroy(struct mtd_info *mtd)
-{
-	printk("sharp_destroy()\n");
-
-}
-
-static int __init sharp_probe_init(void)
-{
-	printk("MTD Sharp chip driver <ds@lineo.com>\n");
-
-	register_mtd_chip_driver(&sharp_chipdrv);
-
-	return 0;
-}
-
-static void __exit sharp_probe_exit(void)
-{
-	unregister_mtd_chip_driver(&sharp_chipdrv);
-}
-
-module_init(sharp_probe_init);
-module_exit(sharp_probe_exit);
-
-
-MODULE_LICENSE("GPL");
-MODULE_AUTHOR("David Schleef <ds@schleef.org>");
-MODULE_DESCRIPTION("Old MTD chip driver for pre-CFI Sharp flash chips");
diff --git a/drivers/mtd/devices/block2mtd.c b/drivers/mtd/devices/block2mtd.c
index fc4cc8ba9e29..be4b9948c762 100644
--- a/drivers/mtd/devices/block2mtd.c
+++ b/drivers/mtd/devices/block2mtd.c
@@ -373,7 +373,7 @@ static inline void kill_final_newline(char *str)
 
 #ifndef MODULE
 static int block2mtd_init_called = 0;
-static __initdata char block2mtd_paramline[80 + 12]; /* 80 for device, 12 for erase size */
+static char block2mtd_paramline[80 + 12]; /* 80 for device, 12 for erase size */
 #endif
 
 
diff --git a/drivers/mtd/maps/Kconfig b/drivers/mtd/maps/Kconfig
index d990d8141ef5..b665e4ac2208 100644
--- a/drivers/mtd/maps/Kconfig
+++ b/drivers/mtd/maps/Kconfig
@@ -60,7 +60,7 @@ config MTD_PHYSMAP_BANKWIDTH
 	  (i.e., run-time calling physmap_configure()).
 
 config MTD_PHYSMAP_OF
-	tristate "Flash device in physical memory map based on OF descirption"
+	tristate "Flash device in physical memory map based on OF description"
 	depends on PPC_OF && (MTD_CFI || MTD_JEDECPROBE || MTD_ROM)
 	help
 	  This provides a 'mapping' driver which allows the NOR Flash and
@@ -358,22 +358,6 @@ config MTD_CFI_FLAGADM
 	  Mapping for the Flaga digital module. If you don't have one, ignore
 	  this setting.
 
-config MTD_BEECH
-	tristate "CFI Flash device mapped on IBM 405LP Beech"
-	depends on MTD_CFI && BEECH
-	help
-	  This enables access routines for the flash chips on the IBM
-	  405LP Beech board. If you have one of these boards and would like
-	  to use the flash chips on it, say 'Y'.
-
-config MTD_ARCTIC
-	tristate "CFI Flash device mapped on IBM 405LP Arctic"
-	depends on MTD_CFI && ARCTIC2
-	help
-	  This enables access routines for the flash chips on the IBM 405LP
-	  Arctic board. If you have one of these boards and would like to
-	  use the flash chips on it, say 'Y'.
-
 config MTD_WALNUT
 	tristate "Flash device mapped on IBM 405GP Walnut"
 	depends on MTD_JEDECPROBE && WALNUT
diff --git a/drivers/mtd/maps/Makefile b/drivers/mtd/maps/Makefile
index de036c5e6139..3acbb5d01ca4 100644
--- a/drivers/mtd/maps/Makefile
+++ b/drivers/mtd/maps/Makefile
@@ -58,8 +58,6 @@ obj-$(CONFIG_MTD_NETtel)	+= nettel.o
 obj-$(CONFIG_MTD_SCB2_FLASH)	+= scb2_flash.o
 obj-$(CONFIG_MTD_EBONY)		+= ebony.o
 obj-$(CONFIG_MTD_OCOTEA)	+= ocotea.o
-obj-$(CONFIG_MTD_BEECH)		+= beech-mtd.o
-obj-$(CONFIG_MTD_ARCTIC)	+= arctic-mtd.o
 obj-$(CONFIG_MTD_WALNUT)        += walnut.o
 obj-$(CONFIG_MTD_H720X)		+= h720x-flash.o
 obj-$(CONFIG_MTD_SBC8240)	+= sbc8240.o
diff --git a/drivers/mtd/maps/arctic-mtd.c b/drivers/mtd/maps/arctic-mtd.c
deleted file mode 100644
index 2cc902436275..000000000000
--- a/drivers/mtd/maps/arctic-mtd.c
+++ /dev/null
@@ -1,145 +0,0 @@
-/*
- * $Id: arctic-mtd.c,v 1.14 2005/11/07 11:14:26 gleixner Exp $
- *
- * drivers/mtd/maps/arctic-mtd.c MTD mappings and partition tables for
- *                              IBM 405LP Arctic boards.
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
- *
- * Copyright (C) 2002, International Business Machines Corporation
- * All Rights Reserved.
- *
- * Bishop Brock
- * IBM Research, Austin Center for Low-Power Computing
- * bcbrock@us.ibm.com
- * March 2002
- *
- * modified for Arctic by,
- * David Gibson
- * IBM OzLabs, Canberra, Australia
- * <arctic@gibson.dropbear.id.au>
- */
-
-#include <linux/kernel.h>
-#include <linux/module.h>
-#include <linux/types.h>
-#include <linux/init.h>
-
-#include <linux/mtd/mtd.h>
-#include <linux/mtd/map.h>
-#include <linux/mtd/partitions.h>
-
-#include <asm/io.h>
-#include <asm/ibm4xx.h>
-
-/*
- * 0 : 0xFE00 0000 - 0xFEFF FFFF : Filesystem 1 (16MiB)
- * 1 : 0xFF00 0000 - 0xFF4F FFFF : kernel (5.12MiB)
- * 2 : 0xFF50 0000 - 0xFFF5 FFFF : Filesystem 2 (10.624MiB) (if non-XIP)
- * 3 : 0xFFF6 0000 - 0xFFFF FFFF : PIBS Firmware (640KiB)
- */
-
-#define FFS1_SIZE	0x01000000 /* 16MiB */
-#define KERNEL_SIZE	0x00500000 /* 5.12MiB */
-#define FFS2_SIZE	0x00a60000 /* 10.624MiB */
-#define FIRMWARE_SIZE	0x000a0000 /* 640KiB */
-
-
-#define NAME     	"Arctic Linux Flash"
-#define PADDR    	SUBZERO_BOOTFLASH_PADDR
-#define BUSWIDTH	2
-#define SIZE		SUBZERO_BOOTFLASH_SIZE
-#define PARTITIONS	4
-
-/* Flash memories on these boards are memory resources, accessed big-endian. */
-
-{
-  /* do nothing for now */
-}
-
-static struct map_info arctic_mtd_map = {
-	.name		= NAME,
-	.size		= SIZE,
-	.bankwidth	= BUSWIDTH,
-	.phys		= PADDR,
-};
-
-static struct mtd_info *arctic_mtd;
-
-static struct mtd_partition arctic_partitions[PARTITIONS] = {
-	{ .name		= "Filesystem",
-	  .size		= FFS1_SIZE,
-	  .offset	= 0,},
-        { .name		= "Kernel",
-	  .size		= KERNEL_SIZE,
-	  .offset	= FFS1_SIZE,},
-	{ .name		= "Filesystem",
-	  .size		= FFS2_SIZE,
-	  .offset	= FFS1_SIZE + KERNEL_SIZE,},
-	{ .name		= "Firmware",
-	  .size		= FIRMWARE_SIZE,
-	  .offset	= SUBZERO_BOOTFLASH_SIZE - FIRMWARE_SIZE,},
-};
-
-static int __init
-init_arctic_mtd(void)
-{
-	int err;
-
-	printk("%s: 0x%08x at 0x%08x\n", NAME, SIZE, PADDR);
-
-	arctic_mtd_map.virt = ioremap(PADDR, SIZE);
-
-	if (!arctic_mtd_map.virt) {
-		printk("%s: failed to ioremap 0x%x\n", NAME, PADDR);
-		return -EIO;
-	}
-	simple_map_init(&arctic_mtd_map);
-
-	printk("%s: probing %d-bit flash bus\n", NAME, BUSWIDTH * 8);
-	arctic_mtd = do_map_probe("cfi_probe", &arctic_mtd_map);
-
-	if (!arctic_mtd) {
-		iounmap(arctic_mtd_map.virt);
-		return -ENXIO;
-	}
-
-	arctic_mtd->owner = THIS_MODULE;
-
-	err = add_mtd_partitions(arctic_mtd, arctic_partitions, PARTITIONS);
-	if (err) {
-		printk("%s: add_mtd_partitions failed\n", NAME);
-		iounmap(arctic_mtd_map.virt);
-	}
-
-	return err;
-}
-
-static void __exit
-cleanup_arctic_mtd(void)
-{
-	if (arctic_mtd) {
-		del_mtd_partitions(arctic_mtd);
-		map_destroy(arctic_mtd);
-		iounmap((void *) arctic_mtd_map.virt);
-	}
-}
-
-module_init(init_arctic_mtd);
-module_exit(cleanup_arctic_mtd);
-
-MODULE_LICENSE("GPL");
-MODULE_AUTHOR("David Gibson <arctic@gibson.dropbear.id.au>");
-MODULE_DESCRIPTION("MTD map and partitions for IBM 405LP Arctic boards");
diff --git a/drivers/mtd/maps/beech-mtd.c b/drivers/mtd/maps/beech-mtd.c
deleted file mode 100644
index d76d5981b863..000000000000
--- a/drivers/mtd/maps/beech-mtd.c
+++ /dev/null
@@ -1,122 +0,0 @@
-/*
- * $Id: beech-mtd.c,v 1.11 2005/11/07 11:14:26 gleixner Exp $
- *
- * drivers/mtd/maps/beech-mtd.c MTD mappings and partition tables for
- *                              IBM 405LP Beech boards.
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
- *
- * Copyright (C) 2002, International Business Machines Corporation
- * All Rights Reserved.
- *
- * Bishop Brock
- * IBM Research, Austin Center for Low-Power Computing
- * bcbrock@us.ibm.com
- * March 2002
- *
- */
-
-#include <linux/kernel.h>
-#include <linux/module.h>
-#include <linux/types.h>
-#include <linux/init.h>
-
-#include <linux/mtd/mtd.h>
-#include <linux/mtd/map.h>
-#include <linux/mtd/partitions.h>
-
-#include <asm/io.h>
-#include <asm/ibm4xx.h>
-
-#define NAME     "Beech Linux Flash"
-#define PADDR    BEECH_BIGFLASH_PADDR
-#define SIZE     BEECH_BIGFLASH_SIZE
-#define BUSWIDTH 1
-
-/* Flash memories on these boards are memory resources, accessed big-endian. */
-
-
-static struct map_info beech_mtd_map = {
-	.name =		NAME,
-	.size =		SIZE,
-	.bankwidth =	BUSWIDTH,
-	.phys =		PADDR
-};
-
-static struct mtd_info *beech_mtd;
-
-static struct mtd_partition beech_partitions[2] = {
-	{
-	      .name = "Linux Kernel",
-	      .size = BEECH_KERNEL_SIZE,
-	      .offset = BEECH_KERNEL_OFFSET
-	}, {
-	      .name = "Free Area",
-	      .size = BEECH_FREE_AREA_SIZE,
-	      .offset = BEECH_FREE_AREA_OFFSET
-	}
-};
-
-static int __init
-init_beech_mtd(void)
-{
-	int err;
-
-	printk("%s: 0x%08x at 0x%08x\n", NAME, SIZE, PADDR);
-
-	beech_mtd_map.virt = ioremap(PADDR, SIZE);
-
-	if (!beech_mtd_map.virt) {
-		printk("%s: failed to ioremap 0x%x\n", NAME, PADDR);
-		return -EIO;
-	}
-
-	simple_map_init(&beech_mtd_map);
-
-	printk("%s: probing %d-bit flash bus\n", NAME, BUSWIDTH * 8);
-	beech_mtd = do_map_probe("cfi_probe", &beech_mtd_map);
-
-	if (!beech_mtd) {
-		iounmap(beech_mtd_map.virt);
-		return -ENXIO;
-	}
-
-	beech_mtd->owner = THIS_MODULE;
-
-	err = add_mtd_partitions(beech_mtd, beech_partitions, 2);
-	if (err) {
-		printk("%s: add_mtd_partitions failed\n", NAME);
-		iounmap(beech_mtd_map.virt);
-	}
-
-	return err;
-}
-
-static void __exit
-cleanup_beech_mtd(void)
-{
-	if (beech_mtd) {
-		del_mtd_partitions(beech_mtd);
-		map_destroy(beech_mtd);
-		iounmap((void *) beech_mtd_map.virt);
-	}
-}
-
-module_init(init_beech_mtd);
-module_exit(cleanup_beech_mtd);
-
-MODULE_LICENSE("GPL");
-MODULE_AUTHOR("Bishop Brock <bcbrock@us.ibm.com>");
-MODULE_DESCRIPTION("MTD map and partitions for IBM 405LP Beech boards");
diff --git a/drivers/mtd/maps/nettel.c b/drivers/mtd/maps/nettel.c
index 9f53c655af3a..7b96cd02f82b 100644
--- a/drivers/mtd/maps/nettel.c
+++ b/drivers/mtd/maps/nettel.c
@@ -358,7 +358,7 @@ int __init nettel_init(void)
 	/* Turn other PAR off so the first probe doesn't find it */
 	*intel1par = 0;
 
-	/* Probe for the the size of the first Intel flash */
+	/* Probe for the size of the first Intel flash */
 	nettel_intel_map.size = maxsize;
 	nettel_intel_map.phys = intel0addr;
 	nettel_intel_map.virt = ioremap_nocache(intel0addr, maxsize);
diff --git a/drivers/mtd/maps/physmap_of.c b/drivers/mtd/maps/physmap_of.c
index 72107dc06d67..bbb42c35b69b 100644
--- a/drivers/mtd/maps/physmap_of.c
+++ b/drivers/mtd/maps/physmap_of.c
@@ -186,7 +186,7 @@ static int __devinit of_physmap_probe(struct of_device *dev, const struct of_dev
 	else {
  		if (strcmp(of_probe, "ROM"))
 			dev_dbg(&dev->dev, "map_probe: don't know probe type "
-			"'%s', mapping as rom\n");
+			"'%s', mapping as rom\n", of_probe);
 		info->mtd = do_map_probe("mtd_rom", &info->map);
 	}
 	if (info->mtd == NULL) {
diff --git a/drivers/mtd/mtdpart.c b/drivers/mtd/mtdpart.c
index 1af989023c66..9c6236852942 100644
--- a/drivers/mtd/mtdpart.c
+++ b/drivers/mtd/mtdpart.c
@@ -347,7 +347,6 @@ int add_mtd_partitions(struct mtd_info *master,
 		slave->mtd.subpage_sft = master->subpage_sft;
 
 		slave->mtd.name = parts[i].name;
-		slave->mtd.bank_size = master->bank_size;
 		slave->mtd.owner = master->owner;
 
 		slave->mtd.read = part_read;
diff --git a/drivers/mtd/nand/Kconfig b/drivers/mtd/nand/Kconfig
index d05873b8c155..f1d60b6f048e 100644
--- a/drivers/mtd/nand/Kconfig
+++ b/drivers/mtd/nand/Kconfig
@@ -232,11 +232,13 @@ config MTD_NAND_BASLER_EXCITE
           will be named "excite_nandflash.ko".
 
 config MTD_NAND_CAFE
-       tristate "NAND support for OLPC CAFÉ chip"
-       depends on PCI
-       help
-	 Use NAND flash attached to the CAFÉ chip designed for the $100
-	 laptop.
+	tristate "NAND support for OLPC CAFÉ chip"
+	depends on PCI
+	select REED_SOLOMON
+	select REED_SOLOMON_DEC16
+	help
+	  Use NAND flash attached to the CAFÉ chip designed for the $100
+	  laptop.
 
 config MTD_NAND_CS553X
 	tristate "NAND support for CS5535/CS5536 (AMD Geode companion chip)"
@@ -270,4 +272,13 @@ config MTD_NAND_NANDSIM
 	  The simulator may simulate various NAND flash chips for the
 	  MTD nand layer.
 
+config MTD_NAND_PLATFORM
+	tristate "Support for generic platform NAND driver"
+	depends on MTD_NAND
+	help
+	  This implements a generic NAND driver for on-SOC platform
+	  devices. You will need to provide platform-specific functions
+	  via platform_data.
+
+
 endif # MTD_NAND
diff --git a/drivers/mtd/nand/Makefile b/drivers/mtd/nand/Makefile
index 6872031a3fb2..edba1db14bfa 100644
--- a/drivers/mtd/nand/Makefile
+++ b/drivers/mtd/nand/Makefile
@@ -26,6 +26,6 @@ obj-$(CONFIG_MTD_NAND_NDFC)		+= ndfc.o
 obj-$(CONFIG_MTD_NAND_AT91)		+= at91_nand.o
 obj-$(CONFIG_MTD_NAND_CM_X270)		+= cmx270_nand.o
 obj-$(CONFIG_MTD_NAND_BASLER_EXCITE)	+= excite_nandflash.o
+obj-$(CONFIG_MTD_NAND_PLATFORM)		+= plat_nand.o
 
 nand-objs := nand_base.o nand_bbt.o
-cafe_nand-objs := cafe.o cafe_ecc.o
diff --git a/drivers/mtd/nand/at91_nand.c b/drivers/mtd/nand/at91_nand.c
index 14b80cc90a7b..512e999177f7 100644
--- a/drivers/mtd/nand/at91_nand.c
+++ b/drivers/mtd/nand/at91_nand.c
@@ -82,6 +82,10 @@ static void at91_nand_disable(struct at91_nand_host *host)
 		at91_set_gpio_value(host->board->enable_pin, 1);
 }
 
+#ifdef CONFIG_MTD_PARTITIONS
+const char *part_probes[] = { "cmdlinepart", NULL };
+#endif
+
 /*
  * Probe for the NAND device.
  */
@@ -151,6 +155,12 @@ static int __init at91_nand_probe(struct platform_device *pdev)
 #ifdef CONFIG_MTD_PARTITIONS
 	if (host->board->partition_info)
 		partitions = host->board->partition_info(mtd->size, &num_partitions);
+#ifdef CONFIG_MTD_CMDLINE_PARTS
+	else {
+		mtd->name = "at91_nand";
+		num_partitions = parse_mtd_partitions(mtd, part_probes, &partitions, 0);
+	}
+#endif
 
 	if ((!partitions) || (num_partitions == 0)) {
 		printk(KERN_ERR "at91_nand: No parititions defined, or unsupported device.\n");
diff --git a/drivers/mtd/nand/cafe_ecc.c b/drivers/mtd/nand/cafe_ecc.c
deleted file mode 100644
index ea5c8491d2c5..000000000000
--- a/drivers/mtd/nand/cafe_ecc.c
+++ /dev/null
@@ -1,1381 +0,0 @@
-/* Error correction for CAFÉ NAND controller
- *
- * © 2006 Marvell, Inc.
- * Author: Tom Chiou
- *
- * This program is free software; you can redistribute it and/or modify it
- * under the terms of the GNU General Public License as published by the Free
- * Software Foundation; either version 2 of the License, or (at your option)
- * any later version.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
- * more details.
- *
- * You should have received a copy of the GNU General Public License along with
- * this program; if not, write to the Free Software Foundation, Inc., 59
- * Temple Place - Suite 330, Boston, MA  02111-1307, USA.
- */
-
-#include <linux/kernel.h>
-#include <linux/module.h>
-#include <linux/errno.h>
-
-static unsigned short gf4096_mul(unsigned short, unsigned short);
-static unsigned short gf64_mul(unsigned short, unsigned short);
-static unsigned short gf4096_inv(unsigned short);
-static unsigned short err_pos(unsigned short);
-static void find_4bit_err_coefs(unsigned short, unsigned short, unsigned short,
-				unsigned short, unsigned short, unsigned short,
-				unsigned short, unsigned short, unsigned short *);
-static void zero_4x5_col3(unsigned short[4][5]);
-static void zero_4x5_col2(unsigned short[4][5]);
-static void zero_4x5_col1(unsigned short[4][5]);
-static void swap_4x5_rows(unsigned short[4][5], int, int, int);
-static void swap_2x3_rows(unsigned short m[2][3]);
-static void solve_4x5(unsigned short m[4][5], unsigned short *, int *);
-static void sort_coefs(int *, unsigned short *, int);
-static void find_4bit_err_pats(unsigned short, unsigned short, unsigned short,
-			       unsigned short, unsigned short, unsigned short,
-			       unsigned short, unsigned short, unsigned short *);
-static void find_3bit_err_coefs(unsigned short, unsigned short, unsigned short,
-				unsigned short, unsigned short, unsigned short,
-				unsigned short *);
-static void zero_3x4_col2(unsigned short[3][4]);
-static void zero_3x4_col1(unsigned short[3][4]);
-static void swap_3x4_rows(unsigned short[3][4], int, int, int);
-static void solve_3x4(unsigned short[3][4], unsigned short *, int *);
-static void find_3bit_err_pats(unsigned short, unsigned short, unsigned short,
-			       unsigned short, unsigned short, unsigned short,
-			       unsigned short *);
-
-static void find_2bit_err_pats(unsigned short, unsigned short, unsigned short,
-			       unsigned short, unsigned short *);
-static void find_2x2_soln(unsigned short, unsigned short, unsigned short,
-			  unsigned short, unsigned short, unsigned short,
-			  unsigned short *);
-static void solve_2x3(unsigned short[2][3], unsigned short *);
-static int chk_no_err_only(unsigned short *, unsigned short *);
-static int chk_1_err_only(unsigned short *, unsigned short *);
-static int chk_2_err_only(unsigned short *, unsigned short *);
-static int chk_3_err_only(unsigned short *, unsigned short *);
-static int chk_4_err_only(unsigned short *, unsigned short *);
-
-static unsigned short gf64_mul(unsigned short a, unsigned short b)
-{
-	unsigned short tmp1, tmp2, tmp3, tmp4, tmp5;
-	unsigned short c_bit0, c_bit1, c_bit2, c_bit3, c_bit4, c_bit5, c;
-
-	tmp1 = ((a) ^ (a >> 5));
-	tmp2 = ((a >> 4) ^ (a >> 5));
-	tmp3 = ((a >> 3) ^ (a >> 4));
-	tmp4 = ((a >> 2) ^ (a >> 3));
-	tmp5 = ((a >> 1) ^ (a >> 2));
-
-	c_bit0 = ((a & b) ^ ((a >> 5) & (b >> 1)) ^ ((a >> 4) & (b >> 2)) ^
-		  ((a >> 3) & (b >> 3)) ^ ((a >> 2) & (b >> 4)) ^ ((a >> 1) & (b >> 5))) & 0x1;
-
-	c_bit1 = (((a >> 1) & b) ^ (tmp1 & (b >> 1)) ^ (tmp2 & (b >> 2)) ^
-		  (tmp3 & (b >> 3)) ^ (tmp4 & (b >> 4)) ^ (tmp5 & (b >> 5))) & 0x1;
-
-	c_bit2 = (((a >> 2) & b) ^ ((a >> 1) & (b >> 1)) ^ (tmp1 & (b >> 2)) ^
-		  (tmp2 & (b >> 3)) ^ (tmp3 & (b >> 4)) ^ (tmp4 & (b >> 5))) & 0x1;
-
-	c_bit3 = (((a >> 3) & b) ^ ((a >> 2) & (b >> 1)) ^ ((a >> 1) & (b >> 2)) ^
-		  (tmp1 & (b >> 3)) ^ (tmp2 & (b >> 4)) ^ (tmp3 & (b >> 5))) & 0x1;
-
-	c_bit4 = (((a >> 4) & b) ^ ((a >> 3) & (b >> 1)) ^ ((a >> 2) & (b >> 2)) ^
-		  ((a >> 1) & (b >> 3)) ^ (tmp1 & (b >> 4)) ^ (tmp2 & (b >> 5))) & 0x1;
-
-	c_bit5 = (((a >> 5) & b) ^ ((a >> 4) & (b >> 1)) ^ ((a >> 3) & (b >> 2)) ^
-		  ((a >> 2) & (b >> 3)) ^ ((a >> 1) & (b >> 4)) ^ (tmp1 & (b >> 5))) & 0x1;
-
-	c = c_bit0 | (c_bit1 << 1) | (c_bit2 << 2) | (c_bit3 << 3) | (c_bit4 << 4) | (c_bit5 << 5);
-
-	return c;
-}
-
-static unsigned short gf4096_mul(unsigned short a, unsigned short b)
-{
-	unsigned short ah, al, bh, bl, alxah, blxbh, ablh, albl, ahbh, ahbhB, c;
-
-	ah = (a >> 6) & 0x3f;
-	al = a & 0x3f;
-	bh = (b >> 6) & 0x3f;
-	bl = b & 0x3f;
-	alxah = al ^ ah;
-	blxbh = bl ^ bh;
-
-	ablh = gf64_mul(alxah, blxbh);
-	albl = gf64_mul(al, bl);
-	ahbh = gf64_mul(ah, bh);
-
-	ahbhB = ((ahbh & 0x1) << 5) |
-	    ((ahbh & 0x20) >> 1) |
-	    ((ahbh & 0x10) >> 1) | ((ahbh & 0x8) >> 1) | ((ahbh & 0x4) >> 1) | (((ahbh >> 1) ^ ahbh) & 0x1);
-
-	c = ((ablh ^ albl) << 6) | (ahbhB ^ albl);
-	return c;
-}
-
-static void find_2bit_err_pats(unsigned short s0, unsigned short s1, unsigned short r0, unsigned short r1, unsigned short *pats)
-{
-	find_2x2_soln(0x1, 0x1, r0, r1, s0, s1, pats);
-}
-
-static void find_3bit_err_coefs(unsigned short s0, unsigned short s1,
-				unsigned short s2, unsigned short s3, unsigned short s4, unsigned short s5, unsigned short *coefs)
-{
-	unsigned short m[3][4];
-	int row_order[3];
-
-	row_order[0] = 0;
-	row_order[1] = 1;
-	row_order[2] = 2;
-	m[0][0] = s2;
-	m[0][1] = s1;
-	m[0][2] = s0;
-	m[0][3] = s3;
-	m[1][0] = s3;
-	m[1][1] = s2;
-	m[1][2] = s1;
-	m[1][3] = s4;
-	m[2][0] = s4;
-	m[2][1] = s3;
-	m[2][2] = s2;
-	m[2][3] = s5;
-
-	if (m[0][2] != 0x0) {
-		zero_3x4_col2(m);
-	} else if (m[1][2] != 0x0) {
-		swap_3x4_rows(m, 0, 1, 4);
-		zero_3x4_col2(m);
-	} else if (m[2][2] != 0x0) {
-		swap_3x4_rows(m, 0, 2, 4);
-		zero_3x4_col2(m);
-	} else {
-		printk(KERN_ERR "Error: find_3bit_err_coefs, s0,s1,s2 all zeros!\n");
-	}
-
-	if (m[1][1] != 0x0) {
-		zero_3x4_col1(m);
-	} else if (m[2][1] != 0x0) {
-		swap_3x4_rows(m, 1, 2, 4);
-		zero_3x4_col1(m);
-	} else {
-		printk(KERN_ERR "Error: find_3bit_err_coefs, cannot resolve col 1!\n");
-	}
-
-	/* solve coefs */
-	solve_3x4(m, coefs, row_order);
-}
-
-static void zero_3x4_col2(unsigned short m[3][4])
-{
-	unsigned short minv1, minv2;
-
-	minv1 = gf4096_mul(m[1][2], gf4096_inv(m[0][2]));
-	minv2 = gf4096_mul(m[2][2], gf4096_inv(m[0][2]));
-	m[1][0] = m[1][0] ^ gf4096_mul(m[0][0], minv1);
-	m[1][1] = m[1][1] ^ gf4096_mul(m[0][1], minv1);
-	m[1][3] = m[1][3] ^ gf4096_mul(m[0][3], minv1);
-	m[2][0] = m[2][0] ^ gf4096_mul(m[0][0], minv2);
-	m[2][1] = m[2][1] ^ gf4096_mul(m[0][1], minv2);
-	m[2][3] = m[2][3] ^ gf4096_mul(m[0][3], minv2);
-}
-
-static void zero_3x4_col1(unsigned short m[3][4])
-{
-	unsigned short minv;
-	minv = gf4096_mul(m[2][1], gf4096_inv(m[1][1]));
-	m[2][0] = m[2][0] ^ gf4096_mul(m[1][0], minv);
-	m[2][3] = m[2][3] ^ gf4096_mul(m[1][3], minv);
-}
-
-static void swap_3x4_rows(unsigned short m[3][4], int i, int j, int col_width)
-{
-	unsigned short tmp0;
-	int cnt;
-	for (cnt = 0; cnt < col_width; cnt++) {
-		tmp0 = m[i][cnt];
-		m[i][cnt] = m[j][cnt];
-		m[j][cnt] = tmp0;
-	}
-}
-
-static void solve_3x4(unsigned short m[3][4], unsigned short *coefs, int *row_order)
-{
-	unsigned short tmp[3];
-	tmp[0] = gf4096_mul(m[2][3], gf4096_inv(m[2][0]));
-	tmp[1] = gf4096_mul((gf4096_mul(tmp[0], m[1][0]) ^ m[1][3]), gf4096_inv(m[1][1]));
-	tmp[2] = gf4096_mul((gf4096_mul(tmp[0], m[0][0]) ^ gf4096_mul(tmp[1], m[0][1]) ^ m[0][3]), gf4096_inv(m[0][2]));
-	sort_coefs(row_order, tmp, 3);
-	coefs[0] = tmp[0];
-	coefs[1] = tmp[1];
-	coefs[2] = tmp[2];
-}
-
-static void find_3bit_err_pats(unsigned short s0, unsigned short s1,
-			       unsigned short s2, unsigned short r0,
-			       unsigned short r1, unsigned short r2,
-			       unsigned short *pats)
-{
-	find_2x2_soln(r0 ^ r2, r1 ^ r2,
-		      gf4096_mul(r0, r0 ^ r2), gf4096_mul(r1, r1 ^ r2),
-		      gf4096_mul(s0, r2) ^ s1, gf4096_mul(s1, r2) ^ s2, pats);
-	pats[2] = s0 ^ pats[0] ^ pats[1];
-}
-
-static void find_4bit_err_coefs(unsigned short s0, unsigned short s1,
-				unsigned short s2, unsigned short s3,
-				unsigned short s4, unsigned short s5,
-				unsigned short s6, unsigned short s7,
-				unsigned short *coefs)
-{
-	unsigned short m[4][5];
-	int row_order[4];
-
-	row_order[0] = 0;
-	row_order[1] = 1;
-	row_order[2] = 2;
-	row_order[3] = 3;
-
-	m[0][0] = s3;
-	m[0][1] = s2;
-	m[0][2] = s1;
-	m[0][3] = s0;
-	m[0][4] = s4;
-	m[1][0] = s4;
-	m[1][1] = s3;
-	m[1][2] = s2;
-	m[1][3] = s1;
-	m[1][4] = s5;
-	m[2][0] = s5;
-	m[2][1] = s4;
-	m[2][2] = s3;
-	m[2][3] = s2;
-	m[2][4] = s6;
-	m[3][0] = s6;
-	m[3][1] = s5;
-	m[3][2] = s4;
-	m[3][3] = s3;
-	m[3][4] = s7;
-
-	if (m[0][3] != 0x0) {
-		zero_4x5_col3(m);
-	} else if (m[1][3] != 0x0) {
-		swap_4x5_rows(m, 0, 1, 5);
-		zero_4x5_col3(m);
-	} else if (m[2][3] != 0x0) {
-		swap_4x5_rows(m, 0, 2, 5);
-		zero_4x5_col3(m);
-	} else if (m[3][3] != 0x0) {
-		swap_4x5_rows(m, 0, 3, 5);
-		zero_4x5_col3(m);
-	} else {
-		printk(KERN_ERR "Error: find_4bit_err_coefs, s0,s1,s2,s3 all zeros!\n");
-	}
-
-	if (m[1][2] != 0x0) {
-		zero_4x5_col2(m);
-	} else if (m[2][2] != 0x0) {
-		swap_4x5_rows(m, 1, 2, 5);
-		zero_4x5_col2(m);
-	} else if (m[3][2] != 0x0) {
-		swap_4x5_rows(m, 1, 3, 5);
-		zero_4x5_col2(m);
-	} else {
-		printk(KERN_ERR "Error: find_4bit_err_coefs, cannot resolve col 2!\n");
-	}
-
-	if (m[2][1] != 0x0) {
-		zero_4x5_col1(m);
-	} else if (m[3][1] != 0x0) {
-		swap_4x5_rows(m, 2, 3, 5);
-		zero_4x5_col1(m);
-	} else {
-		printk(KERN_ERR "Error: find_4bit_err_coefs, cannot resolve col 1!\n");
-	}
-
-	solve_4x5(m, coefs, row_order);
-}
-
-static void zero_4x5_col3(unsigned short m[4][5])
-{
-	unsigned short minv1, minv2, minv3;
-
-	minv1 = gf4096_mul(m[1][3], gf4096_inv(m[0][3]));
-	minv2 = gf4096_mul(m[2][3], gf4096_inv(m[0][3]));
-	minv3 = gf4096_mul(m[3][3], gf4096_inv(m[0][3]));
-
-	m[1][0] = m[1][0] ^ gf4096_mul(m[0][0], minv1);
-	m[1][1] = m[1][1] ^ gf4096_mul(m[0][1], minv1);
-	m[1][2] = m[1][2] ^ gf4096_mul(m[0][2], minv1);
-	m[1][4] = m[1][4] ^ gf4096_mul(m[0][4], minv1);
-	m[2][0] = m[2][0] ^ gf4096_mul(m[0][0], minv2);
-	m[2][1] = m[2][1] ^ gf4096_mul(m[0][1], minv2);
-	m[2][2] = m[2][2] ^ gf4096_mul(m[0][2], minv2);
-	m[2][4] = m[2][4] ^ gf4096_mul(m[0][4], minv2);
-	m[3][0] = m[3][0] ^ gf4096_mul(m[0][0], minv3);
-	m[3][1] = m[3][1] ^ gf4096_mul(m[0][1], minv3);
-	m[3][2] = m[3][2] ^ gf4096_mul(m[0][2], minv3);
-	m[3][4] = m[3][4] ^ gf4096_mul(m[0][4], minv3);
-}
-
-static void zero_4x5_col2(unsigned short m[4][5])
-{
-	unsigned short minv2, minv3;
-
-	minv2 = gf4096_mul(m[2][2], gf4096_inv(m[1][2]));
-	minv3 = gf4096_mul(m[3][2], gf4096_inv(m[1][2]));
-
-	m[2][0] = m[2][0] ^ gf4096_mul(m[1][0], minv2);
-	m[2][1] = m[2][1] ^ gf4096_mul(m[1][1], minv2);
-	m[2][4] = m[2][4] ^ gf4096_mul(m[1][4], minv2);
-	m[3][0] = m[3][0] ^ gf4096_mul(m[1][0], minv3);
-	m[3][1] = m[3][1] ^ gf4096_mul(m[1][1], minv3);
-	m[3][4] = m[3][4] ^ gf4096_mul(m[1][4], minv3);
-}
-
-static void zero_4x5_col1(unsigned short m[4][5])
-{
-	unsigned short minv;
-
-	minv = gf4096_mul(m[3][1], gf4096_inv(m[2][1]));
-
-	m[3][0] = m[3][0] ^ gf4096_mul(m[2][0], minv);
-	m[3][4] = m[3][4] ^ gf4096_mul(m[2][4], minv);
-}
-
-static void swap_4x5_rows(unsigned short m[4][5], int i, int j, int col_width)
-{
-	unsigned short tmp0;
-	int cnt;
-
-	for (cnt = 0; cnt < col_width; cnt++) {
-		tmp0 = m[i][cnt];
-		m[i][cnt] = m[j][cnt];
-		m[j][cnt] = tmp0;
-	}
-}
-
-static void solve_4x5(unsigned short m[4][5], unsigned short *coefs, int *row_order)
-{
-	unsigned short tmp[4];
-
-	tmp[0] = gf4096_mul(m[3][4], gf4096_inv(m[3][0]));
-	tmp[1] = gf4096_mul((gf4096_mul(tmp[0], m[2][0]) ^ m[2][4]), gf4096_inv(m[2][1]));
-	tmp[2] = gf4096_mul((gf4096_mul(tmp[0], m[1][0]) ^ gf4096_mul(tmp[1], m[1][1]) ^ m[1][4]), gf4096_inv(m[1][2]));
-	tmp[3] = gf4096_mul((gf4096_mul(tmp[0], m[0][0]) ^
-			gf4096_mul(tmp[1], m[0][1]) ^ gf4096_mul(tmp[2], m[0][2]) ^ m[0][4]), gf4096_inv(m[0][3]));
-	sort_coefs(row_order, tmp, 4);
-	coefs[0] = tmp[0];
-	coefs[1] = tmp[1];
-	coefs[2] = tmp[2];
-	coefs[3] = tmp[3];
-}
-
-static void sort_coefs(int *order, unsigned short *soln, int len)
-{
-	int cnt, start_cnt, least_ord, least_cnt;
-	unsigned short tmp0;
-	for (start_cnt = 0; start_cnt < len; start_cnt++) {
-		for (cnt = start_cnt; cnt < len; cnt++) {
-			if (cnt == start_cnt) {
-				least_ord = order[cnt];
-				least_cnt = start_cnt;
-			} else {
-				if (least_ord > order[cnt]) {
-					least_ord = order[cnt];
-					least_cnt = cnt;
-				}
-			}
-		}
-		if (least_cnt != start_cnt) {
-			tmp0 = order[least_cnt];
-			order[least_cnt] = order[start_cnt];
-			order[start_cnt] = tmp0;
-			tmp0 = soln[least_cnt];
-			soln[least_cnt] = soln[start_cnt];
-			soln[start_cnt] = tmp0;
-		}
-	}
-}
-
-static void find_4bit_err_pats(unsigned short s0, unsigned short s1,
-			       unsigned short s2, unsigned short s3,
-			       unsigned short z1, unsigned short z2,
-			       unsigned short z3, unsigned short z4,
-			       unsigned short *pats)
-{
-	unsigned short z4_z1, z3z4_z3z3, z4_z2, s0z4_s1, z1z4_z1z1,
-		z4_z3, z2z4_z2z2, s1z4_s2, z3z3z4_z3z3z3, z1z1z4_z1z1z1, z2z2z4_z2z2z2, s2z4_s3;
-	unsigned short tmp0, tmp1, tmp2, tmp3;
-
-	z4_z1 = z4 ^ z1;
-	z3z4_z3z3 = gf4096_mul(z3, z4) ^ gf4096_mul(z3, z3);
-	z4_z2 = z4 ^ z2;
-	s0z4_s1 = gf4096_mul(s0, z4) ^ s1;
-	z1z4_z1z1 = gf4096_mul(z1, z4) ^ gf4096_mul(z1, z1);
-	z4_z3 = z4 ^ z3;
-	z2z4_z2z2 = gf4096_mul(z2, z4) ^ gf4096_mul(z2, z2);
-	s1z4_s2 = gf4096_mul(s1, z4) ^ s2;
-	z3z3z4_z3z3z3 = gf4096_mul(gf4096_mul(z3, z3), z4) ^ gf4096_mul(gf4096_mul(z3, z3), z3);
-	z1z1z4_z1z1z1 = gf4096_mul(gf4096_mul(z1, z1), z4) ^ gf4096_mul(gf4096_mul(z1, z1), z1);
-	z2z2z4_z2z2z2 = gf4096_mul(gf4096_mul(z2, z2), z4) ^ gf4096_mul(gf4096_mul(z2, z2), z2);
-	s2z4_s3 = gf4096_mul(s2, z4) ^ s3;
-
-	//find err pat 0,1
-	find_2x2_soln(gf4096_mul(z4_z1, z3z4_z3z3) ^
-		      gf4096_mul(z1z4_z1z1, z4_z3), gf4096_mul(z4_z2,
-							       z3z4_z3z3) ^
-		      gf4096_mul(z2z4_z2z2, z4_z3), gf4096_mul(z1z4_z1z1,
-							       z3z3z4_z3z3z3) ^
-		      gf4096_mul(z1z1z4_z1z1z1, z3z4_z3z3),
-		      gf4096_mul(z2z4_z2z2,
-				 z3z3z4_z3z3z3) ^ gf4096_mul(z2z2z4_z2z2z2,
-							     z3z4_z3z3),
-		      gf4096_mul(s0z4_s1, z3z4_z3z3) ^ gf4096_mul(s1z4_s2,
-								  z4_z3),
-		      gf4096_mul(s1z4_s2, z3z3z4_z3z3z3) ^ gf4096_mul(s2z4_s3, z3z4_z3z3), pats);
-	tmp0 = pats[0];
-	tmp1 = pats[1];
-	tmp2 = pats[0] ^ pats[1] ^ s0;
-	tmp3 = gf4096_mul(pats[0], z1) ^ gf4096_mul(pats[1], z2) ^ s1;
-
-	//find err pat 2,3
-	find_2x2_soln(0x1, 0x1, z3, z4, tmp2, tmp3, pats);
-	pats[2] = pats[0];
-	pats[3] = pats[1];
-	pats[0] = tmp0;
-	pats[1] = tmp1;
-}
-
-static void find_2x2_soln(unsigned short c00, unsigned short c01,
-			  unsigned short c10, unsigned short c11,
-			  unsigned short lval0, unsigned short lval1,
-			  unsigned short *soln)
-{
-	unsigned short m[2][3];
-	m[0][0] = c00;
-	m[0][1] = c01;
-	m[0][2] = lval0;
-	m[1][0] = c10;
-	m[1][1] = c11;
-	m[1][2] = lval1;
-
-	if (m[0][1] != 0x0) {
-		/* */
-	} else if (m[1][1] != 0x0) {
-		swap_2x3_rows(m);
-	} else {
-		printk(KERN_ERR "Warning: find_2bit_err_coefs, s0,s1 all zeros!\n");
-	}
-
-	solve_2x3(m, soln);
-}
-
-static void swap_2x3_rows(unsigned short m[2][3])
-{
-	unsigned short tmp0;
-	int cnt;
-
-	for (cnt = 0; cnt < 3; cnt++) {
-		tmp0 = m[0][cnt];
-		m[0][cnt] = m[1][cnt];
-		m[1][cnt] = tmp0;
-	}
-}
-
-static void solve_2x3(unsigned short m[2][3], unsigned short *coefs)
-{
-	unsigned short minv;
-
-	minv = gf4096_mul(m[1][1], gf4096_inv(m[0][1]));
-	m[1][0] = m[1][0] ^ gf4096_mul(m[0][0], minv);
-	m[1][2] = m[1][2] ^ gf4096_mul(m[0][2], minv);
-	coefs[0] = gf4096_mul(m[1][2], gf4096_inv(m[1][0]));
-	coefs[1] = gf4096_mul((gf4096_mul(coefs[0], m[0][0]) ^ m[0][2]), gf4096_inv(m[0][1]));
-}
-
-static unsigned char gf64_inv[64] = {
-	 0,  1, 33, 62, 49, 43, 31, 44, 57, 37, 52, 28, 46, 40, 22, 25,
-	61, 54, 51, 39, 26, 35, 14, 24, 23, 15, 20, 34, 11, 53, 45,  6,
-	63,  2, 27, 21, 56,  9, 50, 19, 13, 47, 48,  5,  7, 30, 12, 41,
-	42,  4, 38, 18, 10, 29, 17, 60, 36,  8, 59, 58, 55, 16,  3, 32
-};
-
-static unsigned short gf4096_inv(unsigned short din)
-{
-	unsigned short alahxal, ah2B, deno, inv, bl, bh;
-	unsigned short ah, al, ahxal;
-	unsigned short dout;
-
-	ah = (din >> 6) & 0x3f;
-	al = din & 0x3f;
-	ahxal = ah ^ al;
-	ah2B = (((ah ^ (ah >> 3)) & 0x1) << 5) |
-		((ah >> 1) & 0x10) |
-		((((ah >> 5) ^ (ah >> 2)) & 0x1) << 3) |
-		((ah >> 2) & 0x4) | ((((ah >> 4) ^ (ah >> 1)) & 0x1) << 1) | (ah & 0x1);
-	alahxal = gf64_mul(ahxal, al);
-	deno = alahxal ^ ah2B;
-	inv = gf64_inv[deno];
-	bl = gf64_mul(inv, ahxal);
-	bh = gf64_mul(inv, ah);
-	dout = ((bh & 0x3f) << 6) | (bl & 0x3f);
-	return (((bh & 0x3f) << 6) | (bl & 0x3f));
-}
-
-static unsigned short err_pos_lut[4096] = {
-	0xfff, 0x000, 0x451, 0xfff, 0xfff, 0x3cf, 0xfff, 0x041,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x28a, 0xfff, 0x492, 0xfff,
-	0x145, 0xfff, 0xfff, 0x514, 0xfff, 0x082, 0xfff, 0xfff,
-	0xfff, 0x249, 0x38e, 0x410, 0xfff, 0x104, 0x208, 0x1c7,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x2cb, 0xfff, 0xfff, 0xfff,
-	0x0c3, 0x34d, 0x4d3, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x186, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x30c, 0x555, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x166, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x385, 0x14e, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x4e1,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x538, 0xfff, 0x16d, 0xfff,
-	0xfff, 0xfff, 0x45b, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x29c, 0x2cc, 0x30b, 0x2b3, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x0b3, 0xfff, 0x2f7,
-	0xfff, 0x32b, 0xfff, 0xfff, 0xfff, 0xfff, 0x0a7, 0xfff,
-	0xfff, 0x2da, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x07e, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x11c, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x22f, 0xfff, 0x1f4, 0xfff, 0xfff,
-	0x2b0, 0x504, 0xfff, 0x114, 0xfff, 0xfff, 0xfff, 0x21d,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x00d, 0x3c4, 0x340, 0x10f,
-	0xfff, 0xfff, 0x266, 0x02e, 0xfff, 0xfff, 0xfff, 0x4f8,
-	0x337, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x07b, 0x168, 0xfff, 0xfff, 0x0fe,
-	0xfff, 0xfff, 0x51a, 0xfff, 0x458, 0xfff, 0x36d, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x073, 0x37d, 0x415, 0x550, 0xfff,
-	0xfff, 0xfff, 0x23b, 0x4b4, 0xfff, 0xfff, 0xfff, 0x1a1,
-	0xfff, 0xfff, 0x3aa, 0xfff, 0x117, 0x04d, 0x341, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x518, 0x03e, 0x0f2, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x363, 0xfff, 0x0b9, 0xfff, 0xfff,
-	0x241, 0xfff, 0xfff, 0x049, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x15f, 0x52d, 0xfff, 0xfff, 0xfff, 0x29e, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x4cf, 0x0fc, 0xfff, 0x36f, 0x3d3, 0xfff,
-	0x228, 0xfff, 0xfff, 0x45e, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x238, 0xfff, 0xfff, 0xfff, 0xfff, 0x47f, 0xfff, 0xfff,
-	0x43a, 0x265, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x3e8,
-	0xfff, 0xfff, 0x01a, 0xfff, 0xfff, 0xfff, 0xfff, 0x21e,
-	0x1fc, 0x40b, 0xfff, 0xfff, 0xfff, 0x2d0, 0x159, 0xfff,
-	0xfff, 0x313, 0xfff, 0xfff, 0x05c, 0x4cc, 0xfff, 0xfff,
-	0x0f6, 0x3d5, 0xfff, 0xfff, 0xfff, 0x54f, 0xfff, 0xfff,
-	0xfff, 0x172, 0x1e4, 0x07c, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x53c, 0x1ad, 0x535,
-	0x19b, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x092, 0xfff, 0x2be, 0xfff, 0xfff, 0x482,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x0e6, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x476, 0xfff, 0x51d, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x342, 0x2b5, 0x22e, 0x09a, 0xfff, 0x08d,
-	0x44f, 0x3ed, 0xfff, 0xfff, 0xfff, 0xfff, 0x3d1, 0xfff,
-	0xfff, 0x543, 0xfff, 0x48f, 0xfff, 0x3d2, 0xfff, 0x0d5,
-	0x113, 0x0ec, 0x427, 0xfff, 0xfff, 0xfff, 0x4c4, 0xfff,
-	0xfff, 0x50a, 0xfff, 0x144, 0xfff, 0x105, 0x39f, 0x294,
-	0x164, 0xfff, 0x31a, 0xfff, 0xfff, 0x49a, 0xfff, 0x130,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x1be, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x49e, 0x371, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x0e8, 0x49c, 0x0f4, 0xfff,
-	0x338, 0x1a7, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x36c, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x1ae, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x31b, 0xfff, 0xfff, 0x2dd, 0x522, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x2f4,
-	0x3c6, 0x30d, 0xfff, 0xfff, 0xfff, 0xfff, 0x34c, 0x18f,
-	0x30a, 0xfff, 0x01f, 0x079, 0xfff, 0xfff, 0x54d, 0x46b,
-	0x28c, 0x37f, 0xfff, 0xfff, 0xfff, 0xfff, 0x355, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x14f, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x359, 0x3fe, 0x3c5, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x423, 0xfff, 0xfff, 0x34a, 0x22c, 0xfff,
-	0x25a, 0xfff, 0xfff, 0x4ad, 0xfff, 0x28d, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x547, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x2e2, 0xfff, 0xfff, 0x1d5, 0xfff, 0x2a8, 0xfff, 0xfff,
-	0x03f, 0xfff, 0xfff, 0xfff, 0xfff, 0x3eb, 0x0fa, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x55b, 0xfff,
-	0x08e, 0xfff, 0x3ae, 0xfff, 0x3a4, 0xfff, 0x282, 0x158,
-	0xfff, 0x382, 0xfff, 0xfff, 0x499, 0xfff, 0xfff, 0x08a,
-	0xfff, 0xfff, 0xfff, 0x456, 0x3be, 0xfff, 0x1e2, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x559, 0xfff, 0x1a0, 0xfff,
-	0xfff, 0x0b4, 0xfff, 0xfff, 0xfff, 0x2df, 0xfff, 0xfff,
-	0xfff, 0x07f, 0x4f5, 0xfff, 0xfff, 0x27c, 0x133, 0x017,
-	0xfff, 0x3fd, 0xfff, 0xfff, 0xfff, 0x44d, 0x4cd, 0x17a,
-	0x0d7, 0x537, 0xfff, 0xfff, 0x353, 0xfff, 0xfff, 0x351,
-	0x366, 0xfff, 0x44a, 0xfff, 0x1a6, 0xfff, 0xfff, 0xfff,
-	0x291, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x1e3,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x389, 0xfff, 0x07a, 0xfff,
-	0x1b6, 0x2ed, 0xfff, 0xfff, 0xfff, 0xfff, 0x24e, 0x074,
-	0xfff, 0xfff, 0x3dc, 0xfff, 0x4e3, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x4eb, 0xfff, 0xfff, 0x3b8, 0x4de, 0xfff, 0x19c,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x262,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x076, 0x4e8, 0x3da,
-	0xfff, 0x531, 0xfff, 0xfff, 0x14a, 0xfff, 0x0a2, 0x433,
-	0x3df, 0x1e9, 0xfff, 0xfff, 0xfff, 0xfff, 0x3e7, 0x285,
-	0x2d8, 0xfff, 0xfff, 0xfff, 0x349, 0x18d, 0x098, 0xfff,
-	0x0df, 0x4bf, 0xfff, 0xfff, 0x0b2, 0xfff, 0x346, 0x24d,
-	0xfff, 0xfff, 0xfff, 0x24f, 0x4fa, 0x2f9, 0xfff, 0xfff,
-	0x3c9, 0xfff, 0x2b4, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x056, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x179, 0xfff, 0x0e9, 0x3f0, 0x33d, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x1fd, 0xfff, 0xfff, 0x526, 0xfff,
-	0xfff, 0xfff, 0x53d, 0xfff, 0xfff, 0xfff, 0x170, 0x331,
-	0xfff, 0x068, 0xfff, 0xfff, 0xfff, 0x3f7, 0xfff, 0x3d8,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x09f, 0x556, 0xfff, 0xfff, 0x02d, 0xfff, 0xfff,
-	0x553, 0xfff, 0xfff, 0xfff, 0x1f0, 0xfff, 0xfff, 0x4d6,
-	0x41e, 0xfff, 0xfff, 0xfff, 0xfff, 0x4d5, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x248, 0xfff, 0xfff, 0xfff, 0x0a3,
-	0xfff, 0x217, 0xfff, 0xfff, 0xfff, 0x4f1, 0x209, 0xfff,
-	0xfff, 0x475, 0x234, 0x52b, 0x398, 0xfff, 0x08b, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x2c2, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x268, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x4a3, 0xfff, 0x0aa, 0xfff, 0x1d9, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x155, 0xfff, 0xfff, 0xfff, 0xfff, 0x0bf,
-	0x539, 0xfff, 0xfff, 0x2f1, 0x545, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x2a7, 0x06f, 0xfff, 0x378, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x25e, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x15d, 0x02a, 0xfff, 0xfff, 0x0bc,
-	0x235, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x150, 0xfff, 0x1a9, 0xfff, 0xfff, 0xfff, 0xfff, 0x381,
-	0xfff, 0x04e, 0x270, 0x13f, 0xfff, 0xfff, 0x405, 0xfff,
-	0x3cd, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x2ef, 0xfff, 0x06a, 0xfff, 0xfff, 0xfff, 0x34f,
-	0x212, 0xfff, 0xfff, 0x0e2, 0xfff, 0x083, 0x298, 0xfff,
-	0xfff, 0xfff, 0x0c2, 0xfff, 0xfff, 0x52e, 0xfff, 0x488,
-	0xfff, 0xfff, 0xfff, 0x36b, 0xfff, 0xfff, 0xfff, 0x442,
-	0x091, 0xfff, 0x41c, 0xfff, 0xfff, 0x3a5, 0xfff, 0x4e6,
-	0xfff, 0xfff, 0x40d, 0x31d, 0xfff, 0xfff, 0xfff, 0x4c1,
-	0x053, 0xfff, 0x418, 0x13c, 0xfff, 0x350, 0xfff, 0x0ae,
-	0xfff, 0xfff, 0x41f, 0xfff, 0x470, 0xfff, 0x4ca, 0xfff,
-	0xfff, 0xfff, 0x02b, 0x450, 0xfff, 0x1f8, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x293, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x411, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x0b8, 0xfff, 0xfff, 0xfff,
-	0x3e1, 0xfff, 0xfff, 0xfff, 0xfff, 0x43c, 0xfff, 0x2b2,
-	0x2ab, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x1ec,
-	0xfff, 0xfff, 0xfff, 0x3f8, 0x034, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x11a, 0xfff, 0x541, 0x45c, 0x134,
-	0x1cc, 0xfff, 0xfff, 0xfff, 0x469, 0xfff, 0xfff, 0x44b,
-	0x161, 0xfff, 0xfff, 0xfff, 0x055, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x307, 0xfff, 0xfff, 0xfff, 0xfff, 0x2d1, 0xfff,
-	0xfff, 0xfff, 0x124, 0x37b, 0x26b, 0x336, 0xfff, 0xfff,
-	0x2e4, 0x3cb, 0xfff, 0xfff, 0x0f8, 0x3c8, 0xfff, 0xfff,
-	0xfff, 0x461, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x4b5,
-	0x2cf, 0xfff, 0xfff, 0xfff, 0x20f, 0xfff, 0x35a, 0xfff,
-	0x490, 0xfff, 0x185, 0xfff, 0xfff, 0xfff, 0xfff, 0x42e,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x54b, 0xfff, 0xfff, 0xfff,
-	0x146, 0xfff, 0x412, 0xfff, 0xfff, 0xfff, 0x1ff, 0xfff,
-	0xfff, 0x3e0, 0xfff, 0xfff, 0xfff, 0xfff, 0x2d5, 0xfff,
-	0x4df, 0x505, 0xfff, 0x413, 0xfff, 0x1a5, 0xfff, 0x3b2,
-	0xfff, 0xfff, 0xfff, 0x35b, 0xfff, 0x116, 0xfff, 0xfff,
-	0x171, 0x4d0, 0xfff, 0x154, 0x12d, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x468, 0x4db, 0xfff,
-	0xfff, 0x1df, 0xfff, 0xfff, 0xfff, 0xfff, 0x05a, 0xfff,
-	0x0f1, 0x403, 0xfff, 0x22b, 0x2e0, 0xfff, 0xfff, 0xfff,
-	0x2b7, 0x373, 0xfff, 0xfff, 0xfff, 0xfff, 0x13e, 0xfff,
-	0xfff, 0xfff, 0x0d0, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x329, 0x1d2, 0x3fa, 0x047, 0xfff, 0x2f2, 0xfff, 0xfff,
-	0x141, 0x0ac, 0x1d7, 0xfff, 0x07d, 0xfff, 0xfff, 0xfff,
-	0x1c1, 0xfff, 0x487, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x045, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x288, 0x0cd, 0xfff, 0xfff, 0xfff, 0xfff, 0x226, 0x1d8,
-	0xfff, 0x153, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x4cb,
-	0x528, 0xfff, 0xfff, 0xfff, 0x20a, 0x343, 0x3a1, 0xfff,
-	0xfff, 0xfff, 0x2d7, 0x2d3, 0x1aa, 0x4c5, 0xfff, 0xfff,
-	0xfff, 0x42b, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x3e9, 0xfff, 0x20b, 0x260,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x37c, 0x2fd,
-	0xfff, 0xfff, 0x2c8, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x31e, 0xfff, 0x335, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x135, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x35c, 0x4dd, 0x129, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x1ef, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x34e, 0xfff, 0xfff, 0xfff, 0xfff, 0x407, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x3ad, 0xfff, 0xfff, 0xfff,
-	0x379, 0xfff, 0xfff, 0x1d0, 0x38d, 0xfff, 0xfff, 0x1e8,
-	0x184, 0x3c1, 0x1c4, 0xfff, 0x1f9, 0xfff, 0xfff, 0x424,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x1d3, 0x0d4, 0xfff, 0x4e9,
-	0xfff, 0xfff, 0xfff, 0x530, 0x107, 0xfff, 0x106, 0x04f,
-	0xfff, 0xfff, 0x4c7, 0x503, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x15c, 0xfff, 0x23f, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x4f3, 0xfff, 0xfff, 0x3c7,
-	0xfff, 0x278, 0xfff, 0xfff, 0x0a6, 0xfff, 0xfff, 0xfff,
-	0x122, 0x1cf, 0xfff, 0x327, 0xfff, 0x2e5, 0xfff, 0x29d,
-	0xfff, 0xfff, 0x3f1, 0xfff, 0xfff, 0x48d, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x054, 0xfff, 0xfff, 0xfff, 0xfff, 0x178,
-	0x27e, 0x4e0, 0x352, 0x02f, 0x09c, 0xfff, 0x2a0, 0xfff,
-	0xfff, 0x46a, 0x457, 0xfff, 0xfff, 0x501, 0xfff, 0x2ba,
-	0xfff, 0xfff, 0xfff, 0x54e, 0x2e7, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x551, 0xfff, 0xfff, 0x1db, 0x2aa, 0xfff,
-	0xfff, 0x4bc, 0xfff, 0xfff, 0x395, 0xfff, 0x0de, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x455, 0xfff, 0x17e,
-	0xfff, 0x221, 0x4a7, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x388, 0xfff, 0xfff, 0xfff, 0x308, 0xfff, 0xfff, 0xfff,
-	0x20e, 0x4b9, 0xfff, 0x273, 0x20c, 0x09e, 0xfff, 0x057,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x3f2, 0xfff, 0x1a8, 0x3a6,
-	0x14c, 0xfff, 0xfff, 0x071, 0xfff, 0xfff, 0x53a, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x109, 0xfff, 0xfff, 0x399, 0xfff,
-	0x061, 0x4f0, 0x39e, 0x244, 0xfff, 0x035, 0xfff, 0xfff,
-	0x305, 0x47e, 0x297, 0xfff, 0xfff, 0x2b8, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x1bc, 0xfff, 0x2fc,
-	0xfff, 0xfff, 0x554, 0xfff, 0xfff, 0xfff, 0xfff, 0x3b6,
-	0xfff, 0xfff, 0xfff, 0x515, 0x397, 0xfff, 0xfff, 0x12f,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x4e5,
-	0xfff, 0x4fc, 0xfff, 0xfff, 0x05e, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x0a8, 0x3af, 0x015, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x138, 0xfff, 0xfff, 0xfff, 0x540, 0xfff, 0xfff,
-	0xfff, 0x027, 0x523, 0x2f0, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x16c, 0xfff, 0x27d, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x04c, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x4dc,
-	0xfff, 0xfff, 0x059, 0x301, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x1a3, 0xfff, 0x15a, 0xfff, 0xfff,
-	0x0a5, 0xfff, 0x435, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x051, 0xfff, 0xfff, 0x131, 0xfff, 0x4f4, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x441, 0xfff, 0x4fb, 0xfff, 0x03b,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x1ed, 0x274,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x0d3, 0x55e, 0x1b3,
-	0xfff, 0x0bd, 0xfff, 0xfff, 0xfff, 0xfff, 0x225, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x4b7, 0xfff, 0xfff, 0x2ff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x4c3, 0xfff,
-	0x383, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x2f6,
-	0xfff, 0xfff, 0x1ee, 0xfff, 0x03d, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x26f, 0x1dc, 0xfff, 0x0db, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x0ce, 0xfff, 0xfff, 0x127, 0x03a,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x311, 0xfff,
-	0xfff, 0x13d, 0x09d, 0x47b, 0x2a6, 0x50d, 0x510, 0x19a,
-	0xfff, 0x354, 0x414, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x44c, 0x3b0, 0xfff, 0x23d, 0x429, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x4c0, 0x416, 0xfff, 0x05b, 0xfff, 0xfff, 0x137, 0xfff,
-	0x25f, 0x49f, 0xfff, 0x279, 0x013, 0xfff, 0xfff, 0xfff,
-	0x269, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x3d0, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x077, 0xfff, 0xfff, 0x3fb,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x271, 0x3a0, 0xfff, 0xfff,
-	0x40f, 0xfff, 0xfff, 0x3de, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x1ab, 0x26a,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x489, 0xfff, 0xfff,
-	0x252, 0xfff, 0xfff, 0xfff, 0xfff, 0x1b7, 0x42f, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x3b7,
-	0xfff, 0x2bb, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x0f7, 0x01d, 0xfff, 0x067, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x4e2, 0xfff, 0xfff, 0x4bb, 0xfff,
-	0xfff, 0xfff, 0x17b, 0xfff, 0x0ee, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x36e, 0xfff, 0xfff, 0xfff, 0x533, 0xfff,
-	0xfff, 0xfff, 0x4d4, 0x356, 0xfff, 0xfff, 0x375, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x4a4, 0x513, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x4ff, 0xfff, 0x2af,
-	0xfff, 0xfff, 0x026, 0xfff, 0x0ad, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x26e, 0xfff, 0xfff, 0xfff, 0xfff, 0x493, 0xfff,
-	0x463, 0x4d2, 0x4be, 0xfff, 0xfff, 0xfff, 0xfff, 0x4f2,
-	0x0b6, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x32d, 0x315, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x13a, 0x4a1, 0xfff, 0x27a, 0xfff, 0xfff, 0xfff,
-	0x47a, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x334, 0xfff, 0xfff, 0xfff, 0xfff, 0x54c, 0xfff, 0xfff,
-	0xfff, 0x0c9, 0x007, 0xfff, 0xfff, 0x12e, 0xfff, 0x0ff,
-	0xfff, 0xfff, 0x3f5, 0x509, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x1c3, 0x2ad, 0xfff, 0xfff, 0x47c, 0x261, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x152, 0xfff, 0xfff, 0xfff, 0x339,
-	0xfff, 0x243, 0x1c0, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x063, 0xfff, 0xfff, 0x254, 0xfff, 0xfff, 0x173, 0xfff,
-	0x0c7, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x362, 0x259, 0x485, 0x374, 0x0dc, 0x3ab, 0xfff,
-	0x1c5, 0x534, 0x544, 0xfff, 0xfff, 0x508, 0xfff, 0x402,
-	0x408, 0xfff, 0x0e7, 0xfff, 0xfff, 0x00a, 0x205, 0xfff,
-	0xfff, 0x2b9, 0xfff, 0xfff, 0xfff, 0x465, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x23a, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x147, 0x19d, 0x115, 0x214, 0xfff, 0x090, 0x368,
-	0xfff, 0x210, 0xfff, 0xfff, 0x280, 0x52a, 0x163, 0x148,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x326, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x2de, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x206, 0x2c1, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x189, 0xfff, 0xfff, 0xfff, 0xfff, 0x367, 0xfff, 0x1a4,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x443, 0xfff, 0x27b,
-	0xfff, 0xfff, 0x251, 0x549, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x188, 0x04b, 0xfff, 0xfff, 0xfff, 0x31f,
-	0x4a6, 0xfff, 0x246, 0x1de, 0x156, 0xfff, 0xfff, 0xfff,
-	0x3a9, 0xfff, 0xfff, 0xfff, 0x2fa, 0xfff, 0x128, 0x0d1,
-	0x449, 0x255, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x258, 0xfff, 0xfff, 0xfff,
-	0x532, 0xfff, 0xfff, 0xfff, 0x303, 0x517, 0xfff, 0xfff,
-	0x2a9, 0x24a, 0xfff, 0xfff, 0x231, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x4b6, 0x516, 0xfff, 0xfff, 0x0e4, 0x0eb,
-	0xfff, 0x4e4, 0xfff, 0x275, 0xfff, 0xfff, 0x031, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x025, 0x21a, 0xfff, 0x0cc,
-	0x45f, 0x3d9, 0x289, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x23e, 0xfff, 0xfff, 0xfff, 0x438, 0x097,
-	0x419, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x0a9, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x37e, 0x0e0, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x431,
-	0x372, 0xfff, 0xfff, 0xfff, 0x1ba, 0x06e, 0xfff, 0x1b1,
-	0xfff, 0xfff, 0x12a, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x193, 0xfff, 0xfff, 0xfff, 0xfff, 0x10a,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x048, 0x1b4,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x295, 0x140, 0x108, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x16f, 0xfff, 0x0a4, 0x37a, 0xfff,
-	0x29a, 0xfff, 0x284, 0xfff, 0xfff, 0xfff, 0xfff, 0x4c6,
-	0x2a2, 0x3a3, 0xfff, 0x201, 0xfff, 0xfff, 0xfff, 0x4bd,
-	0x005, 0x54a, 0x3b5, 0x204, 0x2ee, 0x11d, 0x436, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x3ec, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x11f, 0x498, 0x21c, 0xfff,
-	0xfff, 0xfff, 0x3d6, 0xfff, 0x4ab, 0xfff, 0x432, 0x2eb,
-	0x542, 0x4fd, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x4ce, 0xfff, 0xfff, 0x2fb, 0xfff,
-	0xfff, 0x2e1, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x1b9, 0x037, 0x0dd,
-	0xfff, 0xfff, 0xfff, 0x2bf, 0x521, 0x496, 0x095, 0xfff,
-	0xfff, 0x328, 0x070, 0x1bf, 0xfff, 0x393, 0xfff, 0xfff,
-	0x102, 0xfff, 0xfff, 0x21b, 0xfff, 0x142, 0x263, 0x519,
-	0xfff, 0x2a5, 0x177, 0xfff, 0x14d, 0x471, 0x4ae, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x1f6, 0xfff, 0x481, 0xfff, 0xfff, 0xfff, 0x151, 0xfff,
-	0xfff, 0xfff, 0x085, 0x33f, 0xfff, 0xfff, 0xfff, 0x084,
-	0xfff, 0xfff, 0xfff, 0x345, 0x3a2, 0xfff, 0xfff, 0x0a0,
-	0x0da, 0x024, 0xfff, 0xfff, 0xfff, 0x1bd, 0xfff, 0x55c,
-	0x467, 0x445, 0xfff, 0xfff, 0xfff, 0x052, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x51e, 0xfff, 0xfff, 0x39d, 0xfff, 0x35f,
-	0xfff, 0x376, 0x3ee, 0xfff, 0xfff, 0xfff, 0xfff, 0x448,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x16a,
-	0xfff, 0x036, 0x38f, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x211,
-	0xfff, 0xfff, 0xfff, 0x230, 0xfff, 0xfff, 0x3ba, 0xfff,
-	0xfff, 0xfff, 0x3ce, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x229, 0xfff, 0x176, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x00b, 0xfff, 0x162, 0x018, 0xfff,
-	0xfff, 0x233, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x400, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x12b, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x3f4, 0xfff, 0x0f0, 0xfff, 0x1ac, 0xfff, 0xfff,
-	0x119, 0xfff, 0x2c0, 0xfff, 0xfff, 0xfff, 0x49b, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x23c, 0xfff,
-	0x4b3, 0x010, 0x064, 0xfff, 0xfff, 0x4ba, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x3c2, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x006, 0x196, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x100, 0x191, 0xfff,
-	0x1ea, 0x29f, 0xfff, 0xfff, 0xfff, 0x276, 0xfff, 0xfff,
-	0x2b1, 0x3b9, 0xfff, 0x03c, 0xfff, 0xfff, 0xfff, 0x180,
-	0xfff, 0x08f, 0xfff, 0xfff, 0x19e, 0x019, 0xfff, 0x0b0,
-	0x0fd, 0x332, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x06b, 0x2e8, 0xfff, 0x446, 0xfff, 0xfff, 0x004,
-	0x247, 0x197, 0xfff, 0x112, 0x169, 0x292, 0xfff, 0x302,
-	0xfff, 0xfff, 0x33b, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x287, 0x21f, 0xfff, 0x3ea, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x4e7, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x3a8, 0xfff, 0xfff, 0x2bc, 0xfff,
-	0x484, 0x296, 0xfff, 0x1c9, 0x08c, 0x1e5, 0x48a, 0xfff,
-	0x360, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x1ca, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x10d, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x066, 0x2ea, 0x28b, 0x25b, 0xfff, 0x072,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x2b6, 0xfff, 0xfff, 0x272,
-	0xfff, 0xfff, 0x525, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x2ca, 0xfff, 0xfff, 0xfff, 0x299, 0xfff, 0xfff, 0xfff,
-	0x558, 0x41a, 0xfff, 0x4f7, 0x557, 0xfff, 0x4a0, 0x344,
-	0x12c, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x125,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x40e, 0xfff, 0xfff, 0x502, 0xfff, 0x103, 0x3e6, 0xfff,
-	0x527, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x45d, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x44e, 0xfff, 0xfff, 0xfff, 0xfff, 0x0d2, 0x4c9, 0x35e,
-	0x459, 0x2d9, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x17d,
-	0x0c4, 0xfff, 0xfff, 0xfff, 0x3ac, 0x390, 0x094, 0xfff,
-	0x483, 0x0ab, 0xfff, 0x253, 0xfff, 0x391, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x123, 0x0ef, 0xfff, 0xfff, 0xfff, 0x330,
-	0x38c, 0xfff, 0xfff, 0x2ae, 0xfff, 0xfff, 0xfff, 0x042,
-	0x012, 0x06d, 0xfff, 0xfff, 0xfff, 0x32a, 0x3db, 0x364,
-	0x2dc, 0xfff, 0x30f, 0x3d7, 0x4a5, 0x050, 0xfff, 0xfff,
-	0x029, 0xfff, 0xfff, 0xfff, 0xfff, 0x1d1, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x480, 0xfff,
-	0x4ed, 0x081, 0x0a1, 0xfff, 0xfff, 0xfff, 0x30e, 0x52f,
-	0x257, 0xfff, 0xfff, 0x447, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x401, 0x3cc, 0xfff, 0xfff, 0x0fb,
-	0x2c9, 0x42a, 0x314, 0x33e, 0x3bd, 0x318, 0xfff, 0x10e,
-	0x2a1, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x24c,
-	0x506, 0xfff, 0x267, 0xfff, 0xfff, 0x219, 0xfff, 0x1eb,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x309, 0x3e2, 0x46c, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x384, 0xfff, 0xfff, 0xfff, 0xfff, 0x50c, 0xfff, 0x24b,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x038,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x194,
-	0x143, 0x3e3, 0xfff, 0xfff, 0xfff, 0x4c2, 0xfff, 0xfff,
-	0x0e1, 0x25c, 0xfff, 0x237, 0xfff, 0x1fe, 0xfff, 0xfff,
-	0xfff, 0x065, 0x2a4, 0xfff, 0x386, 0x55a, 0x11b, 0xfff,
-	0xfff, 0x192, 0xfff, 0x183, 0x00e, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x4b2, 0x18e, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x486, 0x4ef, 0x0c6, 0x380, 0xfff, 0x4a8, 0xfff,
-	0x0c5, 0xfff, 0xfff, 0xfff, 0xfff, 0x093, 0x1b8, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x2e6,
-	0xfff, 0x0f3, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x28e, 0xfff, 0x53b, 0x420, 0x22a, 0x33a, 0xfff, 0x387,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x2a3, 0xfff, 0xfff,
-	0xfff, 0x428, 0x500, 0xfff, 0xfff, 0x120, 0x2c6, 0x290,
-	0x2f5, 0x0e3, 0xfff, 0x0b7, 0xfff, 0x319, 0x474, 0xfff,
-	0xfff, 0xfff, 0x529, 0x014, 0xfff, 0x41b, 0x40a, 0x18b,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x0d9,
-	0xfff, 0x38a, 0xfff, 0xfff, 0xfff, 0xfff, 0x1ce, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x3b1, 0xfff, 0xfff, 0x05d,
-	0x2c4, 0xfff, 0xfff, 0x4af, 0xfff, 0x030, 0xfff, 0xfff,
-	0x203, 0xfff, 0x277, 0x256, 0xfff, 0xfff, 0xfff, 0x4f9,
-	0xfff, 0x2c7, 0xfff, 0x466, 0x016, 0x1cd, 0xfff, 0x167,
-	0xfff, 0xfff, 0x0c8, 0xfff, 0x43d, 0xfff, 0xfff, 0x020,
-	0xfff, 0xfff, 0x232, 0x1cb, 0x1e0, 0xfff, 0xfff, 0x347,
-	0xfff, 0x478, 0xfff, 0x365, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x358, 0xfff, 0x10b, 0xfff, 0x35d, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x452, 0x22d, 0xfff, 0xfff, 0x47d, 0xfff,
-	0x2f3, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x460, 0xfff,
-	0xfff, 0xfff, 0x50b, 0xfff, 0xfff, 0xfff, 0x2ec, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x4b1, 0x422, 0xfff, 0xfff,
-	0xfff, 0x2d4, 0xfff, 0x239, 0xfff, 0xfff, 0xfff, 0x439,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x491, 0x075, 0xfff, 0xfff, 0xfff, 0x06c, 0xfff,
-	0xfff, 0x0f9, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x139, 0xfff, 0x4f6, 0xfff, 0xfff, 0x409, 0xfff,
-	0xfff, 0x15b, 0xfff, 0xfff, 0x348, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x4a2, 0x49d, 0xfff, 0x033, 0x175, 0xfff, 0x039,
-	0xfff, 0x312, 0x40c, 0xfff, 0xfff, 0x325, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x4aa, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0x165, 0x3bc, 0x48c, 0x310, 0x096,
-	0xfff, 0xfff, 0x250, 0x1a2, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x20d, 0x2ac, 0xfff, 0xfff, 0x39b, 0xfff, 0x377, 0xfff,
-	0x512, 0x495, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x357, 0x4ea, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x198, 0xfff, 0xfff, 0xfff, 0x434, 0x04a,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x062, 0xfff, 0x1d6, 0x1c8,
-	0xfff, 0x1f3, 0x281, 0xfff, 0x462, 0xfff, 0xfff, 0xfff,
-	0x4b0, 0xfff, 0x207, 0xfff, 0xfff, 0xfff, 0xfff, 0x3dd,
-	0xfff, 0xfff, 0x55d, 0xfff, 0x552, 0x494, 0x1af, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x227, 0xfff, 0xfff, 0x069,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x43e,
-	0x0b5, 0xfff, 0x524, 0x2d2, 0xfff, 0xfff, 0xfff, 0x28f,
-	0xfff, 0x01b, 0x50e, 0xfff, 0xfff, 0x1bb, 0xfff, 0xfff,
-	0x41d, 0xfff, 0x32e, 0x48e, 0xfff, 0x1f7, 0x224, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x394, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x52c, 0xfff, 0xfff, 0xfff, 0x392, 0xfff, 0x1e7,
-	0xfff, 0xfff, 0x3f9, 0x3a7, 0xfff, 0x51f, 0xfff, 0x0bb,
-	0x118, 0x3ca, 0xfff, 0x1dd, 0xfff, 0x48b, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x50f, 0xfff, 0x0d6, 0xfff, 0x1fa, 0xfff,
-	0x11e, 0xfff, 0xfff, 0xfff, 0xfff, 0x4d7, 0xfff, 0x078,
-	0x008, 0xfff, 0x25d, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x032, 0x33c, 0xfff, 0x4d9, 0x160, 0xfff, 0xfff, 0x300,
-	0x0b1, 0xfff, 0x322, 0xfff, 0x4ec, 0xfff, 0xfff, 0x200,
-	0x00c, 0x369, 0x473, 0xfff, 0xfff, 0x32c, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x53e, 0x3d4, 0x417, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x34b, 0x001, 0x39a, 0x02c, 0xfff, 0xfff, 0x2ce, 0x00f,
-	0xfff, 0x0ba, 0xfff, 0xfff, 0xfff, 0xfff, 0x060, 0xfff,
-	0x406, 0xfff, 0xfff, 0xfff, 0x4ee, 0x4ac, 0xfff, 0x43f,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x29b, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x216,
-	0x190, 0xfff, 0x396, 0x464, 0xfff, 0xfff, 0x323, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x2e9, 0xfff, 0x26d,
-	0x2cd, 0x040, 0xfff, 0xfff, 0xfff, 0xfff, 0x38b, 0x3c0,
-	0xfff, 0xfff, 0xfff, 0x1f2, 0xfff, 0x0ea, 0xfff, 0xfff,
-	0x472, 0xfff, 0x1fb, 0xfff, 0xfff, 0x0af, 0x27f, 0xfff,
-	0xfff, 0xfff, 0x479, 0x023, 0xfff, 0x0d8, 0x3b3, 0xfff,
-	0xfff, 0xfff, 0x121, 0xfff, 0xfff, 0x3bf, 0xfff, 0xfff,
-	0x16b, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x45a, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x0be, 0xfff, 0xfff, 0xfff, 0x111, 0xfff, 0x220,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x09b, 0x218, 0xfff, 0x022, 0x202, 0xfff,
-	0x4c8, 0xfff, 0x0ed, 0xfff, 0xfff, 0x182, 0xfff, 0xfff,
-	0xfff, 0x17f, 0x213, 0xfff, 0x321, 0x36a, 0xfff, 0x086,
-	0xfff, 0xfff, 0xfff, 0x43b, 0x088, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x26c, 0xfff, 0x2f8, 0x3b4, 0xfff, 0xfff, 0xfff,
-	0x132, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x333, 0x444,
-	0x0c1, 0x4d8, 0x46d, 0x264, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x426, 0xfff, 0xfff, 0xfff, 0xfff, 0x2fe, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x011, 0xfff, 0x05f, 0xfff, 0xfff, 0xfff,
-	0xfff, 0x10c, 0x101, 0xfff, 0xfff, 0xfff, 0xfff, 0x110,
-	0xfff, 0x044, 0x304, 0x361, 0x404, 0xfff, 0x51b, 0x099,
-	0xfff, 0x440, 0xfff, 0xfff, 0xfff, 0x222, 0xfff, 0xfff,
-	0xfff, 0xfff, 0x1b5, 0xfff, 0x136, 0x430, 0xfff, 0x1da,
-	0xfff, 0xfff, 0xfff, 0x043, 0xfff, 0x17c, 0xfff, 0xfff,
-	0xfff, 0x01c, 0xfff, 0xfff, 0xfff, 0x425, 0x236, 0xfff,
-	0x317, 0xfff, 0xfff, 0x437, 0x3fc, 0xfff, 0x1f1, 0xfff,
-	0x324, 0xfff, 0xfff, 0x0ca, 0x306, 0xfff, 0x548, 0xfff,
-	0x46e, 0xfff, 0xfff, 0xfff, 0x4b8, 0x1c2, 0x286, 0xfff,
-	0xfff, 0x087, 0x18a, 0x19f, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x18c, 0xfff, 0x215, 0xfff, 0xfff, 0xfff, 0xfff, 0x283,
-	0xfff, 0xfff, 0xfff, 0x126, 0xfff, 0xfff, 0x370, 0xfff,
-	0x53f, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0x31c, 0xfff,
-	0x4d1, 0xfff, 0xfff, 0xfff, 0x021, 0xfff, 0x157, 0xfff,
-	0xfff, 0x028, 0x16e, 0xfff, 0x421, 0xfff, 0x1c6, 0xfff,
-	0xfff, 0x511, 0xfff, 0xfff, 0x39c, 0x46f, 0x1b2, 0xfff,
-	0xfff, 0x316, 0xfff, 0xfff, 0x009, 0xfff, 0xfff, 0x195,
-	0xfff, 0x240, 0x546, 0xfff, 0xfff, 0x520, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x454, 0xfff, 0xfff, 0xfff,
-	0x3f3, 0xfff, 0xfff, 0x187, 0xfff, 0x4a9, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x51c, 0x453, 0x1e6, 0xfff,
-	0xfff, 0xfff, 0x1b0, 0xfff, 0x477, 0xfff, 0xfff, 0xfff,
-	0x4fe, 0xfff, 0x32f, 0xfff, 0xfff, 0x15e, 0x1d4, 0xfff,
-	0x0e5, 0xfff, 0xfff, 0xfff, 0x242, 0x14b, 0x046, 0xfff,
-	0x3f6, 0x3bb, 0x3e4, 0xfff, 0xfff, 0x2e3, 0xfff, 0x245,
-	0xfff, 0x149, 0xfff, 0xfff, 0xfff, 0x2db, 0xfff, 0xfff,
-	0x181, 0xfff, 0x089, 0x2c5, 0xfff, 0x1f5, 0xfff, 0x2d6,
-	0x507, 0xfff, 0x42d, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0x080, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff, 0xfff,
-	0xfff, 0xfff, 0xfff, 0xfff, 0x3c3, 0x320, 0xfff, 0x1e1,
-	0xfff, 0x0f5, 0x13b, 0xfff, 0xfff, 0xfff, 0x003, 0x4da,
-	0xfff, 0xfff, 0xfff, 0x42c, 0xfff, 0xfff, 0x0cb, 0xfff,
-	0x536, 0x2c3, 0xfff, 0xfff, 0xfff, 0xfff, 0x199, 0xfff,
-	0xfff, 0x0c0, 0xfff, 0x01e, 0x497, 0xfff, 0xfff, 0x3e5,
-	0xfff, 0xfff, 0xfff, 0x0cf, 0xfff, 0x2bd, 0xfff, 0x223,
-	0xfff, 0x3ff, 0xfff, 0x058, 0x174, 0x3ef, 0xfff, 0x002
-};
-
-static unsigned short err_pos(unsigned short din)
-{
-	BUG_ON(din >= ARRAY_SIZE(err_pos_lut));
-	return err_pos_lut[din];
-}
-static int chk_no_err_only(unsigned short *chk_syndrome_list, unsigned short *err_info)
-{
-	if ((chk_syndrome_list[0] | chk_syndrome_list[1] |
-	     chk_syndrome_list[2] | chk_syndrome_list[3] |
-	     chk_syndrome_list[4] | chk_syndrome_list[5] |
-	     chk_syndrome_list[6] | chk_syndrome_list[7]) != 0x0) {
-		return -EINVAL;
-	} else {
-		err_info[0] = 0x0;
-		return 0;
-	}
-}
-static int chk_1_err_only(unsigned short *chk_syndrome_list, unsigned short *err_info)
-{
-	unsigned short tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6;
-	tmp0 = gf4096_mul(chk_syndrome_list[1], gf4096_inv(chk_syndrome_list[0]));
-	tmp1 = gf4096_mul(chk_syndrome_list[2], gf4096_inv(chk_syndrome_list[1]));
-	tmp2 = gf4096_mul(chk_syndrome_list[3], gf4096_inv(chk_syndrome_list[2]));
-	tmp3 = gf4096_mul(chk_syndrome_list[4], gf4096_inv(chk_syndrome_list[3]));
-	tmp4 = gf4096_mul(chk_syndrome_list[5], gf4096_inv(chk_syndrome_list[4]));
-	tmp5 = gf4096_mul(chk_syndrome_list[6], gf4096_inv(chk_syndrome_list[5]));
-	tmp6 = gf4096_mul(chk_syndrome_list[7], gf4096_inv(chk_syndrome_list[6]));
-	if ((tmp0 == tmp1) & (tmp1 == tmp2) & (tmp2 == tmp3) & (tmp3 == tmp4) & (tmp4 == tmp5) & (tmp5 == tmp6)) {
-		err_info[0] = 0x1;	// encode 1-symbol error as 0x1
-		err_info[1] = err_pos(tmp0);
-		err_info[1] = (unsigned short)(0x55e - err_info[1]);
-		err_info[5] = chk_syndrome_list[0];
-		return 0;
-	} else
-		return -EINVAL;
-}
-static int chk_2_err_only(unsigned short *chk_syndrome_list, unsigned short *err_info)
-{
-	unsigned short tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
-	unsigned short coefs[4];
-	unsigned short err_pats[4];
-	int found_num_root = 0;
-	unsigned short bit2_root0, bit2_root1;
-	unsigned short bit2_root0_inv, bit2_root1_inv;
-	unsigned short err_loc_eqn, test_root;
-	unsigned short bit2_loc0, bit2_loc1;
-	unsigned short bit2_pat0, bit2_pat1;
-
-	find_2x2_soln(chk_syndrome_list[1],
-		      chk_syndrome_list[0],
-		      chk_syndrome_list[2], chk_syndrome_list[1], chk_syndrome_list[2], chk_syndrome_list[3], coefs);
-	for (test_root = 0x1; test_root < 0xfff; test_root++) {
-		err_loc_eqn =
-		    gf4096_mul(coefs[1], gf4096_mul(test_root, test_root)) ^ gf4096_mul(coefs[0], test_root) ^ 0x1;
-		if (err_loc_eqn == 0x0) {
-			if (found_num_root == 0) {
-				bit2_root0 = test_root;
-				found_num_root = 1;
-			} else if (found_num_root == 1) {
-				bit2_root1 = test_root;
-				found_num_root = 2;
-				break;
-			}
-		}
-	}
-	if (found_num_root != 2)
-		return -EINVAL;
-	else {
-		bit2_root0_inv = gf4096_inv(bit2_root0);
-		bit2_root1_inv = gf4096_inv(bit2_root1);
-		find_2bit_err_pats(chk_syndrome_list[0],
-				   chk_syndrome_list[1], bit2_root0_inv, bit2_root1_inv, err_pats);
-		bit2_pat0 = err_pats[0];
-		bit2_pat1 = err_pats[1];
-		//for(x+1)
-		tmp0 = gf4096_mul(gf4096_mul(bit2_root0_inv, bit2_root0_inv), gf4096_mul(bit2_root0_inv, bit2_root0_inv));	//rinv0^4
-		tmp1 = gf4096_mul(bit2_root0_inv, tmp0);	//rinv0^5
-		tmp2 = gf4096_mul(bit2_root0_inv, tmp1);	//rinv0^6
-		tmp3 = gf4096_mul(bit2_root0_inv, tmp2);	//rinv0^7
-		tmp4 = gf4096_mul(gf4096_mul(bit2_root1_inv, bit2_root1_inv), gf4096_mul(bit2_root1_inv, bit2_root1_inv));	//rinv1^4
-		tmp5 = gf4096_mul(bit2_root1_inv, tmp4);	//rinv1^5
-		tmp6 = gf4096_mul(bit2_root1_inv, tmp5);	//rinv1^6
-		tmp7 = gf4096_mul(bit2_root1_inv, tmp6);	//rinv1^7
-		//check if only 2-bit error
-		if ((chk_syndrome_list[4] ==
-		     (gf4096_mul(bit2_pat0, tmp0) ^
-		      gf4096_mul(bit2_pat1,
-				 tmp4))) & (chk_syndrome_list[5] ==
-					    (gf4096_mul(bit2_pat0, tmp1) ^
-					     gf4096_mul(bit2_pat1,
-							tmp5))) &
-		    (chk_syndrome_list[6] ==
-		     (gf4096_mul(bit2_pat0, tmp2) ^
-		      gf4096_mul(bit2_pat1,
-				 tmp6))) & (chk_syndrome_list[7] ==
-					    (gf4096_mul(bit2_pat0, tmp3) ^ gf4096_mul(bit2_pat1, tmp7)))) {
-			if ((err_pos(bit2_root0_inv) == 0xfff) | (err_pos(bit2_root1_inv) == 0xfff)) {
-				return -EINVAL;
-			} else {
-				bit2_loc0 = 0x55e - err_pos(bit2_root0_inv);
-				bit2_loc1 = 0x55e - err_pos(bit2_root1_inv);
-				err_info[0] = 0x2;	// encode 2-symbol error as 0x2
-				err_info[1] = bit2_loc0;
-				err_info[2] = bit2_loc1;
-				err_info[5] = bit2_pat0;
-				err_info[6] = bit2_pat1;
-				return 0;
-			}
-		} else
-			return -EINVAL;
-	}
-}
-static int chk_3_err_only(unsigned short *chk_syndrome_list, unsigned short *err_info)
-{
-	unsigned short tmp0, tmp1, tmp2, tmp3, tmp4, tmp5;
-	unsigned short coefs[4];
-	unsigned short err_pats[4];
-	int found_num_root = 0;
-	unsigned short bit3_root0, bit3_root1, bit3_root2;
-	unsigned short bit3_root0_inv, bit3_root1_inv, bit3_root2_inv;
-	unsigned short err_loc_eqn, test_root;
-
-	find_3bit_err_coefs(chk_syndrome_list[0], chk_syndrome_list[1],
-			    chk_syndrome_list[2], chk_syndrome_list[3],
-			    chk_syndrome_list[4], chk_syndrome_list[5], coefs);
-
-	for (test_root = 0x1; test_root < 0xfff; test_root++) {
-		err_loc_eqn = gf4096_mul(coefs[2],
-					 gf4096_mul(gf4096_mul(test_root, test_root),
-						    test_root)) ^ gf4096_mul(coefs[1], gf4096_mul(test_root, test_root))
-			^ gf4096_mul(coefs[0], test_root) ^ 0x1;
-
-		if (err_loc_eqn == 0x0) {
-			if (found_num_root == 0) {
-				bit3_root0 = test_root;
-				found_num_root = 1;
-			} else if (found_num_root == 1) {
-				bit3_root1 = test_root;
-				found_num_root = 2;
-			} else if (found_num_root == 2) {
-				bit3_root2 = test_root;
-				found_num_root = 3;
-				break;
-			}
-		}
-	}
-	if (found_num_root != 3)
-		return -EINVAL;
-	else {
-		bit3_root0_inv = gf4096_inv(bit3_root0);
-		bit3_root1_inv = gf4096_inv(bit3_root1);
-		bit3_root2_inv = gf4096_inv(bit3_root2);
-
-		find_3bit_err_pats(chk_syndrome_list[0], chk_syndrome_list[1],
-				   chk_syndrome_list[2], bit3_root0_inv,
-				   bit3_root1_inv, bit3_root2_inv, err_pats);
-
-		//check if only 3-bit error
-		tmp0 = gf4096_mul(bit3_root0_inv, bit3_root0_inv);
-		tmp0 = gf4096_mul(tmp0, tmp0);
-		tmp0 = gf4096_mul(tmp0, bit3_root0_inv);
-		tmp0 = gf4096_mul(tmp0, bit3_root0_inv);	//rinv0^6
-		tmp1 = gf4096_mul(tmp0, bit3_root0_inv);	//rinv0^7
-		tmp2 = gf4096_mul(bit3_root1_inv, bit3_root1_inv);
-		tmp2 = gf4096_mul(tmp2, tmp2);
-		tmp2 = gf4096_mul(tmp2, bit3_root1_inv);
-		tmp2 = gf4096_mul(tmp2, bit3_root1_inv);	//rinv1^6
-		tmp3 = gf4096_mul(tmp2, bit3_root1_inv);	//rinv1^7
-		tmp4 = gf4096_mul(bit3_root2_inv, bit3_root2_inv);
-		tmp4 = gf4096_mul(tmp4, tmp4);
-		tmp4 = gf4096_mul(tmp4, bit3_root2_inv);
-		tmp4 = gf4096_mul(tmp4, bit3_root2_inv);	//rinv2^6
-		tmp5 = gf4096_mul(tmp4, bit3_root2_inv);	//rinv2^7
-
-		//check if only 3 errors
-		if ((chk_syndrome_list[6] == (gf4096_mul(err_pats[0], tmp0) ^
-					      gf4096_mul(err_pats[1], tmp2) ^
-					      gf4096_mul(err_pats[2], tmp4))) &
-		    (chk_syndrome_list[7] == (gf4096_mul(err_pats[0], tmp1) ^
-					      gf4096_mul(err_pats[1], tmp3) ^ gf4096_mul(err_pats[2], tmp5)))) {
-			if ((err_pos(bit3_root0_inv) == 0xfff) |
-			    (err_pos(bit3_root1_inv) == 0xfff) | (err_pos(bit3_root2_inv) == 0xfff)) {
-				return -EINVAL;
-			} else {
-				err_info[0] = 0x3;
-				err_info[1] = (0x55e - err_pos(bit3_root0_inv));
-				err_info[2] = (0x55e - err_pos(bit3_root1_inv));
-				err_info[3] = (0x55e - err_pos(bit3_root2_inv));
-				err_info[5] = err_pats[0];
-				err_info[6] = err_pats[1];
-				err_info[7] = err_pats[2];
-				return 0;
-			}
-		} else
-			return -EINVAL;
-	}
-}
-static int chk_4_err_only(unsigned short *chk_syndrome_list, unsigned short *err_info)
-{
-	unsigned short coefs[4];
-	unsigned short err_pats[4];
-	int found_num_root = 0;
-	unsigned short bit4_root0, bit4_root1, bit4_root2, bit4_root3;
-	unsigned short bit4_root0_inv, bit4_root1_inv, bit4_root2_inv, bit4_root3_inv;
-	unsigned short err_loc_eqn, test_root;
-
-	find_4bit_err_coefs(chk_syndrome_list[0],
-			    chk_syndrome_list[1],
-			    chk_syndrome_list[2],
-			    chk_syndrome_list[3],
-			    chk_syndrome_list[4],
-			    chk_syndrome_list[5], chk_syndrome_list[6], chk_syndrome_list[7], coefs);
-
-	for (test_root = 0x1; test_root < 0xfff; test_root++) {
-		err_loc_eqn =
-		    gf4096_mul(coefs[3],
-			       gf4096_mul(gf4096_mul
-					  (gf4096_mul(test_root, test_root),
-					   test_root),
-					  test_root)) ^ gf4096_mul(coefs[2],
-								   gf4096_mul
-								   (gf4096_mul(test_root, test_root), test_root))
-		    ^ gf4096_mul(coefs[1], gf4096_mul(test_root, test_root)) ^ gf4096_mul(coefs[0], test_root)
-		    ^ 0x1;
-		if (err_loc_eqn == 0x0) {
-			if (found_num_root == 0) {
-				bit4_root0 = test_root;
-				found_num_root = 1;
-			} else if (found_num_root == 1) {
-				bit4_root1 = test_root;
-				found_num_root = 2;
-			} else if (found_num_root == 2) {
-				bit4_root2 = test_root;
-				found_num_root = 3;
-			} else {
-				found_num_root = 4;
-				bit4_root3 = test_root;
-				break;
-			}
-		}
-	}
-	if (found_num_root != 4) {
-		return -EINVAL;
-	} else {
-		bit4_root0_inv = gf4096_inv(bit4_root0);
-		bit4_root1_inv = gf4096_inv(bit4_root1);
-		bit4_root2_inv = gf4096_inv(bit4_root2);
-		bit4_root3_inv = gf4096_inv(bit4_root3);
-		find_4bit_err_pats(chk_syndrome_list[0],
-				   chk_syndrome_list[1],
-				   chk_syndrome_list[2],
-				   chk_syndrome_list[3],
-				   bit4_root0_inv, bit4_root1_inv, bit4_root2_inv, bit4_root3_inv, err_pats);
-		err_info[0] = 0x4;
-		err_info[1] = (0x55e - err_pos(bit4_root0_inv));
-		err_info[2] = (0x55e - err_pos(bit4_root1_inv));
-		err_info[3] = (0x55e - err_pos(bit4_root2_inv));
-		err_info[4] = (0x55e - err_pos(bit4_root3_inv));
-		err_info[5] = err_pats[0];
-		err_info[6] = err_pats[1];
-		err_info[7] = err_pats[2];
-		err_info[8] = err_pats[3];
-		return 0;
-	}
-}
-
-void correct_12bit_symbol(unsigned char *buf, unsigned short sym,
-			  unsigned short val)
-{
-	if (unlikely(sym > 1366)) {
-		printk(KERN_ERR "Error: symbol %d out of range; cannot correct\n", sym);
-	} else if (sym == 0) {
-		buf[0] ^= val;
-	} else if (sym & 1) {
-		buf[1+(3*(sym-1))/2] ^= (val >> 4);
-		buf[2+(3*(sym-1))/2] ^= ((val & 0xf) << 4);
-	} else {
-		buf[2+(3*(sym-2))/2] ^= (val >> 8);
-		buf[3+(3*(sym-2))/2] ^= (val & 0xff);
-	}
-}
-
-static int debugecc = 0;
-module_param(debugecc, int, 0644);
-
-int cafe_correct_ecc(unsigned char *buf,
-		     unsigned short *chk_syndrome_list)
-{
-	unsigned short err_info[9];
-	int i;
-
-	if (debugecc) {
-		printk(KERN_WARNING "cafe_correct_ecc invoked. Syndromes %x %x %x %x %x %x %x %x\n",
-		       chk_syndrome_list[0], chk_syndrome_list[1],
-		       chk_syndrome_list[2], chk_syndrome_list[3],
-		       chk_syndrome_list[4], chk_syndrome_list[5],
-		       chk_syndrome_list[6], chk_syndrome_list[7]);
-		for (i=0; i < 2048; i+=16) {
-			printk(KERN_WARNING "D %04x: %02x %02x %02x %02x %02x %02x %02x %02x  %02x %02x %02x %02x %02x %02x %02x %02x\n",
-			       i,
-			       buf[i], buf[i+1], buf[i+2], buf[i+3],
-			       buf[i+4], buf[i+5], buf[i+6], buf[i+7],
-			       buf[i+8], buf[i+9], buf[i+10], buf[i+11],
-			       buf[i+12], buf[i+13], buf[i+14], buf[i+15]);
-		}
-		for ( ; i < 2112; i+=16) {
-			printk(KERN_WARNING "O   %02x: %02x %02x %02x %02x %02x %02x %02x %02x  %02x %02x %02x %02x %02x %02x %02x %02x\n",
-			       i - 2048,
-			       buf[i], buf[i+1], buf[i+2], buf[i+3],
-			       buf[i+4], buf[i+5], buf[i+6], buf[i+7],
-			       buf[i+8], buf[i+9], buf[i+10], buf[i+11],
-			       buf[i+12], buf[i+13], buf[i+14], buf[i+15]);
-		}
-	}
-
-
-
-	if (chk_no_err_only(chk_syndrome_list, err_info) &&
-	    chk_1_err_only(chk_syndrome_list, err_info) &&
-	    chk_2_err_only(chk_syndrome_list, err_info) &&
-	    chk_3_err_only(chk_syndrome_list, err_info) &&
-	    chk_4_err_only(chk_syndrome_list, err_info)) {
-		return -EIO;
-	}
-
-	for (i=0; i < err_info[0]; i++) {
-		if (debugecc)
-			printk(KERN_WARNING "Correct symbol %d with 0x%03x\n",
-			       err_info[1+i], err_info[5+i]);
-
-		correct_12bit_symbol(buf, err_info[1+i], err_info[5+i]);
-	}
-
-	return err_info[0];
-}
-
diff --git a/drivers/mtd/nand/cafe.c b/drivers/mtd/nand/cafe_nand.c
index c328a7514510..cff969d05d4a 100644
--- a/drivers/mtd/nand/cafe.c
+++ b/drivers/mtd/nand/cafe_nand.c
@@ -11,6 +11,7 @@
 #undef DEBUG
 #include <linux/mtd/mtd.h>
 #include <linux/mtd/nand.h>
+#include <linux/rslib.h>
 #include <linux/pci.h>
 #include <linux/delay.h>
 #include <linux/interrupt.h>
@@ -46,13 +47,14 @@
 #define CAFE_GLOBAL_IRQ_MASK	0x300c
 #define CAFE_NAND_RESET		0x3034
 
-int cafe_correct_ecc(unsigned char *buf,
-		     unsigned short *chk_syndrome_list);
+/* Missing from the datasheet: bit 19 of CTRL1 sets CE0 vs. CE1 */
+#define CTRL1_CHIPSELECT	(1<<19)
 
 struct cafe_priv {
 	struct nand_chip nand;
 	struct pci_dev *pdev;
 	void __iomem *mmio;
+	struct rs_control *rs;
 	uint32_t ctl1;
 	uint32_t ctl2;
 	int datalen;
@@ -195,8 +197,8 @@ static void cafe_nand_cmdfunc(struct mtd_info *mtd, unsigned command,
 
 	cafe->data_pos = cafe->datalen = 0;
 
-	/* Set command valid bit */
-	ctl1 = 0x80000000 | command;
+	/* Set command valid bit, mask in the chip select bit  */
+	ctl1 = 0x80000000 | command | (cafe->ctl1 & CTRL1_CHIPSELECT);
 
 	/* Set RD or WR bits as appropriate */
 	if (command == NAND_CMD_READID || command == NAND_CMD_STATUS) {
@@ -309,8 +311,16 @@ static void cafe_nand_cmdfunc(struct mtd_info *mtd, unsigned command,
 
 static void cafe_select_chip(struct mtd_info *mtd, int chipnr)
 {
-	//struct cafe_priv *cafe = mtd->priv;
-	//	cafe_dev_dbg(&cafe->pdev->dev, "select_chip %d\n", chipnr);
+	struct cafe_priv *cafe = mtd->priv;
+
+	cafe_dev_dbg(&cafe->pdev->dev, "select_chip %d\n", chipnr);
+
+	/* Mask the appropriate bit into the stored value of ctl1
+	   which will be used by cafe_nand_cmdfunc() */
+	if (chipnr)
+		cafe->ctl1 |= CTRL1_CHIPSELECT;
+	else
+		cafe->ctl1 &= ~CTRL1_CHIPSELECT;
 }
 
 static int cafe_nand_interrupt(int irq, void *id)
@@ -374,28 +384,66 @@ static int cafe_nand_read_page(struct mtd_info *mtd, struct nand_chip *chip,
 	chip->read_buf(mtd, chip->oob_poi, mtd->oobsize);
 
 	if (checkecc && cafe_readl(cafe, NAND_ECC_RESULT) & (1<<18)) {
-		unsigned short syn[8];
-		int i;
+		unsigned short syn[8], pat[4];
+		int pos[4];
+		u8 *oob = chip->oob_poi;
+		int i, n;
 
 		for (i=0; i<8; i+=2) {
 			uint32_t tmp = cafe_readl(cafe, NAND_ECC_SYN01 + (i*2));
-			syn[i] = tmp & 0xfff;
-			syn[i+1] = (tmp >> 16) & 0xfff;
+			syn[i] = cafe->rs->index_of[tmp & 0xfff];
+			syn[i+1] = cafe->rs->index_of[(tmp >> 16) & 0xfff];
+		}
+
+		n = decode_rs16(cafe->rs, NULL, NULL, 1367, syn, 0, pos, 0,
+		                pat);
+
+		for (i = 0; i < n; i++) {
+			int p = pos[i];
+
+			/* The 12-bit symbols are mapped to bytes here */
+
+			if (p > 1374) {
+				/* out of range */
+				n = -1374;
+			} else if (p == 0) {
+				/* high four bits do not correspond to data */
+				if (pat[i] > 0xff)
+					n = -2048;
+				else
+					buf[0] ^= pat[i];
+			} else if (p == 1365) {
+				buf[2047] ^= pat[i] >> 4;
+				oob[0] ^= pat[i] << 4;
+			} else if (p > 1365) {
+				if ((p & 1) == 1) {
+					oob[3*p/2 - 2048] ^= pat[i] >> 4;
+					oob[3*p/2 - 2047] ^= pat[i] << 4;
+				} else {
+					oob[3*p/2 - 2049] ^= pat[i] >> 8;
+					oob[3*p/2 - 2048] ^= pat[i];
+				}
+			} else if ((p & 1) == 1) {
+				buf[3*p/2] ^= pat[i] >> 4;
+				buf[3*p/2 + 1] ^= pat[i] << 4;
+			} else {
+				buf[3*p/2 - 1] ^= pat[i] >> 8;
+				buf[3*p/2] ^= pat[i];
+			}
 		}
 
-		if ((i = cafe_correct_ecc(buf, syn)) < 0) {
+		if (n < 0) {
 			dev_dbg(&cafe->pdev->dev, "Failed to correct ECC at %08x\n",
 				cafe_readl(cafe, NAND_ADDR2) * 2048);
-			for (i=0; i< 0x5c; i+=4)
+			for (i = 0; i < 0x5c; i += 4)
 				printk("Register %x: %08x\n", i, readl(cafe->mmio + i));
 			mtd->ecc_stats.failed++;
 		} else {
-			dev_dbg(&cafe->pdev->dev, "Corrected %d symbol errors\n", i);
-			mtd->ecc_stats.corrected += i;
+			dev_dbg(&cafe->pdev->dev, "Corrected %d symbol errors\n", n);
+			mtd->ecc_stats.corrected += n;
 		}
 	}
 
-
 	return 0;
 }
 
@@ -416,7 +464,7 @@ static uint8_t cafe_mirror_pattern_512[] = { 0xBC };
 
 static struct nand_bbt_descr cafe_bbt_main_descr_2048 = {
 	.options = NAND_BBT_LASTBLOCK | NAND_BBT_CREATE | NAND_BBT_WRITE
-		| NAND_BBT_2BIT | NAND_BBT_VERSION | NAND_BBT_PERCHIP,
+		| NAND_BBT_2BIT | NAND_BBT_VERSION,
 	.offs =	14,
 	.len = 4,
 	.veroffs = 18,
@@ -426,7 +474,7 @@ static struct nand_bbt_descr cafe_bbt_main_descr_2048 = {
 
 static struct nand_bbt_descr cafe_bbt_mirror_descr_2048 = {
 	.options = NAND_BBT_LASTBLOCK | NAND_BBT_CREATE | NAND_BBT_WRITE
-		| NAND_BBT_2BIT | NAND_BBT_VERSION | NAND_BBT_PERCHIP,
+		| NAND_BBT_2BIT | NAND_BBT_VERSION,
 	.offs =	14,
 	.len = 4,
 	.veroffs = 18,
@@ -442,7 +490,7 @@ static struct nand_ecclayout cafe_oobinfo_512 = {
 
 static struct nand_bbt_descr cafe_bbt_main_descr_512 = {
 	.options = NAND_BBT_LASTBLOCK | NAND_BBT_CREATE | NAND_BBT_WRITE
-		| NAND_BBT_2BIT | NAND_BBT_VERSION | NAND_BBT_PERCHIP,
+		| NAND_BBT_2BIT | NAND_BBT_VERSION,
 	.offs =	14,
 	.len = 1,
 	.veroffs = 15,
@@ -452,7 +500,7 @@ static struct nand_bbt_descr cafe_bbt_main_descr_512 = {
 
 static struct nand_bbt_descr cafe_bbt_mirror_descr_512 = {
 	.options = NAND_BBT_LASTBLOCK | NAND_BBT_CREATE | NAND_BBT_WRITE
-		| NAND_BBT_2BIT | NAND_BBT_VERSION | NAND_BBT_PERCHIP,
+		| NAND_BBT_2BIT | NAND_BBT_VERSION,
 	.offs =	14,
 	.len = 1,
 	.veroffs = 15,
@@ -525,6 +573,48 @@ static int cafe_nand_block_bad(struct mtd_info *mtd, loff_t ofs, int getchip)
 	return 0;
 }
 
+/* F_2[X]/(X**6+X+1)  */
+static unsigned short __devinit gf64_mul(u8 a, u8 b)
+{
+	u8 c;
+	unsigned int i;
+
+	c = 0;
+	for (i = 0; i < 6; i++) {
+		if (a & 1)
+			c ^= b;
+		a >>= 1;
+		b <<= 1;
+		if ((b & 0x40) != 0)
+			b ^= 0x43;
+	}
+
+	return c;
+}
+
+/* F_64[X]/(X**2+X+A**-1) with A the generator of F_64[X]  */
+static u16 __devinit gf4096_mul(u16 a, u16 b)
+{
+	u8 ah, al, bh, bl, ch, cl;
+
+	ah = a >> 6;
+	al = a & 0x3f;
+	bh = b >> 6;
+	bl = b & 0x3f;
+
+	ch = gf64_mul(ah ^ al, bh ^ bl) ^ gf64_mul(al, bl);
+	cl = gf64_mul(gf64_mul(ah, bh), 0x21) ^ gf64_mul(al, bl);
+
+	return (ch << 6) ^ cl;
+}
+
+static int __devinit cafe_mul(int x)
+{
+	if (x == 0)
+		return 1;
+	return gf4096_mul(x, 0xe01);
+}
+
 static int __devinit cafe_nand_probe(struct pci_dev *pdev,
 				     const struct pci_device_id *ent)
 {
@@ -564,6 +654,12 @@ static int __devinit cafe_nand_probe(struct pci_dev *pdev,
 	}
 	cafe->nand.buffers = (void *)cafe->dmabuf + 2112;
 
+	cafe->rs = init_rs_non_canonical(12, &cafe_mul, 0, 1, 8);
+	if (!cafe->rs) {
+		err = -ENOMEM;
+		goto out_ior;
+	}
+
 	cafe->nand.cmdfunc = cafe_nand_cmdfunc;
 	cafe->nand.dev_ready = cafe_device_ready;
 	cafe->nand.read_byte = cafe_read_byte;
@@ -646,7 +742,7 @@ static int __devinit cafe_nand_probe(struct pci_dev *pdev,
 		cafe_readl(cafe, GLOBAL_CTRL), cafe_readl(cafe, GLOBAL_IRQ_MASK));
 
 	/* Scan to find existence of the device */
-	if (nand_scan_ident(mtd, 1)) {
+	if (nand_scan_ident(mtd, 2)) {
 		err = -ENXIO;
 		goto out_irq;
 	}
@@ -713,6 +809,7 @@ static void __devexit cafe_nand_remove(struct pci_dev *pdev)
 	cafe_writel(cafe, ~1 & cafe_readl(cafe, GLOBAL_IRQ_MASK), GLOBAL_IRQ_MASK);
 	free_irq(pdev->irq, mtd);
 	nand_release(mtd);
+	free_rs(cafe->rs);
 	pci_iounmap(pdev, cafe->mmio);
 	dma_free_coherent(&cafe->pdev->dev, 2112, cafe->dmabuf, cafe->dmaaddr);
 	kfree(mtd);
diff --git a/drivers/mtd/nand/nand_base.c b/drivers/mtd/nand/nand_base.c
index 04de315e4937..7e68203fe1ba 100644
--- a/drivers/mtd/nand/nand_base.c
+++ b/drivers/mtd/nand/nand_base.c
@@ -303,28 +303,27 @@ static int nand_block_bad(struct mtd_info *mtd, loff_t ofs, int getchip)
 	struct nand_chip *chip = mtd->priv;
 	u16 bad;
 
+	page = (int)(ofs >> chip->page_shift) & chip->pagemask;
+
 	if (getchip) {
-		page = (int)(ofs >> chip->page_shift);
 		chipnr = (int)(ofs >> chip->chip_shift);
 
 		nand_get_device(chip, mtd, FL_READING);
 
 		/* Select the NAND device */
 		chip->select_chip(mtd, chipnr);
-	} else
-		page = (int)(ofs >> chip->page_shift);
+	}
 
 	if (chip->options & NAND_BUSWIDTH_16) {
 		chip->cmdfunc(mtd, NAND_CMD_READOOB, chip->badblockpos & 0xFE,
-			      page & chip->pagemask);
+			      page);
 		bad = cpu_to_le16(chip->read_word(mtd));
 		if (chip->badblockpos & 0x1)
 			bad >>= 8;
 		if ((bad & 0xFF) != 0xff)
 			res = 1;
 	} else {
-		chip->cmdfunc(mtd, NAND_CMD_READOOB, chip->badblockpos,
-			      page & chip->pagemask);
+		chip->cmdfunc(mtd, NAND_CMD_READOOB, chip->badblockpos, page);
 		if (chip->read_byte(mtd) != 0xff)
 			res = 1;
 	}
diff --git a/drivers/mtd/nand/plat_nand.c b/drivers/mtd/nand/plat_nand.c
new file mode 100644
index 000000000000..cd725fc5e813
--- /dev/null
+++ b/drivers/mtd/nand/plat_nand.c
@@ -0,0 +1,150 @@
+/*
+ * Generic NAND driver
+ *
+ * Author: Vitaly Wool <vitalywool@gmail.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ */
+
+#include <linux/io.h>
+#include <linux/module.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+#include <linux/mtd/mtd.h>
+#include <linux/mtd/nand.h>
+#include <linux/mtd/partitions.h>
+
+struct plat_nand_data {
+	struct nand_chip	chip;
+	struct mtd_info		mtd;
+	void __iomem		*io_base;
+#ifdef CONFIG_MTD_PARTITIONS
+	int			nr_parts;
+	struct mtd_partition	*parts;
+#endif
+};
+
+/*
+ * Probe for the NAND device.
+ */
+static int __init plat_nand_probe(struct platform_device *pdev)
+{
+	struct platform_nand_data *pdata = pdev->dev.platform_data;
+	struct plat_nand_data *data;
+	int res = 0;
+
+	/* Allocate memory for the device structure (and zero it) */
+	data = kzalloc(sizeof(struct plat_nand_data), GFP_KERNEL);
+	if (!data) {
+		dev_err(&pdev->dev, "failed to allocate device structure.\n");
+		return -ENOMEM;
+	}
+
+	data->io_base = ioremap(pdev->resource[0].start,
+				pdev->resource[0].end - pdev->resource[0].start + 1);
+	if (data->io_base == NULL) {
+		dev_err(&pdev->dev, "ioremap failed\n");
+		kfree(data);
+		return -EIO;
+	}
+
+	data->chip.priv = &data;
+	data->mtd.priv = &data->chip;
+	data->mtd.owner = THIS_MODULE;
+
+	data->chip.IO_ADDR_R = data->io_base;
+	data->chip.IO_ADDR_W = data->io_base;
+	data->chip.cmd_ctrl = pdata->ctrl.cmd_ctrl;
+	data->chip.dev_ready = pdata->ctrl.dev_ready;
+	data->chip.select_chip = pdata->ctrl.select_chip;
+	data->chip.chip_delay = pdata->chip.chip_delay;
+	data->chip.options |= pdata->chip.options;
+
+	data->chip.ecc.hwctl = pdata->ctrl.hwcontrol;
+	data->chip.ecc.layout = pdata->chip.ecclayout;
+	data->chip.ecc.mode = NAND_ECC_SOFT;
+
+	platform_set_drvdata(pdev, data);
+
+	/* Scan to find existance of the device */
+	if (nand_scan(&data->mtd, 1)) {
+		res = -ENXIO;
+		goto out;
+	}
+
+#ifdef CONFIG_MTD_PARTITIONS
+	if (pdata->chip.part_probe_types) {
+		res = parse_mtd_partitions(&data->mtd,
+					pdata->chip.part_probe_types,
+					&data->parts, 0);
+		if (res > 0) {
+			add_mtd_partitions(&data->mtd, data->parts, res);
+			return 0;
+		}
+	}
+	if (pdata->chip.partitions) {
+		data->parts = pdata->chip.partitions;
+		res = add_mtd_partitions(&data->mtd, data->parts,
+			pdata->chip.nr_partitions);
+	} else
+#endif
+	res = add_mtd_device(&data->mtd);
+
+	if (!res)
+		return res;
+
+	nand_release(&data->mtd);
+out:
+	platform_set_drvdata(pdev, NULL);
+	iounmap(data->io_base);
+	kfree(data);
+	return res;
+}
+
+/*
+ * Remove a NAND device.
+ */
+static int __devexit plat_nand_remove(struct platform_device *pdev)
+{
+	struct plat_nand_data *data = platform_get_drvdata(pdev);
+	struct platform_nand_data *pdata = pdev->dev.platform_data;
+
+	nand_release(&data->mtd);
+#ifdef CONFIG_MTD_PARTITIONS
+	if (data->parts && data->parts != pdata->chip.partitions)
+		kfree(data->parts);
+#endif
+	iounmap(data->io_base);
+	kfree(data);
+
+	return 0;
+}
+
+static struct platform_driver plat_nand_driver = {
+	.probe		= plat_nand_probe,
+	.remove		= plat_nand_remove,
+	.driver		= {
+		.name	= "gen_nand",
+		.owner	= THIS_MODULE,
+	},
+};
+
+static int __init plat_nand_init(void)
+{
+	return platform_driver_register(&plat_nand_driver);
+}
+
+static void __exit plat_nand_exit(void)
+{
+	platform_driver_unregister(&plat_nand_driver);
+}
+
+module_init(plat_nand_init);
+module_exit(plat_nand_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Vitaly Wool");
+MODULE_DESCRIPTION("Simple generic NAND driver");
diff --git a/drivers/mtd/onenand/onenand_base.c b/drivers/mtd/onenand/onenand_base.c
index 000794c6caf5..0537fac8de74 100644
--- a/drivers/mtd/onenand/onenand_base.c
+++ b/drivers/mtd/onenand/onenand_base.c
@@ -2192,7 +2192,7 @@ static int onenand_check_maf(int manuf)
  * @param mtd		MTD device structure
  *
  * OneNAND detection method:
- *   Compare the the values from command with ones from register
+ *   Compare the values from command with ones from register
  */
 static int onenand_probe(struct mtd_info *mtd)
 {
diff --git a/drivers/net/3c509.c b/drivers/net/3c509.c
index 9588da3a30e7..127f60841b10 100644
--- a/drivers/net/3c509.c
+++ b/drivers/net/3c509.c
@@ -95,8 +95,7 @@ static int max_interrupt_work = 10;
 #include <asm/io.h>
 #include <asm/irq.h>
 
-static char versionA[] __initdata = DRV_NAME ".c:" DRV_VERSION " " DRV_RELDATE " becker@scyld.com\n";
-static char versionB[] __initdata = "http://www.scyld.com/network/3c509.html\n";
+static char version[] __initdata = DRV_NAME ".c:" DRV_VERSION " " DRV_RELDATE " becker@scyld.com\n";
 
 #if defined(CONFIG_PM) && (defined(CONFIG_MCA) || defined(CONFIG_EISA))
 #define EL3_SUSPEND
@@ -360,7 +359,7 @@ static int __init el3_common_init(struct net_device *dev)
 	printk(", IRQ %d.\n", dev->irq);
 
 	if (el3_debug > 0)
-		printk(KERN_INFO "%s" KERN_INFO "%s", versionA, versionB);
+		printk(KERN_INFO "%s", version);
 	return 0;
 
 }
diff --git a/drivers/net/3c59x.c b/drivers/net/3c59x.c
index 80924f76dee8..f26ca331615e 100644
--- a/drivers/net/3c59x.c
+++ b/drivers/net/3c59x.c
@@ -103,7 +103,7 @@ static int vortex_debug = 1;
 
 
 static char version[] __devinitdata =
-DRV_NAME ": Donald Becker and others. www.scyld.com/network/vortex.html\n";
+DRV_NAME ": Donald Becker and others.\n";
 
 MODULE_AUTHOR("Donald Becker <becker@scyld.com>");
 MODULE_DESCRIPTION("3Com 3c59x/3c9xx ethernet driver ");
diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index b86ccd2ecd5b..fa489b10c38c 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -2493,12 +2493,28 @@ config PASEMI_MAC
 	  This driver supports the on-chip 1/10Gbit Ethernet controller on
 	  PA Semi's PWRficient line of chips.
 
+config MLX4_CORE
+	tristate
+	depends on PCI
+	default n
+
+config MLX4_DEBUG
+	bool "Verbose debugging output" if (MLX4_CORE && EMBEDDED)
+	default y
+	---help---
+	  This option causes debugging code to be compiled into the
+	  mlx4_core driver.  The output can be turned on via the
+	  debug_level module parameter (which can also be set after
+	  the driver is loaded through sysfs).
+
 endmenu
 
 source "drivers/net/tokenring/Kconfig"
 
 source "drivers/net/wireless/Kconfig"
 
+source "drivers/net/usb/Kconfig"
+
 source "drivers/net/pcmcia/Kconfig"
 
 source "drivers/net/wan/Kconfig"
diff --git a/drivers/net/Makefile b/drivers/net/Makefile
index 59c0459a037c..a77affa4f6e6 100644
--- a/drivers/net/Makefile
+++ b/drivers/net/Makefile
@@ -197,6 +197,7 @@ obj-$(CONFIG_SMC911X) += smc911x.o
 obj-$(CONFIG_DM9000) += dm9000.o
 obj-$(CONFIG_FEC_8XX) += fec_8xx/
 obj-$(CONFIG_PASEMI_MAC) += pasemi_mac.o
+obj-$(CONFIG_MLX4_CORE) += mlx4/
 
 obj-$(CONFIG_MACB) += macb.o
 
@@ -206,6 +207,14 @@ obj-$(CONFIG_TR) += tokenring/
 obj-$(CONFIG_WAN) += wan/
 obj-$(CONFIG_ARCNET) += arcnet/
 obj-$(CONFIG_NET_PCMCIA) += pcmcia/
+
+obj-$(CONFIG_USB_CATC)          += usb/
+obj-$(CONFIG_USB_KAWETH)        += usb/
+obj-$(CONFIG_USB_PEGASUS)       += usb/
+obj-$(CONFIG_USB_RTL8150)       += usb/
+obj-$(CONFIG_USB_USBNET)        += usb/
+obj-$(CONFIG_USB_ZD1201)        += usb/
+
 obj-y += wireless/
 obj-$(CONFIG_NET_TULIP) += tulip/
 obj-$(CONFIG_HAMRADIO) += hamradio/
diff --git a/drivers/net/atl1/atl1_main.c b/drivers/net/atl1/atl1_main.c
index d28f88bbdd5f..78cf00ff3d38 100644
--- a/drivers/net/atl1/atl1_main.c
+++ b/drivers/net/atl1/atl1_main.c
@@ -2038,6 +2038,15 @@ static int atl1_close(struct net_device *netdev)
 	return 0;
 }
 
+#ifdef CONFIG_NET_POLL_CONTROLLER
+static void atl1_poll_controller(struct net_device *netdev)
+{
+	disable_irq(netdev->irq);
+	atl1_intr(netdev->irq, netdev);
+	enable_irq(netdev->irq);
+}
+#endif
+
 /*
  * If TPD Buffer size equal to 0, PCIE DMAR_TO_INT
  * will assert. We do soft reset <0x1400=1> according
@@ -2190,6 +2199,9 @@ static int __devinit atl1_probe(struct pci_dev *pdev,
 	netdev->do_ioctl = &atl1_ioctl;
 	netdev->tx_timeout = &atl1_tx_timeout;
 	netdev->watchdog_timeo = 5 * HZ;
+#ifdef CONFIG_NET_POLL_CONTROLLER
+	netdev->poll_controller = atl1_poll_controller;
+#endif
 	netdev->vlan_rx_register = atl1_vlan_rx_register;
 	netdev->vlan_rx_add_vid = atl1_vlan_rx_add_vid;
 	netdev->vlan_rx_kill_vid = atl1_vlan_rx_kill_vid;
diff --git a/drivers/net/atp.c b/drivers/net/atp.c
index 18aba838c1ff..82d78ff8399b 100644
--- a/drivers/net/atp.c
+++ b/drivers/net/atp.c
@@ -31,10 +31,8 @@
 
 */
 
-static const char versionA[] =
+static const char version[] =
 "atp.c:v1.09=ac 2002/10/01 Donald Becker <becker@scyld.com>\n";
-static const char versionB[] =
-"  http://www.scyld.com/network/atp.html\n";
 
 /* The user-configurable values.
    These may be modified when a driver module is loaded.*/
@@ -324,7 +322,7 @@ static int __init atp_probe1(long ioaddr)
 
 #ifndef MODULE
 	if (net_debug)
-		printk(KERN_INFO "%s" KERN_INFO "%s", versionA, versionB);
+		printk(KERN_INFO "%s", version);
 #endif
 
 	printk(KERN_NOTICE "%s: Pocket adapter found at %#3lx, IRQ %d, SAPROM "
@@ -926,7 +924,7 @@ static void set_rx_mode_8012(struct net_device *dev)
 
 static int __init atp_init_module(void) {
 	if (debug)					/* Emit version even if no cards detected. */
-		printk(KERN_INFO "%s" KERN_INFO "%s", versionA, versionB);
+		printk(KERN_INFO "%s", version);
 	return atp_init();
 }
 
diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index 724bce51f936..223517dcbcfd 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -3461,7 +3461,7 @@ void bond_unregister_arp(struct bonding *bond)
 /*---------------------------- Hashing Policies -----------------------------*/
 
 /*
- * Hash for the the output device based upon layer 3 and layer 4 data. If
+ * Hash for the output device based upon layer 3 and layer 4 data. If
  * the packet is a frag or not TCP or UDP, just use layer 3 data.  If it is
  * altogether not IP, mimic bond_xmit_hash_policy_l2()
  */
diff --git a/drivers/net/e1000/e1000_main.c b/drivers/net/e1000/e1000_main.c
index 3a03a74c0609..637ae8f68791 100644
--- a/drivers/net/e1000/e1000_main.c
+++ b/drivers/net/e1000/e1000_main.c
@@ -1214,7 +1214,7 @@ e1000_remove(struct pci_dev *pdev)
 	int i;
 #endif
 
-	flush_scheduled_work();
+	cancel_work_sync(&adapter->reset_task);
 
 	e1000_release_manageability(adapter);
 
diff --git a/drivers/net/eepro.c b/drivers/net/eepro.c
index 39654e1e2bed..47680237f783 100644
--- a/drivers/net/eepro.c
+++ b/drivers/net/eepro.c
@@ -1126,7 +1126,7 @@ static void eepro_tx_timeout (struct net_device *dev)
 	printk (KERN_ERR "%s: transmit timed out, %s?\n", dev->name,
 		"network cable problem");
 	/* This is not a duplicate. One message for the console,
-	   one for the the log file  */
+	   one for the log file  */
 	printk (KERN_DEBUG "%s: transmit timed out, %s?\n", dev->name,
 		"network cable problem");
 	eepro_complete_selreset(ioaddr);
diff --git a/drivers/net/eepro100.c b/drivers/net/eepro100.c
index 6c267c38df97..9800341956a2 100644
--- a/drivers/net/eepro100.c
+++ b/drivers/net/eepro100.c
@@ -28,7 +28,7 @@
 */
 
 static const char * const version =
-"eepro100.c:v1.09j-t 9/29/99 Donald Becker http://www.scyld.com/network/eepro100.html\n"
+"eepro100.c:v1.09j-t 9/29/99 Donald Becker\n"
 "eepro100.c: $Revision: 1.36 $ 2000/11/17 Modified by Andrey V. Savochkin <saw@saw.sw.com.sg> and others\n";
 
 /* A few user-configurable values that apply to all boards.
diff --git a/drivers/net/epic100.c b/drivers/net/epic100.c
index 4e3f14c9c717..5e517946f46a 100644
--- a/drivers/net/epic100.c
+++ b/drivers/net/epic100.c
@@ -93,8 +93,6 @@ static int rx_copybreak;
 static char version[] __devinitdata =
 DRV_NAME ".c:v1.11 1/7/2001 Written by Donald Becker <becker@scyld.com>\n";
 static char version2[] __devinitdata =
-"  http://www.scyld.com/network/epic100.html\n";
-static char version3[] __devinitdata =
 "  (unofficial 2.4.x kernel port, version " DRV_VERSION ", " DRV_RELDATE ")\n";
 
 MODULE_AUTHOR("Donald Becker <becker@scyld.com>");
@@ -323,8 +321,8 @@ static int __devinit epic_init_one (struct pci_dev *pdev,
 #ifndef MODULE
 	static int printed_version;
 	if (!printed_version++)
-		printk (KERN_INFO "%s" KERN_INFO "%s" KERN_INFO "%s",
-			version, version2, version3);
+		printk (KERN_INFO "%s" KERN_INFO "%s",
+			version, version2);
 #endif
 
 	card_idx++;
@@ -1596,8 +1594,8 @@ static int __init epic_init (void)
 {
 /* when a module, this is printed whether or not devices are found in probe */
 #ifdef MODULE
-	printk (KERN_INFO "%s" KERN_INFO "%s" KERN_INFO "%s",
-		version, version2, version3);
+	printk (KERN_INFO "%s" KERN_INFO "%s",
+		version, version2);
 #endif
 
 	return pci_register_driver(&epic_driver);
diff --git a/drivers/net/hamradio/Kconfig b/drivers/net/hamradio/Kconfig
index 6e90619b3b41..36d2c7d4f4d0 100644
--- a/drivers/net/hamradio/Kconfig
+++ b/drivers/net/hamradio/Kconfig
@@ -140,7 +140,7 @@ config BAYCOM_SER_HDX
 	  modems that connect to a serial interface. The driver supports the
 	  ser12 design in half-duplex mode. This is the old driver.  It is
 	  still provided in case your serial interface chip does not work with
-	  the full-duplex driver. This driver is depreciated.  To configure
+	  the full-duplex driver. This driver is deprecated.  To configure
 	  the driver, use the sethdlc utility available in the standard ax25
 	  utilities package. For information on the modems, see
 	  <http://www.baycom.de/> and
diff --git a/drivers/net/irda/donauboe.h b/drivers/net/irda/donauboe.h
index 2ab173d9a0e4..1e67720f1066 100644
--- a/drivers/net/irda/donauboe.h
+++ b/drivers/net/irda/donauboe.h
@@ -113,7 +113,7 @@
 /* RxOver  overflow in Recv FIFO                                */
 /* SipRcv  received serial gap  (or other condition you set)    */
 /* Interrupts are enabled by writing a one to the IER register  */
-/* Interrupts are cleared by writting a one to the ISR register */
+/* Interrupts are cleared by writing a one to the ISR register */
 /*                                                              */
 /* 6. The remaining registers: 0x6 and 0x3 appear to be         */
 /*    reserved parts of 16 or 32 bit registersthe remainder     */
diff --git a/drivers/net/ixgb/ixgb_ee.c b/drivers/net/ixgb/ixgb_ee.c
index f15aebde7b90..52c99d01d568 100644
--- a/drivers/net/ixgb/ixgb_ee.c
+++ b/drivers/net/ixgb/ixgb_ee.c
@@ -315,7 +315,7 @@ ixgb_wait_eeprom_command(struct ixgb_hw *hw)
  * hw - Struct containing variables accessed by shared code
  *
  * Reads the first 64 16 bit words of the EEPROM and sums the values read.
- * If the the sum of the 64 16 bit words is 0xBABA, the EEPROM's checksum is
+ * If the sum of the 64 16 bit words is 0xBABA, the EEPROM's checksum is
  * valid.
  *
  * Returns:
diff --git a/drivers/net/meth.h b/drivers/net/meth.h
index 84960dae2a22..ea3b8fc86d1e 100644
--- a/drivers/net/meth.h
+++ b/drivers/net/meth.h
@@ -126,7 +126,7 @@ typedef struct rx_packet {
 				       /*   Note: when loopback is set this bit becomes collision control.  Setting this bit will */
 				       /*         cause a collision to be reported. */
 
-				       /* Bits 5 and 6 are used to determine the the Destination address filter mode */
+				       /* Bits 5 and 6 are used to determine the Destination address filter mode */
 #define METH_ACCEPT_MY 0			/* 00: Accept PHY address only */
 #define METH_ACCEPT_MCAST 0x20	/* 01: Accept physical, broadcast, and multicast filter matches only */
 #define METH_ACCEPT_AMCAST 0x40	/* 10: Accept physical, broadcast, and all multicast packets */
diff --git a/drivers/net/mlx4/Makefile b/drivers/net/mlx4/Makefile
new file mode 100644
index 000000000000..0952a6528f58
--- /dev/null
+++ b/drivers/net/mlx4/Makefile
@@ -0,0 +1,4 @@
+obj-$(CONFIG_MLX4_CORE)		+= mlx4_core.o
+
+mlx4_core-y :=	alloc.o catas.o cmd.o cq.o eq.o fw.o icm.o intf.o main.o mcg.o \
+		mr.o pd.o profile.o qp.o reset.o srq.o
diff --git a/drivers/net/mlx4/alloc.c b/drivers/net/mlx4/alloc.c
new file mode 100644
index 000000000000..9ffdb9d29da9
--- /dev/null
+++ b/drivers/net/mlx4/alloc.c
@@ -0,0 +1,179 @@
+/*
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/errno.h>
+#include <linux/slab.h>
+#include <linux/bitmap.h>
+
+#include "mlx4.h"
+
+u32 mlx4_bitmap_alloc(struct mlx4_bitmap *bitmap)
+{
+	u32 obj;
+
+	spin_lock(&bitmap->lock);
+
+	obj = find_next_zero_bit(bitmap->table, bitmap->max, bitmap->last);
+	if (obj >= bitmap->max) {
+		bitmap->top = (bitmap->top + bitmap->max) & bitmap->mask;
+		obj = find_first_zero_bit(bitmap->table, bitmap->max);
+	}
+
+	if (obj < bitmap->max) {
+		set_bit(obj, bitmap->table);
+		obj |= bitmap->top;
+		bitmap->last = obj + 1;
+	} else
+		obj = -1;
+
+	spin_unlock(&bitmap->lock);
+
+	return obj;
+}
+
+void mlx4_bitmap_free(struct mlx4_bitmap *bitmap, u32 obj)
+{
+	obj &= bitmap->max - 1;
+
+	spin_lock(&bitmap->lock);
+	clear_bit(obj, bitmap->table);
+	bitmap->last = min(bitmap->last, obj);
+	bitmap->top = (bitmap->top + bitmap->max) & bitmap->mask;
+	spin_unlock(&bitmap->lock);
+}
+
+int mlx4_bitmap_init(struct mlx4_bitmap *bitmap, u32 num, u32 mask, u32 reserved)
+{
+	int i;
+
+	/* num must be a power of 2 */
+	if (num != roundup_pow_of_two(num))
+		return -EINVAL;
+
+	bitmap->last = 0;
+	bitmap->top  = 0;
+	bitmap->max  = num;
+	bitmap->mask = mask;
+	spin_lock_init(&bitmap->lock);
+	bitmap->table = kzalloc(BITS_TO_LONGS(num) * sizeof (long), GFP_KERNEL);
+	if (!bitmap->table)
+		return -ENOMEM;
+
+	for (i = 0; i < reserved; ++i)
+		set_bit(i, bitmap->table);
+
+	return 0;
+}
+
+void mlx4_bitmap_cleanup(struct mlx4_bitmap *bitmap)
+{
+	kfree(bitmap->table);
+}
+
+/*
+ * Handling for queue buffers -- we allocate a bunch of memory and
+ * register it in a memory region at HCA virtual address 0.  If the
+ * requested size is > max_direct, we split the allocation into
+ * multiple pages, so we don't require too much contiguous memory.
+ */
+
+int mlx4_buf_alloc(struct mlx4_dev *dev, int size, int max_direct,
+		   struct mlx4_buf *buf)
+{
+	dma_addr_t t;
+
+	if (size <= max_direct) {
+		buf->nbufs        = 1;
+		buf->npages       = 1;
+		buf->page_shift   = get_order(size) + PAGE_SHIFT;
+		buf->u.direct.buf = dma_alloc_coherent(&dev->pdev->dev,
+						       size, &t, GFP_KERNEL);
+		if (!buf->u.direct.buf)
+			return -ENOMEM;
+
+		buf->u.direct.map = t;
+
+		while (t & ((1 << buf->page_shift) - 1)) {
+			--buf->page_shift;
+			buf->npages *= 2;
+		}
+
+		memset(buf->u.direct.buf, 0, size);
+	} else {
+		int i;
+
+		buf->nbufs       = (size + PAGE_SIZE - 1) / PAGE_SIZE;
+		buf->npages      = buf->nbufs;
+		buf->page_shift  = PAGE_SHIFT;
+		buf->u.page_list = kzalloc(buf->nbufs * sizeof *buf->u.page_list,
+					   GFP_KERNEL);
+		if (!buf->u.page_list)
+			return -ENOMEM;
+
+		for (i = 0; i < buf->nbufs; ++i) {
+			buf->u.page_list[i].buf =
+				dma_alloc_coherent(&dev->pdev->dev, PAGE_SIZE,
+						   &t, GFP_KERNEL);
+			if (!buf->u.page_list[i].buf)
+				goto err_free;
+
+			buf->u.page_list[i].map = t;
+
+			memset(buf->u.page_list[i].buf, 0, PAGE_SIZE);
+		}
+	}
+
+	return 0;
+
+err_free:
+	mlx4_buf_free(dev, size, buf);
+
+	return -ENOMEM;
+}
+EXPORT_SYMBOL_GPL(mlx4_buf_alloc);
+
+void mlx4_buf_free(struct mlx4_dev *dev, int size, struct mlx4_buf *buf)
+{
+	int i;
+
+	if (buf->nbufs == 1)
+		dma_free_coherent(&dev->pdev->dev, size, buf->u.direct.buf,
+				  buf->u.direct.map);
+	else {
+		for (i = 0; i < buf->nbufs; ++i)
+			dma_free_coherent(&dev->pdev->dev, PAGE_SIZE,
+					  buf->u.page_list[i].buf,
+					  buf->u.page_list[i].map);
+		kfree(buf->u.page_list);
+	}
+}
+EXPORT_SYMBOL_GPL(mlx4_buf_free);
diff --git a/drivers/net/mlx4/catas.c b/drivers/net/mlx4/catas.c
new file mode 100644
index 000000000000..1bb088aeaf71
--- /dev/null
+++ b/drivers/net/mlx4/catas.c
@@ -0,0 +1,70 @@
+/*
+ * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "mlx4.h"
+
+void mlx4_handle_catas_err(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	int i;
+
+	mlx4_err(dev, "Catastrophic error detected:\n");
+	for (i = 0; i < priv->fw.catas_size; ++i)
+		mlx4_err(dev, "  buf[%02x]: %08x\n",
+			 i, swab32(readl(priv->catas_err.map + i)));
+
+	mlx4_dispatch_event(dev, MLX4_EVENT_TYPE_LOCAL_CATAS_ERROR, 0, 0);
+}
+
+void mlx4_map_catas_buf(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	unsigned long addr;
+
+	addr = pci_resource_start(dev->pdev, priv->fw.catas_bar) +
+		priv->fw.catas_offset;
+
+	priv->catas_err.map = ioremap(addr, priv->fw.catas_size * 4);
+	if (!priv->catas_err.map)
+		mlx4_warn(dev, "Failed to map catastrophic error buffer at 0x%lx\n",
+			  addr);
+
+}
+
+void mlx4_unmap_catas_buf(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	if (priv->catas_err.map)
+		iounmap(priv->catas_err.map);
+}
diff --git a/drivers/net/mlx4/cmd.c b/drivers/net/mlx4/cmd.c
new file mode 100644
index 000000000000..c1f81a993f5d
--- /dev/null
+++ b/drivers/net/mlx4/cmd.c
@@ -0,0 +1,429 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2005, 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/sched.h>
+#include <linux/pci.h>
+#include <linux/errno.h>
+
+#include <linux/mlx4/cmd.h>
+
+#include <asm/io.h>
+
+#include "mlx4.h"
+
+#define CMD_POLL_TOKEN 0xffff
+
+enum {
+	/* command completed successfully: */
+	CMD_STAT_OK		= 0x00,
+	/* Internal error (such as a bus error) occurred while processing command: */
+	CMD_STAT_INTERNAL_ERR	= 0x01,
+	/* Operation/command not supported or opcode modifier not supported: */
+	CMD_STAT_BAD_OP		= 0x02,
+	/* Parameter not supported or parameter out of range: */
+	CMD_STAT_BAD_PARAM	= 0x03,
+	/* System not enabled or bad system state: */
+	CMD_STAT_BAD_SYS_STATE	= 0x04,
+	/* Attempt to access reserved or unallocaterd resource: */
+	CMD_STAT_BAD_RESOURCE	= 0x05,
+	/* Requested resource is currently executing a command, or is otherwise busy: */
+	CMD_STAT_RESOURCE_BUSY	= 0x06,
+	/* Required capability exceeds device limits: */
+	CMD_STAT_EXCEED_LIM	= 0x08,
+	/* Resource is not in the appropriate state or ownership: */
+	CMD_STAT_BAD_RES_STATE	= 0x09,
+	/* Index out of range: */
+	CMD_STAT_BAD_INDEX	= 0x0a,
+	/* FW image corrupted: */
+	CMD_STAT_BAD_NVMEM	= 0x0b,
+	/* Attempt to modify a QP/EE which is not in the presumed state: */
+	CMD_STAT_BAD_QP_STATE   = 0x10,
+	/* Bad segment parameters (Address/Size): */
+	CMD_STAT_BAD_SEG_PARAM	= 0x20,
+	/* Memory Region has Memory Windows bound to: */
+	CMD_STAT_REG_BOUND	= 0x21,
+	/* HCA local attached memory not present: */
+	CMD_STAT_LAM_NOT_PRE	= 0x22,
+	/* Bad management packet (silently discarded): */
+	CMD_STAT_BAD_PKT	= 0x30,
+	/* More outstanding CQEs in CQ than new CQ size: */
+	CMD_STAT_BAD_SIZE	= 0x40
+};
+
+enum {
+	HCR_IN_PARAM_OFFSET	= 0x00,
+	HCR_IN_MODIFIER_OFFSET	= 0x08,
+	HCR_OUT_PARAM_OFFSET	= 0x0c,
+	HCR_TOKEN_OFFSET	= 0x14,
+	HCR_STATUS_OFFSET	= 0x18,
+
+	HCR_OPMOD_SHIFT		= 12,
+	HCR_T_BIT		= 21,
+	HCR_E_BIT		= 22,
+	HCR_GO_BIT		= 23
+};
+
+enum {
+	GO_BIT_TIMEOUT		= 10000
+};
+
+struct mlx4_cmd_context {
+	struct completion	done;
+	int			result;
+	int			next;
+	u64			out_param;
+	u16			token;
+};
+
+static int mlx4_status_to_errno(u8 status) {
+	static const int trans_table[] = {
+		[CMD_STAT_INTERNAL_ERR]	  = -EIO,
+		[CMD_STAT_BAD_OP]	  = -EPERM,
+		[CMD_STAT_BAD_PARAM]	  = -EINVAL,
+		[CMD_STAT_BAD_SYS_STATE]  = -ENXIO,
+		[CMD_STAT_BAD_RESOURCE]	  = -EBADF,
+		[CMD_STAT_RESOURCE_BUSY]  = -EBUSY,
+		[CMD_STAT_EXCEED_LIM]	  = -ENOMEM,
+		[CMD_STAT_BAD_RES_STATE]  = -EBADF,
+		[CMD_STAT_BAD_INDEX]	  = -EBADF,
+		[CMD_STAT_BAD_NVMEM]	  = -EFAULT,
+		[CMD_STAT_BAD_QP_STATE]   = -EINVAL,
+		[CMD_STAT_BAD_SEG_PARAM]  = -EFAULT,
+		[CMD_STAT_REG_BOUND]	  = -EBUSY,
+		[CMD_STAT_LAM_NOT_PRE]	  = -EAGAIN,
+		[CMD_STAT_BAD_PKT]	  = -EINVAL,
+		[CMD_STAT_BAD_SIZE]	  = -ENOMEM,
+	};
+
+	if (status >= ARRAY_SIZE(trans_table) ||
+	    (status != CMD_STAT_OK && trans_table[status] == 0))
+		return -EIO;
+
+	return trans_table[status];
+}
+
+static int cmd_pending(struct mlx4_dev *dev)
+{
+	u32 status = readl(mlx4_priv(dev)->cmd.hcr + HCR_STATUS_OFFSET);
+
+	return (status & swab32(1 << HCR_GO_BIT)) ||
+		(mlx4_priv(dev)->cmd.toggle ==
+		 !!(status & swab32(1 << HCR_T_BIT)));
+}
+
+static int mlx4_cmd_post(struct mlx4_dev *dev, u64 in_param, u64 out_param,
+			 u32 in_modifier, u8 op_modifier, u16 op, u16 token,
+			 int event)
+{
+	struct mlx4_cmd *cmd = &mlx4_priv(dev)->cmd;
+	u32 __iomem *hcr = cmd->hcr;
+	int ret = -EAGAIN;
+	unsigned long end;
+
+	mutex_lock(&cmd->hcr_mutex);
+
+	end = jiffies;
+	if (event)
+		end += HZ * 10;
+
+	while (cmd_pending(dev)) {
+		if (time_after_eq(jiffies, end))
+			goto out;
+		cond_resched();
+	}
+
+	/*
+	 * We use writel (instead of something like memcpy_toio)
+	 * because writes of less than 32 bits to the HCR don't work
+	 * (and some architectures such as ia64 implement memcpy_toio
+	 * in terms of writeb).
+	 */
+	__raw_writel((__force u32) cpu_to_be32(in_param >> 32),		  hcr + 0);
+	__raw_writel((__force u32) cpu_to_be32(in_param & 0xfffffffful),  hcr + 1);
+	__raw_writel((__force u32) cpu_to_be32(in_modifier),		  hcr + 2);
+	__raw_writel((__force u32) cpu_to_be32(out_param >> 32),	  hcr + 3);
+	__raw_writel((__force u32) cpu_to_be32(out_param & 0xfffffffful), hcr + 4);
+	__raw_writel((__force u32) cpu_to_be32(token << 16),		  hcr + 5);
+
+	/* __raw_writel may not order writes. */
+	wmb();
+
+	__raw_writel((__force u32) cpu_to_be32((1 << HCR_GO_BIT)		|
+					       (cmd->toggle << HCR_T_BIT)	|
+					       (event ? (1 << HCR_E_BIT) : 0)	|
+					       (op_modifier << HCR_OPMOD_SHIFT) |
+					       op),			  hcr + 6);
+	cmd->toggle = cmd->toggle ^ 1;
+
+	ret = 0;
+
+out:
+	mutex_unlock(&cmd->hcr_mutex);
+	return ret;
+}
+
+static int mlx4_cmd_poll(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
+			 int out_is_imm, u32 in_modifier, u8 op_modifier,
+			 u16 op, unsigned long timeout)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	void __iomem *hcr = priv->cmd.hcr;
+	int err = 0;
+	unsigned long end;
+
+	down(&priv->cmd.poll_sem);
+
+	err = mlx4_cmd_post(dev, in_param, out_param ? *out_param : 0,
+			    in_modifier, op_modifier, op, CMD_POLL_TOKEN, 0);
+	if (err)
+		goto out;
+
+	end = msecs_to_jiffies(timeout) + jiffies;
+	while (cmd_pending(dev) && time_before(jiffies, end))
+		cond_resched();
+
+	if (cmd_pending(dev)) {
+		err = -ETIMEDOUT;
+		goto out;
+	}
+
+	if (out_is_imm)
+		*out_param =
+			(u64) be32_to_cpu((__force __be32)
+					  __raw_readl(hcr + HCR_OUT_PARAM_OFFSET)) << 32 |
+			(u64) be32_to_cpu((__force __be32)
+					  __raw_readl(hcr + HCR_OUT_PARAM_OFFSET + 4));
+
+	err = mlx4_status_to_errno(be32_to_cpu((__force __be32)
+					       __raw_readl(hcr + HCR_STATUS_OFFSET)) >> 24);
+
+out:
+	up(&priv->cmd.poll_sem);
+	return err;
+}
+
+void mlx4_cmd_event(struct mlx4_dev *dev, u16 token, u8 status, u64 out_param)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cmd_context *context =
+		&priv->cmd.context[token & priv->cmd.token_mask];
+
+	/* previously timed out command completing at long last */
+	if (token != context->token)
+		return;
+
+	context->result    = mlx4_status_to_errno(status);
+	context->out_param = out_param;
+
+	context->token += priv->cmd.token_mask + 1;
+
+	complete(&context->done);
+}
+
+static int mlx4_cmd_wait(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
+			 int out_is_imm, u32 in_modifier, u8 op_modifier,
+			 u16 op, unsigned long timeout)
+{
+	struct mlx4_cmd *cmd = &mlx4_priv(dev)->cmd;
+	struct mlx4_cmd_context *context;
+	int err = 0;
+
+	down(&cmd->event_sem);
+
+	spin_lock(&cmd->context_lock);
+	BUG_ON(cmd->free_head < 0);
+	context = &cmd->context[cmd->free_head];
+	cmd->free_head = context->next;
+	spin_unlock(&cmd->context_lock);
+
+	init_completion(&context->done);
+
+	mlx4_cmd_post(dev, in_param, out_param ? *out_param : 0,
+		      in_modifier, op_modifier, op, context->token, 1);
+
+	if (!wait_for_completion_timeout(&context->done, msecs_to_jiffies(timeout))) {
+		err = -EBUSY;
+		goto out;
+	}
+
+	err = context->result;
+	if (err)
+		goto out;
+
+	if (out_is_imm)
+		*out_param = context->out_param;
+
+out:
+	spin_lock(&cmd->context_lock);
+	context->next = cmd->free_head;
+	cmd->free_head = context - cmd->context;
+	spin_unlock(&cmd->context_lock);
+
+	up(&cmd->event_sem);
+	return err;
+}
+
+int __mlx4_cmd(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
+	       int out_is_imm, u32 in_modifier, u8 op_modifier,
+	       u16 op, unsigned long timeout)
+{
+	if (mlx4_priv(dev)->cmd.use_events)
+		return mlx4_cmd_wait(dev, in_param, out_param, out_is_imm,
+				     in_modifier, op_modifier, op, timeout);
+	else
+		return mlx4_cmd_poll(dev, in_param, out_param, out_is_imm,
+				     in_modifier, op_modifier, op, timeout);
+}
+EXPORT_SYMBOL_GPL(__mlx4_cmd);
+
+int mlx4_cmd_init(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	mutex_init(&priv->cmd.hcr_mutex);
+	sema_init(&priv->cmd.poll_sem, 1);
+	priv->cmd.use_events = 0;
+	priv->cmd.toggle     = 1;
+
+	priv->cmd.hcr = ioremap(pci_resource_start(dev->pdev, 0) + MLX4_HCR_BASE,
+				MLX4_HCR_SIZE);
+	if (!priv->cmd.hcr) {
+		mlx4_err(dev, "Couldn't map command register.");
+		return -ENOMEM;
+	}
+
+	priv->cmd.pool = pci_pool_create("mlx4_cmd", dev->pdev,
+					 MLX4_MAILBOX_SIZE,
+					 MLX4_MAILBOX_SIZE, 0);
+	if (!priv->cmd.pool) {
+		iounmap(priv->cmd.hcr);
+		return -ENOMEM;
+	}
+
+	return 0;
+}
+
+void mlx4_cmd_cleanup(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	pci_pool_destroy(priv->cmd.pool);
+	iounmap(priv->cmd.hcr);
+}
+
+/*
+ * Switch to using events to issue FW commands (can only be called
+ * after event queue for command events has been initialized).
+ */
+int mlx4_cmd_use_events(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int i;
+
+	priv->cmd.context = kmalloc(priv->cmd.max_cmds *
+				   sizeof (struct mlx4_cmd_context),
+				   GFP_KERNEL);
+	if (!priv->cmd.context)
+		return -ENOMEM;
+
+	for (i = 0; i < priv->cmd.max_cmds; ++i) {
+		priv->cmd.context[i].token = i;
+		priv->cmd.context[i].next  = i + 1;
+	}
+
+	priv->cmd.context[priv->cmd.max_cmds - 1].next = -1;
+	priv->cmd.free_head = 0;
+
+	sema_init(&priv->cmd.event_sem, priv->cmd.max_cmds);
+	spin_lock_init(&priv->cmd.context_lock);
+
+	for (priv->cmd.token_mask = 1;
+	     priv->cmd.token_mask < priv->cmd.max_cmds;
+	     priv->cmd.token_mask <<= 1)
+		; /* nothing */
+	--priv->cmd.token_mask;
+
+	priv->cmd.use_events = 1;
+
+	down(&priv->cmd.poll_sem);
+
+	return 0;
+}
+
+/*
+ * Switch back to polling (used when shutting down the device)
+ */
+void mlx4_cmd_use_polling(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int i;
+
+	priv->cmd.use_events = 0;
+
+	for (i = 0; i < priv->cmd.max_cmds; ++i)
+		down(&priv->cmd.event_sem);
+
+	kfree(priv->cmd.context);
+
+	up(&priv->cmd.poll_sem);
+}
+
+struct mlx4_cmd_mailbox *mlx4_alloc_cmd_mailbox(struct mlx4_dev *dev)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+
+	mailbox = kmalloc(sizeof *mailbox, GFP_KERNEL);
+	if (!mailbox)
+		return ERR_PTR(-ENOMEM);
+
+	mailbox->buf = pci_pool_alloc(mlx4_priv(dev)->cmd.pool, GFP_KERNEL,
+				      &mailbox->dma);
+	if (!mailbox->buf) {
+		kfree(mailbox);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	return mailbox;
+}
+EXPORT_SYMBOL_GPL(mlx4_alloc_cmd_mailbox);
+
+void mlx4_free_cmd_mailbox(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox)
+{
+	if (!mailbox)
+		return;
+
+	pci_pool_free(mlx4_priv(dev)->cmd.pool, mailbox->buf, mailbox->dma);
+	kfree(mailbox);
+}
+EXPORT_SYMBOL_GPL(mlx4_free_cmd_mailbox);
diff --git a/drivers/net/mlx4/cq.c b/drivers/net/mlx4/cq.c
new file mode 100644
index 000000000000..437d78ad0912
--- /dev/null
+++ b/drivers/net/mlx4/cq.c
@@ -0,0 +1,254 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
+ * Copyright (c) 2005, 2006, 2007 Cisco Systems, Inc. All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2004 Voltaire, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/init.h>
+#include <linux/hardirq.h>
+
+#include <linux/mlx4/cmd.h>
+
+#include "mlx4.h"
+#include "icm.h"
+
+struct mlx4_cq_context {
+	__be32			flags;
+	u16			reserved1[3];
+	__be16			page_offset;
+	__be32			logsize_usrpage;
+	u8			reserved2;
+	u8			cq_period;
+	u8			reserved3;
+	u8			cq_max_count;
+	u8			reserved4[3];
+	u8			comp_eqn;
+	u8			log_page_size;
+	u8			reserved5[2];
+	u8			mtt_base_addr_h;
+	__be32			mtt_base_addr_l;
+	__be32			last_notified_index;
+	__be32			solicit_producer_index;
+	__be32			consumer_index;
+	__be32			producer_index;
+	u8			reserved6[2];
+	__be64			db_rec_addr;
+};
+
+#define MLX4_CQ_STATUS_OK		( 0 << 28)
+#define MLX4_CQ_STATUS_OVERFLOW		( 9 << 28)
+#define MLX4_CQ_STATUS_WRITE_FAIL	(10 << 28)
+#define MLX4_CQ_FLAG_CC			( 1 << 18)
+#define MLX4_CQ_FLAG_OI			( 1 << 17)
+#define MLX4_CQ_STATE_ARMED		( 9 <<  8)
+#define MLX4_CQ_STATE_ARMED_SOL		( 6 <<  8)
+#define MLX4_EQ_STATE_FIRED		(10 <<  8)
+
+void mlx4_cq_completion(struct mlx4_dev *dev, u32 cqn)
+{
+	struct mlx4_cq *cq;
+
+	cq = radix_tree_lookup(&mlx4_priv(dev)->cq_table.tree,
+			       cqn & (dev->caps.num_cqs - 1));
+	if (!cq) {
+		mlx4_warn(dev, "Completion event for bogus CQ %08x\n", cqn);
+		return;
+	}
+
+	++cq->arm_sn;
+
+	cq->comp(cq);
+}
+
+void mlx4_cq_event(struct mlx4_dev *dev, u32 cqn, int event_type)
+{
+	struct mlx4_cq_table *cq_table = &mlx4_priv(dev)->cq_table;
+	struct mlx4_cq *cq;
+
+	spin_lock(&cq_table->lock);
+
+	cq = radix_tree_lookup(&cq_table->tree, cqn & (dev->caps.num_cqs - 1));
+	if (cq)
+		atomic_inc(&cq->refcount);
+
+	spin_unlock(&cq_table->lock);
+
+	if (!cq) {
+		mlx4_warn(dev, "Async event for bogus CQ %08x\n", cqn);
+		return;
+	}
+
+	cq->event(cq, event_type);
+
+	if (atomic_dec_and_test(&cq->refcount))
+		complete(&cq->free);
+}
+
+static int mlx4_SW2HW_CQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			 int cq_num)
+{
+	return mlx4_cmd(dev, mailbox->dma, cq_num, 0, MLX4_CMD_SW2HW_CQ,
+			MLX4_CMD_TIME_CLASS_A);
+}
+
+static int mlx4_HW2SW_CQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			 int cq_num)
+{
+	return mlx4_cmd_box(dev, 0, mailbox ? mailbox->dma : 0, cq_num,
+			    mailbox ? 0 : 1, MLX4_CMD_HW2SW_CQ,
+			    MLX4_CMD_TIME_CLASS_A);
+}
+
+int mlx4_cq_alloc(struct mlx4_dev *dev, int nent, struct mlx4_mtt *mtt,
+		  struct mlx4_uar *uar, u64 db_rec, struct mlx4_cq *cq)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cq_table *cq_table = &priv->cq_table;
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_cq_context *cq_context;
+	u64 mtt_addr;
+	int err;
+
+	cq->cqn = mlx4_bitmap_alloc(&cq_table->bitmap);
+	if (cq->cqn == -1)
+		return -ENOMEM;
+
+	err = mlx4_table_get(dev, &cq_table->table, cq->cqn);
+	if (err)
+		goto err_out;
+
+	err = mlx4_table_get(dev, &cq_table->cmpt_table, cq->cqn);
+	if (err)
+		goto err_put;
+
+	spin_lock_irq(&cq_table->lock);
+	err = radix_tree_insert(&cq_table->tree, cq->cqn, cq);
+	spin_unlock_irq(&cq_table->lock);
+	if (err)
+		goto err_cmpt_put;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox)) {
+		err = PTR_ERR(mailbox);
+		goto err_radix;
+	}
+
+	cq_context = mailbox->buf;
+	memset(cq_context, 0, sizeof *cq_context);
+
+	cq_context->logsize_usrpage = cpu_to_be32((ilog2(nent) << 24) | uar->index);
+	cq_context->comp_eqn        = priv->eq_table.eq[MLX4_EQ_COMP].eqn;
+	cq_context->log_page_size   = mtt->page_shift - MLX4_ICM_PAGE_SHIFT;
+
+	mtt_addr = mlx4_mtt_addr(dev, mtt);
+	cq_context->mtt_base_addr_h = mtt_addr >> 32;
+	cq_context->mtt_base_addr_l = cpu_to_be32(mtt_addr & 0xffffffff);
+	cq_context->db_rec_addr     = cpu_to_be64(db_rec);
+
+	err = mlx4_SW2HW_CQ(dev, mailbox, cq->cqn);
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	if (err)
+		goto err_radix;
+
+	cq->cons_index = 0;
+	cq->arm_sn     = 1;
+	cq->uar        = uar;
+	atomic_set(&cq->refcount, 1);
+	init_completion(&cq->free);
+
+	return 0;
+
+err_radix:
+	spin_lock_irq(&cq_table->lock);
+	radix_tree_delete(&cq_table->tree, cq->cqn);
+	spin_unlock_irq(&cq_table->lock);
+
+err_cmpt_put:
+	mlx4_table_put(dev, &cq_table->cmpt_table, cq->cqn);
+
+err_put:
+	mlx4_table_put(dev, &cq_table->table, cq->cqn);
+
+err_out:
+	mlx4_bitmap_free(&cq_table->bitmap, cq->cqn);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_cq_alloc);
+
+void mlx4_cq_free(struct mlx4_dev *dev, struct mlx4_cq *cq)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cq_table *cq_table = &priv->cq_table;
+	int err;
+
+	err = mlx4_HW2SW_CQ(dev, NULL, cq->cqn);
+	if (err)
+		mlx4_warn(dev, "HW2SW_CQ failed (%d) for CQN %06x\n", err, cq->cqn);
+
+	synchronize_irq(priv->eq_table.eq[MLX4_EQ_COMP].irq);
+
+	spin_lock_irq(&cq_table->lock);
+	radix_tree_delete(&cq_table->tree, cq->cqn);
+	spin_unlock_irq(&cq_table->lock);
+
+	if (atomic_dec_and_test(&cq->refcount))
+		complete(&cq->free);
+	wait_for_completion(&cq->free);
+
+	mlx4_table_put(dev, &cq_table->table, cq->cqn);
+	mlx4_bitmap_free(&cq_table->bitmap, cq->cqn);
+}
+EXPORT_SYMBOL_GPL(mlx4_cq_free);
+
+int __devinit mlx4_init_cq_table(struct mlx4_dev *dev)
+{
+	struct mlx4_cq_table *cq_table = &mlx4_priv(dev)->cq_table;
+	int err;
+
+	spin_lock_init(&cq_table->lock);
+	INIT_RADIX_TREE(&cq_table->tree, GFP_ATOMIC);
+
+	err = mlx4_bitmap_init(&cq_table->bitmap, dev->caps.num_cqs,
+			       dev->caps.num_cqs - 1, dev->caps.reserved_cqs);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+void mlx4_cleanup_cq_table(struct mlx4_dev *dev)
+{
+	/* Nothing to do to clean up radix_tree */
+	mlx4_bitmap_cleanup(&mlx4_priv(dev)->cq_table.bitmap);
+}
diff --git a/drivers/net/mlx4/eq.c b/drivers/net/mlx4/eq.c
new file mode 100644
index 000000000000..acf1c801a1b8
--- /dev/null
+++ b/drivers/net/mlx4/eq.c
@@ -0,0 +1,696 @@
+/*
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2005, 2006, 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *	- Redistributions of source code must retain the above
+ *	  copyright notice, this list of conditions and the following
+ *	  disclaimer.
+ *
+ *	- Redistributions in binary form must reproduce the above
+ *	  copyright notice, this list of conditions and the following
+ *	  disclaimer in the documentation and/or other materials
+ *	  provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/init.h>
+#include <linux/interrupt.h>
+
+#include <linux/mlx4/cmd.h>
+
+#include "mlx4.h"
+#include "fw.h"
+
+enum {
+	MLX4_NUM_ASYNC_EQE	= 0x100,
+	MLX4_NUM_SPARE_EQE	= 0x80,
+	MLX4_EQ_ENTRY_SIZE	= 0x20
+};
+
+/*
+ * Must be packed because start is 64 bits but only aligned to 32 bits.
+ */
+struct mlx4_eq_context {
+	__be32			flags;
+	u16			reserved1[3];
+	__be16			page_offset;
+	u8			log_eq_size;
+	u8			reserved2[4];
+	u8			eq_period;
+	u8			reserved3;
+	u8			eq_max_count;
+	u8			reserved4[3];
+	u8			intr;
+	u8			log_page_size;
+	u8			reserved5[2];
+	u8			mtt_base_addr_h;
+	__be32			mtt_base_addr_l;
+	u32			reserved6[2];
+	__be32			consumer_index;
+	__be32			producer_index;
+	u32			reserved7[4];
+};
+
+#define MLX4_EQ_STATUS_OK	   ( 0 << 28)
+#define MLX4_EQ_STATUS_WRITE_FAIL  (10 << 28)
+#define MLX4_EQ_OWNER_SW	   ( 0 << 24)
+#define MLX4_EQ_OWNER_HW	   ( 1 << 24)
+#define MLX4_EQ_FLAG_EC		   ( 1 << 18)
+#define MLX4_EQ_FLAG_OI		   ( 1 << 17)
+#define MLX4_EQ_STATE_ARMED	   ( 9 <<  8)
+#define MLX4_EQ_STATE_FIRED	   (10 <<  8)
+#define MLX4_EQ_STATE_ALWAYS_ARMED (11 <<  8)
+
+#define MLX4_ASYNC_EVENT_MASK ((1ull << MLX4_EVENT_TYPE_PATH_MIG)	    | \
+			       (1ull << MLX4_EVENT_TYPE_COMM_EST)	    | \
+			       (1ull << MLX4_EVENT_TYPE_SQ_DRAINED)	    | \
+			       (1ull << MLX4_EVENT_TYPE_CQ_ERROR)	    | \
+			       (1ull << MLX4_EVENT_TYPE_WQ_CATAS_ERROR)	    | \
+			       (1ull << MLX4_EVENT_TYPE_EEC_CATAS_ERROR)    | \
+			       (1ull << MLX4_EVENT_TYPE_PATH_MIG_FAILED)    | \
+			       (1ull << MLX4_EVENT_TYPE_WQ_INVAL_REQ_ERROR) | \
+			       (1ull << MLX4_EVENT_TYPE_WQ_ACCESS_ERROR)    | \
+			       (1ull << MLX4_EVENT_TYPE_LOCAL_CATAS_ERROR)  | \
+			       (1ull << MLX4_EVENT_TYPE_PORT_CHANGE)	    | \
+			       (1ull << MLX4_EVENT_TYPE_ECC_DETECT)	    | \
+			       (1ull << MLX4_EVENT_TYPE_SRQ_CATAS_ERROR)    | \
+			       (1ull << MLX4_EVENT_TYPE_SRQ_QP_LAST_WQE)    | \
+			       (1ull << MLX4_EVENT_TYPE_SRQ_LIMIT)	    | \
+			       (1ull << MLX4_EVENT_TYPE_CMD))
+#define MLX4_CATAS_EVENT_MASK  (1ull << MLX4_EVENT_TYPE_LOCAL_CATAS_ERROR)
+
+struct mlx4_eqe {
+	u8			reserved1;
+	u8			type;
+	u8			reserved2;
+	u8			subtype;
+	union {
+		u32		raw[6];
+		struct {
+			__be32	cqn;
+		} __attribute__((packed)) comp;
+		struct {
+			u16	reserved1;
+			__be16	token;
+			u32	reserved2;
+			u8	reserved3[3];
+			u8	status;
+			__be64	out_param;
+		} __attribute__((packed)) cmd;
+		struct {
+			__be32	qpn;
+		} __attribute__((packed)) qp;
+		struct {
+			__be32	srqn;
+		} __attribute__((packed)) srq;
+		struct {
+			__be32	cqn;
+			u32	reserved1;
+			u8	reserved2[3];
+			u8	syndrome;
+		} __attribute__((packed)) cq_err;
+		struct {
+			u32	reserved1[2];
+			__be32	port;
+		} __attribute__((packed)) port_change;
+	}			event;
+	u8			reserved3[3];
+	u8			owner;
+} __attribute__((packed));
+
+static void eq_set_ci(struct mlx4_eq *eq, int req_not)
+{
+	__raw_writel((__force u32) cpu_to_be32((eq->cons_index & 0xffffff) |
+					       req_not << 31),
+		     eq->doorbell);
+	/* We still want ordering, just not swabbing, so add a barrier */
+	mb();
+}
+
+static struct mlx4_eqe *get_eqe(struct mlx4_eq *eq, u32 entry)
+{
+	unsigned long off = (entry & (eq->nent - 1)) * MLX4_EQ_ENTRY_SIZE;
+	return eq->page_list[off / PAGE_SIZE].buf + off % PAGE_SIZE;
+}
+
+static struct mlx4_eqe *next_eqe_sw(struct mlx4_eq *eq)
+{
+	struct mlx4_eqe *eqe = get_eqe(eq, eq->cons_index);
+	return !!(eqe->owner & 0x80) ^ !!(eq->cons_index & eq->nent) ? NULL : eqe;
+}
+
+static int mlx4_eq_int(struct mlx4_dev *dev, struct mlx4_eq *eq)
+{
+	struct mlx4_eqe *eqe;
+	int cqn;
+	int eqes_found = 0;
+	int set_ci = 0;
+
+	while ((eqe = next_eqe_sw(eq))) {
+		/*
+		 * Make sure we read EQ entry contents after we've
+		 * checked the ownership bit.
+		 */
+		rmb();
+
+		switch (eqe->type) {
+		case MLX4_EVENT_TYPE_COMP:
+			cqn = be32_to_cpu(eqe->event.comp.cqn) & 0xffffff;
+			mlx4_cq_completion(dev, cqn);
+			break;
+
+		case MLX4_EVENT_TYPE_PATH_MIG:
+		case MLX4_EVENT_TYPE_COMM_EST:
+		case MLX4_EVENT_TYPE_SQ_DRAINED:
+		case MLX4_EVENT_TYPE_SRQ_QP_LAST_WQE:
+		case MLX4_EVENT_TYPE_WQ_CATAS_ERROR:
+		case MLX4_EVENT_TYPE_PATH_MIG_FAILED:
+		case MLX4_EVENT_TYPE_WQ_INVAL_REQ_ERROR:
+		case MLX4_EVENT_TYPE_WQ_ACCESS_ERROR:
+			mlx4_qp_event(dev, be32_to_cpu(eqe->event.qp.qpn) & 0xffffff,
+				      eqe->type);
+			break;
+
+		case MLX4_EVENT_TYPE_SRQ_LIMIT:
+		case MLX4_EVENT_TYPE_SRQ_CATAS_ERROR:
+			mlx4_srq_event(dev, be32_to_cpu(eqe->event.srq.srqn) & 0xffffff,
+				      eqe->type);
+			break;
+
+		case MLX4_EVENT_TYPE_CMD:
+			mlx4_cmd_event(dev,
+				       be16_to_cpu(eqe->event.cmd.token),
+				       eqe->event.cmd.status,
+				       be64_to_cpu(eqe->event.cmd.out_param));
+			break;
+
+		case MLX4_EVENT_TYPE_PORT_CHANGE:
+			mlx4_dispatch_event(dev, eqe->type, eqe->subtype,
+					    be32_to_cpu(eqe->event.port_change.port) >> 28);
+			break;
+
+		case MLX4_EVENT_TYPE_CQ_ERROR:
+			mlx4_warn(dev, "CQ %s on CQN %06x\n",
+				  eqe->event.cq_err.syndrome == 1 ?
+				  "overrun" : "access violation",
+				  be32_to_cpu(eqe->event.cq_err.cqn) & 0xffffff);
+			mlx4_cq_event(dev, be32_to_cpu(eqe->event.cq_err.cqn),
+				      eqe->type);
+			break;
+
+		case MLX4_EVENT_TYPE_EQ_OVERFLOW:
+			mlx4_warn(dev, "EQ overrun on EQN %d\n", eq->eqn);
+			break;
+
+		case MLX4_EVENT_TYPE_EEC_CATAS_ERROR:
+		case MLX4_EVENT_TYPE_ECC_DETECT:
+		default:
+			mlx4_warn(dev, "Unhandled event %02x(%02x) on EQ %d at index %u\n",
+				  eqe->type, eqe->subtype, eq->eqn, eq->cons_index);
+			break;
+		};
+
+		++eq->cons_index;
+		eqes_found = 1;
+		++set_ci;
+
+		/*
+		 * The HCA will think the queue has overflowed if we
+		 * don't tell it we've been processing events.  We
+		 * create our EQs with MLX4_NUM_SPARE_EQE extra
+		 * entries, so we must update our consumer index at
+		 * least that often.
+		 */
+		if (unlikely(set_ci >= MLX4_NUM_SPARE_EQE)) {
+			/*
+			 * Conditional on hca_type is OK here because
+			 * this is a rare case, not the fast path.
+			 */
+			eq_set_ci(eq, 0);
+			set_ci = 0;
+		}
+	}
+
+	eq_set_ci(eq, 1);
+
+	return eqes_found;
+}
+
+static irqreturn_t mlx4_interrupt(int irq, void *dev_ptr)
+{
+	struct mlx4_dev *dev = dev_ptr;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int work = 0;
+	int i;
+
+	writel(priv->eq_table.clr_mask, priv->eq_table.clr_int);
+
+	for (i = 0; i < MLX4_EQ_CATAS; ++i)
+		work |= mlx4_eq_int(dev, &priv->eq_table.eq[i]);
+
+	return IRQ_RETVAL(work);
+}
+
+static irqreturn_t mlx4_msi_x_interrupt(int irq, void *eq_ptr)
+{
+	struct mlx4_eq  *eq  = eq_ptr;
+	struct mlx4_dev *dev = eq->dev;
+
+	mlx4_eq_int(dev, eq);
+
+	/* MSI-X vectors always belong to us */
+	return IRQ_HANDLED;
+}
+
+static irqreturn_t mlx4_catas_interrupt(int irq, void *dev_ptr)
+{
+	mlx4_handle_catas_err(dev_ptr);
+
+	/* MSI-X vectors always belong to us */
+	return IRQ_HANDLED;
+}
+
+static int mlx4_MAP_EQ(struct mlx4_dev *dev, u64 event_mask, int unmap,
+			int eq_num)
+{
+	return mlx4_cmd(dev, event_mask, (unmap << 31) | eq_num,
+			0, MLX4_CMD_MAP_EQ, MLX4_CMD_TIME_CLASS_B);
+}
+
+static int mlx4_SW2HW_EQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			 int eq_num)
+{
+	return mlx4_cmd(dev, mailbox->dma, eq_num, 0, MLX4_CMD_SW2HW_EQ,
+			MLX4_CMD_TIME_CLASS_A);
+}
+
+static int mlx4_HW2SW_EQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			 int eq_num)
+{
+	return mlx4_cmd_box(dev, 0, mailbox->dma, eq_num, 0, MLX4_CMD_HW2SW_EQ,
+			    MLX4_CMD_TIME_CLASS_A);
+}
+
+static void __devinit __iomem *mlx4_get_eq_uar(struct mlx4_dev *dev,
+					       struct mlx4_eq *eq)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int index;
+
+	index = eq->eqn / 4 - dev->caps.reserved_eqs / 4;
+
+	if (!priv->eq_table.uar_map[index]) {
+		priv->eq_table.uar_map[index] =
+			ioremap(pci_resource_start(dev->pdev, 2) +
+				((eq->eqn / 4) << PAGE_SHIFT),
+				PAGE_SIZE);
+		if (!priv->eq_table.uar_map[index]) {
+			mlx4_err(dev, "Couldn't map EQ doorbell for EQN 0x%06x\n",
+				 eq->eqn);
+			return NULL;
+		}
+	}
+
+	return priv->eq_table.uar_map[index] + 0x800 + 8 * (eq->eqn % 4);
+}
+
+static int __devinit mlx4_create_eq(struct mlx4_dev *dev, int nent,
+				    u8 intr, struct mlx4_eq *eq)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_eq_context *eq_context;
+	int npages;
+	u64 *dma_list = NULL;
+	dma_addr_t t;
+	u64 mtt_addr;
+	int err = -ENOMEM;
+	int i;
+
+	eq->dev   = dev;
+	eq->nent  = roundup_pow_of_two(max(nent, 2));
+	npages = PAGE_ALIGN(eq->nent * MLX4_EQ_ENTRY_SIZE) / PAGE_SIZE;
+
+	eq->page_list = kmalloc(npages * sizeof *eq->page_list,
+				GFP_KERNEL);
+	if (!eq->page_list)
+		goto err_out;
+
+	for (i = 0; i < npages; ++i)
+		eq->page_list[i].buf = NULL;
+
+	dma_list = kmalloc(npages * sizeof *dma_list, GFP_KERNEL);
+	if (!dma_list)
+		goto err_out_free;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		goto err_out_free;
+	eq_context = mailbox->buf;
+
+	for (i = 0; i < npages; ++i) {
+		eq->page_list[i].buf = dma_alloc_coherent(&dev->pdev->dev,
+							  PAGE_SIZE, &t, GFP_KERNEL);
+		if (!eq->page_list[i].buf)
+			goto err_out_free_pages;
+
+		dma_list[i] = t;
+		eq->page_list[i].map = t;
+
+		memset(eq->page_list[i].buf, 0, PAGE_SIZE);
+	}
+
+	eq->eqn = mlx4_bitmap_alloc(&priv->eq_table.bitmap);
+	if (eq->eqn == -1)
+		goto err_out_free_pages;
+
+	eq->doorbell = mlx4_get_eq_uar(dev, eq);
+	if (!eq->doorbell) {
+		err = -ENOMEM;
+		goto err_out_free_eq;
+	}
+
+	err = mlx4_mtt_init(dev, npages, PAGE_SHIFT, &eq->mtt);
+	if (err)
+		goto err_out_free_eq;
+
+	err = mlx4_write_mtt(dev, &eq->mtt, 0, npages, dma_list);
+	if (err)
+		goto err_out_free_mtt;
+
+	memset(eq_context, 0, sizeof *eq_context);
+	eq_context->flags	  = cpu_to_be32(MLX4_EQ_STATUS_OK   |
+						MLX4_EQ_STATE_ARMED);
+	eq_context->log_eq_size	  = ilog2(eq->nent);
+	eq_context->intr	  = intr;
+	eq_context->log_page_size = PAGE_SHIFT - MLX4_ICM_PAGE_SHIFT;
+
+	mtt_addr = mlx4_mtt_addr(dev, &eq->mtt);
+	eq_context->mtt_base_addr_h = mtt_addr >> 32;
+	eq_context->mtt_base_addr_l = cpu_to_be32(mtt_addr & 0xffffffff);
+
+	err = mlx4_SW2HW_EQ(dev, mailbox, eq->eqn);
+	if (err) {
+		mlx4_warn(dev, "SW2HW_EQ failed (%d)\n", err);
+		goto err_out_free_mtt;
+	}
+
+	kfree(dma_list);
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	eq->cons_index = 0;
+
+	return err;
+
+err_out_free_mtt:
+	mlx4_mtt_cleanup(dev, &eq->mtt);
+
+err_out_free_eq:
+	mlx4_bitmap_free(&priv->eq_table.bitmap, eq->eqn);
+
+err_out_free_pages:
+	for (i = 0; i < npages; ++i)
+		if (eq->page_list[i].buf)
+			dma_free_coherent(&dev->pdev->dev, PAGE_SIZE,
+					  eq->page_list[i].buf,
+					  eq->page_list[i].map);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+err_out_free:
+	kfree(eq->page_list);
+	kfree(dma_list);
+
+err_out:
+	return err;
+}
+
+static void mlx4_free_eq(struct mlx4_dev *dev,
+			 struct mlx4_eq *eq)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cmd_mailbox *mailbox;
+	int err;
+	int npages = PAGE_ALIGN(MLX4_EQ_ENTRY_SIZE * eq->nent) / PAGE_SIZE;
+	int i;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return;
+
+	err = mlx4_HW2SW_EQ(dev, mailbox, eq->eqn);
+	if (err)
+		mlx4_warn(dev, "HW2SW_EQ failed (%d)\n", err);
+
+	if (0) {
+		mlx4_dbg(dev, "Dumping EQ context %02x:\n", eq->eqn);
+		for (i = 0; i < sizeof (struct mlx4_eq_context) / 4; ++i) {
+			if (i % 4 == 0)
+				printk("[%02x] ", i * 4);
+			printk(" %08x", be32_to_cpup(mailbox->buf + i * 4));
+			if ((i + 1) % 4 == 0)
+				printk("\n");
+		}
+	}
+
+	mlx4_mtt_cleanup(dev, &eq->mtt);
+	for (i = 0; i < npages; ++i)
+		pci_free_consistent(dev->pdev, PAGE_SIZE,
+				    eq->page_list[i].buf,
+				    eq->page_list[i].map);
+
+	kfree(eq->page_list);
+	mlx4_bitmap_free(&priv->eq_table.bitmap, eq->eqn);
+	mlx4_free_cmd_mailbox(dev, mailbox);
+}
+
+static void mlx4_free_irqs(struct mlx4_dev *dev)
+{
+	struct mlx4_eq_table *eq_table = &mlx4_priv(dev)->eq_table;
+	int i;
+
+	if (eq_table->have_irq)
+		free_irq(dev->pdev->irq, dev);
+	for (i = 0; i < MLX4_NUM_EQ; ++i)
+		if (eq_table->eq[i].have_irq)
+			free_irq(eq_table->eq[i].irq, eq_table->eq + i);
+}
+
+static int __devinit mlx4_map_clr_int(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	priv->clr_base = ioremap(pci_resource_start(dev->pdev, priv->fw.clr_int_bar) +
+				 priv->fw.clr_int_base, MLX4_CLR_INT_SIZE);
+	if (!priv->clr_base) {
+		mlx4_err(dev, "Couldn't map interrupt clear register, aborting.\n");
+		return -ENOMEM;
+	}
+
+	return 0;
+}
+
+static void mlx4_unmap_clr_int(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	iounmap(priv->clr_base);
+}
+
+int __devinit mlx4_map_eq_icm(struct mlx4_dev *dev, u64 icm_virt)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int ret;
+
+	/*
+	 * We assume that mapping one page is enough for the whole EQ
+	 * context table.  This is fine with all current HCAs, because
+	 * we only use 32 EQs and each EQ uses 64 bytes of context
+	 * memory, or 1 KB total.
+	 */
+	priv->eq_table.icm_virt = icm_virt;
+	priv->eq_table.icm_page = alloc_page(GFP_HIGHUSER);
+	if (!priv->eq_table.icm_page)
+		return -ENOMEM;
+	priv->eq_table.icm_dma  = pci_map_page(dev->pdev, priv->eq_table.icm_page, 0,
+					       PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
+	if (pci_dma_mapping_error(priv->eq_table.icm_dma)) {
+		__free_page(priv->eq_table.icm_page);
+		return -ENOMEM;
+	}
+
+	ret = mlx4_MAP_ICM_page(dev, priv->eq_table.icm_dma, icm_virt);
+	if (ret) {
+		pci_unmap_page(dev->pdev, priv->eq_table.icm_dma, PAGE_SIZE,
+			       PCI_DMA_BIDIRECTIONAL);
+		__free_page(priv->eq_table.icm_page);
+	}
+
+	return ret;
+}
+
+void mlx4_unmap_eq_icm(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	mlx4_UNMAP_ICM(dev, priv->eq_table.icm_virt, 1);
+	pci_unmap_page(dev->pdev, priv->eq_table.icm_dma, PAGE_SIZE,
+		       PCI_DMA_BIDIRECTIONAL);
+	__free_page(priv->eq_table.icm_page);
+}
+
+int __devinit mlx4_init_eq_table(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int err;
+	int i;
+
+	err = mlx4_bitmap_init(&priv->eq_table.bitmap, dev->caps.num_eqs,
+			       dev->caps.num_eqs - 1, dev->caps.reserved_eqs);
+	if (err)
+		return err;
+
+	for (i = 0; i < ARRAY_SIZE(priv->eq_table.uar_map); ++i)
+		priv->eq_table.uar_map[i] = NULL;
+
+	err = mlx4_map_clr_int(dev);
+	if (err)
+		goto err_out_free;
+
+	priv->eq_table.clr_mask =
+		swab32(1 << (priv->eq_table.inta_pin & 31));
+	priv->eq_table.clr_int  = priv->clr_base +
+		(priv->eq_table.inta_pin < 32 ? 4 : 0);
+
+	err = mlx4_create_eq(dev, dev->caps.num_cqs + MLX4_NUM_SPARE_EQE,
+			     (dev->flags & MLX4_FLAG_MSI_X) ? MLX4_EQ_COMP : 0,
+			     &priv->eq_table.eq[MLX4_EQ_COMP]);
+	if (err)
+		goto err_out_unmap;
+
+	err = mlx4_create_eq(dev, MLX4_NUM_ASYNC_EQE + MLX4_NUM_SPARE_EQE,
+			     (dev->flags & MLX4_FLAG_MSI_X) ? MLX4_EQ_ASYNC : 0,
+			     &priv->eq_table.eq[MLX4_EQ_ASYNC]);
+	if (err)
+		goto err_out_comp;
+
+	if (dev->flags & MLX4_FLAG_MSI_X) {
+		static const char *eq_name[] = {
+			[MLX4_EQ_COMP]  = DRV_NAME " (comp)",
+			[MLX4_EQ_ASYNC] = DRV_NAME " (async)",
+			[MLX4_EQ_CATAS] = DRV_NAME " (catas)"
+		};
+
+		err = mlx4_create_eq(dev, 1, MLX4_EQ_CATAS,
+				     &priv->eq_table.eq[MLX4_EQ_CATAS]);
+		if (err)
+			goto err_out_async;
+
+		for (i = 0; i < MLX4_EQ_CATAS; ++i) {
+			err = request_irq(priv->eq_table.eq[i].irq,
+					  mlx4_msi_x_interrupt,
+					  0, eq_name[i], priv->eq_table.eq + i);
+			if (err)
+				goto err_out_catas;
+
+			priv->eq_table.eq[i].have_irq = 1;
+		}
+
+		err = request_irq(priv->eq_table.eq[MLX4_EQ_CATAS].irq,
+				  mlx4_catas_interrupt, 0,
+				  eq_name[MLX4_EQ_CATAS], dev);
+		if (err)
+			goto err_out_catas;
+
+		priv->eq_table.eq[MLX4_EQ_CATAS].have_irq = 1;
+	} else {
+		err = request_irq(dev->pdev->irq, mlx4_interrupt,
+				  SA_SHIRQ, DRV_NAME, dev);
+		if (err)
+			goto err_out_async;
+
+		priv->eq_table.have_irq = 1;
+	}
+
+	err = mlx4_MAP_EQ(dev, MLX4_ASYNC_EVENT_MASK, 0,
+			  priv->eq_table.eq[MLX4_EQ_ASYNC].eqn);
+	if (err)
+		mlx4_warn(dev, "MAP_EQ for async EQ %d failed (%d)\n",
+			   priv->eq_table.eq[MLX4_EQ_ASYNC].eqn, err);
+
+	for (i = 0; i < MLX4_EQ_CATAS; ++i)
+		eq_set_ci(&priv->eq_table.eq[i], 1);
+
+	if (dev->flags & MLX4_FLAG_MSI_X) {
+		err = mlx4_MAP_EQ(dev, MLX4_CATAS_EVENT_MASK, 0,
+				  priv->eq_table.eq[MLX4_EQ_CATAS].eqn);
+		if (err)
+			mlx4_warn(dev, "MAP_EQ for catas EQ %d failed (%d)\n",
+				  priv->eq_table.eq[MLX4_EQ_CATAS].eqn, err);
+	}
+
+	return 0;
+
+err_out_catas:
+	mlx4_free_eq(dev, &priv->eq_table.eq[MLX4_EQ_CATAS]);
+
+err_out_async:
+	mlx4_free_eq(dev, &priv->eq_table.eq[MLX4_EQ_ASYNC]);
+
+err_out_comp:
+	mlx4_free_eq(dev, &priv->eq_table.eq[MLX4_EQ_COMP]);
+
+err_out_unmap:
+	mlx4_unmap_clr_int(dev);
+	mlx4_free_irqs(dev);
+
+err_out_free:
+	mlx4_bitmap_cleanup(&priv->eq_table.bitmap);
+	return err;
+}
+
+void mlx4_cleanup_eq_table(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int i;
+
+	if (dev->flags & MLX4_FLAG_MSI_X)
+		mlx4_MAP_EQ(dev, MLX4_CATAS_EVENT_MASK, 1,
+			    priv->eq_table.eq[MLX4_EQ_CATAS].eqn);
+
+	mlx4_MAP_EQ(dev, MLX4_ASYNC_EVENT_MASK, 1,
+		    priv->eq_table.eq[MLX4_EQ_ASYNC].eqn);
+
+	mlx4_free_irqs(dev);
+
+	for (i = 0; i < MLX4_EQ_CATAS; ++i)
+		mlx4_free_eq(dev, &priv->eq_table.eq[i]);
+	if (dev->flags & MLX4_FLAG_MSI_X)
+		mlx4_free_eq(dev, &priv->eq_table.eq[MLX4_EQ_CATAS]);
+
+	mlx4_unmap_clr_int(dev);
+
+	for (i = 0; i < ARRAY_SIZE(priv->eq_table.uar_map); ++i)
+		if (priv->eq_table.uar_map[i])
+			iounmap(priv->eq_table.uar_map[i]);
+
+	mlx4_bitmap_cleanup(&priv->eq_table.bitmap);
+}
diff --git a/drivers/net/mlx4/fw.c b/drivers/net/mlx4/fw.c
new file mode 100644
index 000000000000..c42717313663
--- /dev/null
+++ b/drivers/net/mlx4/fw.c
@@ -0,0 +1,775 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2005, 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx4/cmd.h>
+
+#include "fw.h"
+#include "icm.h"
+
+extern void __buggy_use_of_MLX4_GET(void);
+extern void __buggy_use_of_MLX4_PUT(void);
+
+#define MLX4_GET(dest, source, offset)				      \
+	do {							      \
+		void *__p = (char *) (source) + (offset);	      \
+		switch (sizeof (dest)) {			      \
+		case 1: (dest) = *(u8 *) __p;	    break;	      \
+		case 2: (dest) = be16_to_cpup(__p); break;	      \
+		case 4: (dest) = be32_to_cpup(__p); break;	      \
+		case 8: (dest) = be64_to_cpup(__p); break;	      \
+		default: __buggy_use_of_MLX4_GET();		      \
+		}						      \
+	} while (0)
+
+#define MLX4_PUT(dest, source, offset)				      \
+	do {							      \
+		void *__d = ((char *) (dest) + (offset));	      \
+		switch (sizeof(source)) {			      \
+		case 1: *(u8 *) __d = (source);		       break; \
+		case 2:	*(__be16 *) __d = cpu_to_be16(source); break; \
+		case 4:	*(__be32 *) __d = cpu_to_be32(source); break; \
+		case 8:	*(__be64 *) __d = cpu_to_be64(source); break; \
+		default: __buggy_use_of_MLX4_PUT();		      \
+		}						      \
+	} while (0)
+
+static void dump_dev_cap_flags(struct mlx4_dev *dev, u32 flags)
+{
+	static const char *fname[] = {
+		[ 0] = "RC transport",
+		[ 1] = "UC transport",
+		[ 2] = "UD transport",
+		[ 3] = "SRC transport",
+		[ 4] = "reliable multicast",
+		[ 5] = "FCoIB support",
+		[ 6] = "SRQ support",
+		[ 7] = "IPoIB checksum offload",
+		[ 8] = "P_Key violation counter",
+		[ 9] = "Q_Key violation counter",
+		[10] = "VMM",
+		[16] = "MW support",
+		[17] = "APM support",
+		[18] = "Atomic ops support",
+		[19] = "Raw multicast support",
+		[20] = "Address vector port checking support",
+		[21] = "UD multicast support",
+		[24] = "Demand paging support",
+		[25] = "Router support"
+	};
+	int i;
+
+	mlx4_dbg(dev, "DEV_CAP flags:\n");
+	for (i = 0; i < 32; ++i)
+		if (fname[i] && (flags & (1 << i)))
+			mlx4_dbg(dev, "    %s\n", fname[i]);
+}
+
+int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	u32 *outbox;
+	u8 field;
+	u16 size;
+	u16 stat_rate;
+	int err;
+
+#define QUERY_DEV_CAP_OUT_SIZE		       0x100
+#define QUERY_DEV_CAP_MAX_SRQ_SZ_OFFSET		0x10
+#define QUERY_DEV_CAP_MAX_QP_SZ_OFFSET		0x11
+#define QUERY_DEV_CAP_RSVD_QP_OFFSET		0x12
+#define QUERY_DEV_CAP_MAX_QP_OFFSET		0x13
+#define QUERY_DEV_CAP_RSVD_SRQ_OFFSET		0x14
+#define QUERY_DEV_CAP_MAX_SRQ_OFFSET		0x15
+#define QUERY_DEV_CAP_RSVD_EEC_OFFSET		0x16
+#define QUERY_DEV_CAP_MAX_EEC_OFFSET		0x17
+#define QUERY_DEV_CAP_MAX_CQ_SZ_OFFSET		0x19
+#define QUERY_DEV_CAP_RSVD_CQ_OFFSET		0x1a
+#define QUERY_DEV_CAP_MAX_CQ_OFFSET		0x1b
+#define QUERY_DEV_CAP_MAX_MPT_OFFSET		0x1d
+#define QUERY_DEV_CAP_RSVD_EQ_OFFSET		0x1e
+#define QUERY_DEV_CAP_MAX_EQ_OFFSET		0x1f
+#define QUERY_DEV_CAP_RSVD_MTT_OFFSET		0x20
+#define QUERY_DEV_CAP_MAX_MRW_SZ_OFFSET		0x21
+#define QUERY_DEV_CAP_RSVD_MRW_OFFSET		0x22
+#define QUERY_DEV_CAP_MAX_MTT_SEG_OFFSET	0x23
+#define QUERY_DEV_CAP_MAX_AV_OFFSET		0x27
+#define QUERY_DEV_CAP_MAX_REQ_QP_OFFSET		0x29
+#define QUERY_DEV_CAP_MAX_RES_QP_OFFSET		0x2b
+#define QUERY_DEV_CAP_MAX_RDMA_OFFSET		0x2f
+#define QUERY_DEV_CAP_RSZ_SRQ_OFFSET		0x33
+#define QUERY_DEV_CAP_ACK_DELAY_OFFSET		0x35
+#define QUERY_DEV_CAP_MTU_WIDTH_OFFSET		0x36
+#define QUERY_DEV_CAP_VL_PORT_OFFSET		0x37
+#define QUERY_DEV_CAP_MAX_GID_OFFSET		0x3b
+#define QUERY_DEV_CAP_RATE_SUPPORT_OFFSET	0x3c
+#define QUERY_DEV_CAP_MAX_PKEY_OFFSET		0x3f
+#define QUERY_DEV_CAP_FLAGS_OFFSET		0x44
+#define QUERY_DEV_CAP_RSVD_UAR_OFFSET		0x48
+#define QUERY_DEV_CAP_UAR_SZ_OFFSET		0x49
+#define QUERY_DEV_CAP_PAGE_SZ_OFFSET		0x4b
+#define QUERY_DEV_CAP_BF_OFFSET			0x4c
+#define QUERY_DEV_CAP_LOG_BF_REG_SZ_OFFSET	0x4d
+#define QUERY_DEV_CAP_LOG_MAX_BF_REGS_PER_PAGE_OFFSET	0x4e
+#define QUERY_DEV_CAP_LOG_MAX_BF_PAGES_OFFSET	0x4f
+#define QUERY_DEV_CAP_MAX_SG_SQ_OFFSET		0x51
+#define QUERY_DEV_CAP_MAX_DESC_SZ_SQ_OFFSET	0x52
+#define QUERY_DEV_CAP_MAX_SG_RQ_OFFSET		0x55
+#define QUERY_DEV_CAP_MAX_DESC_SZ_RQ_OFFSET	0x56
+#define QUERY_DEV_CAP_MAX_QP_MCG_OFFSET		0x61
+#define QUERY_DEV_CAP_RSVD_MCG_OFFSET		0x62
+#define QUERY_DEV_CAP_MAX_MCG_OFFSET		0x63
+#define QUERY_DEV_CAP_RSVD_PD_OFFSET		0x64
+#define QUERY_DEV_CAP_MAX_PD_OFFSET		0x65
+#define QUERY_DEV_CAP_RDMARC_ENTRY_SZ_OFFSET	0x80
+#define QUERY_DEV_CAP_QPC_ENTRY_SZ_OFFSET	0x82
+#define QUERY_DEV_CAP_AUX_ENTRY_SZ_OFFSET	0x84
+#define QUERY_DEV_CAP_ALTC_ENTRY_SZ_OFFSET	0x86
+#define QUERY_DEV_CAP_EQC_ENTRY_SZ_OFFSET	0x88
+#define QUERY_DEV_CAP_CQC_ENTRY_SZ_OFFSET	0x8a
+#define QUERY_DEV_CAP_SRQ_ENTRY_SZ_OFFSET	0x8c
+#define QUERY_DEV_CAP_C_MPT_ENTRY_SZ_OFFSET	0x8e
+#define QUERY_DEV_CAP_MTT_ENTRY_SZ_OFFSET	0x90
+#define QUERY_DEV_CAP_D_MPT_ENTRY_SZ_OFFSET	0x92
+#define QUERY_DEV_CAP_BMME_FLAGS_OFFSET		0x97
+#define QUERY_DEV_CAP_RSVD_LKEY_OFFSET		0x98
+#define QUERY_DEV_CAP_MAX_ICM_SZ_OFFSET		0xa0
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	outbox = mailbox->buf;
+
+	err = mlx4_cmd_box(dev, 0, mailbox->dma, 0, 0, MLX4_CMD_QUERY_DEV_CAP,
+			   MLX4_CMD_TIME_CLASS_A);
+
+	if (err)
+		goto out;
+
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_QP_OFFSET);
+	dev_cap->reserved_qps = 1 << (field & 0xf);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_QP_OFFSET);
+	dev_cap->max_qps = 1 << (field & 0x1f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_SRQ_OFFSET);
+	dev_cap->reserved_srqs = 1 << (field >> 4);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_SRQ_OFFSET);
+	dev_cap->max_srqs = 1 << (field & 0x1f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_CQ_SZ_OFFSET);
+	dev_cap->max_cq_sz = 1 << field;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_CQ_OFFSET);
+	dev_cap->reserved_cqs = 1 << (field & 0xf);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_CQ_OFFSET);
+	dev_cap->max_cqs = 1 << (field & 0x1f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_MPT_OFFSET);
+	dev_cap->max_mpts = 1 << (field & 0x3f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_EQ_OFFSET);
+	dev_cap->reserved_eqs = 1 << (field & 0xf);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_EQ_OFFSET);
+	dev_cap->max_eqs = 1 << (field & 0x7);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_MTT_OFFSET);
+	dev_cap->reserved_mtts = 1 << (field >> 4);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_MRW_SZ_OFFSET);
+	dev_cap->max_mrw_sz = 1 << field;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_MRW_OFFSET);
+	dev_cap->reserved_mrws = 1 << (field & 0xf);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_MTT_SEG_OFFSET);
+	dev_cap->max_mtt_seg = 1 << (field & 0x3f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_REQ_QP_OFFSET);
+	dev_cap->max_requester_per_qp = 1 << (field & 0x3f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_RES_QP_OFFSET);
+	dev_cap->max_responder_per_qp = 1 << (field & 0x3f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_RDMA_OFFSET);
+	dev_cap->max_rdma_global = 1 << (field & 0x3f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_ACK_DELAY_OFFSET);
+	dev_cap->local_ca_ack_delay = field & 0x1f;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MTU_WIDTH_OFFSET);
+	dev_cap->max_mtu	= field >> 4;
+	dev_cap->max_port_width = field & 0xf;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_VL_PORT_OFFSET);
+	dev_cap->max_vl    = field >> 4;
+	dev_cap->num_ports = field & 0xf;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_GID_OFFSET);
+	dev_cap->max_gids = 1 << (field & 0xf);
+	MLX4_GET(stat_rate, outbox, QUERY_DEV_CAP_RATE_SUPPORT_OFFSET);
+	dev_cap->stat_rate_support = stat_rate;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_PKEY_OFFSET);
+	dev_cap->max_pkeys = 1 << (field & 0xf);
+	MLX4_GET(dev_cap->flags, outbox, QUERY_DEV_CAP_FLAGS_OFFSET);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_UAR_OFFSET);
+	dev_cap->reserved_uars = field >> 4;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_UAR_SZ_OFFSET);
+	dev_cap->uar_size = 1 << ((field & 0x3f) + 20);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_PAGE_SZ_OFFSET);
+	dev_cap->min_page_sz = 1 << field;
+
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_BF_OFFSET);
+	if (field & 0x80) {
+		MLX4_GET(field, outbox, QUERY_DEV_CAP_LOG_BF_REG_SZ_OFFSET);
+		dev_cap->bf_reg_size = 1 << (field & 0x1f);
+		MLX4_GET(field, outbox, QUERY_DEV_CAP_LOG_MAX_BF_REGS_PER_PAGE_OFFSET);
+		dev_cap->bf_regs_per_page = 1 << (field & 0x3f);
+		mlx4_dbg(dev, "BlueFlame available (reg size %d, regs/page %d)\n",
+			 dev_cap->bf_reg_size, dev_cap->bf_regs_per_page);
+	} else {
+		dev_cap->bf_reg_size = 0;
+		mlx4_dbg(dev, "BlueFlame not available\n");
+	}
+
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_SG_SQ_OFFSET);
+	dev_cap->max_sq_sg = field;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_MAX_DESC_SZ_SQ_OFFSET);
+	dev_cap->max_sq_desc_sz = size;
+
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_QP_MCG_OFFSET);
+	dev_cap->max_qp_per_mcg = 1 << field;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_MCG_OFFSET);
+	dev_cap->reserved_mgms = field & 0xf;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_MCG_OFFSET);
+	dev_cap->max_mcgs = 1 << field;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_PD_OFFSET);
+	dev_cap->reserved_pds = field >> 4;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_PD_OFFSET);
+	dev_cap->max_pds = 1 << (field & 0x3f);
+
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_RDMARC_ENTRY_SZ_OFFSET);
+	dev_cap->rdmarc_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_QPC_ENTRY_SZ_OFFSET);
+	dev_cap->qpc_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_AUX_ENTRY_SZ_OFFSET);
+	dev_cap->aux_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_ALTC_ENTRY_SZ_OFFSET);
+	dev_cap->altc_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_EQC_ENTRY_SZ_OFFSET);
+	dev_cap->eqc_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_CQC_ENTRY_SZ_OFFSET);
+	dev_cap->cqc_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_SRQ_ENTRY_SZ_OFFSET);
+	dev_cap->srq_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_C_MPT_ENTRY_SZ_OFFSET);
+	dev_cap->cmpt_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_MTT_ENTRY_SZ_OFFSET);
+	dev_cap->mtt_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_D_MPT_ENTRY_SZ_OFFSET);
+	dev_cap->dmpt_entry_sz = size;
+
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_SRQ_SZ_OFFSET);
+	dev_cap->max_srq_sz = 1 << field;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_QP_SZ_OFFSET);
+	dev_cap->max_qp_sz = 1 << field;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSZ_SRQ_OFFSET);
+	dev_cap->resize_srq = field & 1;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_SG_RQ_OFFSET);
+	dev_cap->max_rq_sg = field;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_MAX_DESC_SZ_RQ_OFFSET);
+	dev_cap->max_rq_desc_sz = size;
+
+	MLX4_GET(dev_cap->bmme_flags, outbox,
+		 QUERY_DEV_CAP_BMME_FLAGS_OFFSET);
+	MLX4_GET(dev_cap->reserved_lkey, outbox,
+		 QUERY_DEV_CAP_RSVD_LKEY_OFFSET);
+	MLX4_GET(dev_cap->max_icm_sz, outbox,
+		 QUERY_DEV_CAP_MAX_ICM_SZ_OFFSET);
+
+	if (dev_cap->bmme_flags & 1)
+		mlx4_dbg(dev, "Base MM extensions: yes "
+			 "(flags %d, rsvd L_Key %08x)\n",
+			 dev_cap->bmme_flags, dev_cap->reserved_lkey);
+	else
+		mlx4_dbg(dev, "Base MM extensions: no\n");
+
+	/*
+	 * Each UAR has 4 EQ doorbells; so if a UAR is reserved, then
+	 * we can't use any EQs whose doorbell falls on that page,
+	 * even if the EQ itself isn't reserved.
+	 */
+	dev_cap->reserved_eqs = max(dev_cap->reserved_uars * 4,
+				    dev_cap->reserved_eqs);
+
+	mlx4_dbg(dev, "Max ICM size %lld MB\n",
+		 (unsigned long long) dev_cap->max_icm_sz >> 20);
+	mlx4_dbg(dev, "Max QPs: %d, reserved QPs: %d, entry size: %d\n",
+		 dev_cap->max_qps, dev_cap->reserved_qps, dev_cap->qpc_entry_sz);
+	mlx4_dbg(dev, "Max SRQs: %d, reserved SRQs: %d, entry size: %d\n",
+		 dev_cap->max_srqs, dev_cap->reserved_srqs, dev_cap->srq_entry_sz);
+	mlx4_dbg(dev, "Max CQs: %d, reserved CQs: %d, entry size: %d\n",
+		 dev_cap->max_cqs, dev_cap->reserved_cqs, dev_cap->cqc_entry_sz);
+	mlx4_dbg(dev, "Max EQs: %d, reserved EQs: %d, entry size: %d\n",
+		 dev_cap->max_eqs, dev_cap->reserved_eqs, dev_cap->eqc_entry_sz);
+	mlx4_dbg(dev, "reserved MPTs: %d, reserved MTTs: %d\n",
+		 dev_cap->reserved_mrws, dev_cap->reserved_mtts);
+	mlx4_dbg(dev, "Max PDs: %d, reserved PDs: %d, reserved UARs: %d\n",
+		 dev_cap->max_pds, dev_cap->reserved_pds, dev_cap->reserved_uars);
+	mlx4_dbg(dev, "Max QP/MCG: %d, reserved MGMs: %d\n",
+		 dev_cap->max_pds, dev_cap->reserved_mgms);
+	mlx4_dbg(dev, "Max CQEs: %d, max WQEs: %d, max SRQ WQEs: %d\n",
+		 dev_cap->max_cq_sz, dev_cap->max_qp_sz, dev_cap->max_srq_sz);
+	mlx4_dbg(dev, "Local CA ACK delay: %d, max MTU: %d, port width cap: %d\n",
+		 dev_cap->local_ca_ack_delay, 128 << dev_cap->max_mtu,
+		 dev_cap->max_port_width);
+	mlx4_dbg(dev, "Max SQ desc size: %d, max SQ S/G: %d\n",
+		 dev_cap->max_sq_desc_sz, dev_cap->max_sq_sg);
+	mlx4_dbg(dev, "Max RQ desc size: %d, max RQ S/G: %d\n",
+		 dev_cap->max_rq_desc_sz, dev_cap->max_rq_sg);
+
+	dump_dev_cap_flags(dev, dev_cap->flags);
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+int mlx4_map_cmd(struct mlx4_dev *dev, u16 op, struct mlx4_icm *icm, u64 virt)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_icm_iter iter;
+	__be64 *pages;
+	int lg;
+	int nent = 0;
+	int i;
+	int err = 0;
+	int ts = 0, tc = 0;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	memset(mailbox->buf, 0, MLX4_MAILBOX_SIZE);
+	pages = mailbox->buf;
+
+	for (mlx4_icm_first(icm, &iter);
+	     !mlx4_icm_last(&iter);
+	     mlx4_icm_next(&iter)) {
+		/*
+		 * We have to pass pages that are aligned to their
+		 * size, so find the least significant 1 in the
+		 * address or size and use that as our log2 size.
+		 */
+		lg = ffs(mlx4_icm_addr(&iter) | mlx4_icm_size(&iter)) - 1;
+		if (lg < MLX4_ICM_PAGE_SHIFT) {
+			mlx4_warn(dev, "Got FW area not aligned to %d (%llx/%lx).\n",
+				   MLX4_ICM_PAGE_SIZE,
+				   (unsigned long long) mlx4_icm_addr(&iter),
+				   mlx4_icm_size(&iter));
+			err = -EINVAL;
+			goto out;
+		}
+
+		for (i = 0; i < mlx4_icm_size(&iter) >> lg; ++i) {
+			if (virt != -1) {
+				pages[nent * 2] = cpu_to_be64(virt);
+				virt += 1 << lg;
+			}
+
+			pages[nent * 2 + 1] =
+				cpu_to_be64((mlx4_icm_addr(&iter) + (i << lg)) |
+					    (lg - MLX4_ICM_PAGE_SHIFT));
+			ts += 1 << (lg - 10);
+			++tc;
+
+			if (++nent == MLX4_MAILBOX_SIZE / 16) {
+				err = mlx4_cmd(dev, mailbox->dma, nent, 0, op,
+						MLX4_CMD_TIME_CLASS_B);
+				if (err)
+					goto out;
+				nent = 0;
+			}
+		}
+	}
+
+	if (nent)
+		err = mlx4_cmd(dev, mailbox->dma, nent, 0, op, MLX4_CMD_TIME_CLASS_B);
+	if (err)
+		goto out;
+
+	switch (op) {
+	case MLX4_CMD_MAP_FA:
+		mlx4_dbg(dev, "Mapped %d chunks/%d KB for FW.\n", tc, ts);
+		break;
+	case MLX4_CMD_MAP_ICM_AUX:
+		mlx4_dbg(dev, "Mapped %d chunks/%d KB for ICM aux.\n", tc, ts);
+		break;
+	case MLX4_CMD_MAP_ICM:
+		mlx4_dbg(dev, "Mapped %d chunks/%d KB at %llx for ICM.\n",
+			  tc, ts, (unsigned long long) virt - (ts << 10));
+		break;
+	}
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+int mlx4_MAP_FA(struct mlx4_dev *dev, struct mlx4_icm *icm)
+{
+	return mlx4_map_cmd(dev, MLX4_CMD_MAP_FA, icm, -1);
+}
+
+int mlx4_UNMAP_FA(struct mlx4_dev *dev)
+{
+	return mlx4_cmd(dev, 0, 0, 0, MLX4_CMD_UNMAP_FA, MLX4_CMD_TIME_CLASS_B);
+}
+
+
+int mlx4_RUN_FW(struct mlx4_dev *dev)
+{
+	return mlx4_cmd(dev, 0, 0, 0, MLX4_CMD_RUN_FW, MLX4_CMD_TIME_CLASS_A);
+}
+
+int mlx4_QUERY_FW(struct mlx4_dev *dev)
+{
+	struct mlx4_fw  *fw  = &mlx4_priv(dev)->fw;
+	struct mlx4_cmd *cmd = &mlx4_priv(dev)->cmd;
+	struct mlx4_cmd_mailbox *mailbox;
+	u32 *outbox;
+	int err = 0;
+	u64 fw_ver;
+	u8 lg;
+
+#define QUERY_FW_OUT_SIZE             0x100
+#define QUERY_FW_VER_OFFSET            0x00
+#define QUERY_FW_MAX_CMD_OFFSET        0x0f
+#define QUERY_FW_ERR_START_OFFSET      0x30
+#define QUERY_FW_ERR_SIZE_OFFSET       0x38
+#define QUERY_FW_ERR_BAR_OFFSET        0x3c
+
+#define QUERY_FW_SIZE_OFFSET           0x00
+#define QUERY_FW_CLR_INT_BASE_OFFSET   0x20
+#define QUERY_FW_CLR_INT_BAR_OFFSET    0x28
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	outbox = mailbox->buf;
+
+	err = mlx4_cmd_box(dev, 0, mailbox->dma, 0, 0, MLX4_CMD_QUERY_FW,
+			    MLX4_CMD_TIME_CLASS_A);
+	if (err)
+		goto out;
+
+	MLX4_GET(fw_ver, outbox, QUERY_FW_VER_OFFSET);
+	/*
+	 * FW subminor version is at more signifant bits than minor
+	 * version, so swap here.
+	 */
+	dev->caps.fw_ver = (fw_ver & 0xffff00000000ull) |
+		((fw_ver & 0xffff0000ull) >> 16) |
+		((fw_ver & 0x0000ffffull) << 16);
+
+	MLX4_GET(lg, outbox, QUERY_FW_MAX_CMD_OFFSET);
+	cmd->max_cmds = 1 << lg;
+
+	mlx4_dbg(dev, "FW version %d.%d.%03d, max commands %d\n",
+		 (int) (dev->caps.fw_ver >> 32),
+		 (int) (dev->caps.fw_ver >> 16) & 0xffff,
+		 (int) dev->caps.fw_ver & 0xffff,
+		 cmd->max_cmds);
+
+	MLX4_GET(fw->catas_offset, outbox, QUERY_FW_ERR_START_OFFSET);
+	MLX4_GET(fw->catas_size,   outbox, QUERY_FW_ERR_SIZE_OFFSET);
+	MLX4_GET(fw->catas_bar,    outbox, QUERY_FW_ERR_BAR_OFFSET);
+	fw->catas_bar = (fw->catas_bar >> 6) * 2;
+
+	mlx4_dbg(dev, "Catastrophic error buffer at 0x%llx, size 0x%x, BAR %d\n",
+		 (unsigned long long) fw->catas_offset, fw->catas_size, fw->catas_bar);
+
+	MLX4_GET(fw->fw_pages,     outbox, QUERY_FW_SIZE_OFFSET);
+	MLX4_GET(fw->clr_int_base, outbox, QUERY_FW_CLR_INT_BASE_OFFSET);
+	MLX4_GET(fw->clr_int_bar,  outbox, QUERY_FW_CLR_INT_BAR_OFFSET);
+	fw->clr_int_bar = (fw->clr_int_bar >> 6) * 2;
+
+	mlx4_dbg(dev, "FW size %d KB\n", fw->fw_pages >> 2);
+
+	/*
+	 * Round up number of system pages needed in case
+	 * MLX4_ICM_PAGE_SIZE < PAGE_SIZE.
+	 */
+	fw->fw_pages =
+		ALIGN(fw->fw_pages, PAGE_SIZE / MLX4_ICM_PAGE_SIZE) >>
+		(PAGE_SHIFT - MLX4_ICM_PAGE_SHIFT);
+
+	mlx4_dbg(dev, "Clear int @ %llx, BAR %d\n",
+		 (unsigned long long) fw->clr_int_base, fw->clr_int_bar);
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+static void get_board_id(void *vsd, char *board_id)
+{
+	int i;
+
+#define VSD_OFFSET_SIG1		0x00
+#define VSD_OFFSET_SIG2		0xde
+#define VSD_OFFSET_MLX_BOARD_ID	0xd0
+#define VSD_OFFSET_TS_BOARD_ID	0x20
+
+#define VSD_SIGNATURE_TOPSPIN	0x5ad
+
+	memset(board_id, 0, MLX4_BOARD_ID_LEN);
+
+	if (be16_to_cpup(vsd + VSD_OFFSET_SIG1) == VSD_SIGNATURE_TOPSPIN &&
+	    be16_to_cpup(vsd + VSD_OFFSET_SIG2) == VSD_SIGNATURE_TOPSPIN) {
+		strlcpy(board_id, vsd + VSD_OFFSET_TS_BOARD_ID, MLX4_BOARD_ID_LEN);
+	} else {
+		/*
+		 * The board ID is a string but the firmware byte
+		 * swaps each 4-byte word before passing it back to
+		 * us.  Therefore we need to swab it before printing.
+		 */
+		for (i = 0; i < 4; ++i)
+			((u32 *) board_id)[i] =
+				swab32(*(u32 *) (vsd + VSD_OFFSET_MLX_BOARD_ID + i * 4));
+	}
+}
+
+int mlx4_QUERY_ADAPTER(struct mlx4_dev *dev, struct mlx4_adapter *adapter)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	u32 *outbox;
+	int err;
+
+#define QUERY_ADAPTER_OUT_SIZE             0x100
+#define QUERY_ADAPTER_VENDOR_ID_OFFSET     0x00
+#define QUERY_ADAPTER_DEVICE_ID_OFFSET     0x04
+#define QUERY_ADAPTER_REVISION_ID_OFFSET   0x08
+#define QUERY_ADAPTER_INTA_PIN_OFFSET      0x10
+#define QUERY_ADAPTER_VSD_OFFSET           0x20
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	outbox = mailbox->buf;
+
+	err = mlx4_cmd_box(dev, 0, mailbox->dma, 0, 0, MLX4_CMD_QUERY_ADAPTER,
+			   MLX4_CMD_TIME_CLASS_A);
+	if (err)
+		goto out;
+
+	MLX4_GET(adapter->vendor_id, outbox,   QUERY_ADAPTER_VENDOR_ID_OFFSET);
+	MLX4_GET(adapter->device_id, outbox,   QUERY_ADAPTER_DEVICE_ID_OFFSET);
+	MLX4_GET(adapter->revision_id, outbox, QUERY_ADAPTER_REVISION_ID_OFFSET);
+	MLX4_GET(adapter->inta_pin, outbox,    QUERY_ADAPTER_INTA_PIN_OFFSET);
+
+	get_board_id(outbox + QUERY_ADAPTER_VSD_OFFSET / 4,
+		     adapter->board_id);
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+int mlx4_INIT_HCA(struct mlx4_dev *dev, struct mlx4_init_hca_param *param)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	__be32 *inbox;
+	int err;
+
+#define INIT_HCA_IN_SIZE		 0x200
+#define INIT_HCA_VERSION_OFFSET		 0x000
+#define	 INIT_HCA_VERSION		 2
+#define INIT_HCA_FLAGS_OFFSET		 0x014
+#define INIT_HCA_QPC_OFFSET		 0x020
+#define	 INIT_HCA_QPC_BASE_OFFSET	 (INIT_HCA_QPC_OFFSET + 0x10)
+#define	 INIT_HCA_LOG_QP_OFFSET		 (INIT_HCA_QPC_OFFSET + 0x17)
+#define	 INIT_HCA_SRQC_BASE_OFFSET	 (INIT_HCA_QPC_OFFSET + 0x28)
+#define	 INIT_HCA_LOG_SRQ_OFFSET	 (INIT_HCA_QPC_OFFSET + 0x2f)
+#define	 INIT_HCA_CQC_BASE_OFFSET	 (INIT_HCA_QPC_OFFSET + 0x30)
+#define	 INIT_HCA_LOG_CQ_OFFSET		 (INIT_HCA_QPC_OFFSET + 0x37)
+#define	 INIT_HCA_ALTC_BASE_OFFSET	 (INIT_HCA_QPC_OFFSET + 0x40)
+#define	 INIT_HCA_AUXC_BASE_OFFSET	 (INIT_HCA_QPC_OFFSET + 0x50)
+#define	 INIT_HCA_EQC_BASE_OFFSET	 (INIT_HCA_QPC_OFFSET + 0x60)
+#define	 INIT_HCA_LOG_EQ_OFFSET		 (INIT_HCA_QPC_OFFSET + 0x67)
+#define	 INIT_HCA_RDMARC_BASE_OFFSET	 (INIT_HCA_QPC_OFFSET + 0x70)
+#define	 INIT_HCA_LOG_RD_OFFSET		 (INIT_HCA_QPC_OFFSET + 0x77)
+#define INIT_HCA_MCAST_OFFSET		 0x0c0
+#define	 INIT_HCA_MC_BASE_OFFSET	 (INIT_HCA_MCAST_OFFSET + 0x00)
+#define	 INIT_HCA_LOG_MC_ENTRY_SZ_OFFSET (INIT_HCA_MCAST_OFFSET + 0x12)
+#define	 INIT_HCA_LOG_MC_HASH_SZ_OFFSET	 (INIT_HCA_MCAST_OFFSET + 0x16)
+#define	 INIT_HCA_LOG_MC_TABLE_SZ_OFFSET (INIT_HCA_MCAST_OFFSET + 0x1b)
+#define INIT_HCA_TPT_OFFSET		 0x0f0
+#define	 INIT_HCA_DMPT_BASE_OFFSET	 (INIT_HCA_TPT_OFFSET + 0x00)
+#define	 INIT_HCA_LOG_MPT_SZ_OFFSET	 (INIT_HCA_TPT_OFFSET + 0x0b)
+#define	 INIT_HCA_MTT_BASE_OFFSET	 (INIT_HCA_TPT_OFFSET + 0x10)
+#define	 INIT_HCA_CMPT_BASE_OFFSET	 (INIT_HCA_TPT_OFFSET + 0x18)
+#define INIT_HCA_UAR_OFFSET		 0x120
+#define	 INIT_HCA_LOG_UAR_SZ_OFFSET	 (INIT_HCA_UAR_OFFSET + 0x0a)
+#define  INIT_HCA_UAR_PAGE_SZ_OFFSET     (INIT_HCA_UAR_OFFSET + 0x0b)
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	inbox = mailbox->buf;
+
+	memset(inbox, 0, INIT_HCA_IN_SIZE);
+
+	*((u8 *) mailbox->buf + INIT_HCA_VERSION_OFFSET) = INIT_HCA_VERSION;
+
+#if defined(__LITTLE_ENDIAN)
+	*(inbox + INIT_HCA_FLAGS_OFFSET / 4) &= ~cpu_to_be32(1 << 1);
+#elif defined(__BIG_ENDIAN)
+	*(inbox + INIT_HCA_FLAGS_OFFSET / 4) |= cpu_to_be32(1 << 1);
+#else
+#error Host endianness not defined
+#endif
+	/* Check port for UD address vector: */
+	*(inbox + INIT_HCA_FLAGS_OFFSET / 4) |= cpu_to_be32(1);
+
+	/* QPC/EEC/CQC/EQC/RDMARC attributes */
+
+	MLX4_PUT(inbox, param->qpc_base,      INIT_HCA_QPC_BASE_OFFSET);
+	MLX4_PUT(inbox, param->log_num_qps,   INIT_HCA_LOG_QP_OFFSET);
+	MLX4_PUT(inbox, param->srqc_base,     INIT_HCA_SRQC_BASE_OFFSET);
+	MLX4_PUT(inbox, param->log_num_srqs,  INIT_HCA_LOG_SRQ_OFFSET);
+	MLX4_PUT(inbox, param->cqc_base,      INIT_HCA_CQC_BASE_OFFSET);
+	MLX4_PUT(inbox, param->log_num_cqs,   INIT_HCA_LOG_CQ_OFFSET);
+	MLX4_PUT(inbox, param->altc_base,     INIT_HCA_ALTC_BASE_OFFSET);
+	MLX4_PUT(inbox, param->auxc_base,     INIT_HCA_AUXC_BASE_OFFSET);
+	MLX4_PUT(inbox, param->eqc_base,      INIT_HCA_EQC_BASE_OFFSET);
+	MLX4_PUT(inbox, param->log_num_eqs,   INIT_HCA_LOG_EQ_OFFSET);
+	MLX4_PUT(inbox, param->rdmarc_base,   INIT_HCA_RDMARC_BASE_OFFSET);
+	MLX4_PUT(inbox, param->log_rd_per_qp, INIT_HCA_LOG_RD_OFFSET);
+
+	/* multicast attributes */
+
+	MLX4_PUT(inbox, param->mc_base,		INIT_HCA_MC_BASE_OFFSET);
+	MLX4_PUT(inbox, param->log_mc_entry_sz, INIT_HCA_LOG_MC_ENTRY_SZ_OFFSET);
+	MLX4_PUT(inbox, param->log_mc_hash_sz,  INIT_HCA_LOG_MC_HASH_SZ_OFFSET);
+	MLX4_PUT(inbox, param->log_mc_table_sz, INIT_HCA_LOG_MC_TABLE_SZ_OFFSET);
+
+	/* TPT attributes */
+
+	MLX4_PUT(inbox, param->dmpt_base,  INIT_HCA_DMPT_BASE_OFFSET);
+	MLX4_PUT(inbox, param->log_mpt_sz, INIT_HCA_LOG_MPT_SZ_OFFSET);
+	MLX4_PUT(inbox, param->mtt_base,   INIT_HCA_MTT_BASE_OFFSET);
+	MLX4_PUT(inbox, param->cmpt_base,  INIT_HCA_CMPT_BASE_OFFSET);
+
+	/* UAR attributes */
+
+	MLX4_PUT(inbox, (u8) (PAGE_SHIFT - 12), INIT_HCA_UAR_PAGE_SZ_OFFSET);
+	MLX4_PUT(inbox, param->log_uar_sz,      INIT_HCA_LOG_UAR_SZ_OFFSET);
+
+	err = mlx4_cmd(dev, mailbox->dma, 0, 0, MLX4_CMD_INIT_HCA, 1000);
+
+	if (err)
+		mlx4_err(dev, "INIT_HCA returns %d\n", err);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+int mlx4_INIT_PORT(struct mlx4_dev *dev, struct mlx4_init_port_param *param, int port)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	u32 *inbox;
+	int err;
+	u32 flags;
+
+#define INIT_PORT_IN_SIZE          256
+#define INIT_PORT_FLAGS_OFFSET     0x00
+#define INIT_PORT_FLAG_SIG         (1 << 18)
+#define INIT_PORT_FLAG_NG          (1 << 17)
+#define INIT_PORT_FLAG_G0          (1 << 16)
+#define INIT_PORT_VL_SHIFT         4
+#define INIT_PORT_PORT_WIDTH_SHIFT 8
+#define INIT_PORT_MTU_OFFSET       0x04
+#define INIT_PORT_MAX_GID_OFFSET   0x06
+#define INIT_PORT_MAX_PKEY_OFFSET  0x0a
+#define INIT_PORT_GUID0_OFFSET     0x10
+#define INIT_PORT_NODE_GUID_OFFSET 0x18
+#define INIT_PORT_SI_GUID_OFFSET   0x20
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	inbox = mailbox->buf;
+
+	memset(inbox, 0, INIT_PORT_IN_SIZE);
+
+	flags = 0;
+	flags |= param->set_guid0     ? INIT_PORT_FLAG_G0  : 0;
+	flags |= param->set_node_guid ? INIT_PORT_FLAG_NG  : 0;
+	flags |= param->set_si_guid   ? INIT_PORT_FLAG_SIG : 0;
+	flags |= (param->vl_cap & 0xf) << INIT_PORT_VL_SHIFT;
+	flags |= (param->port_width_cap & 0xf) << INIT_PORT_PORT_WIDTH_SHIFT;
+	MLX4_PUT(inbox, flags,            INIT_PORT_FLAGS_OFFSET);
+
+	MLX4_PUT(inbox, param->mtu,       INIT_PORT_MTU_OFFSET);
+	MLX4_PUT(inbox, param->max_gid,   INIT_PORT_MAX_GID_OFFSET);
+	MLX4_PUT(inbox, param->max_pkey,  INIT_PORT_MAX_PKEY_OFFSET);
+	MLX4_PUT(inbox, param->guid0,     INIT_PORT_GUID0_OFFSET);
+	MLX4_PUT(inbox, param->node_guid, INIT_PORT_NODE_GUID_OFFSET);
+	MLX4_PUT(inbox, param->si_guid,   INIT_PORT_SI_GUID_OFFSET);
+
+	err = mlx4_cmd(dev, mailbox->dma, port, 0, MLX4_CMD_INIT_PORT,
+		       MLX4_CMD_TIME_CLASS_A);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_INIT_PORT);
+
+int mlx4_CLOSE_PORT(struct mlx4_dev *dev, int port)
+{
+	return mlx4_cmd(dev, 0, port, 0, MLX4_CMD_CLOSE_PORT, 1000);
+}
+EXPORT_SYMBOL_GPL(mlx4_CLOSE_PORT);
+
+int mlx4_CLOSE_HCA(struct mlx4_dev *dev, int panic)
+{
+	return mlx4_cmd(dev, 0, 0, panic, MLX4_CMD_CLOSE_HCA, 1000);
+}
+
+int mlx4_SET_ICM_SIZE(struct mlx4_dev *dev, u64 icm_size, u64 *aux_pages)
+{
+	int ret = mlx4_cmd_imm(dev, icm_size, aux_pages, 0, 0,
+			       MLX4_CMD_SET_ICM_SIZE,
+			       MLX4_CMD_TIME_CLASS_A);
+	if (ret)
+		return ret;
+
+	/*
+	 * Round up number of system pages needed in case
+	 * MLX4_ICM_PAGE_SIZE < PAGE_SIZE.
+	 */
+	*aux_pages = ALIGN(*aux_pages, PAGE_SIZE / MLX4_ICM_PAGE_SIZE) >>
+		(PAGE_SHIFT - MLX4_ICM_PAGE_SHIFT);
+
+	return 0;
+}
+
+int mlx4_NOP(struct mlx4_dev *dev)
+{
+	/* Input modifier of 0x1f means "finish as soon as possible." */
+	return mlx4_cmd(dev, 0, 0x1f, 0, MLX4_CMD_NOP, 100);
+}
diff --git a/drivers/net/mlx4/fw.h b/drivers/net/mlx4/fw.h
new file mode 100644
index 000000000000..2616fa53d4d0
--- /dev/null
+++ b/drivers/net/mlx4/fw.h
@@ -0,0 +1,167 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2006, 2007 Cisco Systems.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef MLX4_FW_H
+#define MLX4_FW_H
+
+#include "mlx4.h"
+#include "icm.h"
+
+struct mlx4_dev_cap {
+	int max_srq_sz;
+	int max_qp_sz;
+	int reserved_qps;
+	int max_qps;
+	int reserved_srqs;
+	int max_srqs;
+	int max_cq_sz;
+	int reserved_cqs;
+	int max_cqs;
+	int max_mpts;
+	int reserved_eqs;
+	int max_eqs;
+	int reserved_mtts;
+	int max_mrw_sz;
+	int reserved_mrws;
+	int max_mtt_seg;
+	int max_requester_per_qp;
+	int max_responder_per_qp;
+	int max_rdma_global;
+	int local_ca_ack_delay;
+	int max_mtu;
+	int max_port_width;
+	int max_vl;
+	int num_ports;
+	int max_gids;
+	u16 stat_rate_support;
+	int max_pkeys;
+	u32 flags;
+	int reserved_uars;
+	int uar_size;
+	int min_page_sz;
+	int bf_reg_size;
+	int bf_regs_per_page;
+	int max_sq_sg;
+	int max_sq_desc_sz;
+	int max_rq_sg;
+	int max_rq_desc_sz;
+	int max_qp_per_mcg;
+	int reserved_mgms;
+	int max_mcgs;
+	int reserved_pds;
+	int max_pds;
+	int qpc_entry_sz;
+	int rdmarc_entry_sz;
+	int altc_entry_sz;
+	int aux_entry_sz;
+	int srq_entry_sz;
+	int cqc_entry_sz;
+	int eqc_entry_sz;
+	int dmpt_entry_sz;
+	int cmpt_entry_sz;
+	int mtt_entry_sz;
+	int resize_srq;
+	u8  bmme_flags;
+	u32 reserved_lkey;
+	u64 max_icm_sz;
+};
+
+struct mlx4_adapter {
+	u32  vendor_id;
+	u32  device_id;
+	u32  revision_id;
+	char board_id[MLX4_BOARD_ID_LEN];
+	u8   inta_pin;
+};
+
+struct mlx4_init_hca_param {
+	u64 qpc_base;
+	u64 rdmarc_base;
+	u64 auxc_base;
+	u64 altc_base;
+	u64 srqc_base;
+	u64 cqc_base;
+	u64 eqc_base;
+	u64 mc_base;
+	u64 dmpt_base;
+	u64 cmpt_base;
+	u64 mtt_base;
+	u16 log_mc_entry_sz;
+	u16 log_mc_hash_sz;
+	u8  log_num_qps;
+	u8  log_num_srqs;
+	u8  log_num_cqs;
+	u8  log_num_eqs;
+	u8  log_rd_per_qp;
+	u8  log_mc_table_sz;
+	u8  log_mpt_sz;
+	u8  log_uar_sz;
+};
+
+struct mlx4_init_ib_param {
+	int port_width;
+	int vl_cap;
+	int mtu_cap;
+	u16 gid_cap;
+	u16 pkey_cap;
+	int set_guid0;
+	u64 guid0;
+	int set_node_guid;
+	u64 node_guid;
+	int set_si_guid;
+	u64 si_guid;
+};
+
+struct mlx4_set_ib_param {
+	int set_si_guid;
+	int reset_qkey_viol;
+	u64 si_guid;
+	u32 cap_mask;
+};
+
+int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap);
+int mlx4_MAP_FA(struct mlx4_dev *dev, struct mlx4_icm *icm);
+int mlx4_UNMAP_FA(struct mlx4_dev *dev);
+int mlx4_RUN_FW(struct mlx4_dev *dev);
+int mlx4_QUERY_FW(struct mlx4_dev *dev);
+int mlx4_QUERY_ADAPTER(struct mlx4_dev *dev, struct mlx4_adapter *adapter);
+int mlx4_INIT_HCA(struct mlx4_dev *dev, struct mlx4_init_hca_param *param);
+int mlx4_CLOSE_HCA(struct mlx4_dev *dev, int panic);
+int mlx4_map_cmd(struct mlx4_dev *dev, u16 op, struct mlx4_icm *icm, u64 virt);
+int mlx4_SET_ICM_SIZE(struct mlx4_dev *dev, u64 icm_size, u64 *aux_pages);
+int mlx4_MAP_ICM_AUX(struct mlx4_dev *dev, struct mlx4_icm *icm);
+int mlx4_UNMAP_ICM_AUX(struct mlx4_dev *dev);
+int mlx4_NOP(struct mlx4_dev *dev);
+
+#endif /* MLX4_FW_H */
diff --git a/drivers/net/mlx4/icm.c b/drivers/net/mlx4/icm.c
new file mode 100644
index 000000000000..e96feaed6ed4
--- /dev/null
+++ b/drivers/net/mlx4/icm.c
@@ -0,0 +1,379 @@
+/*
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/init.h>
+#include <linux/errno.h>
+
+#include <linux/mlx4/cmd.h>
+
+#include "mlx4.h"
+#include "icm.h"
+#include "fw.h"
+
+/*
+ * We allocate in as big chunks as we can, up to a maximum of 256 KB
+ * per chunk.
+ */
+enum {
+	MLX4_ICM_ALLOC_SIZE	= 1 << 18,
+	MLX4_TABLE_CHUNK_SIZE	= 1 << 18
+};
+
+void mlx4_free_icm(struct mlx4_dev *dev, struct mlx4_icm *icm)
+{
+	struct mlx4_icm_chunk *chunk, *tmp;
+	int i;
+
+	list_for_each_entry_safe(chunk, tmp, &icm->chunk_list, list) {
+		if (chunk->nsg > 0)
+			pci_unmap_sg(dev->pdev, chunk->mem, chunk->npages,
+				     PCI_DMA_BIDIRECTIONAL);
+
+		for (i = 0; i < chunk->npages; ++i)
+			__free_pages(chunk->mem[i].page,
+				     get_order(chunk->mem[i].length));
+
+		kfree(chunk);
+	}
+
+	kfree(icm);
+}
+
+struct mlx4_icm *mlx4_alloc_icm(struct mlx4_dev *dev, int npages,
+				gfp_t gfp_mask)
+{
+	struct mlx4_icm *icm;
+	struct mlx4_icm_chunk *chunk = NULL;
+	int cur_order;
+
+	icm = kmalloc(sizeof *icm, gfp_mask & ~(__GFP_HIGHMEM | __GFP_NOWARN));
+	if (!icm)
+		return icm;
+
+	icm->refcount = 0;
+	INIT_LIST_HEAD(&icm->chunk_list);
+
+	cur_order = get_order(MLX4_ICM_ALLOC_SIZE);
+
+	while (npages > 0) {
+		if (!chunk) {
+			chunk = kmalloc(sizeof *chunk,
+					gfp_mask & ~(__GFP_HIGHMEM | __GFP_NOWARN));
+			if (!chunk)
+				goto fail;
+
+			chunk->npages = 0;
+			chunk->nsg    = 0;
+			list_add_tail(&chunk->list, &icm->chunk_list);
+		}
+
+		while (1 << cur_order > npages)
+			--cur_order;
+
+		chunk->mem[chunk->npages].page = alloc_pages(gfp_mask, cur_order);
+		if (chunk->mem[chunk->npages].page) {
+			chunk->mem[chunk->npages].length = PAGE_SIZE << cur_order;
+			chunk->mem[chunk->npages].offset = 0;
+
+			if (++chunk->npages == MLX4_ICM_CHUNK_LEN) {
+				chunk->nsg = pci_map_sg(dev->pdev, chunk->mem,
+							chunk->npages,
+							PCI_DMA_BIDIRECTIONAL);
+
+				if (chunk->nsg <= 0)
+					goto fail;
+
+				chunk = NULL;
+			}
+
+			npages -= 1 << cur_order;
+		} else {
+			--cur_order;
+			if (cur_order < 0)
+				goto fail;
+		}
+	}
+
+	if (chunk) {
+		chunk->nsg = pci_map_sg(dev->pdev, chunk->mem,
+					chunk->npages,
+					PCI_DMA_BIDIRECTIONAL);
+
+		if (chunk->nsg <= 0)
+			goto fail;
+	}
+
+	return icm;
+
+fail:
+	mlx4_free_icm(dev, icm);
+	return NULL;
+}
+
+static int mlx4_MAP_ICM(struct mlx4_dev *dev, struct mlx4_icm *icm, u64 virt)
+{
+	return mlx4_map_cmd(dev, MLX4_CMD_MAP_ICM, icm, virt);
+}
+
+int mlx4_UNMAP_ICM(struct mlx4_dev *dev, u64 virt, u32 page_count)
+{
+	return mlx4_cmd(dev, virt, page_count, 0, MLX4_CMD_UNMAP_ICM,
+			MLX4_CMD_TIME_CLASS_B);
+}
+
+int mlx4_MAP_ICM_page(struct mlx4_dev *dev, u64 dma_addr, u64 virt)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	__be64 *inbox;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	inbox = mailbox->buf;
+
+	inbox[0] = cpu_to_be64(virt);
+	inbox[1] = cpu_to_be64(dma_addr);
+
+	err = mlx4_cmd(dev, mailbox->dma, 1, 0, MLX4_CMD_MAP_ICM,
+		       MLX4_CMD_TIME_CLASS_B);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	if (!err)
+		mlx4_dbg(dev, "Mapped page at %llx to %llx for ICM.\n",
+			  (unsigned long long) dma_addr, (unsigned long long) virt);
+
+	return err;
+}
+
+int mlx4_MAP_ICM_AUX(struct mlx4_dev *dev, struct mlx4_icm *icm)
+{
+	return mlx4_map_cmd(dev, MLX4_CMD_MAP_ICM_AUX, icm, -1);
+}
+
+int mlx4_UNMAP_ICM_AUX(struct mlx4_dev *dev)
+{
+	return mlx4_cmd(dev, 0, 0, 0, MLX4_CMD_UNMAP_ICM_AUX, MLX4_CMD_TIME_CLASS_B);
+}
+
+int mlx4_table_get(struct mlx4_dev *dev, struct mlx4_icm_table *table, int obj)
+{
+	int i = (obj & (table->num_obj - 1)) / (MLX4_TABLE_CHUNK_SIZE / table->obj_size);
+	int ret = 0;
+
+	mutex_lock(&table->mutex);
+
+	if (table->icm[i]) {
+		++table->icm[i]->refcount;
+		goto out;
+	}
+
+	table->icm[i] = mlx4_alloc_icm(dev, MLX4_TABLE_CHUNK_SIZE >> PAGE_SHIFT,
+				       (table->lowmem ? GFP_KERNEL : GFP_HIGHUSER) |
+				       __GFP_NOWARN);
+	if (!table->icm[i]) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	if (mlx4_MAP_ICM(dev, table->icm[i], table->virt +
+			 (u64) i * MLX4_TABLE_CHUNK_SIZE)) {
+		mlx4_free_icm(dev, table->icm[i]);
+		table->icm[i] = NULL;
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	++table->icm[i]->refcount;
+
+out:
+	mutex_unlock(&table->mutex);
+	return ret;
+}
+
+void mlx4_table_put(struct mlx4_dev *dev, struct mlx4_icm_table *table, int obj)
+{
+	int i;
+
+	i = (obj & (table->num_obj - 1)) / (MLX4_TABLE_CHUNK_SIZE / table->obj_size);
+
+	mutex_lock(&table->mutex);
+
+	if (--table->icm[i]->refcount == 0) {
+		mlx4_UNMAP_ICM(dev, table->virt + i * MLX4_TABLE_CHUNK_SIZE,
+			       MLX4_TABLE_CHUNK_SIZE / MLX4_ICM_PAGE_SIZE);
+		mlx4_free_icm(dev, table->icm[i]);
+		table->icm[i] = NULL;
+	}
+
+	mutex_unlock(&table->mutex);
+}
+
+void *mlx4_table_find(struct mlx4_icm_table *table, int obj)
+{
+	int idx, offset, i;
+	struct mlx4_icm_chunk *chunk;
+	struct mlx4_icm *icm;
+	struct page *page = NULL;
+
+	if (!table->lowmem)
+		return NULL;
+
+	mutex_lock(&table->mutex);
+
+	idx = obj & (table->num_obj - 1);
+	icm = table->icm[idx / (MLX4_TABLE_CHUNK_SIZE / table->obj_size)];
+	offset = idx % (MLX4_TABLE_CHUNK_SIZE / table->obj_size);
+
+	if (!icm)
+		goto out;
+
+	list_for_each_entry(chunk, &icm->chunk_list, list) {
+		for (i = 0; i < chunk->npages; ++i) {
+			if (chunk->mem[i].length > offset) {
+				page = chunk->mem[i].page;
+				goto out;
+			}
+			offset -= chunk->mem[i].length;
+		}
+	}
+
+out:
+	mutex_unlock(&table->mutex);
+	return page ? lowmem_page_address(page) + offset : NULL;
+}
+
+int mlx4_table_get_range(struct mlx4_dev *dev, struct mlx4_icm_table *table,
+			 int start, int end)
+{
+	int inc = MLX4_TABLE_CHUNK_SIZE / table->obj_size;
+	int i, err;
+
+	for (i = start; i <= end; i += inc) {
+		err = mlx4_table_get(dev, table, i);
+		if (err)
+			goto fail;
+	}
+
+	return 0;
+
+fail:
+	while (i > start) {
+		i -= inc;
+		mlx4_table_put(dev, table, i);
+	}
+
+	return err;
+}
+
+void mlx4_table_put_range(struct mlx4_dev *dev, struct mlx4_icm_table *table,
+			  int start, int end)
+{
+	int i;
+
+	for (i = start; i <= end; i += MLX4_TABLE_CHUNK_SIZE / table->obj_size)
+		mlx4_table_put(dev, table, i);
+}
+
+int mlx4_init_icm_table(struct mlx4_dev *dev, struct mlx4_icm_table *table,
+			u64 virt, int obj_size,	int nobj, int reserved,
+			int use_lowmem)
+{
+	int obj_per_chunk;
+	int num_icm;
+	unsigned chunk_size;
+	int i;
+
+	obj_per_chunk = MLX4_TABLE_CHUNK_SIZE / obj_size;
+	num_icm = (nobj + obj_per_chunk - 1) / obj_per_chunk;
+
+	table->icm      = kcalloc(num_icm, sizeof *table->icm, GFP_KERNEL);
+	if (!table->icm)
+		return -ENOMEM;
+	table->virt     = virt;
+	table->num_icm  = num_icm;
+	table->num_obj  = nobj;
+	table->obj_size = obj_size;
+	table->lowmem   = use_lowmem;
+	mutex_init(&table->mutex);
+
+	for (i = 0; i * MLX4_TABLE_CHUNK_SIZE < reserved * obj_size; ++i) {
+		chunk_size = MLX4_TABLE_CHUNK_SIZE;
+		if ((i + 1) * MLX4_TABLE_CHUNK_SIZE > nobj * obj_size)
+			chunk_size = PAGE_ALIGN(nobj * obj_size - i * MLX4_TABLE_CHUNK_SIZE);
+
+		table->icm[i] = mlx4_alloc_icm(dev, chunk_size >> PAGE_SHIFT,
+					       (use_lowmem ? GFP_KERNEL : GFP_HIGHUSER) |
+					       __GFP_NOWARN);
+		if (!table->icm[i])
+			goto err;
+		if (mlx4_MAP_ICM(dev, table->icm[i], virt + i * MLX4_TABLE_CHUNK_SIZE)) {
+			mlx4_free_icm(dev, table->icm[i]);
+			table->icm[i] = NULL;
+			goto err;
+		}
+
+		/*
+		 * Add a reference to this ICM chunk so that it never
+		 * gets freed (since it contains reserved firmware objects).
+		 */
+		++table->icm[i]->refcount;
+	}
+
+	return 0;
+
+err:
+	for (i = 0; i < num_icm; ++i)
+		if (table->icm[i]) {
+			mlx4_UNMAP_ICM(dev, virt + i * MLX4_TABLE_CHUNK_SIZE,
+				       MLX4_TABLE_CHUNK_SIZE / MLX4_ICM_PAGE_SIZE);
+			mlx4_free_icm(dev, table->icm[i]);
+		}
+
+	return -ENOMEM;
+}
+
+void mlx4_cleanup_icm_table(struct mlx4_dev *dev, struct mlx4_icm_table *table)
+{
+	int i;
+
+	for (i = 0; i < table->num_icm; ++i)
+		if (table->icm[i]) {
+			mlx4_UNMAP_ICM(dev, table->virt + i * MLX4_TABLE_CHUNK_SIZE,
+				       MLX4_TABLE_CHUNK_SIZE / MLX4_ICM_PAGE_SIZE);
+			mlx4_free_icm(dev, table->icm[i]);
+		}
+
+	kfree(table->icm);
+}
diff --git a/drivers/net/mlx4/icm.h b/drivers/net/mlx4/icm.h
new file mode 100644
index 000000000000..bea223d879a5
--- /dev/null
+++ b/drivers/net/mlx4/icm.h
@@ -0,0 +1,135 @@
+/*
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef MLX4_ICM_H
+#define MLX4_ICM_H
+
+#include <linux/list.h>
+#include <linux/pci.h>
+#include <linux/mutex.h>
+
+#define MLX4_ICM_CHUNK_LEN						\
+	((256 - sizeof (struct list_head) - 2 * sizeof (int)) /		\
+	 (sizeof (struct scatterlist)))
+
+enum {
+	MLX4_ICM_PAGE_SHIFT	= 12,
+	MLX4_ICM_PAGE_SIZE	= 1 << MLX4_ICM_PAGE_SHIFT,
+};
+
+struct mlx4_icm_chunk {
+	struct list_head	list;
+	int			npages;
+	int			nsg;
+	struct scatterlist	mem[MLX4_ICM_CHUNK_LEN];
+};
+
+struct mlx4_icm {
+	struct list_head	chunk_list;
+	int			refcount;
+};
+
+struct mlx4_icm_iter {
+	struct mlx4_icm	       *icm;
+	struct mlx4_icm_chunk  *chunk;
+	int			page_idx;
+};
+
+struct mlx4_dev;
+
+struct mlx4_icm *mlx4_alloc_icm(struct mlx4_dev *dev, int npages, gfp_t gfp_mask);
+void mlx4_free_icm(struct mlx4_dev *dev, struct mlx4_icm *icm);
+
+int mlx4_table_get(struct mlx4_dev *dev, struct mlx4_icm_table *table, int obj);
+void mlx4_table_put(struct mlx4_dev *dev, struct mlx4_icm_table *table, int obj);
+int mlx4_table_get_range(struct mlx4_dev *dev, struct mlx4_icm_table *table,
+			 int start, int end);
+void mlx4_table_put_range(struct mlx4_dev *dev, struct mlx4_icm_table *table,
+			  int start, int end);
+int mlx4_init_icm_table(struct mlx4_dev *dev, struct mlx4_icm_table *table,
+			u64 virt, int obj_size,	int nobj, int reserved,
+			int use_lowmem);
+void mlx4_cleanup_icm_table(struct mlx4_dev *dev, struct mlx4_icm_table *table);
+int mlx4_table_get(struct mlx4_dev *dev, struct mlx4_icm_table *table, int obj);
+void mlx4_table_put(struct mlx4_dev *dev, struct mlx4_icm_table *table, int obj);
+void *mlx4_table_find(struct mlx4_icm_table *table, int obj);
+int mlx4_table_get_range(struct mlx4_dev *dev, struct mlx4_icm_table *table,
+			 int start, int end);
+void mlx4_table_put_range(struct mlx4_dev *dev, struct mlx4_icm_table *table,
+			  int start, int end);
+
+static inline void mlx4_icm_first(struct mlx4_icm *icm,
+				  struct mlx4_icm_iter *iter)
+{
+	iter->icm      = icm;
+	iter->chunk    = list_empty(&icm->chunk_list) ?
+		NULL : list_entry(icm->chunk_list.next,
+				  struct mlx4_icm_chunk, list);
+	iter->page_idx = 0;
+}
+
+static inline int mlx4_icm_last(struct mlx4_icm_iter *iter)
+{
+	return !iter->chunk;
+}
+
+static inline void mlx4_icm_next(struct mlx4_icm_iter *iter)
+{
+	if (++iter->page_idx >= iter->chunk->nsg) {
+		if (iter->chunk->list.next == &iter->icm->chunk_list) {
+			iter->chunk = NULL;
+			return;
+		}
+
+		iter->chunk = list_entry(iter->chunk->list.next,
+					 struct mlx4_icm_chunk, list);
+		iter->page_idx = 0;
+	}
+}
+
+static inline dma_addr_t mlx4_icm_addr(struct mlx4_icm_iter *iter)
+{
+	return sg_dma_address(&iter->chunk->mem[iter->page_idx]);
+}
+
+static inline unsigned long mlx4_icm_size(struct mlx4_icm_iter *iter)
+{
+	return sg_dma_len(&iter->chunk->mem[iter->page_idx]);
+}
+
+int mlx4_UNMAP_ICM(struct mlx4_dev *dev, u64 virt, u32 page_count);
+int mlx4_MAP_ICM_page(struct mlx4_dev *dev, u64 dma_addr, u64 virt);
+int mlx4_MAP_ICM_AUX(struct mlx4_dev *dev, struct mlx4_icm *icm);
+int mlx4_UNMAP_ICM_AUX(struct mlx4_dev *dev);
+
+#endif /* MLX4_ICM_H */
diff --git a/drivers/net/mlx4/intf.c b/drivers/net/mlx4/intf.c
new file mode 100644
index 000000000000..65854f9e9c76
--- /dev/null
+++ b/drivers/net/mlx4/intf.c
@@ -0,0 +1,165 @@
+/*
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx4/driver.h>
+
+#include "mlx4.h"
+
+struct mlx4_device_context {
+	struct list_head	list;
+	struct mlx4_interface  *intf;
+	void		       *context;
+};
+
+static LIST_HEAD(intf_list);
+static LIST_HEAD(dev_list);
+static DEFINE_MUTEX(intf_mutex);
+
+static void mlx4_add_device(struct mlx4_interface *intf, struct mlx4_priv *priv)
+{
+	struct mlx4_device_context *dev_ctx;
+
+	dev_ctx = kmalloc(sizeof *dev_ctx, GFP_KERNEL);
+	if (!dev_ctx)
+		return;
+
+	dev_ctx->intf    = intf;
+	dev_ctx->context = intf->add(&priv->dev);
+
+	if (dev_ctx->context) {
+		spin_lock_irq(&priv->ctx_lock);
+		list_add_tail(&dev_ctx->list, &priv->ctx_list);
+		spin_unlock_irq(&priv->ctx_lock);
+	} else
+		kfree(dev_ctx);
+}
+
+static void mlx4_remove_device(struct mlx4_interface *intf, struct mlx4_priv *priv)
+{
+	struct mlx4_device_context *dev_ctx;
+
+	list_for_each_entry(dev_ctx, &priv->ctx_list, list)
+		if (dev_ctx->intf == intf) {
+			spin_lock_irq(&priv->ctx_lock);
+			list_del(&dev_ctx->list);
+			spin_unlock_irq(&priv->ctx_lock);
+
+			intf->remove(&priv->dev, dev_ctx->context);
+			kfree(dev_ctx);
+			return;
+		}
+}
+
+int mlx4_register_interface(struct mlx4_interface *intf)
+{
+	struct mlx4_priv *priv;
+
+	if (!intf->add || !intf->remove)
+		return -EINVAL;
+
+	mutex_lock(&intf_mutex);
+
+	list_add_tail(&intf->list, &intf_list);
+	list_for_each_entry(priv, &dev_list, dev_list)
+		mlx4_add_device(intf, priv);
+
+	mutex_unlock(&intf_mutex);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_register_interface);
+
+void mlx4_unregister_interface(struct mlx4_interface *intf)
+{
+	struct mlx4_priv *priv;
+
+	mutex_lock(&intf_mutex);
+
+	list_for_each_entry(priv, &dev_list, dev_list)
+		mlx4_remove_device(intf, priv);
+
+	list_del(&intf->list);
+
+	mutex_unlock(&intf_mutex);
+}
+EXPORT_SYMBOL_GPL(mlx4_unregister_interface);
+
+void mlx4_dispatch_event(struct mlx4_dev *dev, enum mlx4_event type,
+			 int subtype, int port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_device_context *dev_ctx;
+	unsigned long flags;
+
+	spin_lock_irqsave(&priv->ctx_lock, flags);
+
+	list_for_each_entry(dev_ctx, &priv->ctx_list, list)
+		if (dev_ctx->intf->event)
+			dev_ctx->intf->event(dev, dev_ctx->context, type,
+					     subtype, port);
+
+	spin_unlock_irqrestore(&priv->ctx_lock, flags);
+}
+
+int mlx4_register_device(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_interface *intf;
+
+	INIT_LIST_HEAD(&priv->ctx_list);
+	spin_lock_init(&priv->ctx_lock);
+
+	mutex_lock(&intf_mutex);
+
+	list_add_tail(&priv->dev_list, &dev_list);
+	list_for_each_entry(intf, &intf_list, list)
+		mlx4_add_device(intf, priv);
+
+	mutex_unlock(&intf_mutex);
+
+	return 0;
+}
+
+void mlx4_unregister_device(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_interface *intf;
+
+	mutex_lock(&intf_mutex);
+
+	list_for_each_entry(intf, &intf_list, list)
+		mlx4_remove_device(intf, priv);
+
+	list_del(&priv->dev_list);
+
+	mutex_unlock(&intf_mutex);
+}
diff --git a/drivers/net/mlx4/main.c b/drivers/net/mlx4/main.c
new file mode 100644
index 000000000000..4debb024eaf9
--- /dev/null
+++ b/drivers/net/mlx4/main.c
@@ -0,0 +1,936 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/errno.h>
+#include <linux/pci.h>
+#include <linux/dma-mapping.h>
+
+#include <linux/mlx4/device.h>
+#include <linux/mlx4/doorbell.h>
+
+#include "mlx4.h"
+#include "fw.h"
+#include "icm.h"
+
+MODULE_AUTHOR("Roland Dreier");
+MODULE_DESCRIPTION("Mellanox ConnectX HCA low-level driver");
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_VERSION(DRV_VERSION);
+
+#ifdef CONFIG_MLX4_DEBUG
+
+int mlx4_debug_level = 0;
+module_param_named(debug_level, mlx4_debug_level, int, 0644);
+MODULE_PARM_DESC(debug_level, "Enable debug tracing if > 0");
+
+#endif /* CONFIG_MLX4_DEBUG */
+
+#ifdef CONFIG_PCI_MSI
+
+static int msi_x;
+module_param(msi_x, int, 0444);
+MODULE_PARM_DESC(msi_x, "attempt to use MSI-X if nonzero");
+
+#else /* CONFIG_PCI_MSI */
+
+#define msi_x (0)
+
+#endif /* CONFIG_PCI_MSI */
+
+static const char mlx4_version[] __devinitdata =
+	DRV_NAME ": Mellanox ConnectX core driver v"
+	DRV_VERSION " (" DRV_RELDATE ")\n";
+
+static struct mlx4_profile default_profile = {
+	.num_qp		= 1 << 16,
+	.num_srq	= 1 << 16,
+	.rdmarc_per_qp	= 4,
+	.num_cq		= 1 << 16,
+	.num_mcg	= 1 << 13,
+	.num_mpt	= 1 << 17,
+	.num_mtt	= 1 << 20,
+};
+
+static int __devinit mlx4_dev_cap(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
+{
+	int err;
+
+	err = mlx4_QUERY_DEV_CAP(dev, dev_cap);
+	if (err) {
+		mlx4_err(dev, "QUERY_DEV_CAP command failed, aborting.\n");
+		return err;
+	}
+
+	if (dev_cap->min_page_sz > PAGE_SIZE) {
+		mlx4_err(dev, "HCA minimum page size of %d bigger than "
+			 "kernel PAGE_SIZE of %ld, aborting.\n",
+			 dev_cap->min_page_sz, PAGE_SIZE);
+		return -ENODEV;
+	}
+	if (dev_cap->num_ports > MLX4_MAX_PORTS) {
+		mlx4_err(dev, "HCA has %d ports, but we only support %d, "
+			 "aborting.\n",
+			 dev_cap->num_ports, MLX4_MAX_PORTS);
+		return -ENODEV;
+	}
+
+	if (dev_cap->uar_size > pci_resource_len(dev->pdev, 2)) {
+		mlx4_err(dev, "HCA reported UAR size of 0x%x bigger than "
+			 "PCI resource 2 size of 0x%llx, aborting.\n",
+			 dev_cap->uar_size,
+			 (unsigned long long) pci_resource_len(dev->pdev, 2));
+		return -ENODEV;
+	}
+
+	dev->caps.num_ports	     = dev_cap->num_ports;
+	dev->caps.num_uars	     = dev_cap->uar_size / PAGE_SIZE;
+	dev->caps.vl_cap	     = dev_cap->max_vl;
+	dev->caps.mtu_cap	     = dev_cap->max_mtu;
+	dev->caps.gid_table_len	     = dev_cap->max_gids;
+	dev->caps.pkey_table_len     = dev_cap->max_pkeys;
+	dev->caps.local_ca_ack_delay = dev_cap->local_ca_ack_delay;
+	dev->caps.bf_reg_size	     = dev_cap->bf_reg_size;
+	dev->caps.bf_regs_per_page   = dev_cap->bf_regs_per_page;
+	dev->caps.max_sq_sg	     = dev_cap->max_sq_sg;
+	dev->caps.max_rq_sg	     = dev_cap->max_rq_sg;
+	dev->caps.max_wqes	     = dev_cap->max_qp_sz;
+	dev->caps.max_qp_init_rdma   = dev_cap->max_requester_per_qp;
+	dev->caps.reserved_qps	     = dev_cap->reserved_qps;
+	dev->caps.max_srq_wqes	     = dev_cap->max_srq_sz;
+	dev->caps.max_srq_sge	     = dev_cap->max_rq_sg - 1;
+	dev->caps.reserved_srqs	     = dev_cap->reserved_srqs;
+	dev->caps.max_sq_desc_sz     = dev_cap->max_sq_desc_sz;
+	dev->caps.max_rq_desc_sz     = dev_cap->max_rq_desc_sz;
+	dev->caps.num_qp_per_mgm     = MLX4_QP_PER_MGM;
+	/*
+	 * Subtract 1 from the limit because we need to allocate a
+	 * spare CQE so the HCA HW can tell the difference between an
+	 * empty CQ and a full CQ.
+	 */
+	dev->caps.max_cqes	     = dev_cap->max_cq_sz - 1;
+	dev->caps.reserved_cqs	     = dev_cap->reserved_cqs;
+	dev->caps.reserved_eqs	     = dev_cap->reserved_eqs;
+	dev->caps.reserved_mtts	     = dev_cap->reserved_mtts;
+	dev->caps.reserved_mrws	     = dev_cap->reserved_mrws;
+	dev->caps.reserved_uars	     = dev_cap->reserved_uars;
+	dev->caps.reserved_pds	     = dev_cap->reserved_pds;
+	dev->caps.port_width_cap     = dev_cap->max_port_width;
+	dev->caps.mtt_entry_sz	     = MLX4_MTT_ENTRY_PER_SEG * dev_cap->mtt_entry_sz;
+	dev->caps.page_size_cap	     = ~(u32) (dev_cap->min_page_sz - 1);
+	dev->caps.flags		     = dev_cap->flags;
+	dev->caps.stat_rate_support  = dev_cap->stat_rate_support;
+
+	return 0;
+}
+
+static int __devinit mlx4_load_fw(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int err;
+
+	priv->fw.fw_icm = mlx4_alloc_icm(dev, priv->fw.fw_pages,
+					 GFP_HIGHUSER | __GFP_NOWARN);
+	if (!priv->fw.fw_icm) {
+		mlx4_err(dev, "Couldn't allocate FW area, aborting.\n");
+		return -ENOMEM;
+	}
+
+	err = mlx4_MAP_FA(dev, priv->fw.fw_icm);
+	if (err) {
+		mlx4_err(dev, "MAP_FA command failed, aborting.\n");
+		goto err_free;
+	}
+
+	err = mlx4_RUN_FW(dev);
+	if (err) {
+		mlx4_err(dev, "RUN_FW command failed, aborting.\n");
+		goto err_unmap_fa;
+	}
+
+	return 0;
+
+err_unmap_fa:
+	mlx4_UNMAP_FA(dev);
+
+err_free:
+	mlx4_free_icm(dev, priv->fw.fw_icm);
+	return err;
+}
+
+static int __devinit mlx4_init_cmpt_table(struct mlx4_dev *dev, u64 cmpt_base,
+					  int cmpt_entry_sz)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int err;
+
+	err = mlx4_init_icm_table(dev, &priv->qp_table.cmpt_table,
+				  cmpt_base +
+				  ((u64) (MLX4_CMPT_TYPE_QP *
+					  cmpt_entry_sz) << MLX4_CMPT_SHIFT),
+				  cmpt_entry_sz, dev->caps.num_qps,
+				  dev->caps.reserved_qps, 0);
+	if (err)
+		goto err;
+
+	err = mlx4_init_icm_table(dev, &priv->srq_table.cmpt_table,
+				  cmpt_base +
+				  ((u64) (MLX4_CMPT_TYPE_SRQ *
+					  cmpt_entry_sz) << MLX4_CMPT_SHIFT),
+				  cmpt_entry_sz, dev->caps.num_srqs,
+				  dev->caps.reserved_srqs, 0);
+	if (err)
+		goto err_qp;
+
+	err = mlx4_init_icm_table(dev, &priv->cq_table.cmpt_table,
+				  cmpt_base +
+				  ((u64) (MLX4_CMPT_TYPE_CQ *
+					  cmpt_entry_sz) << MLX4_CMPT_SHIFT),
+				  cmpt_entry_sz, dev->caps.num_cqs,
+				  dev->caps.reserved_cqs, 0);
+	if (err)
+		goto err_srq;
+
+	err = mlx4_init_icm_table(dev, &priv->eq_table.cmpt_table,
+				  cmpt_base +
+				  ((u64) (MLX4_CMPT_TYPE_EQ *
+					  cmpt_entry_sz) << MLX4_CMPT_SHIFT),
+				  cmpt_entry_sz,
+				  roundup_pow_of_two(MLX4_NUM_EQ +
+						     dev->caps.reserved_eqs),
+				  MLX4_NUM_EQ + dev->caps.reserved_eqs, 0);
+	if (err)
+		goto err_cq;
+
+	return 0;
+
+err_cq:
+	mlx4_cleanup_icm_table(dev, &priv->cq_table.cmpt_table);
+
+err_srq:
+	mlx4_cleanup_icm_table(dev, &priv->srq_table.cmpt_table);
+
+err_qp:
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.cmpt_table);
+
+err:
+	return err;
+}
+
+static int __devinit mlx4_init_icm(struct mlx4_dev *dev,
+				   struct mlx4_dev_cap *dev_cap,
+				   struct mlx4_init_hca_param *init_hca,
+				   u64 icm_size)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	u64 aux_pages;
+	int err;
+
+	err = mlx4_SET_ICM_SIZE(dev, icm_size, &aux_pages);
+	if (err) {
+		mlx4_err(dev, "SET_ICM_SIZE command failed, aborting.\n");
+		return err;
+	}
+
+	mlx4_dbg(dev, "%lld KB of HCA context requires %lld KB aux memory.\n",
+		 (unsigned long long) icm_size >> 10,
+		 (unsigned long long) aux_pages << 2);
+
+	priv->fw.aux_icm = mlx4_alloc_icm(dev, aux_pages,
+					  GFP_HIGHUSER | __GFP_NOWARN);
+	if (!priv->fw.aux_icm) {
+		mlx4_err(dev, "Couldn't allocate aux memory, aborting.\n");
+		return -ENOMEM;
+	}
+
+	err = mlx4_MAP_ICM_AUX(dev, priv->fw.aux_icm);
+	if (err) {
+		mlx4_err(dev, "MAP_ICM_AUX command failed, aborting.\n");
+		goto err_free_aux;
+	}
+
+	err = mlx4_init_cmpt_table(dev, init_hca->cmpt_base, dev_cap->cmpt_entry_sz);
+	if (err) {
+		mlx4_err(dev, "Failed to map cMPT context memory, aborting.\n");
+		goto err_unmap_aux;
+	}
+
+	err = mlx4_map_eq_icm(dev, init_hca->eqc_base);
+	if (err) {
+		mlx4_err(dev, "Failed to map EQ context memory, aborting.\n");
+		goto err_unmap_cmpt;
+	}
+
+	err = mlx4_init_icm_table(dev, &priv->mr_table.mtt_table,
+				  init_hca->mtt_base,
+				  dev->caps.mtt_entry_sz,
+				  dev->caps.num_mtt_segs,
+				  dev->caps.reserved_mtts, 1);
+	if (err) {
+		mlx4_err(dev, "Failed to map MTT context memory, aborting.\n");
+		goto err_unmap_eq;
+	}
+
+	err = mlx4_init_icm_table(dev, &priv->mr_table.dmpt_table,
+				  init_hca->dmpt_base,
+				  dev_cap->dmpt_entry_sz,
+				  dev->caps.num_mpts,
+				  dev->caps.reserved_mrws, 1);
+	if (err) {
+		mlx4_err(dev, "Failed to map dMPT context memory, aborting.\n");
+		goto err_unmap_mtt;
+	}
+
+	err = mlx4_init_icm_table(dev, &priv->qp_table.qp_table,
+				  init_hca->qpc_base,
+				  dev_cap->qpc_entry_sz,
+				  dev->caps.num_qps,
+				  dev->caps.reserved_qps, 0);
+	if (err) {
+		mlx4_err(dev, "Failed to map QP context memory, aborting.\n");
+		goto err_unmap_dmpt;
+	}
+
+	err = mlx4_init_icm_table(dev, &priv->qp_table.auxc_table,
+				  init_hca->auxc_base,
+				  dev_cap->aux_entry_sz,
+				  dev->caps.num_qps,
+				  dev->caps.reserved_qps, 0);
+	if (err) {
+		mlx4_err(dev, "Failed to map AUXC context memory, aborting.\n");
+		goto err_unmap_qp;
+	}
+
+	err = mlx4_init_icm_table(dev, &priv->qp_table.altc_table,
+				  init_hca->altc_base,
+				  dev_cap->altc_entry_sz,
+				  dev->caps.num_qps,
+				  dev->caps.reserved_qps, 0);
+	if (err) {
+		mlx4_err(dev, "Failed to map ALTC context memory, aborting.\n");
+		goto err_unmap_auxc;
+	}
+
+	err = mlx4_init_icm_table(dev, &priv->qp_table.rdmarc_table,
+				  init_hca->rdmarc_base,
+				  dev_cap->rdmarc_entry_sz << priv->qp_table.rdmarc_shift,
+				  dev->caps.num_qps,
+				  dev->caps.reserved_qps, 0);
+	if (err) {
+		mlx4_err(dev, "Failed to map RDMARC context memory, aborting\n");
+		goto err_unmap_altc;
+	}
+
+	err = mlx4_init_icm_table(dev, &priv->cq_table.table,
+				  init_hca->cqc_base,
+				  dev_cap->cqc_entry_sz,
+				  dev->caps.num_cqs,
+				  dev->caps.reserved_cqs, 0);
+	if (err) {
+		mlx4_err(dev, "Failed to map CQ context memory, aborting.\n");
+		goto err_unmap_rdmarc;
+	}
+
+	err = mlx4_init_icm_table(dev, &priv->srq_table.table,
+				  init_hca->srqc_base,
+				  dev_cap->srq_entry_sz,
+				  dev->caps.num_srqs,
+				  dev->caps.reserved_srqs, 0);
+	if (err) {
+		mlx4_err(dev, "Failed to map SRQ context memory, aborting.\n");
+		goto err_unmap_cq;
+	}
+
+	/*
+	 * It's not strictly required, but for simplicity just map the
+	 * whole multicast group table now.  The table isn't very big
+	 * and it's a lot easier than trying to track ref counts.
+	 */
+	err = mlx4_init_icm_table(dev, &priv->mcg_table.table,
+				  init_hca->mc_base, MLX4_MGM_ENTRY_SIZE,
+				  dev->caps.num_mgms + dev->caps.num_amgms,
+				  dev->caps.num_mgms + dev->caps.num_amgms,
+				  0);
+	if (err) {
+		mlx4_err(dev, "Failed to map MCG context memory, aborting.\n");
+		goto err_unmap_srq;
+	}
+
+	return 0;
+
+err_unmap_srq:
+	mlx4_cleanup_icm_table(dev, &priv->srq_table.table);
+
+err_unmap_cq:
+	mlx4_cleanup_icm_table(dev, &priv->cq_table.table);
+
+err_unmap_rdmarc:
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.rdmarc_table);
+
+err_unmap_altc:
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.altc_table);
+
+err_unmap_auxc:
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.auxc_table);
+
+err_unmap_qp:
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.qp_table);
+
+err_unmap_dmpt:
+	mlx4_cleanup_icm_table(dev, &priv->mr_table.dmpt_table);
+
+err_unmap_mtt:
+	mlx4_cleanup_icm_table(dev, &priv->mr_table.mtt_table);
+
+err_unmap_eq:
+	mlx4_unmap_eq_icm(dev);
+
+err_unmap_cmpt:
+	mlx4_cleanup_icm_table(dev, &priv->eq_table.cmpt_table);
+	mlx4_cleanup_icm_table(dev, &priv->cq_table.cmpt_table);
+	mlx4_cleanup_icm_table(dev, &priv->srq_table.cmpt_table);
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.cmpt_table);
+
+err_unmap_aux:
+	mlx4_UNMAP_ICM_AUX(dev);
+
+err_free_aux:
+	mlx4_free_icm(dev, priv->fw.aux_icm);
+
+	return err;
+}
+
+static void mlx4_free_icms(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	mlx4_cleanup_icm_table(dev, &priv->mcg_table.table);
+	mlx4_cleanup_icm_table(dev, &priv->srq_table.table);
+	mlx4_cleanup_icm_table(dev, &priv->cq_table.table);
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.rdmarc_table);
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.altc_table);
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.auxc_table);
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.qp_table);
+	mlx4_cleanup_icm_table(dev, &priv->mr_table.dmpt_table);
+	mlx4_cleanup_icm_table(dev, &priv->mr_table.mtt_table);
+	mlx4_cleanup_icm_table(dev, &priv->eq_table.cmpt_table);
+	mlx4_cleanup_icm_table(dev, &priv->cq_table.cmpt_table);
+	mlx4_cleanup_icm_table(dev, &priv->srq_table.cmpt_table);
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.cmpt_table);
+	mlx4_unmap_eq_icm(dev);
+
+	mlx4_UNMAP_ICM_AUX(dev);
+	mlx4_free_icm(dev, priv->fw.aux_icm);
+}
+
+static void mlx4_close_hca(struct mlx4_dev *dev)
+{
+	mlx4_CLOSE_HCA(dev, 0);
+	mlx4_free_icms(dev);
+	mlx4_UNMAP_FA(dev);
+	mlx4_free_icm(dev, mlx4_priv(dev)->fw.fw_icm);
+}
+
+static int __devinit mlx4_init_hca(struct mlx4_dev *dev)
+{
+	struct mlx4_priv	  *priv = mlx4_priv(dev);
+	struct mlx4_adapter	   adapter;
+	struct mlx4_dev_cap	   dev_cap;
+	struct mlx4_profile	   profile;
+	struct mlx4_init_hca_param init_hca;
+	u64 icm_size;
+	int err;
+
+	err = mlx4_QUERY_FW(dev);
+	if (err) {
+		mlx4_err(dev, "QUERY_FW command failed, aborting.\n");
+		return err;
+	}
+
+	err = mlx4_load_fw(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to start FW, aborting.\n");
+		return err;
+	}
+
+	err = mlx4_dev_cap(dev, &dev_cap);
+	if (err) {
+		mlx4_err(dev, "QUERY_DEV_CAP command failed, aborting.\n");
+		goto err_stop_fw;
+	}
+
+	profile = default_profile;
+
+	icm_size = mlx4_make_profile(dev, &profile, &dev_cap, &init_hca);
+	if ((long long) icm_size < 0) {
+		err = icm_size;
+		goto err_stop_fw;
+	}
+
+	init_hca.log_uar_sz = ilog2(dev->caps.num_uars);
+
+	err = mlx4_init_icm(dev, &dev_cap, &init_hca, icm_size);
+	if (err)
+		goto err_stop_fw;
+
+	err = mlx4_INIT_HCA(dev, &init_hca);
+	if (err) {
+		mlx4_err(dev, "INIT_HCA command failed, aborting.\n");
+		goto err_free_icm;
+	}
+
+	err = mlx4_QUERY_ADAPTER(dev, &adapter);
+	if (err) {
+		mlx4_err(dev, "QUERY_ADAPTER command failed, aborting.\n");
+		goto err_close;
+	}
+
+	priv->eq_table.inta_pin = adapter.inta_pin;
+	priv->rev_id		= adapter.revision_id;
+	memcpy(priv->board_id, adapter.board_id, sizeof priv->board_id);
+
+	return 0;
+
+err_close:
+	mlx4_close_hca(dev);
+
+err_free_icm:
+	mlx4_free_icms(dev);
+
+err_stop_fw:
+	mlx4_UNMAP_FA(dev);
+	mlx4_free_icm(dev, priv->fw.fw_icm);
+
+	return err;
+}
+
+static int __devinit mlx4_setup_hca(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int err;
+
+	MLX4_INIT_DOORBELL_LOCK(&priv->doorbell_lock);
+
+	err = mlx4_init_uar_table(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to initialize "
+			 "user access region table, aborting.\n");
+		return err;
+	}
+
+	err = mlx4_uar_alloc(dev, &priv->driver_uar);
+	if (err) {
+		mlx4_err(dev, "Failed to allocate driver access region, "
+			 "aborting.\n");
+		goto err_uar_table_free;
+	}
+
+	priv->kar = ioremap(priv->driver_uar.pfn << PAGE_SHIFT, PAGE_SIZE);
+	if (!priv->kar) {
+		mlx4_err(dev, "Couldn't map kernel access region, "
+			 "aborting.\n");
+		err = -ENOMEM;
+		goto err_uar_free;
+	}
+
+	err = mlx4_init_pd_table(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to initialize "
+			 "protection domain table, aborting.\n");
+		goto err_kar_unmap;
+	}
+
+	err = mlx4_init_mr_table(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to initialize "
+			 "memory region table, aborting.\n");
+		goto err_pd_table_free;
+	}
+
+	mlx4_map_catas_buf(dev);
+
+	err = mlx4_init_eq_table(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to initialize "
+			 "event queue table, aborting.\n");
+		goto err_catas_buf;
+	}
+
+	err = mlx4_cmd_use_events(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to switch to event-driven "
+			 "firmware commands, aborting.\n");
+		goto err_eq_table_free;
+	}
+
+	err = mlx4_NOP(dev);
+	if (err) {
+		mlx4_err(dev, "NOP command failed to generate interrupt "
+			 "(IRQ %d), aborting.\n",
+			 priv->eq_table.eq[MLX4_EQ_ASYNC].irq);
+		if (dev->flags & MLX4_FLAG_MSI_X)
+			mlx4_err(dev, "Try again with MSI-X disabled.\n");
+		else
+			mlx4_err(dev, "BIOS or ACPI interrupt routing problem?\n");
+
+		goto err_cmd_poll;
+	}
+
+	mlx4_dbg(dev, "NOP command IRQ test passed\n");
+
+	err = mlx4_init_cq_table(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to initialize "
+			 "completion queue table, aborting.\n");
+		goto err_cmd_poll;
+	}
+
+	err = mlx4_init_srq_table(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to initialize "
+			 "shared receive queue table, aborting.\n");
+		goto err_cq_table_free;
+	}
+
+	err = mlx4_init_qp_table(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to initialize "
+			 "queue pair table, aborting.\n");
+		goto err_srq_table_free;
+	}
+
+	err = mlx4_init_mcg_table(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to initialize "
+			 "multicast group table, aborting.\n");
+		goto err_qp_table_free;
+	}
+
+	return 0;
+
+err_qp_table_free:
+	mlx4_cleanup_qp_table(dev);
+
+err_srq_table_free:
+	mlx4_cleanup_srq_table(dev);
+
+err_cq_table_free:
+	mlx4_cleanup_cq_table(dev);
+
+err_cmd_poll:
+	mlx4_cmd_use_polling(dev);
+
+err_eq_table_free:
+	mlx4_cleanup_eq_table(dev);
+
+err_catas_buf:
+	mlx4_unmap_catas_buf(dev);
+	mlx4_cleanup_mr_table(dev);
+
+err_pd_table_free:
+	mlx4_cleanup_pd_table(dev);
+
+err_kar_unmap:
+	iounmap(priv->kar);
+
+err_uar_free:
+	mlx4_uar_free(dev, &priv->driver_uar);
+
+err_uar_table_free:
+	mlx4_cleanup_uar_table(dev);
+	return err;
+}
+
+static void __devinit mlx4_enable_msi_x(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct msix_entry entries[MLX4_NUM_EQ];
+	int err;
+	int i;
+
+	if (msi_x) {
+		for (i = 0; i < MLX4_NUM_EQ; ++i)
+			entries[i].entry = i;
+
+		err = pci_enable_msix(dev->pdev, entries, ARRAY_SIZE(entries));
+		if (err) {
+			if (err > 0)
+				mlx4_info(dev, "Only %d MSI-X vectors available, "
+					  "not using MSI-X\n", err);
+			goto no_msi;
+		}
+
+		for (i = 0; i < MLX4_NUM_EQ; ++i)
+			priv->eq_table.eq[i].irq = entries[i].vector;
+
+		dev->flags |= MLX4_FLAG_MSI_X;
+		return;
+	}
+
+no_msi:
+	for (i = 0; i < MLX4_NUM_EQ; ++i)
+		priv->eq_table.eq[i].irq = dev->pdev->irq;
+}
+
+static int __devinit mlx4_init_one(struct pci_dev *pdev,
+				   const struct pci_device_id *id)
+{
+	static int mlx4_version_printed;
+	struct mlx4_priv *priv;
+	struct mlx4_dev *dev;
+	int err;
+
+	if (!mlx4_version_printed) {
+		printk(KERN_INFO "%s", mlx4_version);
+		++mlx4_version_printed;
+	}
+
+	printk(KERN_INFO PFX "Initializing %s\n",
+	       pci_name(pdev));
+
+	err = pci_enable_device(pdev);
+	if (err) {
+		dev_err(&pdev->dev, "Cannot enable PCI device, "
+			"aborting.\n");
+		return err;
+	}
+
+	/*
+	 * Check for BARs.  We expect 0: 1MB, 2: 8MB, 4: DDR (may not
+	 * be present)
+	 */
+	if (!(pci_resource_flags(pdev, 0) & IORESOURCE_MEM) ||
+	    pci_resource_len(pdev, 0) != 1 << 20) {
+		dev_err(&pdev->dev, "Missing DCS, aborting.\n");
+		err = -ENODEV;
+		goto err_disable_pdev;
+	}
+	if (!(pci_resource_flags(pdev, 2) & IORESOURCE_MEM)) {
+		dev_err(&pdev->dev, "Missing UAR, aborting.\n");
+		err = -ENODEV;
+		goto err_disable_pdev;
+	}
+
+	err = pci_request_region(pdev, 0, DRV_NAME);
+	if (err) {
+		dev_err(&pdev->dev, "Cannot request control region, aborting.\n");
+		goto err_disable_pdev;
+	}
+
+	err = pci_request_region(pdev, 2, DRV_NAME);
+	if (err) {
+		dev_err(&pdev->dev, "Cannot request UAR region, aborting.\n");
+		goto err_release_bar0;
+	}
+
+	pci_set_master(pdev);
+
+	err = pci_set_dma_mask(pdev, DMA_64BIT_MASK);
+	if (err) {
+		dev_warn(&pdev->dev, "Warning: couldn't set 64-bit PCI DMA mask.\n");
+		err = pci_set_dma_mask(pdev, DMA_32BIT_MASK);
+		if (err) {
+			dev_err(&pdev->dev, "Can't set PCI DMA mask, aborting.\n");
+			goto err_release_bar2;
+		}
+	}
+	err = pci_set_consistent_dma_mask(pdev, DMA_64BIT_MASK);
+	if (err) {
+		dev_warn(&pdev->dev, "Warning: couldn't set 64-bit "
+			 "consistent PCI DMA mask.\n");
+		err = pci_set_consistent_dma_mask(pdev, DMA_32BIT_MASK);
+		if (err) {
+			dev_err(&pdev->dev, "Can't set consistent PCI DMA mask, "
+				"aborting.\n");
+			goto err_release_bar2;
+		}
+	}
+
+	priv = kzalloc(sizeof *priv, GFP_KERNEL);
+	if (!priv) {
+		dev_err(&pdev->dev, "Device struct alloc failed, "
+			"aborting.\n");
+		err = -ENOMEM;
+		goto err_release_bar2;
+	}
+
+	dev       = &priv->dev;
+	dev->pdev = pdev;
+
+	/*
+	 * Now reset the HCA before we touch the PCI capabilities or
+	 * attempt a firmware command, since a boot ROM may have left
+	 * the HCA in an undefined state.
+	 */
+	err = mlx4_reset(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to reset HCA, aborting.\n");
+		goto err_free_dev;
+	}
+
+	mlx4_enable_msi_x(dev);
+
+	if (mlx4_cmd_init(dev)) {
+		mlx4_err(dev, "Failed to init command interface, aborting.\n");
+		goto err_free_dev;
+	}
+
+	err = mlx4_init_hca(dev);
+	if (err)
+		goto err_cmd;
+
+	err = mlx4_setup_hca(dev);
+	if (err)
+		goto err_close;
+
+	err = mlx4_register_device(dev);
+	if (err)
+		goto err_cleanup;
+
+	pci_set_drvdata(pdev, dev);
+
+	return 0;
+
+err_cleanup:
+	mlx4_cleanup_mcg_table(dev);
+	mlx4_cleanup_qp_table(dev);
+	mlx4_cleanup_srq_table(dev);
+	mlx4_cleanup_cq_table(dev);
+	mlx4_cmd_use_polling(dev);
+	mlx4_cleanup_eq_table(dev);
+
+	mlx4_unmap_catas_buf(dev);
+
+	mlx4_cleanup_mr_table(dev);
+	mlx4_cleanup_pd_table(dev);
+	mlx4_cleanup_uar_table(dev);
+
+err_close:
+	mlx4_close_hca(dev);
+
+err_cmd:
+	mlx4_cmd_cleanup(dev);
+
+err_free_dev:
+	if (dev->flags & MLX4_FLAG_MSI_X)
+		pci_disable_msix(pdev);
+
+	kfree(priv);
+
+err_release_bar2:
+	pci_release_region(pdev, 2);
+
+err_release_bar0:
+	pci_release_region(pdev, 0);
+
+err_disable_pdev:
+	pci_disable_device(pdev);
+	pci_set_drvdata(pdev, NULL);
+	return err;
+}
+
+static void __devexit mlx4_remove_one(struct pci_dev *pdev)
+{
+	struct mlx4_dev  *dev  = pci_get_drvdata(pdev);
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int p;
+
+	if (dev) {
+		mlx4_unregister_device(dev);
+
+		for (p = 1; p <= dev->caps.num_ports; ++p)
+			mlx4_CLOSE_PORT(dev, p);
+
+		mlx4_cleanup_mcg_table(dev);
+		mlx4_cleanup_qp_table(dev);
+		mlx4_cleanup_srq_table(dev);
+		mlx4_cleanup_cq_table(dev);
+		mlx4_cmd_use_polling(dev);
+		mlx4_cleanup_eq_table(dev);
+
+		mlx4_unmap_catas_buf(dev);
+
+		mlx4_cleanup_mr_table(dev);
+		mlx4_cleanup_pd_table(dev);
+
+		iounmap(priv->kar);
+		mlx4_uar_free(dev, &priv->driver_uar);
+		mlx4_cleanup_uar_table(dev);
+		mlx4_close_hca(dev);
+		mlx4_cmd_cleanup(dev);
+
+		if (dev->flags & MLX4_FLAG_MSI_X)
+			pci_disable_msix(pdev);
+
+		kfree(priv);
+		pci_release_region(pdev, 2);
+		pci_release_region(pdev, 0);
+		pci_disable_device(pdev);
+		pci_set_drvdata(pdev, NULL);
+	}
+}
+
+static struct pci_device_id mlx4_pci_table[] = {
+	{ PCI_VDEVICE(MELLANOX, 0x6340) }, /* MT25408 "Hermon" SDR */
+	{ PCI_VDEVICE(MELLANOX, 0x634a) }, /* MT25408 "Hermon" DDR */
+	{ PCI_VDEVICE(MELLANOX, 0x6354) }, /* MT25408 "Hermon" QDR */
+	{ 0, }
+};
+
+MODULE_DEVICE_TABLE(pci, mlx4_pci_table);
+
+static struct pci_driver mlx4_driver = {
+	.name		= DRV_NAME,
+	.id_table	= mlx4_pci_table,
+	.probe		= mlx4_init_one,
+	.remove		= __devexit_p(mlx4_remove_one)
+};
+
+static int __init mlx4_init(void)
+{
+	int ret;
+
+	ret = pci_register_driver(&mlx4_driver);
+	return ret < 0 ? ret : 0;
+}
+
+static void __exit mlx4_cleanup(void)
+{
+	pci_unregister_driver(&mlx4_driver);
+}
+
+module_init(mlx4_init);
+module_exit(mlx4_cleanup);
diff --git a/drivers/net/mlx4/mcg.c b/drivers/net/mlx4/mcg.c
new file mode 100644
index 000000000000..672024a0ee71
--- /dev/null
+++ b/drivers/net/mlx4/mcg.c
@@ -0,0 +1,380 @@
+/*
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/init.h>
+#include <linux/string.h>
+#include <linux/slab.h>
+
+#include <linux/mlx4/cmd.h>
+
+#include "mlx4.h"
+
+struct mlx4_mgm {
+	__be32			next_gid_index;
+	__be32			members_count;
+	u32			reserved[2];
+	u8			gid[16];
+	__be32			qp[MLX4_QP_PER_MGM];
+};
+
+static const u8 zero_gid[16];	/* automatically initialized to 0 */
+
+static int mlx4_READ_MCG(struct mlx4_dev *dev, int index,
+			 struct mlx4_cmd_mailbox *mailbox)
+{
+	return mlx4_cmd_box(dev, 0, mailbox->dma, index, 0, MLX4_CMD_READ_MCG,
+			    MLX4_CMD_TIME_CLASS_A);
+}
+
+static int mlx4_WRITE_MCG(struct mlx4_dev *dev, int index,
+			  struct mlx4_cmd_mailbox *mailbox)
+{
+	return mlx4_cmd(dev, mailbox->dma, index, 0, MLX4_CMD_WRITE_MCG,
+			MLX4_CMD_TIME_CLASS_A);
+}
+
+static int mlx4_MGID_HASH(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			  u16 *hash)
+{
+	u64 imm;
+	int err;
+
+	err = mlx4_cmd_imm(dev, mailbox->dma, &imm, 0, 0, MLX4_CMD_MGID_HASH,
+			   MLX4_CMD_TIME_CLASS_A);
+
+	if (!err)
+		*hash = imm;
+
+	return err;
+}
+
+/*
+ * Caller must hold MCG table semaphore.  gid and mgm parameters must
+ * be properly aligned for command interface.
+ *
+ *  Returns 0 unless a firmware command error occurs.
+ *
+ * If GID is found in MGM or MGM is empty, *index = *hash, *prev = -1
+ * and *mgm holds MGM entry.
+ *
+ * if GID is found in AMGM, *index = index in AMGM, *prev = index of
+ * previous entry in hash chain and *mgm holds AMGM entry.
+ *
+ * If no AMGM exists for given gid, *index = -1, *prev = index of last
+ * entry in hash chain and *mgm holds end of hash chain.
+ */
+static int find_mgm(struct mlx4_dev *dev,
+		    u8 *gid, struct mlx4_cmd_mailbox *mgm_mailbox,
+		    u16 *hash, int *prev, int *index)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_mgm *mgm = mgm_mailbox->buf;
+	u8 *mgid;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return -ENOMEM;
+	mgid = mailbox->buf;
+
+	memcpy(mgid, gid, 16);
+
+	err = mlx4_MGID_HASH(dev, mailbox, hash);
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	if (err)
+		return err;
+
+	if (0)
+		mlx4_dbg(dev, "Hash for %04x:%04x:%04x:%04x:"
+			  "%04x:%04x:%04x:%04x is %04x\n",
+			  be16_to_cpu(((__be16 *) gid)[0]),
+			  be16_to_cpu(((__be16 *) gid)[1]),
+			  be16_to_cpu(((__be16 *) gid)[2]),
+			  be16_to_cpu(((__be16 *) gid)[3]),
+			  be16_to_cpu(((__be16 *) gid)[4]),
+			  be16_to_cpu(((__be16 *) gid)[5]),
+			  be16_to_cpu(((__be16 *) gid)[6]),
+			  be16_to_cpu(((__be16 *) gid)[7]),
+			  *hash);
+
+	*index = *hash;
+	*prev  = -1;
+
+	do {
+		err = mlx4_READ_MCG(dev, *index, mgm_mailbox);
+		if (err)
+			return err;
+
+		if (!memcmp(mgm->gid, zero_gid, 16)) {
+			if (*index != *hash) {
+				mlx4_err(dev, "Found zero MGID in AMGM.\n");
+				err = -EINVAL;
+			}
+			return err;
+		}
+
+		if (!memcmp(mgm->gid, gid, 16))
+			return err;
+
+		*prev = *index;
+		*index = be32_to_cpu(mgm->next_gid_index) >> 6;
+	} while (*index);
+
+	*index = -1;
+	return err;
+}
+
+int mlx4_multicast_attach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16])
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_mgm *mgm;
+	u32 members_count;
+	u16 hash;
+	int index, prev;
+	int link = 0;
+	int i;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	mgm = mailbox->buf;
+
+	mutex_lock(&priv->mcg_table.mutex);
+
+	err = find_mgm(dev, gid, mailbox, &hash, &prev, &index);
+	if (err)
+		goto out;
+
+	if (index != -1) {
+		if (!memcmp(mgm->gid, zero_gid, 16))
+			memcpy(mgm->gid, gid, 16);
+	} else {
+		link = 1;
+
+		index = mlx4_bitmap_alloc(&priv->mcg_table.bitmap);
+		if (index == -1) {
+			mlx4_err(dev, "No AMGM entries left\n");
+			err = -ENOMEM;
+			goto out;
+		}
+		index += dev->caps.num_mgms;
+
+		err = mlx4_READ_MCG(dev, index, mailbox);
+		if (err)
+			goto out;
+
+		memset(mgm, 0, sizeof *mgm);
+		memcpy(mgm->gid, gid, 16);
+	}
+
+	members_count = be32_to_cpu(mgm->members_count);
+	if (members_count == MLX4_QP_PER_MGM) {
+		mlx4_err(dev, "MGM at index %x is full.\n", index);
+		err = -ENOMEM;
+		goto out;
+	}
+
+	for (i = 0; i < members_count; ++i)
+		if (mgm->qp[i] == cpu_to_be32(qp->qpn)) {
+			mlx4_dbg(dev, "QP %06x already a member of MGM\n", qp->qpn);
+			err = 0;
+			goto out;
+		}
+
+	mgm->qp[members_count++] = cpu_to_be32(qp->qpn);
+	mgm->members_count       = cpu_to_be32(members_count);
+
+	err = mlx4_WRITE_MCG(dev, index, mailbox);
+	if (err)
+		goto out;
+
+	if (!link)
+		goto out;
+
+	err = mlx4_READ_MCG(dev, prev, mailbox);
+	if (err)
+		goto out;
+
+	mgm->next_gid_index = cpu_to_be32(index << 6);
+
+	err = mlx4_WRITE_MCG(dev, prev, mailbox);
+	if (err)
+		goto out;
+
+out:
+	if (err && link && index != -1) {
+		if (index < dev->caps.num_mgms)
+			mlx4_warn(dev, "Got AMGM index %d < %d",
+				  index, dev->caps.num_mgms);
+		else
+			mlx4_bitmap_free(&priv->mcg_table.bitmap,
+					 index - dev->caps.num_mgms);
+	}
+	mutex_unlock(&priv->mcg_table.mutex);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_multicast_attach);
+
+int mlx4_multicast_detach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16])
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_mgm *mgm;
+	u32 members_count;
+	u16 hash;
+	int prev, index;
+	int i, loc;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	mgm = mailbox->buf;
+
+	mutex_lock(&priv->mcg_table.mutex);
+
+	err = find_mgm(dev, gid, mailbox, &hash, &prev, &index);
+	if (err)
+		goto out;
+
+	if (index == -1) {
+		mlx4_err(dev, "MGID %04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x "
+			  "not found\n",
+			  be16_to_cpu(((__be16 *) gid)[0]),
+			  be16_to_cpu(((__be16 *) gid)[1]),
+			  be16_to_cpu(((__be16 *) gid)[2]),
+			  be16_to_cpu(((__be16 *) gid)[3]),
+			  be16_to_cpu(((__be16 *) gid)[4]),
+			  be16_to_cpu(((__be16 *) gid)[5]),
+			  be16_to_cpu(((__be16 *) gid)[6]),
+			  be16_to_cpu(((__be16 *) gid)[7]));
+		err = -EINVAL;
+		goto out;
+	}
+
+	members_count = be32_to_cpu(mgm->members_count);
+	for (loc = -1, i = 0; i < members_count; ++i)
+		if (mgm->qp[i] == cpu_to_be32(qp->qpn))
+			loc = i;
+
+	if (loc == -1) {
+		mlx4_err(dev, "QP %06x not found in MGM\n", qp->qpn);
+		err = -EINVAL;
+		goto out;
+	}
+
+
+	mgm->members_count = cpu_to_be32(--members_count);
+	mgm->qp[loc]       = mgm->qp[i - 1];
+	mgm->qp[i - 1]     = 0;
+
+	err = mlx4_WRITE_MCG(dev, index, mailbox);
+	if (err)
+		goto out;
+
+	if (i != 1)
+		goto out;
+
+	if (prev == -1) {
+		/* Remove entry from MGM */
+		int amgm_index = be32_to_cpu(mgm->next_gid_index) >> 6;
+		if (amgm_index) {
+			err = mlx4_READ_MCG(dev, amgm_index, mailbox);
+			if (err)
+				goto out;
+		} else
+			memset(mgm->gid, 0, 16);
+
+		err = mlx4_WRITE_MCG(dev, index, mailbox);
+		if (err)
+			goto out;
+
+		if (amgm_index) {
+			if (amgm_index < dev->caps.num_mgms)
+				mlx4_warn(dev, "MGM entry %d had AMGM index %d < %d",
+					  index, amgm_index, dev->caps.num_mgms);
+			else
+				mlx4_bitmap_free(&priv->mcg_table.bitmap,
+						 amgm_index - dev->caps.num_mgms);
+		}
+	} else {
+		/* Remove entry from AMGM */
+		int cur_next_index = be32_to_cpu(mgm->next_gid_index) >> 6;
+		err = mlx4_READ_MCG(dev, prev, mailbox);
+		if (err)
+			goto out;
+
+		mgm->next_gid_index = cpu_to_be32(cur_next_index << 6);
+
+		err = mlx4_WRITE_MCG(dev, prev, mailbox);
+		if (err)
+			goto out;
+
+		if (index < dev->caps.num_mgms)
+			mlx4_warn(dev, "entry %d had next AMGM index %d < %d",
+				  prev, index, dev->caps.num_mgms);
+		else
+			mlx4_bitmap_free(&priv->mcg_table.bitmap,
+					 index - dev->caps.num_mgms);
+	}
+
+out:
+	mutex_unlock(&priv->mcg_table.mutex);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_multicast_detach);
+
+int __devinit mlx4_init_mcg_table(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int err;
+
+	err = mlx4_bitmap_init(&priv->mcg_table.bitmap,
+			       dev->caps.num_amgms, dev->caps.num_amgms - 1, 0);
+	if (err)
+		return err;
+
+	mutex_init(&priv->mcg_table.mutex);
+
+	return 0;
+}
+
+void mlx4_cleanup_mcg_table(struct mlx4_dev *dev)
+{
+	mlx4_bitmap_cleanup(&mlx4_priv(dev)->mcg_table.bitmap);
+}
diff --git a/drivers/net/mlx4/mlx4.h b/drivers/net/mlx4/mlx4.h
new file mode 100644
index 000000000000..9befbae3d196
--- /dev/null
+++ b/drivers/net/mlx4/mlx4.h
@@ -0,0 +1,348 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
+ * Copyright (c) 2005, 2006, 2007 Cisco Systems.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2004 Voltaire, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef MLX4_H
+#define MLX4_H
+
+#include <linux/radix-tree.h>
+
+#include <linux/mlx4/device.h>
+#include <linux/mlx4/doorbell.h>
+
+#define DRV_NAME	"mlx4_core"
+#define PFX		DRV_NAME ": "
+#define DRV_VERSION	"0.01"
+#define DRV_RELDATE	"May 1, 2007"
+
+enum {
+	MLX4_HCR_BASE		= 0x80680,
+	MLX4_HCR_SIZE		= 0x0001c,
+	MLX4_CLR_INT_SIZE	= 0x00008
+};
+
+enum {
+	MLX4_BOARD_ID_LEN	= 64
+};
+
+enum {
+	MLX4_MGM_ENTRY_SIZE	=  0x40,
+	MLX4_QP_PER_MGM		= 4 * (MLX4_MGM_ENTRY_SIZE / 16 - 2),
+	MLX4_MTT_ENTRY_PER_SEG	= 8
+};
+
+enum {
+	MLX4_EQ_ASYNC,
+	MLX4_EQ_COMP,
+	MLX4_EQ_CATAS,
+	MLX4_NUM_EQ
+};
+
+enum {
+	MLX4_NUM_PDS		= 1 << 15
+};
+
+enum {
+	MLX4_CMPT_TYPE_QP	= 0,
+	MLX4_CMPT_TYPE_SRQ	= 1,
+	MLX4_CMPT_TYPE_CQ	= 2,
+	MLX4_CMPT_TYPE_EQ	= 3,
+	MLX4_CMPT_NUM_TYPE
+};
+
+enum {
+	MLX4_CMPT_SHIFT		= 24,
+	MLX4_NUM_CMPTS		= MLX4_CMPT_NUM_TYPE << MLX4_CMPT_SHIFT
+};
+
+#ifdef CONFIG_MLX4_DEBUG
+extern int mlx4_debug_level;
+
+#define mlx4_dbg(mdev, format, arg...)					\
+	do {								\
+		if (mlx4_debug_level)					\
+			dev_printk(KERN_DEBUG, &mdev->pdev->dev, format, ## arg); \
+	} while (0)
+
+#else /* CONFIG_MLX4_DEBUG */
+
+#define mlx4_dbg(mdev, format, arg...) do { (void) mdev; } while (0)
+
+#endif /* CONFIG_MLX4_DEBUG */
+
+#define mlx4_err(mdev, format, arg...) \
+	dev_err(&mdev->pdev->dev, format, ## arg)
+#define mlx4_info(mdev, format, arg...) \
+	dev_info(&mdev->pdev->dev, format, ## arg)
+#define mlx4_warn(mdev, format, arg...) \
+	dev_warn(&mdev->pdev->dev, format, ## arg)
+
+struct mlx4_bitmap {
+	u32			last;
+	u32			top;
+	u32			max;
+	u32			mask;
+	spinlock_t		lock;
+	unsigned long	       *table;
+};
+
+struct mlx4_buddy {
+	unsigned long	      **bits;
+	int			max_order;
+	spinlock_t		lock;
+};
+
+struct mlx4_icm;
+
+struct mlx4_icm_table {
+	u64			virt;
+	int			num_icm;
+	int			num_obj;
+	int			obj_size;
+	int			lowmem;
+	struct mutex		mutex;
+	struct mlx4_icm	      **icm;
+};
+
+struct mlx4_eq {
+	struct mlx4_dev	       *dev;
+	void __iomem	       *doorbell;
+	int			eqn;
+	u32			cons_index;
+	u16			irq;
+	u16			have_irq;
+	int			nent;
+	struct mlx4_buf_list   *page_list;
+	struct mlx4_mtt		mtt;
+};
+
+struct mlx4_profile {
+	int			num_qp;
+	int			rdmarc_per_qp;
+	int			num_srq;
+	int			num_cq;
+	int			num_mcg;
+	int			num_mpt;
+	int			num_mtt;
+};
+
+struct mlx4_fw {
+	u64			clr_int_base;
+	u64			catas_offset;
+	struct mlx4_icm	       *fw_icm;
+	struct mlx4_icm	       *aux_icm;
+	u32			catas_size;
+	u16			fw_pages;
+	u8			clr_int_bar;
+	u8			catas_bar;
+};
+
+struct mlx4_cmd {
+	struct pci_pool	       *pool;
+	void __iomem	       *hcr;
+	struct mutex		hcr_mutex;
+	struct semaphore	poll_sem;
+	struct semaphore	event_sem;
+	int			max_cmds;
+	spinlock_t		context_lock;
+	int			free_head;
+	struct mlx4_cmd_context *context;
+	u16			token_mask;
+	u8			use_events;
+	u8			toggle;
+};
+
+struct mlx4_uar_table {
+	struct mlx4_bitmap	bitmap;
+};
+
+struct mlx4_mr_table {
+	struct mlx4_bitmap	mpt_bitmap;
+	struct mlx4_buddy	mtt_buddy;
+	u64			mtt_base;
+	u64			mpt_base;
+	struct mlx4_icm_table	mtt_table;
+	struct mlx4_icm_table	dmpt_table;
+};
+
+struct mlx4_cq_table {
+	struct mlx4_bitmap	bitmap;
+	spinlock_t		lock;
+	struct radix_tree_root	tree;
+	struct mlx4_icm_table	table;
+	struct mlx4_icm_table	cmpt_table;
+};
+
+struct mlx4_eq_table {
+	struct mlx4_bitmap	bitmap;
+	void __iomem	       *clr_int;
+	void __iomem	       *uar_map[(MLX4_NUM_EQ + 6) / 4];
+	u32			clr_mask;
+	struct mlx4_eq		eq[MLX4_NUM_EQ];
+	u64			icm_virt;
+	struct page	       *icm_page;
+	dma_addr_t		icm_dma;
+	struct mlx4_icm_table	cmpt_table;
+	int			have_irq;
+	u8			inta_pin;
+};
+
+struct mlx4_srq_table {
+	struct mlx4_bitmap	bitmap;
+	spinlock_t		lock;
+	struct radix_tree_root	tree;
+	struct mlx4_icm_table	table;
+	struct mlx4_icm_table	cmpt_table;
+};
+
+struct mlx4_qp_table {
+	struct mlx4_bitmap	bitmap;
+	u32			rdmarc_base;
+	int			rdmarc_shift;
+	spinlock_t		lock;
+	struct mlx4_icm_table	qp_table;
+	struct mlx4_icm_table	auxc_table;
+	struct mlx4_icm_table	altc_table;
+	struct mlx4_icm_table	rdmarc_table;
+	struct mlx4_icm_table	cmpt_table;
+};
+
+struct mlx4_mcg_table {
+	struct mutex		mutex;
+	struct mlx4_bitmap	bitmap;
+	struct mlx4_icm_table	table;
+};
+
+struct mlx4_catas_err {
+	u32 __iomem	       *map;
+	int			size;
+};
+
+struct mlx4_priv {
+	struct mlx4_dev		dev;
+
+	struct list_head	dev_list;
+	struct list_head	ctx_list;
+	spinlock_t		ctx_lock;
+
+	struct mlx4_fw		fw;
+	struct mlx4_cmd		cmd;
+
+	struct mlx4_bitmap	pd_bitmap;
+	struct mlx4_uar_table	uar_table;
+	struct mlx4_mr_table	mr_table;
+	struct mlx4_cq_table	cq_table;
+	struct mlx4_eq_table	eq_table;
+	struct mlx4_srq_table	srq_table;
+	struct mlx4_qp_table	qp_table;
+	struct mlx4_mcg_table	mcg_table;
+
+	struct mlx4_catas_err	catas_err;
+
+	void __iomem	       *clr_base;
+
+	struct mlx4_uar		driver_uar;
+	void __iomem	       *kar;
+	MLX4_DECLARE_DOORBELL_LOCK(doorbell_lock)
+
+	u32			rev_id;
+	char			board_id[MLX4_BOARD_ID_LEN];
+};
+
+static inline struct mlx4_priv *mlx4_priv(struct mlx4_dev *dev)
+{
+	return container_of(dev, struct mlx4_priv, dev);
+}
+
+u32 mlx4_bitmap_alloc(struct mlx4_bitmap *bitmap);
+void mlx4_bitmap_free(struct mlx4_bitmap *bitmap, u32 obj);
+int mlx4_bitmap_init(struct mlx4_bitmap *bitmap, u32 num, u32 mask, u32 reserved);
+void mlx4_bitmap_cleanup(struct mlx4_bitmap *bitmap);
+
+int mlx4_reset(struct mlx4_dev *dev);
+
+int mlx4_init_pd_table(struct mlx4_dev *dev);
+int mlx4_init_uar_table(struct mlx4_dev *dev);
+int mlx4_init_mr_table(struct mlx4_dev *dev);
+int mlx4_init_eq_table(struct mlx4_dev *dev);
+int mlx4_init_cq_table(struct mlx4_dev *dev);
+int mlx4_init_qp_table(struct mlx4_dev *dev);
+int mlx4_init_srq_table(struct mlx4_dev *dev);
+int mlx4_init_mcg_table(struct mlx4_dev *dev);
+
+void mlx4_cleanup_pd_table(struct mlx4_dev *dev);
+void mlx4_cleanup_uar_table(struct mlx4_dev *dev);
+void mlx4_cleanup_mr_table(struct mlx4_dev *dev);
+void mlx4_cleanup_eq_table(struct mlx4_dev *dev);
+void mlx4_cleanup_cq_table(struct mlx4_dev *dev);
+void mlx4_cleanup_qp_table(struct mlx4_dev *dev);
+void mlx4_cleanup_srq_table(struct mlx4_dev *dev);
+void mlx4_cleanup_mcg_table(struct mlx4_dev *dev);
+
+void mlx4_map_catas_buf(struct mlx4_dev *dev);
+void mlx4_unmap_catas_buf(struct mlx4_dev *dev);
+
+int mlx4_register_device(struct mlx4_dev *dev);
+void mlx4_unregister_device(struct mlx4_dev *dev);
+void mlx4_dispatch_event(struct mlx4_dev *dev, enum mlx4_event type,
+			 int subtype, int port);
+
+struct mlx4_dev_cap;
+struct mlx4_init_hca_param;
+
+u64 mlx4_make_profile(struct mlx4_dev *dev,
+		      struct mlx4_profile *request,
+		      struct mlx4_dev_cap *dev_cap,
+		      struct mlx4_init_hca_param *init_hca);
+
+int mlx4_map_eq_icm(struct mlx4_dev *dev, u64 icm_virt);
+void mlx4_unmap_eq_icm(struct mlx4_dev *dev);
+
+int mlx4_cmd_init(struct mlx4_dev *dev);
+void mlx4_cmd_cleanup(struct mlx4_dev *dev);
+void mlx4_cmd_event(struct mlx4_dev *dev, u16 token, u8 status, u64 out_param);
+int mlx4_cmd_use_events(struct mlx4_dev *dev);
+void mlx4_cmd_use_polling(struct mlx4_dev *dev);
+
+void mlx4_cq_completion(struct mlx4_dev *dev, u32 cqn);
+void mlx4_cq_event(struct mlx4_dev *dev, u32 cqn, int event_type);
+
+void mlx4_qp_event(struct mlx4_dev *dev, u32 qpn, int event_type);
+
+void mlx4_srq_event(struct mlx4_dev *dev, u32 srqn, int event_type);
+
+void mlx4_handle_catas_err(struct mlx4_dev *dev);
+
+#endif /* MLX4_H */
diff --git a/drivers/net/mlx4/mr.c b/drivers/net/mlx4/mr.c
new file mode 100644
index 000000000000..b33864dab179
--- /dev/null
+++ b/drivers/net/mlx4/mr.c
@@ -0,0 +1,479 @@
+/*
+ * Copyright (c) 2004 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/init.h>
+#include <linux/errno.h>
+
+#include <linux/mlx4/cmd.h>
+
+#include "mlx4.h"
+#include "icm.h"
+
+/*
+ * Must be packed because mtt_seg is 64 bits but only aligned to 32 bits.
+ */
+struct mlx4_mpt_entry {
+	__be32 flags;
+	__be32 qpn;
+	__be32 key;
+	__be32 pd;
+	__be64 start;
+	__be64 length;
+	__be32 lkey;
+	__be32 win_cnt;
+	u8	reserved1[3];
+	u8	mtt_rep;
+	__be64 mtt_seg;
+	__be32 mtt_sz;
+	__be32 entity_size;
+	__be32 first_byte_offset;
+} __attribute__((packed));
+
+#define MLX4_MPT_FLAG_SW_OWNS	    (0xfUL << 28)
+#define MLX4_MPT_FLAG_MIO	    (1 << 17)
+#define MLX4_MPT_FLAG_BIND_ENABLE   (1 << 15)
+#define MLX4_MPT_FLAG_PHYSICAL	    (1 <<  9)
+#define MLX4_MPT_FLAG_REGION	    (1 <<  8)
+
+#define MLX4_MTT_FLAG_PRESENT		1
+
+static u32 mlx4_buddy_alloc(struct mlx4_buddy *buddy, int order)
+{
+	int o;
+	int m;
+	u32 seg;
+
+	spin_lock(&buddy->lock);
+
+	for (o = order; o <= buddy->max_order; ++o) {
+		m = 1 << (buddy->max_order - o);
+		seg = find_first_bit(buddy->bits[o], m);
+		if (seg < m)
+			goto found;
+	}
+
+	spin_unlock(&buddy->lock);
+	return -1;
+
+ found:
+	clear_bit(seg, buddy->bits[o]);
+
+	while (o > order) {
+		--o;
+		seg <<= 1;
+		set_bit(seg ^ 1, buddy->bits[o]);
+	}
+
+	spin_unlock(&buddy->lock);
+
+	seg <<= order;
+
+	return seg;
+}
+
+static void mlx4_buddy_free(struct mlx4_buddy *buddy, u32 seg, int order)
+{
+	seg >>= order;
+
+	spin_lock(&buddy->lock);
+
+	while (test_bit(seg ^ 1, buddy->bits[order])) {
+		clear_bit(seg ^ 1, buddy->bits[order]);
+		seg >>= 1;
+		++order;
+	}
+
+	set_bit(seg, buddy->bits[order]);
+
+	spin_unlock(&buddy->lock);
+}
+
+static int __devinit mlx4_buddy_init(struct mlx4_buddy *buddy, int max_order)
+{
+	int i, s;
+
+	buddy->max_order = max_order;
+	spin_lock_init(&buddy->lock);
+
+	buddy->bits = kzalloc((buddy->max_order + 1) * sizeof (long *),
+			      GFP_KERNEL);
+	if (!buddy->bits)
+		goto err_out;
+
+	for (i = 0; i <= buddy->max_order; ++i) {
+		s = BITS_TO_LONGS(1 << (buddy->max_order - i));
+		buddy->bits[i] = kmalloc(s * sizeof (long), GFP_KERNEL);
+		if (!buddy->bits[i])
+			goto err_out_free;
+		bitmap_zero(buddy->bits[i], 1 << (buddy->max_order - i));
+	}
+
+	set_bit(0, buddy->bits[buddy->max_order]);
+
+	return 0;
+
+err_out_free:
+	for (i = 0; i <= buddy->max_order; ++i)
+		kfree(buddy->bits[i]);
+
+	kfree(buddy->bits);
+
+err_out:
+	return -ENOMEM;
+}
+
+static void mlx4_buddy_cleanup(struct mlx4_buddy *buddy)
+{
+	int i;
+
+	for (i = 0; i <= buddy->max_order; ++i)
+		kfree(buddy->bits[i]);
+
+	kfree(buddy->bits);
+}
+
+static u32 mlx4_alloc_mtt_range(struct mlx4_dev *dev, int order)
+{
+	struct mlx4_mr_table *mr_table = &mlx4_priv(dev)->mr_table;
+	u32 seg;
+
+	seg = mlx4_buddy_alloc(&mr_table->mtt_buddy, order);
+	if (seg == -1)
+		return -1;
+
+	if (mlx4_table_get_range(dev, &mr_table->mtt_table, seg,
+				 seg + (1 << order) - 1)) {
+		mlx4_buddy_free(&mr_table->mtt_buddy, seg, order);
+		return -1;
+	}
+
+	return seg;
+}
+
+int mlx4_mtt_init(struct mlx4_dev *dev, int npages, int page_shift,
+		  struct mlx4_mtt *mtt)
+{
+	int i;
+
+	if (!npages) {
+		mtt->order      = -1;
+		mtt->page_shift = MLX4_ICM_PAGE_SHIFT;
+		return 0;
+	} else
+		mtt->page_shift = page_shift;
+
+	for (mtt->order = 0, i = MLX4_MTT_ENTRY_PER_SEG; i < npages; i <<= 1)
+		++mtt->order;
+
+	mtt->first_seg = mlx4_alloc_mtt_range(dev, mtt->order);
+	if (mtt->first_seg == -1)
+		return -ENOMEM;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_mtt_init);
+
+void mlx4_mtt_cleanup(struct mlx4_dev *dev, struct mlx4_mtt *mtt)
+{
+	struct mlx4_mr_table *mr_table = &mlx4_priv(dev)->mr_table;
+
+	if (mtt->order < 0)
+		return;
+
+	mlx4_buddy_free(&mr_table->mtt_buddy, mtt->first_seg, mtt->order);
+	mlx4_table_put_range(dev, &mr_table->mtt_table, mtt->first_seg,
+			     mtt->first_seg + (1 << mtt->order) - 1);
+}
+EXPORT_SYMBOL_GPL(mlx4_mtt_cleanup);
+
+u64 mlx4_mtt_addr(struct mlx4_dev *dev, struct mlx4_mtt *mtt)
+{
+	return (u64) mtt->first_seg * dev->caps.mtt_entry_sz;
+}
+EXPORT_SYMBOL_GPL(mlx4_mtt_addr);
+
+static u32 hw_index_to_key(u32 ind)
+{
+	return (ind >> 24) | (ind << 8);
+}
+
+static u32 key_to_hw_index(u32 key)
+{
+	return (key << 24) | (key >> 8);
+}
+
+static int mlx4_SW2HW_MPT(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			  int mpt_index)
+{
+	return mlx4_cmd(dev, mailbox->dma, mpt_index, 0, MLX4_CMD_SW2HW_MPT,
+			MLX4_CMD_TIME_CLASS_B);
+}
+
+static int mlx4_HW2SW_MPT(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			  int mpt_index)
+{
+	return mlx4_cmd_box(dev, 0, mailbox ? mailbox->dma : 0, mpt_index,
+			    !mailbox, MLX4_CMD_HW2SW_MPT, MLX4_CMD_TIME_CLASS_B);
+}
+
+int mlx4_mr_alloc(struct mlx4_dev *dev, u32 pd, u64 iova, u64 size, u32 access,
+		  int npages, int page_shift, struct mlx4_mr *mr)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	u32 index;
+	int err;
+
+	index = mlx4_bitmap_alloc(&priv->mr_table.mpt_bitmap);
+	if (index == -1) {
+		err = -ENOMEM;
+		goto err;
+	}
+
+	mr->iova       = iova;
+	mr->size       = size;
+	mr->pd	       = pd;
+	mr->access     = access;
+	mr->enabled    = 0;
+	mr->key	       = hw_index_to_key(index);
+
+	err = mlx4_mtt_init(dev, npages, page_shift, &mr->mtt);
+	if (err)
+		goto err_index;
+
+	return 0;
+
+err_index:
+	mlx4_bitmap_free(&priv->mr_table.mpt_bitmap, index);
+
+err:
+	kfree(mr);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_mr_alloc);
+
+void mlx4_mr_free(struct mlx4_dev *dev, struct mlx4_mr *mr)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int err;
+
+	if (mr->enabled) {
+		err = mlx4_HW2SW_MPT(dev, NULL,
+				     key_to_hw_index(mr->key) &
+				     (dev->caps.num_mpts - 1));
+		if (err)
+			mlx4_warn(dev, "HW2SW_MPT failed (%d)\n", err);
+	}
+
+	mlx4_mtt_cleanup(dev, &mr->mtt);
+	mlx4_bitmap_free(&priv->mr_table.mpt_bitmap, key_to_hw_index(mr->key));
+}
+EXPORT_SYMBOL_GPL(mlx4_mr_free);
+
+int mlx4_mr_enable(struct mlx4_dev *dev, struct mlx4_mr *mr)
+{
+	struct mlx4_mr_table *mr_table = &mlx4_priv(dev)->mr_table;
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_mpt_entry *mpt_entry;
+	int err;
+
+	err = mlx4_table_get(dev, &mr_table->dmpt_table, key_to_hw_index(mr->key));
+	if (err)
+		return err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox)) {
+		err = PTR_ERR(mailbox);
+		goto err_table;
+	}
+	mpt_entry = mailbox->buf;
+
+	memset(mpt_entry, 0, sizeof *mpt_entry);
+
+	mpt_entry->flags = cpu_to_be32(MLX4_MPT_FLAG_SW_OWNS	 |
+				       MLX4_MPT_FLAG_MIO	 |
+				       MLX4_MPT_FLAG_REGION	 |
+				       mr->access);
+	if (mr->mtt.order < 0)
+		mpt_entry->flags |= cpu_to_be32(MLX4_MPT_FLAG_PHYSICAL);
+
+	mpt_entry->key	       = cpu_to_be32(key_to_hw_index(mr->key));
+	mpt_entry->pd	       = cpu_to_be32(mr->pd);
+	mpt_entry->start       = cpu_to_be64(mr->iova);
+	mpt_entry->length      = cpu_to_be64(mr->size);
+	mpt_entry->entity_size = cpu_to_be32(mr->mtt.page_shift);
+	mpt_entry->mtt_seg     = cpu_to_be64(mlx4_mtt_addr(dev, &mr->mtt));
+
+	err = mlx4_SW2HW_MPT(dev, mailbox,
+			     key_to_hw_index(mr->key) & (dev->caps.num_mpts - 1));
+	if (err) {
+		mlx4_warn(dev, "SW2HW_MPT failed (%d)\n", err);
+		goto err_cmd;
+	}
+
+	mr->enabled = 1;
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	return 0;
+
+err_cmd:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+err_table:
+	mlx4_table_put(dev, &mr_table->dmpt_table, key_to_hw_index(mr->key));
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_mr_enable);
+
+static int mlx4_WRITE_MTT(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			  int num_mtt)
+{
+	return mlx4_cmd(dev, mailbox->dma, num_mtt, 0, MLX4_CMD_WRITE_MTT,
+			MLX4_CMD_TIME_CLASS_B);
+}
+
+int mlx4_write_mtt(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
+		   int start_index, int npages, u64 *page_list)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	__be64 *mtt_entry;
+	int i;
+	int err = 0;
+
+	if (mtt->order < 0)
+		return -EINVAL;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	mtt_entry = mailbox->buf;
+
+	while (npages > 0) {
+		mtt_entry[0] = cpu_to_be64(mlx4_mtt_addr(dev, mtt) + start_index * 8);
+		mtt_entry[1] = 0;
+
+		for (i = 0; i < npages && i < MLX4_MAILBOX_SIZE / 8 - 2; ++i)
+			mtt_entry[i + 2] = cpu_to_be64(page_list[i] |
+						       MLX4_MTT_FLAG_PRESENT);
+
+		/*
+		 * If we have an odd number of entries to write, add
+		 * one more dummy entry for firmware efficiency.
+		 */
+		if (i & 1)
+			mtt_entry[i + 2] = 0;
+
+		err = mlx4_WRITE_MTT(dev, mailbox, (i + 1) & ~1);
+		if (err)
+			goto out;
+
+		npages      -= i;
+		start_index += i;
+		page_list   += i;
+	}
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_write_mtt);
+
+int mlx4_buf_write_mtt(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
+		       struct mlx4_buf *buf)
+{
+	u64 *page_list;
+	int err;
+	int i;
+
+	page_list = kmalloc(buf->npages * sizeof *page_list, GFP_KERNEL);
+	if (!page_list)
+		return -ENOMEM;
+
+	for (i = 0; i < buf->npages; ++i)
+		if (buf->nbufs == 1)
+			page_list[i] = buf->u.direct.map + (i << buf->page_shift);
+		else
+			page_list[i] = buf->u.page_list[i].map;
+
+	err = mlx4_write_mtt(dev, mtt, 0, buf->npages, page_list);
+
+	kfree(page_list);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_buf_write_mtt);
+
+int __devinit mlx4_init_mr_table(struct mlx4_dev *dev)
+{
+	struct mlx4_mr_table *mr_table = &mlx4_priv(dev)->mr_table;
+	int err;
+
+	err = mlx4_bitmap_init(&mr_table->mpt_bitmap, dev->caps.num_mpts,
+			       ~0, dev->caps.reserved_mrws);
+	if (err)
+		return err;
+
+	err = mlx4_buddy_init(&mr_table->mtt_buddy,
+			      ilog2(dev->caps.num_mtt_segs));
+	if (err)
+		goto err_buddy;
+
+	if (dev->caps.reserved_mtts) {
+		if (mlx4_alloc_mtt_range(dev, ilog2(dev->caps.reserved_mtts)) == -1) {
+			mlx4_warn(dev, "MTT table of order %d is too small.\n",
+				  mr_table->mtt_buddy.max_order);
+			err = -ENOMEM;
+			goto err_reserve_mtts;
+		}
+	}
+
+	return 0;
+
+err_reserve_mtts:
+	mlx4_buddy_cleanup(&mr_table->mtt_buddy);
+
+err_buddy:
+	mlx4_bitmap_cleanup(&mr_table->mpt_bitmap);
+
+	return err;
+}
+
+void mlx4_cleanup_mr_table(struct mlx4_dev *dev)
+{
+	struct mlx4_mr_table *mr_table = &mlx4_priv(dev)->mr_table;
+
+	mlx4_buddy_cleanup(&mr_table->mtt_buddy);
+	mlx4_bitmap_cleanup(&mr_table->mpt_bitmap);
+}
diff --git a/drivers/net/mlx4/pd.c b/drivers/net/mlx4/pd.c
new file mode 100644
index 000000000000..23dea1ee7750
--- /dev/null
+++ b/drivers/net/mlx4/pd.c
@@ -0,0 +1,102 @@
+/*
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/init.h>
+#include <linux/errno.h>
+
+#include <asm/page.h>
+
+#include "mlx4.h"
+#include "icm.h"
+
+int mlx4_pd_alloc(struct mlx4_dev *dev, u32 *pdn)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	*pdn = mlx4_bitmap_alloc(&priv->pd_bitmap);
+	if (*pdn == -1)
+		return -ENOMEM;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_pd_alloc);
+
+void mlx4_pd_free(struct mlx4_dev *dev, u32 pdn)
+{
+	mlx4_bitmap_free(&mlx4_priv(dev)->pd_bitmap, pdn);
+}
+EXPORT_SYMBOL_GPL(mlx4_pd_free);
+
+int __devinit mlx4_init_pd_table(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	return mlx4_bitmap_init(&priv->pd_bitmap, dev->caps.num_pds,
+				(1 << 24) - 1, dev->caps.reserved_pds);
+}
+
+void mlx4_cleanup_pd_table(struct mlx4_dev *dev)
+{
+	mlx4_bitmap_cleanup(&mlx4_priv(dev)->pd_bitmap);
+}
+
+
+int mlx4_uar_alloc(struct mlx4_dev *dev, struct mlx4_uar *uar)
+{
+	uar->index = mlx4_bitmap_alloc(&mlx4_priv(dev)->uar_table.bitmap);
+	if (uar->index == -1)
+		return -ENOMEM;
+
+	uar->pfn = (pci_resource_start(dev->pdev, 2) >> PAGE_SHIFT) + uar->index;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_uar_alloc);
+
+void mlx4_uar_free(struct mlx4_dev *dev, struct mlx4_uar *uar)
+{
+	mlx4_bitmap_free(&mlx4_priv(dev)->uar_table.bitmap, uar->index);
+}
+EXPORT_SYMBOL_GPL(mlx4_uar_free);
+
+int mlx4_init_uar_table(struct mlx4_dev *dev)
+{
+	return mlx4_bitmap_init(&mlx4_priv(dev)->uar_table.bitmap,
+				dev->caps.num_uars, dev->caps.num_uars - 1,
+				max(128, dev->caps.reserved_uars));
+}
+
+void mlx4_cleanup_uar_table(struct mlx4_dev *dev)
+{
+	mlx4_bitmap_cleanup(&mlx4_priv(dev)->uar_table.bitmap);
+}
diff --git a/drivers/net/mlx4/profile.c b/drivers/net/mlx4/profile.c
new file mode 100644
index 000000000000..9ca42b213d54
--- /dev/null
+++ b/drivers/net/mlx4/profile.c
@@ -0,0 +1,238 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/init.h>
+
+#include "mlx4.h"
+#include "fw.h"
+
+enum {
+	MLX4_RES_QP,
+	MLX4_RES_RDMARC,
+	MLX4_RES_ALTC,
+	MLX4_RES_AUXC,
+	MLX4_RES_SRQ,
+	MLX4_RES_CQ,
+	MLX4_RES_EQ,
+	MLX4_RES_DMPT,
+	MLX4_RES_CMPT,
+	MLX4_RES_MTT,
+	MLX4_RES_MCG,
+	MLX4_RES_NUM
+};
+
+static const char *res_name[] = {
+	[MLX4_RES_QP]		= "QP",
+	[MLX4_RES_RDMARC]	= "RDMARC",
+	[MLX4_RES_ALTC]		= "ALTC",
+	[MLX4_RES_AUXC]		= "AUXC",
+	[MLX4_RES_SRQ]		= "SRQ",
+	[MLX4_RES_CQ]		= "CQ",
+	[MLX4_RES_EQ]		= "EQ",
+	[MLX4_RES_DMPT]		= "DMPT",
+	[MLX4_RES_CMPT]		= "CMPT",
+	[MLX4_RES_MTT]		= "MTT",
+	[MLX4_RES_MCG]		= "MCG",
+};
+
+u64 mlx4_make_profile(struct mlx4_dev *dev,
+		      struct mlx4_profile *request,
+		      struct mlx4_dev_cap *dev_cap,
+		      struct mlx4_init_hca_param *init_hca)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource {
+		u64 size;
+		u64 start;
+		int type;
+		int num;
+		int log_num;
+	};
+
+	u64 total_size = 0;
+	struct mlx4_resource *profile;
+	struct mlx4_resource tmp;
+	int i, j;
+
+	profile = kzalloc(MLX4_RES_NUM * sizeof *profile, GFP_KERNEL);
+	if (!profile)
+		return -ENOMEM;
+
+	profile[MLX4_RES_QP].size     = dev_cap->qpc_entry_sz;
+	profile[MLX4_RES_RDMARC].size = dev_cap->rdmarc_entry_sz;
+	profile[MLX4_RES_ALTC].size   = dev_cap->altc_entry_sz;
+	profile[MLX4_RES_AUXC].size   = dev_cap->aux_entry_sz;
+	profile[MLX4_RES_SRQ].size    = dev_cap->srq_entry_sz;
+	profile[MLX4_RES_CQ].size     = dev_cap->cqc_entry_sz;
+	profile[MLX4_RES_EQ].size     = dev_cap->eqc_entry_sz;
+	profile[MLX4_RES_DMPT].size   = dev_cap->dmpt_entry_sz;
+	profile[MLX4_RES_CMPT].size   = dev_cap->cmpt_entry_sz;
+	profile[MLX4_RES_MTT].size    = MLX4_MTT_ENTRY_PER_SEG * dev_cap->mtt_entry_sz;
+	profile[MLX4_RES_MCG].size    = MLX4_MGM_ENTRY_SIZE;
+
+	profile[MLX4_RES_QP].num      = request->num_qp;
+	profile[MLX4_RES_RDMARC].num  = request->num_qp * request->rdmarc_per_qp;
+	profile[MLX4_RES_ALTC].num    = request->num_qp;
+	profile[MLX4_RES_AUXC].num    = request->num_qp;
+	profile[MLX4_RES_SRQ].num     = request->num_srq;
+	profile[MLX4_RES_CQ].num      = request->num_cq;
+	profile[MLX4_RES_EQ].num      = MLX4_NUM_EQ + dev_cap->reserved_eqs;
+	profile[MLX4_RES_DMPT].num    = request->num_mpt;
+	profile[MLX4_RES_CMPT].num    = MLX4_NUM_CMPTS;
+	profile[MLX4_RES_MTT].num     = request->num_mtt;
+	profile[MLX4_RES_MCG].num     = request->num_mcg;
+
+	for (i = 0; i < MLX4_RES_NUM; ++i) {
+		profile[i].type     = i;
+		profile[i].num      = roundup_pow_of_two(profile[i].num);
+		profile[i].log_num  = ilog2(profile[i].num);
+		profile[i].size    *= profile[i].num;
+		profile[i].size     = max(profile[i].size, (u64) PAGE_SIZE);
+	}
+
+	/*
+	 * Sort the resources in decreasing order of size.  Since they
+	 * all have sizes that are powers of 2, we'll be able to keep
+	 * resources aligned to their size and pack them without gaps
+	 * using the sorted order.
+	 */
+	for (i = MLX4_RES_NUM; i > 0; --i)
+		for (j = 1; j < i; ++j) {
+			if (profile[j].size > profile[j - 1].size) {
+				tmp	       = profile[j];
+				profile[j]     = profile[j - 1];
+				profile[j - 1] = tmp;
+			}
+		}
+
+	for (i = 0; i < MLX4_RES_NUM; ++i) {
+		if (profile[i].size) {
+			profile[i].start = total_size;
+			total_size	+= profile[i].size;
+		}
+
+		if (total_size > dev_cap->max_icm_sz) {
+			mlx4_err(dev, "Profile requires 0x%llx bytes; "
+				  "won't fit in 0x%llx bytes of context memory.\n",
+				  (unsigned long long) total_size,
+				  (unsigned long long) dev_cap->max_icm_sz);
+			kfree(profile);
+			return -ENOMEM;
+		}
+
+		if (profile[i].size)
+			mlx4_dbg(dev, "  profile[%2d] (%6s): 2^%02d entries @ 0x%10llx, "
+				  "size 0x%10llx\n",
+				 i, res_name[profile[i].type], profile[i].log_num,
+				 (unsigned long long) profile[i].start,
+				 (unsigned long long) profile[i].size);
+	}
+
+	mlx4_dbg(dev, "HCA context memory: reserving %d KB\n",
+		 (int) (total_size >> 10));
+
+	for (i = 0; i < MLX4_RES_NUM; ++i) {
+		switch (profile[i].type) {
+		case MLX4_RES_QP:
+			dev->caps.num_qps     = profile[i].num;
+			init_hca->qpc_base    = profile[i].start;
+			init_hca->log_num_qps = profile[i].log_num;
+			break;
+		case MLX4_RES_RDMARC:
+			for (priv->qp_table.rdmarc_shift = 0;
+			     request->num_qp << priv->qp_table.rdmarc_shift < profile[i].num;
+			     ++priv->qp_table.rdmarc_shift)
+				; /* nothing */
+			dev->caps.max_qp_dest_rdma = 1 << priv->qp_table.rdmarc_shift;
+			priv->qp_table.rdmarc_base   = (u32) profile[i].start;
+			init_hca->rdmarc_base	     = profile[i].start;
+			init_hca->log_rd_per_qp	     = priv->qp_table.rdmarc_shift;
+			break;
+		case MLX4_RES_ALTC:
+			init_hca->altc_base = profile[i].start;
+			break;
+		case MLX4_RES_AUXC:
+			init_hca->auxc_base = profile[i].start;
+			break;
+		case MLX4_RES_SRQ:
+			dev->caps.num_srqs     = profile[i].num;
+			init_hca->srqc_base    = profile[i].start;
+			init_hca->log_num_srqs = profile[i].log_num;
+			break;
+		case MLX4_RES_CQ:
+			dev->caps.num_cqs     = profile[i].num;
+			init_hca->cqc_base    = profile[i].start;
+			init_hca->log_num_cqs = profile[i].log_num;
+			break;
+		case MLX4_RES_EQ:
+			dev->caps.num_eqs     = profile[i].num;
+			init_hca->eqc_base    = profile[i].start;
+			init_hca->log_num_eqs = profile[i].log_num;
+			break;
+		case MLX4_RES_DMPT:
+			dev->caps.num_mpts	= profile[i].num;
+			priv->mr_table.mpt_base = profile[i].start;
+			init_hca->dmpt_base	= profile[i].start;
+			init_hca->log_mpt_sz	= profile[i].log_num;
+			break;
+		case MLX4_RES_CMPT:
+			init_hca->cmpt_base	 = profile[i].start;
+			break;
+		case MLX4_RES_MTT:
+			dev->caps.num_mtt_segs	 = profile[i].num;
+			priv->mr_table.mtt_base	 = profile[i].start;
+			init_hca->mtt_base	 = profile[i].start;
+			break;
+		case MLX4_RES_MCG:
+			dev->caps.num_mgms	  = profile[i].num >> 1;
+			dev->caps.num_amgms	  = profile[i].num >> 1;
+			init_hca->mc_base	  = profile[i].start;
+			init_hca->log_mc_entry_sz = ilog2(MLX4_MGM_ENTRY_SIZE);
+			init_hca->log_mc_table_sz = profile[i].log_num;
+			init_hca->log_mc_hash_sz  = profile[i].log_num - 1;
+			break;
+		default:
+			break;
+		}
+	}
+
+	/*
+	 * PDs don't take any HCA memory, but we assign them as part
+	 * of the HCA profile anyway.
+	 */
+	dev->caps.num_pds = MLX4_NUM_PDS;
+
+	kfree(profile);
+	return total_size;
+}
diff --git a/drivers/net/mlx4/qp.c b/drivers/net/mlx4/qp.c
new file mode 100644
index 000000000000..7f8b7d55b6e1
--- /dev/null
+++ b/drivers/net/mlx4/qp.c
@@ -0,0 +1,280 @@
+/*
+ * Copyright (c) 2004 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005, 2006, 2007 Cisco Systems, Inc. All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2004 Voltaire, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/init.h>
+
+#include <linux/mlx4/cmd.h>
+#include <linux/mlx4/qp.h>
+
+#include "mlx4.h"
+#include "icm.h"
+
+void mlx4_qp_event(struct mlx4_dev *dev, u32 qpn, int event_type)
+{
+	struct mlx4_qp_table *qp_table = &mlx4_priv(dev)->qp_table;
+	struct mlx4_qp *qp;
+
+	spin_lock(&qp_table->lock);
+
+	qp = __mlx4_qp_lookup(dev, qpn);
+	if (qp)
+		atomic_inc(&qp->refcount);
+
+	spin_unlock(&qp_table->lock);
+
+	if (!qp) {
+		mlx4_warn(dev, "Async event for bogus QP %08x\n", qpn);
+		return;
+	}
+
+	qp->event(qp, event_type);
+
+	if (atomic_dec_and_test(&qp->refcount))
+		complete(&qp->free);
+}
+
+int mlx4_qp_modify(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
+		   enum mlx4_qp_state cur_state, enum mlx4_qp_state new_state,
+		   struct mlx4_qp_context *context, enum mlx4_qp_optpar optpar,
+		   int sqd_event, struct mlx4_qp *qp)
+{
+	static const u16 op[MLX4_QP_NUM_STATE][MLX4_QP_NUM_STATE] = {
+		[MLX4_QP_STATE_RST] = {
+			[MLX4_QP_STATE_RST]	= MLX4_CMD_2RST_QP,
+			[MLX4_QP_STATE_ERR]	= MLX4_CMD_2ERR_QP,
+			[MLX4_QP_STATE_INIT]	= MLX4_CMD_RST2INIT_QP,
+		},
+		[MLX4_QP_STATE_INIT]  = {
+			[MLX4_QP_STATE_RST]	= MLX4_CMD_2RST_QP,
+			[MLX4_QP_STATE_ERR]	= MLX4_CMD_2ERR_QP,
+			[MLX4_QP_STATE_INIT]	= MLX4_CMD_INIT2INIT_QP,
+			[MLX4_QP_STATE_RTR]	= MLX4_CMD_INIT2RTR_QP,
+		},
+		[MLX4_QP_STATE_RTR]   = {
+			[MLX4_QP_STATE_RST]	= MLX4_CMD_2RST_QP,
+			[MLX4_QP_STATE_ERR]	= MLX4_CMD_2ERR_QP,
+			[MLX4_QP_STATE_RTS]	= MLX4_CMD_RTR2RTS_QP,
+		},
+		[MLX4_QP_STATE_RTS]   = {
+			[MLX4_QP_STATE_RST]	= MLX4_CMD_2RST_QP,
+			[MLX4_QP_STATE_ERR]	= MLX4_CMD_2ERR_QP,
+			[MLX4_QP_STATE_RTS]	= MLX4_CMD_RTS2RTS_QP,
+			[MLX4_QP_STATE_SQD]	= MLX4_CMD_RTS2SQD_QP,
+		},
+		[MLX4_QP_STATE_SQD] = {
+			[MLX4_QP_STATE_RST]	= MLX4_CMD_2RST_QP,
+			[MLX4_QP_STATE_ERR]	= MLX4_CMD_2ERR_QP,
+			[MLX4_QP_STATE_RTS]	= MLX4_CMD_SQD2RTS_QP,
+			[MLX4_QP_STATE_SQD]	= MLX4_CMD_SQD2SQD_QP,
+		},
+		[MLX4_QP_STATE_SQER] = {
+			[MLX4_QP_STATE_RST]	= MLX4_CMD_2RST_QP,
+			[MLX4_QP_STATE_ERR]	= MLX4_CMD_2ERR_QP,
+			[MLX4_QP_STATE_RTS]	= MLX4_CMD_SQERR2RTS_QP,
+		},
+		[MLX4_QP_STATE_ERR] = {
+			[MLX4_QP_STATE_RST]	= MLX4_CMD_2RST_QP,
+			[MLX4_QP_STATE_ERR]	= MLX4_CMD_2ERR_QP,
+		}
+	};
+
+	struct mlx4_cmd_mailbox *mailbox;
+	int ret = 0;
+
+	if (cur_state < 0 || cur_state >= MLX4_QP_NUM_STATE ||
+	    new_state < 0 || cur_state >= MLX4_QP_NUM_STATE ||
+	    !op[cur_state][new_state])
+		return -EINVAL;
+
+	if (op[cur_state][new_state] == MLX4_CMD_2RST_QP)
+		return mlx4_cmd(dev, 0, qp->qpn, 2,
+				MLX4_CMD_2RST_QP, MLX4_CMD_TIME_CLASS_A);
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	if (cur_state == MLX4_QP_STATE_RST && new_state == MLX4_QP_STATE_INIT) {
+		u64 mtt_addr = mlx4_mtt_addr(dev, mtt);
+		context->mtt_base_addr_h = mtt_addr >> 32;
+		context->mtt_base_addr_l = cpu_to_be32(mtt_addr & 0xffffffff);
+		context->log_page_size   = mtt->page_shift - MLX4_ICM_PAGE_SHIFT;
+	}
+
+	*(__be32 *) mailbox->buf = cpu_to_be32(optpar);
+	memcpy(mailbox->buf + 8, context, sizeof *context);
+
+	((struct mlx4_qp_context *) (mailbox->buf + 8))->local_qpn =
+		cpu_to_be32(qp->qpn);
+
+	ret = mlx4_cmd(dev, mailbox->dma, qp->qpn | (!!sqd_event << 31),
+		       new_state == MLX4_QP_STATE_RST ? 2 : 0,
+		       op[cur_state][new_state], MLX4_CMD_TIME_CLASS_C);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return ret;
+}
+EXPORT_SYMBOL_GPL(mlx4_qp_modify);
+
+int mlx4_qp_alloc(struct mlx4_dev *dev, int sqpn, struct mlx4_qp *qp)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_qp_table *qp_table = &priv->qp_table;
+	int err;
+
+	if (sqpn)
+		qp->qpn = sqpn;
+	else {
+		qp->qpn = mlx4_bitmap_alloc(&qp_table->bitmap);
+		if (qp->qpn == -1)
+			return -ENOMEM;
+	}
+
+	err = mlx4_table_get(dev, &qp_table->qp_table, qp->qpn);
+	if (err)
+		goto err_out;
+
+	err = mlx4_table_get(dev, &qp_table->auxc_table, qp->qpn);
+	if (err)
+		goto err_put_qp;
+
+	err = mlx4_table_get(dev, &qp_table->altc_table, qp->qpn);
+	if (err)
+		goto err_put_auxc;
+
+	err = mlx4_table_get(dev, &qp_table->rdmarc_table, qp->qpn);
+	if (err)
+		goto err_put_altc;
+
+	err = mlx4_table_get(dev, &qp_table->cmpt_table, qp->qpn);
+	if (err)
+		goto err_put_rdmarc;
+
+	spin_lock_irq(&qp_table->lock);
+	err = radix_tree_insert(&dev->qp_table_tree, qp->qpn & (dev->caps.num_qps - 1), qp);
+	spin_unlock_irq(&qp_table->lock);
+	if (err)
+		goto err_put_cmpt;
+
+	atomic_set(&qp->refcount, 1);
+	init_completion(&qp->free);
+
+	return 0;
+
+err_put_cmpt:
+	mlx4_table_put(dev, &qp_table->cmpt_table, qp->qpn);
+
+err_put_rdmarc:
+	mlx4_table_put(dev, &qp_table->rdmarc_table, qp->qpn);
+
+err_put_altc:
+	mlx4_table_put(dev, &qp_table->altc_table, qp->qpn);
+
+err_put_auxc:
+	mlx4_table_put(dev, &qp_table->auxc_table, qp->qpn);
+
+err_put_qp:
+	mlx4_table_put(dev, &qp_table->qp_table, qp->qpn);
+
+err_out:
+	if (!sqpn)
+		mlx4_bitmap_free(&qp_table->bitmap, qp->qpn);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_qp_alloc);
+
+void mlx4_qp_remove(struct mlx4_dev *dev, struct mlx4_qp *qp)
+{
+	struct mlx4_qp_table *qp_table = &mlx4_priv(dev)->qp_table;
+	unsigned long flags;
+
+	spin_lock_irqsave(&qp_table->lock, flags);
+	radix_tree_delete(&dev->qp_table_tree, qp->qpn & (dev->caps.num_qps - 1));
+	spin_unlock_irqrestore(&qp_table->lock, flags);
+}
+EXPORT_SYMBOL_GPL(mlx4_qp_remove);
+
+void mlx4_qp_free(struct mlx4_dev *dev, struct mlx4_qp *qp)
+{
+	struct mlx4_qp_table *qp_table = &mlx4_priv(dev)->qp_table;
+
+	if (atomic_dec_and_test(&qp->refcount))
+		complete(&qp->free);
+	wait_for_completion(&qp->free);
+
+	mlx4_table_put(dev, &qp_table->cmpt_table, qp->qpn);
+	mlx4_table_put(dev, &qp_table->rdmarc_table, qp->qpn);
+	mlx4_table_put(dev, &qp_table->altc_table, qp->qpn);
+	mlx4_table_put(dev, &qp_table->auxc_table, qp->qpn);
+	mlx4_table_put(dev, &qp_table->qp_table, qp->qpn);
+
+	mlx4_bitmap_free(&qp_table->bitmap, qp->qpn);
+}
+EXPORT_SYMBOL_GPL(mlx4_qp_free);
+
+static int mlx4_CONF_SPECIAL_QP(struct mlx4_dev *dev, u32 base_qpn)
+{
+	return mlx4_cmd(dev, 0, base_qpn, 0, MLX4_CMD_CONF_SPECIAL_QP,
+			MLX4_CMD_TIME_CLASS_B);
+}
+
+int __devinit mlx4_init_qp_table(struct mlx4_dev *dev)
+{
+	struct mlx4_qp_table *qp_table = &mlx4_priv(dev)->qp_table;
+	int err;
+
+	spin_lock_init(&qp_table->lock);
+	INIT_RADIX_TREE(&dev->qp_table_tree, GFP_ATOMIC);
+
+	/*
+	 * We reserve 2 extra QPs per port for the special QPs.  The
+	 * block of special QPs must be aligned to a multiple of 8, so
+	 * round up.
+	 */
+	dev->caps.sqp_start = ALIGN(dev->caps.reserved_qps, 8);
+	err = mlx4_bitmap_init(&qp_table->bitmap, dev->caps.num_qps,
+			       (1 << 24) - 1, dev->caps.sqp_start + 8);
+	if (err)
+		return err;
+
+	return mlx4_CONF_SPECIAL_QP(dev, dev->caps.sqp_start);
+}
+
+void mlx4_cleanup_qp_table(struct mlx4_dev *dev)
+{
+	mlx4_CONF_SPECIAL_QP(dev, 0);
+	mlx4_bitmap_cleanup(&mlx4_priv(dev)->qp_table.bitmap);
+}
diff --git a/drivers/net/mlx4/reset.c b/drivers/net/mlx4/reset.c
new file mode 100644
index 000000000000..51eef8492e93
--- /dev/null
+++ b/drivers/net/mlx4/reset.c
@@ -0,0 +1,181 @@
+/*
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/init.h>
+#include <linux/errno.h>
+#include <linux/pci.h>
+#include <linux/delay.h>
+#include <linux/slab.h>
+
+#include "mlx4.h"
+
+int mlx4_reset(struct mlx4_dev *dev)
+{
+	void __iomem *reset;
+	u32 *hca_header = NULL;
+	int pcie_cap;
+	u16 devctl;
+	u16 linkctl;
+	u16 vendor;
+	unsigned long end;
+	u32 sem;
+	int i;
+	int err = 0;
+
+#define MLX4_RESET_BASE		0xf0000
+#define MLX4_RESET_SIZE		  0x400
+#define MLX4_SEM_OFFSET		  0x3fc
+#define MLX4_RESET_OFFSET	   0x10
+#define MLX4_RESET_VALUE	swab32(1)
+
+#define MLX4_SEM_TIMEOUT_JIFFIES	(10 * HZ)
+#define MLX4_RESET_TIMEOUT_JIFFIES	(2 * HZ)
+
+	/*
+	 * Reset the chip.  This is somewhat ugly because we have to
+	 * save off the PCI header before reset and then restore it
+	 * after the chip reboots.  We skip config space offsets 22
+	 * and 23 since those have a special meaning.
+	 */
+
+	/* Do we need to save off the full 4K PCI Express header?? */
+	hca_header = kmalloc(256, GFP_KERNEL);
+	if (!hca_header) {
+		err = -ENOMEM;
+		mlx4_err(dev, "Couldn't allocate memory to save HCA "
+			  "PCI header, aborting.\n");
+		goto out;
+	}
+
+	pcie_cap = pci_find_capability(dev->pdev, PCI_CAP_ID_EXP);
+
+	for (i = 0; i < 64; ++i) {
+		if (i == 22 || i == 23)
+			continue;
+		if (pci_read_config_dword(dev->pdev, i * 4, hca_header + i)) {
+			err = -ENODEV;
+			mlx4_err(dev, "Couldn't save HCA "
+				  "PCI header, aborting.\n");
+			goto out;
+		}
+	}
+
+	reset = ioremap(pci_resource_start(dev->pdev, 0) + MLX4_RESET_BASE,
+			MLX4_RESET_SIZE);
+	if (!reset) {
+		err = -ENOMEM;
+		mlx4_err(dev, "Couldn't map HCA reset register, aborting.\n");
+		goto out;
+	}
+
+	/* grab HW semaphore to lock out flash updates */
+	end = jiffies + MLX4_SEM_TIMEOUT_JIFFIES;
+	do {
+		sem = readl(reset + MLX4_SEM_OFFSET);
+		if (!sem)
+			break;
+
+		msleep(1);
+	} while (time_before(jiffies, end));
+
+	if (sem) {
+		mlx4_err(dev, "Failed to obtain HW semaphore, aborting\n");
+		err = -EAGAIN;
+		iounmap(reset);
+		goto out;
+	}
+
+	/* actually hit reset */
+	writel(MLX4_RESET_VALUE, reset + MLX4_RESET_OFFSET);
+	iounmap(reset);
+
+	end = jiffies + MLX4_RESET_TIMEOUT_JIFFIES;
+	do {
+		if (!pci_read_config_word(dev->pdev, PCI_VENDOR_ID, &vendor) &&
+		    vendor != 0xffff)
+			break;
+
+		msleep(1);
+	} while (time_before(jiffies, end));
+
+	if (vendor == 0xffff) {
+		err = -ENODEV;
+		mlx4_err(dev, "PCI device did not come back after reset, "
+			  "aborting.\n");
+		goto out;
+	}
+
+	/* Now restore the PCI headers */
+	if (pcie_cap) {
+		devctl = hca_header[(pcie_cap + PCI_EXP_DEVCTL) / 4];
+		if (pci_write_config_word(dev->pdev, pcie_cap + PCI_EXP_DEVCTL,
+					   devctl)) {
+			err = -ENODEV;
+			mlx4_err(dev, "Couldn't restore HCA PCI Express "
+				 "Device Control register, aborting.\n");
+			goto out;
+		}
+		linkctl = hca_header[(pcie_cap + PCI_EXP_LNKCTL) / 4];
+		if (pci_write_config_word(dev->pdev, pcie_cap + PCI_EXP_LNKCTL,
+					   linkctl)) {
+			err = -ENODEV;
+			mlx4_err(dev, "Couldn't restore HCA PCI Express "
+				 "Link control register, aborting.\n");
+			goto out;
+		}
+	}
+
+	for (i = 0; i < 16; ++i) {
+		if (i * 4 == PCI_COMMAND)
+			continue;
+
+		if (pci_write_config_dword(dev->pdev, i * 4, hca_header[i])) {
+			err = -ENODEV;
+			mlx4_err(dev, "Couldn't restore HCA reg %x, "
+				  "aborting.\n", i);
+			goto out;
+		}
+	}
+
+	if (pci_write_config_dword(dev->pdev, PCI_COMMAND,
+				   hca_header[PCI_COMMAND / 4])) {
+		err = -ENODEV;
+		mlx4_err(dev, "Couldn't restore HCA COMMAND, "
+			  "aborting.\n");
+		goto out;
+	}
+
+out:
+	kfree(hca_header);
+
+	return err;
+}
diff --git a/drivers/net/mlx4/srq.c b/drivers/net/mlx4/srq.c
new file mode 100644
index 000000000000..2134f83aed87
--- /dev/null
+++ b/drivers/net/mlx4/srq.c
@@ -0,0 +1,227 @@
+/*
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/init.h>
+
+#include <linux/mlx4/cmd.h>
+
+#include "mlx4.h"
+#include "icm.h"
+
+struct mlx4_srq_context {
+	__be32			state_logsize_srqn;
+	u8			logstride;
+	u8			reserved1[3];
+	u8			pg_offset;
+	u8			reserved2[3];
+	u32			reserved3;
+	u8			log_page_size;
+	u8			reserved4[2];
+	u8			mtt_base_addr_h;
+	__be32			mtt_base_addr_l;
+	__be32			pd;
+	__be16			limit_watermark;
+	__be16			wqe_cnt;
+	u16			reserved5;
+	__be16			wqe_counter;
+	u32			reserved6;
+	__be64			db_rec_addr;
+};
+
+void mlx4_srq_event(struct mlx4_dev *dev, u32 srqn, int event_type)
+{
+	struct mlx4_srq_table *srq_table = &mlx4_priv(dev)->srq_table;
+	struct mlx4_srq *srq;
+
+	spin_lock(&srq_table->lock);
+
+	srq = radix_tree_lookup(&srq_table->tree, srqn & (dev->caps.num_srqs - 1));
+	if (srq)
+		atomic_inc(&srq->refcount);
+
+	spin_unlock(&srq_table->lock);
+
+	if (!srq) {
+		mlx4_warn(dev, "Async event for bogus SRQ %08x\n", srqn);
+		return;
+	}
+
+	srq->event(srq, event_type);
+
+	if (atomic_dec_and_test(&srq->refcount))
+		complete(&srq->free);
+}
+
+static int mlx4_SW2HW_SRQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			  int srq_num)
+{
+	return mlx4_cmd(dev, mailbox->dma, srq_num, 0, MLX4_CMD_SW2HW_SRQ,
+			MLX4_CMD_TIME_CLASS_A);
+}
+
+static int mlx4_HW2SW_SRQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			  int srq_num)
+{
+	return mlx4_cmd_box(dev, 0, mailbox ? mailbox->dma : 0, srq_num,
+			    mailbox ? 0 : 1, MLX4_CMD_HW2SW_SRQ,
+			    MLX4_CMD_TIME_CLASS_A);
+}
+
+static int mlx4_ARM_SRQ(struct mlx4_dev *dev, int srq_num, int limit_watermark)
+{
+	return mlx4_cmd(dev, limit_watermark, srq_num, 0, MLX4_CMD_ARM_SRQ,
+			MLX4_CMD_TIME_CLASS_B);
+}
+
+int mlx4_srq_alloc(struct mlx4_dev *dev, u32 pdn, struct mlx4_mtt *mtt,
+		   u64 db_rec, struct mlx4_srq *srq)
+{
+	struct mlx4_srq_table *srq_table = &mlx4_priv(dev)->srq_table;
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_srq_context *srq_context;
+	u64 mtt_addr;
+	int err;
+
+	srq->srqn = mlx4_bitmap_alloc(&srq_table->bitmap);
+	if (srq->srqn == -1)
+		return -ENOMEM;
+
+	err = mlx4_table_get(dev, &srq_table->table, srq->srqn);
+	if (err)
+		goto err_out;
+
+	err = mlx4_table_get(dev, &srq_table->cmpt_table, srq->srqn);
+	if (err)
+		goto err_put;
+
+	spin_lock_irq(&srq_table->lock);
+	err = radix_tree_insert(&srq_table->tree, srq->srqn, srq);
+	spin_unlock_irq(&srq_table->lock);
+	if (err)
+		goto err_cmpt_put;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox)) {
+		err = PTR_ERR(mailbox);
+		goto err_radix;
+	}
+
+	srq_context = mailbox->buf;
+	memset(srq_context, 0, sizeof *srq_context);
+
+	srq_context->state_logsize_srqn = cpu_to_be32((ilog2(srq->max) << 24) |
+						      srq->srqn);
+	srq_context->logstride          = srq->wqe_shift - 4;
+	srq_context->log_page_size      = mtt->page_shift - MLX4_ICM_PAGE_SHIFT;
+
+	mtt_addr = mlx4_mtt_addr(dev, mtt);
+	srq_context->mtt_base_addr_h    = mtt_addr >> 32;
+	srq_context->mtt_base_addr_l    = cpu_to_be32(mtt_addr & 0xffffffff);
+	srq_context->pd			= cpu_to_be32(pdn);
+	srq_context->db_rec_addr        = cpu_to_be64(db_rec);
+
+	err = mlx4_SW2HW_SRQ(dev, mailbox, srq->srqn);
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	if (err)
+		goto err_radix;
+
+	atomic_set(&srq->refcount, 1);
+	init_completion(&srq->free);
+
+	return 0;
+
+err_radix:
+	spin_lock_irq(&srq_table->lock);
+	radix_tree_delete(&srq_table->tree, srq->srqn);
+	spin_unlock_irq(&srq_table->lock);
+
+err_cmpt_put:
+	mlx4_table_put(dev, &srq_table->cmpt_table, srq->srqn);
+
+err_put:
+	mlx4_table_put(dev, &srq_table->table, srq->srqn);
+
+err_out:
+	mlx4_bitmap_free(&srq_table->bitmap, srq->srqn);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_srq_alloc);
+
+void mlx4_srq_free(struct mlx4_dev *dev, struct mlx4_srq *srq)
+{
+	struct mlx4_srq_table *srq_table = &mlx4_priv(dev)->srq_table;
+	int err;
+
+	err = mlx4_HW2SW_SRQ(dev, NULL, srq->srqn);
+	if (err)
+		mlx4_warn(dev, "HW2SW_SRQ failed (%d) for SRQN %06x\n", err, srq->srqn);
+
+	spin_lock_irq(&srq_table->lock);
+	radix_tree_delete(&srq_table->tree, srq->srqn);
+	spin_unlock_irq(&srq_table->lock);
+
+	if (atomic_dec_and_test(&srq->refcount))
+		complete(&srq->free);
+	wait_for_completion(&srq->free);
+
+	mlx4_table_put(dev, &srq_table->table, srq->srqn);
+	mlx4_bitmap_free(&srq_table->bitmap, srq->srqn);
+}
+EXPORT_SYMBOL_GPL(mlx4_srq_free);
+
+int mlx4_srq_arm(struct mlx4_dev *dev, struct mlx4_srq *srq, int limit_watermark)
+{
+	return mlx4_ARM_SRQ(dev, srq->srqn, limit_watermark);
+}
+EXPORT_SYMBOL_GPL(mlx4_srq_arm);
+
+int __devinit mlx4_init_srq_table(struct mlx4_dev *dev)
+{
+	struct mlx4_srq_table *srq_table = &mlx4_priv(dev)->srq_table;
+	int err;
+
+	spin_lock_init(&srq_table->lock);
+	INIT_RADIX_TREE(&srq_table->tree, GFP_ATOMIC);
+
+	err = mlx4_bitmap_init(&srq_table->bitmap, dev->caps.num_srqs,
+			       dev->caps.num_srqs - 1, dev->caps.reserved_srqs);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+void mlx4_cleanup_srq_table(struct mlx4_dev *dev)
+{
+	mlx4_bitmap_cleanup(&mlx4_priv(dev)->srq_table.bitmap);
+}
diff --git a/drivers/net/natsemi.c b/drivers/net/natsemi.c
index 223e0e6264ba..4cf0d3fcb519 100644
--- a/drivers/net/natsemi.c
+++ b/drivers/net/natsemi.c
@@ -131,7 +131,6 @@ static const char version[] __devinitdata =
   KERN_INFO DRV_NAME " dp8381x driver, version "
       DRV_VERSION ", " DRV_RELDATE "\n"
   KERN_INFO "  originally by Donald Becker <becker@scyld.com>\n"
-  KERN_INFO "  http://www.scyld.com/network/natsemi.html\n"
   KERN_INFO "  2.4.x kernel port by Jeff Garzik, Tjeerd Mulder\n";
 
 MODULE_AUTHOR("Donald Becker <becker@scyld.com>");
diff --git a/drivers/net/ne2k-pci.c b/drivers/net/ne2k-pci.c
index 589785d1e762..995c0a5d4066 100644
--- a/drivers/net/ne2k-pci.c
+++ b/drivers/net/ne2k-pci.c
@@ -63,8 +63,7 @@ static int options[MAX_UNITS];
 
 /* These identify the driver base version and may not be removed. */
 static char version[] __devinitdata =
-KERN_INFO DRV_NAME ".c:v" DRV_VERSION " " DRV_RELDATE " D. Becker/P. Gortmaker\n"
-KERN_INFO "  http://www.scyld.com/network/ne2k-pci.html\n";
+KERN_INFO DRV_NAME ".c:v" DRV_VERSION " " DRV_RELDATE " D. Becker/P. Gortmaker\n";
 
 #if defined(__powerpc__)
 #define inl_le(addr)  le32_to_cpu(inl(addr))
diff --git a/drivers/net/pcmcia/ibmtr_cs.c b/drivers/net/pcmcia/ibmtr_cs.c
index 1060154ae750..4ecb8ca5a992 100644
--- a/drivers/net/pcmcia/ibmtr_cs.c
+++ b/drivers/net/pcmcia/ibmtr_cs.c
@@ -189,16 +189,20 @@ static void ibmtr_detach(struct pcmcia_device *link)
 {
     struct ibmtr_dev_t *info = link->priv;
     struct net_device *dev = info->dev;
+     struct tok_info *ti = netdev_priv(dev);
 
     DEBUG(0, "ibmtr_detach(0x%p)\n", link);
+    
+    /* 
+     * When the card removal interrupt hits tok_interrupt(), 
+     * bail out early, so we don't crash the machine 
+     */
+    ti->sram_phys |= 1;
 
     if (link->dev_node)
 	unregister_netdev(dev);
-
-    {
-	struct tok_info *ti = netdev_priv(dev);
-	del_timer_sync(&(ti->tr_timer));
-    }
+    
+    del_timer_sync(&(ti->tr_timer));
 
     ibmtr_release(link);
 
diff --git a/drivers/net/phy/phy.c b/drivers/net/phy/phy.c
index eed433d6056a..f71dab347667 100644
--- a/drivers/net/phy/phy.c
+++ b/drivers/net/phy/phy.c
@@ -662,10 +662,10 @@ int phy_stop_interrupts(struct phy_device *phydev)
 		phy_error(phydev);
 
 	/*
-	 * Finish any pending work; we might have been scheduled
-	 * to be called from keventd ourselves, though.
+	 * Finish any pending work; we might have been scheduled to be called
+	 * from keventd ourselves, but cancel_work_sync() handles that.
 	 */
-	run_scheduled_work(&phydev->phy_queue);
+	cancel_work_sync(&phydev->phy_queue);
 
 	free_irq(phydev->irq, phydev);
 
diff --git a/drivers/net/skge.c b/drivers/net/skge.c
index b07da1054add..e0489578945d 100644
--- a/drivers/net/skge.c
+++ b/drivers/net/skge.c
@@ -3594,7 +3594,9 @@ static struct net_device *skge_devinit(struct skge_hw *hw, int port,
 	skge->duplex = -1;
 	skge->speed = -1;
 	skge->advertising = skge_supported_modes(hw);
-	skge->wol = pci_wake_enabled(hw->pdev) ? wol_supported(hw) : 0;
+
+	if (pci_wake_enabled(hw->pdev))
+		skge->wol = wol_supported(hw) & WAKE_MAGIC;
 
 	hw->dev[port] = dev;
 
diff --git a/drivers/net/sundance.c b/drivers/net/sundance.c
index f51ba31970aa..e1f912d04043 100644
--- a/drivers/net/sundance.c
+++ b/drivers/net/sundance.c
@@ -110,8 +110,7 @@ static char *media[MAX_UNITS];
 
 /* These identify the driver base version and may not be removed. */
 static char version[] =
-KERN_INFO DRV_NAME ".c:v" DRV_VERSION " " DRV_RELDATE "  Written by Donald Becker\n"
-KERN_INFO "  http://www.scyld.com/network/sundance.html\n";
+KERN_INFO DRV_NAME ".c:v" DRV_VERSION " " DRV_RELDATE "  Written by Donald Becker\n";
 
 MODULE_AUTHOR("Donald Becker <becker@scyld.com>");
 MODULE_DESCRIPTION("Sundance Alta Ethernet driver");
diff --git a/drivers/net/tg3.c b/drivers/net/tg3.c
index e5e901ecd808..923b9c725cc3 100644
--- a/drivers/net/tg3.c
+++ b/drivers/net/tg3.c
@@ -3716,10 +3716,8 @@ static void tg3_reset_task(struct work_struct *work)
 	unsigned int restart_timer;
 
 	tg3_full_lock(tp, 0);
-	tp->tg3_flags |= TG3_FLAG_IN_RESET_TASK;
 
 	if (!netif_running(tp->dev)) {
-		tp->tg3_flags &= ~TG3_FLAG_IN_RESET_TASK;
 		tg3_full_unlock(tp);
 		return;
 	}
@@ -3750,8 +3748,6 @@ static void tg3_reset_task(struct work_struct *work)
 		mod_timer(&tp->timer, jiffies + 1);
 
 out:
-	tp->tg3_flags &= ~TG3_FLAG_IN_RESET_TASK;
-
 	tg3_full_unlock(tp);
 }
 
@@ -7390,12 +7386,7 @@ static int tg3_close(struct net_device *dev)
 {
 	struct tg3 *tp = netdev_priv(dev);
 
-	/* Calling flush_scheduled_work() may deadlock because
-	 * linkwatch_event() may be on the workqueue and it will try to get
-	 * the rtnl_lock which we are holding.
-	 */
-	while (tp->tg3_flags & TG3_FLAG_IN_RESET_TASK)
-		msleep(1);
+	cancel_work_sync(&tp->reset_task);
 
 	netif_stop_queue(dev);
 
diff --git a/drivers/net/tg3.h b/drivers/net/tg3.h
index 4d334cf5a243..bd9f4f428e5b 100644
--- a/drivers/net/tg3.h
+++ b/drivers/net/tg3.h
@@ -2228,7 +2228,7 @@ struct tg3 {
 #define TG3_FLAG_JUMBO_RING_ENABLE	0x00800000
 #define TG3_FLAG_10_100_ONLY		0x01000000
 #define TG3_FLAG_PAUSE_AUTONEG		0x02000000
-#define TG3_FLAG_IN_RESET_TASK		0x04000000
+
 #define TG3_FLAG_40BIT_DMA_BUG		0x08000000
 #define TG3_FLAG_BROKEN_CHECKSUMS	0x10000000
 #define TG3_FLAG_SUPPORT_MSI		0x20000000
diff --git a/drivers/net/tulip/interrupt.c b/drivers/net/tulip/interrupt.c
index 9b08afbd1f65..ea896777bcaf 100644
--- a/drivers/net/tulip/interrupt.c
+++ b/drivers/net/tulip/interrupt.c
@@ -269,7 +269,7 @@ done:
             This would turn on IM for devices that is not contributing
             to backlog congestion with unnecessary latency.
 
-             We monitor the the device RX-ring and have:
+             We monitor the device RX-ring and have:
 
              HW Interrupt Mitigation either ON or OFF.
 
diff --git a/drivers/net/tulip/winbond-840.c b/drivers/net/tulip/winbond-840.c
index fa440706fb4a..38f3b99716b8 100644
--- a/drivers/net/tulip/winbond-840.c
+++ b/drivers/net/tulip/winbond-840.c
@@ -1021,7 +1021,7 @@ static int start_tx(struct sk_buff *skb, struct net_device *dev)
 		np->tx_ring[entry].length |= DescEndRing;
 
 	/* Now acquire the irq spinlock.
-	 * The difficult race is the the ordering between
+	 * The difficult race is the ordering between
 	 * increasing np->cur_tx and setting DescOwned:
 	 * - if np->cur_tx is increased first the interrupt
 	 *   handler could consider the packet as transmitted
diff --git a/drivers/net/tulip/xircom_cb.c b/drivers/net/tulip/xircom_cb.c
index 985a1810ca59..2470b1ee33c0 100644
--- a/drivers/net/tulip/xircom_cb.c
+++ b/drivers/net/tulip/xircom_cb.c
@@ -1043,7 +1043,7 @@ static int enable_promisc(struct xircom_private *card)
 
 
 /*
-link_status() checks the the links status and will return 0 for no link, 10 for 10mbit link and 100 for.. guess what.
+link_status() checks the links status and will return 0 for no link, 10 for 10mbit link and 100 for.. guess what.
 
 Must be called in locked state with interrupts disabled
 */
diff --git a/drivers/net/typhoon.c b/drivers/net/typhoon.c
index f2dd7763cd0b..f72573594121 100644
--- a/drivers/net/typhoon.c
+++ b/drivers/net/typhoon.c
@@ -639,7 +639,7 @@ typhoon_issue_command(struct typhoon *tp, int num_cmd, struct cmd_desc *cmd,
 
 	typhoon_inc_cmd_index(&ring->lastWrite, num_cmd);
 
-	/* "I feel a presence... another warrior is on the the mesa."
+	/* "I feel a presence... another warrior is on the mesa."
 	 */
 	wmb();
 	iowrite32(ring->lastWrite, tp->ioaddr + TYPHOON_REG_CMD_READY);
diff --git a/drivers/usb/net/Kconfig b/drivers/net/usb/Kconfig
index 3de564b23147..3de564b23147 100644
--- a/drivers/usb/net/Kconfig
+++ b/drivers/net/usb/Kconfig
diff --git a/drivers/usb/net/Makefile b/drivers/net/usb/Makefile
index 595a539f8384..595a539f8384 100644
--- a/drivers/usb/net/Makefile
+++ b/drivers/net/usb/Makefile
diff --git a/drivers/usb/net/asix.c b/drivers/net/usb/asix.c
index d5ef97bc4d01..d5ef97bc4d01 100644
--- a/drivers/usb/net/asix.c
+++ b/drivers/net/usb/asix.c
diff --git a/drivers/usb/net/catc.c b/drivers/net/usb/catc.c
index 86e90c59d551..86e90c59d551 100644
--- a/drivers/usb/net/catc.c
+++ b/drivers/net/usb/catc.c
diff --git a/drivers/usb/net/cdc_ether.c b/drivers/net/usb/cdc_ether.c
index 5a21f06bf8a5..5a21f06bf8a5 100644
--- a/drivers/usb/net/cdc_ether.c
+++ b/drivers/net/usb/cdc_ether.c
diff --git a/drivers/usb/net/cdc_subset.c b/drivers/net/usb/cdc_subset.c
index bc62b012602b..bc62b012602b 100644
--- a/drivers/usb/net/cdc_subset.c
+++ b/drivers/net/usb/cdc_subset.c
diff --git a/drivers/usb/net/dm9601.c b/drivers/net/usb/dm9601.c
index a67638601477..a67638601477 100644
--- a/drivers/usb/net/dm9601.c
+++ b/drivers/net/usb/dm9601.c
diff --git a/drivers/usb/net/gl620a.c b/drivers/net/usb/gl620a.c
index 031cf5ca4dbb..031cf5ca4dbb 100644
--- a/drivers/usb/net/gl620a.c
+++ b/drivers/net/usb/gl620a.c
diff --git a/drivers/usb/net/kaweth.c b/drivers/net/usb/kaweth.c
index 60d29440f316..60d29440f316 100644
--- a/drivers/usb/net/kaweth.c
+++ b/drivers/net/usb/kaweth.c
diff --git a/drivers/usb/net/kawethfw.h b/drivers/net/usb/kawethfw.h
index cf85fcb0d1a6..cf85fcb0d1a6 100644
--- a/drivers/usb/net/kawethfw.h
+++ b/drivers/net/usb/kawethfw.h
diff --git a/drivers/usb/net/mcs7830.c b/drivers/net/usb/mcs7830.c
index 6240b978fe3d..6240b978fe3d 100644
--- a/drivers/usb/net/mcs7830.c
+++ b/drivers/net/usb/mcs7830.c
diff --git a/drivers/usb/net/net1080.c b/drivers/net/usb/net1080.c
index 19bf8dae70c9..19bf8dae70c9 100644
--- a/drivers/usb/net/net1080.c
+++ b/drivers/net/usb/net1080.c
diff --git a/drivers/usb/net/pegasus.c b/drivers/net/usb/pegasus.c
index a05fd97e5bc2..a05fd97e5bc2 100644
--- a/drivers/usb/net/pegasus.c
+++ b/drivers/net/usb/pegasus.c
diff --git a/drivers/usb/net/pegasus.h b/drivers/net/usb/pegasus.h
index c7467823cd1c..c7467823cd1c 100644
--- a/drivers/usb/net/pegasus.h
+++ b/drivers/net/usb/pegasus.h
diff --git a/drivers/usb/net/plusb.c b/drivers/net/usb/plusb.c
index 45300939d185..45300939d185 100644
--- a/drivers/usb/net/plusb.c
+++ b/drivers/net/usb/plusb.c
diff --git a/drivers/usb/net/rndis_host.c b/drivers/net/usb/rndis_host.c
index 980e4aaa97aa..980e4aaa97aa 100644
--- a/drivers/usb/net/rndis_host.c
+++ b/drivers/net/usb/rndis_host.c
diff --git a/drivers/usb/net/rtl8150.c b/drivers/net/usb/rtl8150.c
index fa598f0340cf..fa598f0340cf 100644
--- a/drivers/usb/net/rtl8150.c
+++ b/drivers/net/usb/rtl8150.c
diff --git a/drivers/usb/net/usbnet.c b/drivers/net/usb/usbnet.c
index f9cd42d058b0..f9cd42d058b0 100644
--- a/drivers/usb/net/usbnet.c
+++ b/drivers/net/usb/usbnet.c
diff --git a/drivers/usb/net/usbnet.h b/drivers/net/usb/usbnet.h
index cbb53e065d6c..82db5a8e528e 100644
--- a/drivers/usb/net/usbnet.h
+++ b/drivers/net/usb/usbnet.h
@@ -129,7 +129,7 @@ extern void usbnet_disconnect(struct usb_interface *);
 
 
 /* Drivers that reuse some of the standard USB CDC infrastructure
- * (notably, using multiple interfaces according to the the CDC
+ * (notably, using multiple interfaces according to the CDC
  * union descriptor) get some helper code.
  */
 struct cdc_state {
diff --git a/drivers/usb/net/zaurus.c b/drivers/net/usb/zaurus.c
index 9f98e8ce487a..9f98e8ce487a 100644
--- a/drivers/usb/net/zaurus.c
+++ b/drivers/net/usb/zaurus.c
diff --git a/drivers/net/wireless/airport.c b/drivers/net/wireless/airport.c
index 38fac3bbcd82..7d5b8c2cc614 100644
--- a/drivers/net/wireless/airport.c
+++ b/drivers/net/wireless/airport.c
@@ -149,7 +149,7 @@ static int airport_hard_reset(struct orinoco_private *priv)
 	/* Vitally important.  If we don't do this it seems we get an
 	 * interrupt somewhere during the power cycle, since
 	 * hw_unavailable is already set it doesn't get ACKed, we get
-	 * into an interrupt loop and the the PMU decides to turn us
+	 * into an interrupt loop and the PMU decides to turn us
 	 * off. */
 	disable_irq(dev->irq);
 
diff --git a/drivers/net/wireless/bcm43xx/bcm43xx.h b/drivers/net/wireless/bcm43xx/bcm43xx.h
index f8483c179e4c..10e07e865426 100644
--- a/drivers/net/wireless/bcm43xx/bcm43xx.h
+++ b/drivers/net/wireless/bcm43xx/bcm43xx.h
@@ -658,12 +658,6 @@ struct bcm43xx_pio {
 
 #define BCM43xx_MAX_80211_CORES		2
 
-#ifdef CONFIG_BCM947XX
-#define core_offset(bcm) (bcm)->current_core_offset
-#else
-#define core_offset(bcm) 0
-#endif
-
 /* Generic information about a core. */
 struct bcm43xx_coreinfo {
 	u8 available:1,
@@ -789,10 +783,6 @@ struct bcm43xx_private {
 
 	/* The currently active core. */
 	struct bcm43xx_coreinfo *current_core;
-#ifdef CONFIG_BCM947XX
-	/** current core memory offset */
-	u32 current_core_offset;
-#endif
 	struct bcm43xx_coreinfo *active_80211_core;
 	/* coreinfo structs for all possible cores follow.
 	 * Note that a core might not exist.
@@ -943,25 +933,25 @@ struct bcm43xx_lopair * bcm43xx_get_lopair(struct bcm43xx_phyinfo *phy,
 static inline
 u16 bcm43xx_read16(struct bcm43xx_private *bcm, u16 offset)
 {
-	return ioread16(bcm->mmio_addr + core_offset(bcm) + offset);
+	return ioread16(bcm->mmio_addr + offset);
 }
 
 static inline
 void bcm43xx_write16(struct bcm43xx_private *bcm, u16 offset, u16 value)
 {
-	iowrite16(value, bcm->mmio_addr + core_offset(bcm) + offset);
+	iowrite16(value, bcm->mmio_addr + offset);
 }
 
 static inline
 u32 bcm43xx_read32(struct bcm43xx_private *bcm, u16 offset)
 {
-	return ioread32(bcm->mmio_addr + core_offset(bcm) + offset);
+	return ioread32(bcm->mmio_addr + offset);
 }
 
 static inline
 void bcm43xx_write32(struct bcm43xx_private *bcm, u16 offset, u32 value)
 {
-	iowrite32(value, bcm->mmio_addr + core_offset(bcm) + offset);
+	iowrite32(value, bcm->mmio_addr + offset);
 }
 
 static inline
diff --git a/drivers/net/wireless/bcm43xx/bcm43xx_dma.c b/drivers/net/wireless/bcm43xx/bcm43xx_dma.c
index e3d2e61a31ee..1f7731fcfbd5 100644
--- a/drivers/net/wireless/bcm43xx/bcm43xx_dma.c
+++ b/drivers/net/wireless/bcm43xx/bcm43xx_dma.c
@@ -660,10 +660,6 @@ struct bcm43xx_dmaring * bcm43xx_setup_dmaring(struct bcm43xx_private *bcm,
 	ring->routing = BCM43xx_DMA32_CLIENTTRANS;
 	if (dma64)
 		ring->routing = BCM43xx_DMA64_CLIENTTRANS;
-#ifdef CONFIG_BCM947XX
-	if (bcm->pci_dev->bus->number == 0)
-		ring->routing = dma64 ? BCM43xx_DMA64_NOTRANS : BCM43xx_DMA32_NOTRANS;
-#endif
 
 	ring->bcm = bcm;
 	ring->nr_slots = nr_slots;
diff --git a/drivers/net/wireless/bcm43xx/bcm43xx_main.c b/drivers/net/wireless/bcm43xx/bcm43xx_main.c
index 5e96bca6730a..ef6b253a92ce 100644
--- a/drivers/net/wireless/bcm43xx/bcm43xx_main.c
+++ b/drivers/net/wireless/bcm43xx/bcm43xx_main.c
@@ -61,10 +61,6 @@ MODULE_AUTHOR("Stefano Brivio");
 MODULE_AUTHOR("Michael Buesch");
 MODULE_LICENSE("GPL");
 
-#ifdef CONFIG_BCM947XX
-extern char *nvram_get(char *name);
-#endif
-
 #if defined(CONFIG_BCM43XX_DMA) && defined(CONFIG_BCM43XX_PIO)
 static int modparam_pio;
 module_param_named(pio, modparam_pio, int, 0444);
@@ -142,10 +138,6 @@ MODULE_PARM_DESC(fwpostfix, "Postfix for .fw files. Useful for using multiple fi
 	{ PCI_VENDOR_ID_BROADCOM, 0x4324, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0 },
 	/* Broadcom 43XG 802.11b/g */
 	{ PCI_VENDOR_ID_BROADCOM, 0x4325, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0 },
-#ifdef CONFIG_BCM947XX
-	/* SB bus on BCM947xx */
-	{ PCI_VENDOR_ID_BROADCOM, 0x0800, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0 },
-#endif
 	{ 0 },
 };
 MODULE_DEVICE_TABLE(pci, bcm43xx_pci_tbl);
@@ -786,9 +778,6 @@ static int bcm43xx_sprom_extract(struct bcm43xx_private *bcm)
 {
 	u16 value;
 	u16 *sprom;
-#ifdef CONFIG_BCM947XX
-	char *c;
-#endif
 
 	sprom = kzalloc(BCM43xx_SPROM_SIZE * sizeof(u16),
 			GFP_KERNEL);
@@ -796,28 +785,7 @@ static int bcm43xx_sprom_extract(struct bcm43xx_private *bcm)
 		printk(KERN_ERR PFX "sprom_extract OOM\n");
 		return -ENOMEM;
 	}
-#ifdef CONFIG_BCM947XX
-	sprom[BCM43xx_SPROM_BOARDFLAGS2] = atoi(nvram_get("boardflags2"));
-	sprom[BCM43xx_SPROM_BOARDFLAGS] = atoi(nvram_get("boardflags"));
-
-	if ((c = nvram_get("il0macaddr")) != NULL)
-		e_aton(c, (char *) &(sprom[BCM43xx_SPROM_IL0MACADDR]));
-
-	if ((c = nvram_get("et1macaddr")) != NULL)
-		e_aton(c, (char *) &(sprom[BCM43xx_SPROM_ET1MACADDR]));
-
-	sprom[BCM43xx_SPROM_PA0B0] = atoi(nvram_get("pa0b0"));
-	sprom[BCM43xx_SPROM_PA0B1] = atoi(nvram_get("pa0b1"));
-	sprom[BCM43xx_SPROM_PA0B2] = atoi(nvram_get("pa0b2"));
-
-	sprom[BCM43xx_SPROM_PA1B0] = atoi(nvram_get("pa1b0"));
-	sprom[BCM43xx_SPROM_PA1B1] = atoi(nvram_get("pa1b1"));
-	sprom[BCM43xx_SPROM_PA1B2] = atoi(nvram_get("pa1b2"));
-
-	sprom[BCM43xx_SPROM_BOARDREV] = atoi(nvram_get("boardrev"));
-#else
 	bcm43xx_sprom_read(bcm, sprom);
-#endif
 
 	/* boardflags2 */
 	value = sprom[BCM43xx_SPROM_BOARDFLAGS2];
@@ -1225,12 +1193,6 @@ static int _switch_core(struct bcm43xx_private *bcm, int core)
 			goto error;
 		udelay(10);
 	}
-#ifdef CONFIG_BCM947XX
-	if (bcm->pci_dev->bus->number == 0)
-		bcm->current_core_offset = 0x1000 * core;
-	else
-		bcm->current_core_offset = 0;
-#endif
 
 	return 0;
 error:
@@ -1387,19 +1349,6 @@ void bcm43xx_wireless_core_reset(struct bcm43xx_private *bcm, int connect_phy)
 
 	if ((bcm43xx_core_enabled(bcm)) &&
 	    !bcm43xx_using_pio(bcm)) {
-//FIXME: Do we _really_ want #ifndef CONFIG_BCM947XX here?
-#if 0
-#ifndef CONFIG_BCM947XX
-		/* reset all used DMA controllers. */
-		bcm43xx_dmacontroller_tx_reset(bcm, BCM43xx_MMIO_DMA1_BASE);
-		bcm43xx_dmacontroller_tx_reset(bcm, BCM43xx_MMIO_DMA2_BASE);
-		bcm43xx_dmacontroller_tx_reset(bcm, BCM43xx_MMIO_DMA3_BASE);
-		bcm43xx_dmacontroller_tx_reset(bcm, BCM43xx_MMIO_DMA4_BASE);
-		bcm43xx_dmacontroller_rx_reset(bcm, BCM43xx_MMIO_DMA1_BASE);
-		if (bcm->current_core->rev < 5)
-			bcm43xx_dmacontroller_rx_reset(bcm, BCM43xx_MMIO_DMA4_BASE);
-#endif
-#endif
 	}
 	if (bcm43xx_status(bcm) == BCM43xx_STAT_SHUTTINGDOWN) {
 		bcm43xx_write32(bcm, BCM43xx_MMIO_STATUS_BITFIELD,
@@ -2140,32 +2089,11 @@ out:
 	return err;
 }
 
-#ifdef CONFIG_BCM947XX
-static struct pci_device_id bcm43xx_47xx_ids[] = {
-	{ PCI_DEVICE(PCI_VENDOR_ID_BROADCOM, 0x4324) },
-	{ 0 }
-};
-#endif
-
 static int bcm43xx_initialize_irq(struct bcm43xx_private *bcm)
 {
 	int err;
 
 	bcm->irq = bcm->pci_dev->irq;
-#ifdef CONFIG_BCM947XX
-	if (bcm->pci_dev->bus->number == 0) {
-		struct pci_dev *d;
-		struct pci_device_id *id;
-		for (id = bcm43xx_47xx_ids; id->vendor; id++) {
-			d = pci_get_device(id->vendor, id->device, NULL);
-			if (d != NULL) {
-				bcm->irq = d->irq;
-				pci_dev_put(d);
-				break;
-			}
-		}
-	}
-#endif
 	err = request_irq(bcm->irq, bcm43xx_interrupt_handler,
 			  IRQF_SHARED, KBUILD_MODNAME, bcm);
 	if (err)
@@ -2645,10 +2573,6 @@ static int bcm43xx_probe_cores(struct bcm43xx_private *bcm)
 			chip_id_16 = 0x4610;
 		else if ((pci_device >= 0x4710) && (pci_device <= 0x4715))
 			chip_id_16 = 0x4710;
-#ifdef CONFIG_BCM947XX
-		else if ((pci_device >= 0x4320) && (pci_device <= 0x4325))
-			chip_id_16 = 0x4309;
-#endif
 		else {
 			printk(KERN_ERR PFX "Could not determine Chip ID\n");
 			return -ENODEV;
@@ -4144,11 +4068,6 @@ static int __devinit bcm43xx_init_one(struct pci_dev *pdev,
 	struct bcm43xx_private *bcm;
 	int err;
 
-#ifdef CONFIG_BCM947XX
-	if ((pdev->bus->number == 0) && (pdev->device != 0x0800))
-		return -ENODEV;
-#endif
-
 #ifdef DEBUG_SINGLE_DEVICE_ONLY
 	if (strcmp(pci_name(pdev), DEBUG_SINGLE_DEVICE_ONLY))
 		return -ENODEV;
diff --git a/drivers/net/wireless/bcm43xx/bcm43xx_main.h b/drivers/net/wireless/bcm43xx/bcm43xx_main.h
index f76357178e4d..c8f3c532bab5 100644
--- a/drivers/net/wireless/bcm43xx/bcm43xx_main.h
+++ b/drivers/net/wireless/bcm43xx/bcm43xx_main.h
@@ -33,25 +33,6 @@
 
 #include "bcm43xx.h"
 
-#ifdef CONFIG_BCM947XX
-#define atoi(str) simple_strtoul(((str != NULL) ? str : ""), NULL, 0)
-
-static inline void e_aton(char *str, char *dest)
-{
-	int i = 0;
-	u16 *d = (u16 *) dest;
-
-	for (;;) {
-		dest[i++] = (char) simple_strtoul(str, NULL, 16);
-		str += 2;
-		if (!*str++ || i == 6)
-			break;
-	}
-	for (i = 0; i < 3; i++)
-		d[i] = cpu_to_be16(d[i]);
-}
-#endif
-
 #define P4D_BYT3S(magic, nr_bytes)	u8 __p4dding##magic[nr_bytes]
 #define P4D_BYTES(line, nr_bytes)	P4D_BYT3S(line, nr_bytes)
 /* Magic helper macro to pad structures. Ignore those above. It's magic. */
diff --git a/drivers/net/wireless/prism54/isl_ioctl.c b/drivers/net/wireless/prism54/isl_ioctl.c
index 841b3c136ad9..283be4a70524 100644
--- a/drivers/net/wireless/prism54/isl_ioctl.c
+++ b/drivers/net/wireless/prism54/isl_ioctl.c
@@ -3054,7 +3054,7 @@ static const iw_handler prism54_handler[] = {
 	(iw_handler) prism54_set_wap,	/* SIOCSIWAP */
 	(iw_handler) prism54_get_wap,	/* SIOCGIWAP */
 	(iw_handler) NULL,	/* -- hole -- */
-	(iw_handler) NULL,	/* SIOCGIWAPLIST depreciated */
+	(iw_handler) NULL,	/* SIOCGIWAPLIST deprecated */
 	(iw_handler) prism54_set_scan,	/* SIOCSIWSCAN */
 	(iw_handler) prism54_get_scan,	/* SIOCGIWSCAN */
 	(iw_handler) prism54_set_essid,	/* SIOCSIWESSID */
diff --git a/drivers/net/wireless/prism54/islpci_dev.c b/drivers/net/wireless/prism54/islpci_dev.c
index a037b11dac9d..084795355b74 100644
--- a/drivers/net/wireless/prism54/islpci_dev.c
+++ b/drivers/net/wireless/prism54/islpci_dev.c
@@ -115,7 +115,7 @@ isl_upload_firmware(islpci_private *priv)
 			    ISL38XX_MEMORY_WINDOW_SIZE : fw_len;
 			u32 __iomem *dev_fw_ptr = device_base + ISL38XX_DIRECT_MEM_WIN;
 
-			/* set the cards base address for writting the data */
+			/* set the card's base address for writing the data */
 			isl38xx_w32_flush(device_base, reg,
 					  ISL38XX_DIR_MEM_BASE_REG);
 			wmb();	/* be paranoid */
diff --git a/drivers/net/wireless/wavelan_cs.c b/drivers/net/wireless/wavelan_cs.c
index 67b867f837ca..5740d4d4267c 100644
--- a/drivers/net/wireless/wavelan_cs.c
+++ b/drivers/net/wireless/wavelan_cs.c
@@ -176,7 +176,7 @@ psa_write(struct net_device *	dev,
   volatile u_char __iomem *verify = lp->mem + PSA_ADDR +
     (psaoff(0, psa_comp_number) << 1);
 
-  /* Authorize writting to PSA */
+  /* Authorize writing to PSA */
   hacr_write(base, HACR_PWR_STAT | HACR_ROM_WEN);
 
   while(n-- > 0)
@@ -1676,7 +1676,7 @@ wv_set_frequency(u_long		base,	/* i/o port of the card */
       fee_write(base, 0x60,
 		dac, 2);
 
-      /* We now should verify here that the EEprom writting was ok */
+      /* We now should verify here that the EEprom writing was ok */
 
       /* ReRead the first area */
       fee_read(base, 0x00,
diff --git a/drivers/net/wireless/wavelan_cs.p.h b/drivers/net/wireless/wavelan_cs.p.h
index 4d1c4905c749..4b9de0093a7b 100644
--- a/drivers/net/wireless/wavelan_cs.p.h
+++ b/drivers/net/wireless/wavelan_cs.p.h
@@ -120,7 +120,7 @@
  * the Wavelan itself (NCR -> AT&T -> Lucent).
  *
  * All started with Anders Klemets <klemets@paul.rutgers.edu>,
- * writting a Wavelan ISA driver for the MACH microkernel. Girish
+ * writing a Wavelan ISA driver for the MACH microkernel. Girish
  * Welling <welling@paul.rutgers.edu> had also worked on it.
  * Keith Moore modify this for the Pcmcia hardware.
  * 
diff --git a/drivers/net/wireless/zd1211rw/zd_usb.c b/drivers/net/wireless/zd1211rw/zd_usb.c
index e04cffc8adf3..8459549d0cee 100644
--- a/drivers/net/wireless/zd1211rw/zd_usb.c
+++ b/drivers/net/wireless/zd1211rw/zd_usb.c
@@ -40,6 +40,7 @@ static struct usb_device_id usb_ids[] = {
 	{ USB_DEVICE(0x126f, 0xa006), .driver_info = DEVICE_ZD1211 },
 	{ USB_DEVICE(0x6891, 0xa727), .driver_info = DEVICE_ZD1211 },
 	{ USB_DEVICE(0x0df6, 0x9071), .driver_info = DEVICE_ZD1211 },
+	{ USB_DEVICE(0x0df6, 0x9075), .driver_info = DEVICE_ZD1211 },
 	{ USB_DEVICE(0x157e, 0x300b), .driver_info = DEVICE_ZD1211 },
 	{ USB_DEVICE(0x079b, 0x004a), .driver_info = DEVICE_ZD1211 },
 	{ USB_DEVICE(0x1740, 0x2000), .driver_info = DEVICE_ZD1211 },
@@ -67,8 +68,11 @@ static struct usb_device_id usb_ids[] = {
 	{ USB_DEVICE(0x0586, 0x3410), .driver_info = DEVICE_ZD1211B },
 	{ USB_DEVICE(0x0baf, 0x0121), .driver_info = DEVICE_ZD1211B },
 	{ USB_DEVICE(0x0586, 0x3412), .driver_info = DEVICE_ZD1211B },
+	{ USB_DEVICE(0x0586, 0x3413), .driver_info = DEVICE_ZD1211B },
+	{ USB_DEVICE(0x0053, 0x5301), .driver_info = DEVICE_ZD1211B },
 	/* "Driverless" devices that need ejecting */
 	{ USB_DEVICE(0x0ace, 0x2011), .driver_info = DEVICE_INSTALLER },
+	{ USB_DEVICE(0x0ace, 0x20ff), .driver_info = DEVICE_INSTALLER },
 	{}
 };
 
diff --git a/drivers/net/yellowfin.c b/drivers/net/yellowfin.c
index 3f4a7cf9efea..f2a90a7fa2d6 100644
--- a/drivers/net/yellowfin.c
+++ b/drivers/net/yellowfin.c
@@ -109,7 +109,6 @@ static int gx_fix;
 /* These identify the driver base version and may not be removed. */
 static char version[] __devinitdata =
 KERN_INFO DRV_NAME ".c:v1.05  1/09/2001  Written by Donald Becker <becker@scyld.com>\n"
-KERN_INFO "  http://www.scyld.com/network/yellowfin.html\n"
 KERN_INFO "  (unofficial 2.4.x port, " DRV_VERSION ", " DRV_RELDATE ")\n";
 
 MODULE_AUTHOR("Donald Becker <becker@scyld.com>");
diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
index 3bb7739d26a5..8e58ea3d95c0 100644
--- a/drivers/pci/pci-driver.c
+++ b/drivers/pci/pci-driver.c
@@ -119,7 +119,7 @@ static inline int pci_create_newid_file(struct pci_driver *drv)
  * system is in its list of supported devices.  Returns the matching
  * pci_device_id structure or %NULL if there is no match.
  *
- * Depreciated, don't use this as it will not catch any dynamic ids
+ * Deprecated, don't use this as it will not catch any dynamic ids
  * that a driver might want to check for.
  */
 const struct pci_device_id *pci_match_id(const struct pci_device_id *ids,
diff --git a/drivers/rtc/Kconfig b/drivers/rtc/Kconfig
index 5e439836db2d..76422eded36e 100644
--- a/drivers/rtc/Kconfig
+++ b/drivers/rtc/Kconfig
@@ -98,7 +98,7 @@ config RTC_INTF_DEV_UIE_EMUL
 	bool "RTC UIE emulation on dev interface"
 	depends on RTC_INTF_DEV
 	help
-	  Provides an emulation for RTC_UIE if the underlaying rtc chip
+	  Provides an emulation for RTC_UIE if the underlying rtc chip
 	  driver does not expose RTC_UIE ioctls.  Those requests generate
 	  once-per-second update interrupts, used for synchronization.
 
diff --git a/drivers/rtc/rtc-sh.c b/drivers/rtc/rtc-sh.c
index 6abf4811958c..e0f91dfce0f5 100644
--- a/drivers/rtc/rtc-sh.c
+++ b/drivers/rtc/rtc-sh.c
@@ -104,7 +104,7 @@ static irqreturn_t sh_rtc_interrupt(int irq, void *dev_id)
 
 	writeb(tmp, rtc->regbase + RCR1);
 
-	rtc_update_irq(&rtc->rtc_dev, 1, events);
+	rtc_update_irq(rtc->rtc_dev, 1, events);
 
 	spin_unlock(&rtc->lock);
 
@@ -139,7 +139,7 @@ static irqreturn_t sh_rtc_alarm(int irq, void *dev_id)
 
 		rtc->rearm_aie = 1;
 
-		rtc_update_irq(&rtc->rtc_dev, 1, events);
+		rtc_update_irq(rtc->rtc_dev, 1, events);
 	}
 
 	spin_unlock(&rtc->lock);
@@ -153,7 +153,7 @@ static irqreturn_t sh_rtc_periodic(int irq, void *dev_id)
 
 	spin_lock(&rtc->lock);
 
-	rtc_update_irq(&rtc->rtc_dev, 1, RTC_PF | RTC_IRQF);
+	rtc_update_irq(rtc->rtc_dev, 1, RTC_PF | RTC_IRQF);
 
 	spin_unlock(&rtc->lock);
 
@@ -341,7 +341,7 @@ static int sh_rtc_read_time(struct device *dev, struct rtc_time *tm)
 		tm->tm_sec--;
 #endif
 
-	dev_dbg(&dev, "%s: tm is secs=%d, mins=%d, hours=%d, "
+	dev_dbg(dev, "%s: tm is secs=%d, mins=%d, hours=%d, "
 		"mday=%d, mon=%d, year=%d, wday=%d\n",
 		__FUNCTION__,
 		tm->tm_sec, tm->tm_min, tm->tm_hour,
diff --git a/drivers/s390/char/sclp_rw.c b/drivers/s390/char/sclp_rw.c
index bbd5b8b66f42..d6b06ab81188 100644
--- a/drivers/s390/char/sclp_rw.c
+++ b/drivers/s390/char/sclp_rw.c
@@ -23,7 +23,7 @@
 
 /*
  * The room for the SCCB (only for writing) is not equal to a pages size
- * (as it is specified as the maximum size in the the SCLP documentation)
+ * (as it is specified as the maximum size in the SCLP documentation)
  * because of the additional data structure described above.
  */
 #define MAX_SCCB_ROOM (PAGE_SIZE - sizeof(struct sclp_buffer))
diff --git a/drivers/s390/net/qeth_main.c b/drivers/s390/net/qeth_main.c
index 29d176036e5c..0b96d49dd636 100644
--- a/drivers/s390/net/qeth_main.c
+++ b/drivers/s390/net/qeth_main.c
@@ -2860,7 +2860,7 @@ qeth_flush_buffers(struct qeth_qdio_out_q *queue, int under_int,
 			if (!atomic_read(&queue->set_pci_flags_count)){
 				/*
 				 * there's no outstanding PCI any more, so we
-				 * have to request a PCI to be sure the the PCI
+				 * have to request a PCI to be sure that the PCI
 				 * will wake at some time in the future then we
 				 * can flush packed buffers that might still be
 				 * hanging around, which can happen if no
diff --git a/drivers/s390/scsi/zfcp_qdio.c b/drivers/s390/scsi/zfcp_qdio.c
index cb08ca3cc0f9..bdf5782b8a7a 100644
--- a/drivers/s390/scsi/zfcp_qdio.c
+++ b/drivers/s390/scsi/zfcp_qdio.c
@@ -222,7 +222,7 @@ zfcp_qdio_handler_error_check(struct zfcp_adapter *adapter, unsigned int status,
                 * Since we have been using this adapter, it is save to assume
                 * that it is not failed but recoverable. The card seems to
                 * report link-up events by self-initiated queue shutdown.
-                * That is why we need to clear the the link-down flag
+                * That is why we need to clear the link-down flag
                 * which is set again in case we have missed by a mile.
                 */
                zfcp_erp_adapter_reopen(
diff --git a/drivers/sbus/char/bpp.c b/drivers/sbus/char/bpp.c
index 74b999d77bbf..4fab0c23814c 100644
--- a/drivers/sbus/char/bpp.c
+++ b/drivers/sbus/char/bpp.c
@@ -156,7 +156,7 @@ static unsigned short get_pins(unsigned minor)
 #define BPP_ICR      0x18
 #define BPP_SIZE     0x1A
 
-/* BPP_CSR.  Bits of type RW1 are cleared with writting '1'. */
+/* BPP_CSR.  Bits of type RW1 are cleared with writing '1'. */
 #define P_DEV_ID_MASK   0xf0000000      /* R   */
 #define P_DEV_ID_ZEBRA  0x40000000
 #define P_DEV_ID_L64854 0xa0000000      /*      == NCR 89C100+89C105. Pity. */
diff --git a/drivers/scsi/aic7xxx/aic79xx_pci.c b/drivers/scsi/aic7xxx/aic79xx_pci.c
index 8d72bbae96ad..0bada0028aa0 100644
--- a/drivers/scsi/aic7xxx/aic79xx_pci.c
+++ b/drivers/scsi/aic7xxx/aic79xx_pci.c
@@ -966,7 +966,7 @@ ahd_aic790X_setup(struct ahd_softc *ahd)
 			      |  AHD_BUSFREEREV_BUG;
 		ahd->bugs |= AHD_LQOOVERRUN_BUG|AHD_EARLY_REQ_BUG;
 
-		/* If the user requested the the SLOWCRC bit to be set. */
+		/* If the user requested that the SLOWCRC bit to be set. */
 		if (aic79xx_slowcrc)
 			ahd->features |= AHD_AIC79XXB_SLOWCRC;
 
diff --git a/drivers/scsi/aic94xx/Makefile b/drivers/scsi/aic94xx/Makefile
index e6b70123940c..e78ce0fa44d2 100644
--- a/drivers/scsi/aic94xx/Makefile
+++ b/drivers/scsi/aic94xx/Makefile
@@ -6,7 +6,7 @@
 #
 # This file is licensed under GPLv2.
 #
-# This file is part of the the aic94xx driver.
+# This file is part of the aic94xx driver.
 #
 # The aic94xx driver is free software; you can redistribute it and/or
 # modify it under the terms of the GNU General Public License as
diff --git a/drivers/scsi/dc395x.c b/drivers/scsi/dc395x.c
index a965ed3548d5..564ea90ed3a0 100644
--- a/drivers/scsi/dc395x.c
+++ b/drivers/scsi/dc395x.c
@@ -541,7 +541,7 @@ static struct ParameterData __devinitdata cfg_data[] = {
 
 
 /*
- * Safe settings. If set to zero the the BIOS/default values with
+ * Safe settings. If set to zero the BIOS/default values with
  * command line overrides will be used. If set to 1 then safe and
  * slow settings will be used.
  */
@@ -617,7 +617,7 @@ static void __devinit fix_settings(void)
 
 /*
  * Mapping from the eeprom delay index value (index into this array)
- * to the the number of actual seconds that the delay should be for.
+ * to the number of actual seconds that the delay should be for.
  */
 static char __devinitdata eeprom_index_to_delay_map[] = 
 	{ 1, 3, 5, 10, 16, 30, 60, 120 };
@@ -4136,7 +4136,7 @@ static void __devinit trms1040_write_all(struct NvRamType *eeprom, unsigned long
  * @io_port:	base I/O address
  * @addr:	offset into SEEPROM
  *
- * Returns the the byte read.
+ * Returns the byte read.
  **/
 static u8 __devinit trms1040_get_data(unsigned long io_port, u8 addr)
 {
diff --git a/drivers/scsi/ide-scsi.c b/drivers/scsi/ide-scsi.c
index 2b5b8a93bc10..8263f752809d 100644
--- a/drivers/scsi/ide-scsi.c
+++ b/drivers/scsi/ide-scsi.c
@@ -721,19 +721,23 @@ static ide_startstop_t idescsi_do_request (ide_drive_t *drive, struct request *r
 	return ide_stopped;
 }
 
+#ifdef CONFIG_IDE_PROC_FS
 static void idescsi_add_settings(ide_drive_t *drive)
 {
 	idescsi_scsi_t *scsi = drive_to_idescsi(drive);
 
 /*
- *			drive	setting name	read/write	ioctl	ioctl		data type	min	max	mul_factor	div_factor	data pointer		set function
+ *			drive	setting name	read/write	data type	min	max	mul_factor	div_factor	data pointer		set function
  */
-	ide_add_setting(drive,	"bios_cyl",	SETTING_RW,	-1,	-1,		TYPE_INT,	0,	1023,	1,		1,		&drive->bios_cyl,	NULL);
-	ide_add_setting(drive,	"bios_head",	SETTING_RW,	-1,	-1,		TYPE_BYTE,	0,	255,	1,		1,		&drive->bios_head,	NULL);
-	ide_add_setting(drive,	"bios_sect",	SETTING_RW,	-1,	-1,		TYPE_BYTE,	0,	63,	1,		1,		&drive->bios_sect,	NULL);
-	ide_add_setting(drive,	"transform",	SETTING_RW,	-1,	-1,		TYPE_INT,	0,	3,	1,		1,		&scsi->transform,	NULL);
-	ide_add_setting(drive,	"log",		SETTING_RW,	-1,	-1,		TYPE_INT,	0,	1,	1,		1,		&scsi->log,		NULL);
+	ide_add_setting(drive,	"bios_cyl",	SETTING_RW,	TYPE_INT,	0,	1023,	1,		1,		&drive->bios_cyl,	NULL);
+	ide_add_setting(drive,	"bios_head",	SETTING_RW,	TYPE_BYTE,	0,	255,	1,		1,		&drive->bios_head,	NULL);
+	ide_add_setting(drive,	"bios_sect",	SETTING_RW,	TYPE_BYTE,	0,	63,	1,		1,		&drive->bios_sect,	NULL);
+	ide_add_setting(drive,	"transform",	SETTING_RW,	TYPE_INT,	0,	3,	1,		1,		&scsi->transform,	NULL);
+	ide_add_setting(drive,	"log",		SETTING_RW,	TYPE_INT,	0,	1,	1,		1,		&scsi->log,		NULL);
 }
+#else
+static inline void idescsi_add_settings(ide_drive_t *drive) { ; }
+#endif
 
 /*
  *	Driver initialization.
@@ -756,7 +760,7 @@ static void ide_scsi_remove(ide_drive_t *drive)
 	struct ide_scsi_obj *scsi = scsihost_to_idescsi(scsihost);
 	struct gendisk *g = scsi->disk;
 
-	ide_unregister_subdriver(drive, scsi->driver);
+	ide_proc_unregister_driver(drive, scsi->driver);
 
 	ide_unregister_region(g);
 
@@ -770,13 +774,11 @@ static void ide_scsi_remove(ide_drive_t *drive)
 
 static int ide_scsi_probe(ide_drive_t *);
 
-#ifdef CONFIG_PROC_FS
+#ifdef CONFIG_IDE_PROC_FS
 static ide_proc_entry_t idescsi_proc[] = {
 	{ "capacity", S_IFREG|S_IRUGO, proc_ide_read_capacity, NULL },
 	{ NULL, 0, NULL, NULL }
 };
-#else
-# define idescsi_proc	NULL
 #endif
 
 static ide_driver_t idescsi_driver = {
@@ -790,11 +792,13 @@ static ide_driver_t idescsi_driver = {
 	.version		= IDESCSI_VERSION,
 	.media			= ide_scsi,
 	.supports_dsc_overlap	= 0,
-	.proc			= idescsi_proc,
 	.do_request		= idescsi_do_request,
 	.end_request		= idescsi_end_request,
 	.error                  = idescsi_atapi_error,
 	.abort                  = idescsi_atapi_abort,
+#ifdef CONFIG_IDE_PROC_FS
+	.proc			= idescsi_proc,
+#endif
 };
 
 static int idescsi_ide_open(struct inode *inode, struct file *filp)
@@ -1153,7 +1157,7 @@ static int ide_scsi_probe(ide_drive_t *drive)
 	idescsi->host = host;
 	idescsi->disk = g;
 	g->private_data = &idescsi->driver;
-	ide_register_subdriver(drive, &idescsi_driver);
+	ide_proc_register_driver(drive, &idescsi_driver);
 	err = 0;
 	idescsi_setup(drive, idescsi);
 	g->fops = &idescsi_ops;
@@ -1165,7 +1169,7 @@ static int ide_scsi_probe(ide_drive_t *drive)
 	}
 	/* fall through on error */
 	ide_unregister_region(g);
-	ide_unregister_subdriver(drive, &idescsi_driver);
+	ide_proc_unregister_driver(drive, &idescsi_driver);
 
 	put_disk(g);
 out_host_put:
diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c
index 61fbcdcbb009..1f5a07bf2a75 100644
--- a/drivers/scsi/scsi_lib.c
+++ b/drivers/scsi/scsi_lib.c
@@ -173,7 +173,7 @@ int scsi_queue_insert(struct scsi_cmnd *cmd, int reason)
  * @retries:	number of times to retry request
  * @flags:	or into request flags;
  *
- * returns the req->errors value which is the the scsi_cmnd result
+ * returns the req->errors value which is the scsi_cmnd result
  * field.
  **/
 int scsi_execute(struct scsi_device *sdev, const unsigned char *cmd,
diff --git a/drivers/spi/atmel_spi.c b/drivers/spi/atmel_spi.c
index 66e7bc985797..1d8a2f6bb8eb 100644
--- a/drivers/spi/atmel_spi.c
+++ b/drivers/spi/atmel_spi.c
@@ -22,10 +22,7 @@
 #include <asm/io.h>
 #include <asm/arch/board.h>
 #include <asm/arch/gpio.h>
-
-#ifdef CONFIG_ARCH_AT91
 #include <asm/arch/cpu.h>
-#endif
 
 #include "atmel_spi.h"
 
@@ -552,10 +549,8 @@ static int __init atmel_spi_probe(struct platform_device *pdev)
 		goto out_free_buffer;
 	as->irq = irq;
 	as->clk = clk;
-#ifdef CONFIG_ARCH_AT91
 	if (!cpu_is_at91rm9200())
 		as->new_1 = 1;
-#endif
 
 	ret = request_irq(irq, atmel_spi_interrupt, 0,
 			pdev->dev.bus_id, master);
diff --git a/drivers/usb/Kconfig b/drivers/usb/Kconfig
index b847bbc8b0e1..278a22cea5bf 100644
--- a/drivers/usb/Kconfig
+++ b/drivers/usb/Kconfig
@@ -87,8 +87,6 @@ source "drivers/usb/storage/Kconfig"
 
 source "drivers/usb/image/Kconfig"
 
-source "drivers/usb/net/Kconfig"
-
 source "drivers/usb/mon/Kconfig"
 
 comment "USB port drivers"
diff --git a/drivers/usb/Makefile b/drivers/usb/Makefile
index 0ef090b1b37c..72464b586990 100644
--- a/drivers/usb/Makefile
+++ b/drivers/usb/Makefile
@@ -23,13 +23,6 @@ obj-$(CONFIG_USB_PRINTER)	+= class/
 obj-$(CONFIG_USB_STORAGE)	+= storage/
 obj-$(CONFIG_USB)		+= storage/
 
-obj-$(CONFIG_USB_CATC)		+= net/
-obj-$(CONFIG_USB_KAWETH)	+= net/
-obj-$(CONFIG_USB_PEGASUS)	+= net/
-obj-$(CONFIG_USB_RTL8150)	+= net/
-obj-$(CONFIG_USB_USBNET)	+= net/
-obj-$(CONFIG_USB_ZD1201)	+= net/
-
 obj-$(CONFIG_USB_MDC800)	+= image/
 obj-$(CONFIG_USB_MICROTEK)	+= image/
 
diff --git a/drivers/usb/atm/usbatm.c b/drivers/usb/atm/usbatm.c
index b082d95bbbaa..11e9b15ca45a 100644
--- a/drivers/usb/atm/usbatm.c
+++ b/drivers/usb/atm/usbatm.c
@@ -1033,7 +1033,7 @@ static int usbatm_do_heavy_init(void *arg)
 
 static int usbatm_heavy_init(struct usbatm_data *instance)
 {
-	int ret = kernel_thread(usbatm_do_heavy_init, instance, CLONE_KERNEL);
+	int ret = kernel_thread(usbatm_do_heavy_init, instance, CLONE_FS | CLONE_FILES);
 
 	if (ret < 0) {
 		usb_err(instance, "%s: failed to create kernel_thread (%d)!\n", __func__, ret);
diff --git a/drivers/usb/misc/auerswald.c b/drivers/usb/misc/auerswald.c
index b5332e679c46..88fb56d5db8f 100644
--- a/drivers/usb/misc/auerswald.c
+++ b/drivers/usb/misc/auerswald.c
@@ -1307,7 +1307,7 @@ static int auerswald_addservice (pauerswald_t cp, pauerscon_t scp)
 }
 
 
-/* remove a service from the the device
+/* remove a service from the device
    scp->id must be set! */
 static void auerswald_removeservice (pauerswald_t cp, pauerscon_t scp)
 {
diff --git a/drivers/usb/serial/Kconfig b/drivers/usb/serial/Kconfig
index ba5d1dc03036..3efe67092f15 100644
--- a/drivers/usb/serial/Kconfig
+++ b/drivers/usb/serial/Kconfig
@@ -558,7 +558,7 @@ config USB_SERIAL_DEBUG
 	tristate "USB Debugging Device"
 	depends on USB_SERIAL
 	help
-	  Say Y here if you have a USB debugging device used to recieve
+	  Say Y here if you have a USB debugging device used to receive
 	  debugging data from another machine.  The most common of these
 	  devices is the NetChip TurboCONNECT device.
 
diff --git a/drivers/usb/serial/aircable.c b/drivers/usb/serial/aircable.c
index b675735bfbee..fbc8c27d5d99 100644
--- a/drivers/usb/serial/aircable.c
+++ b/drivers/usb/serial/aircable.c
@@ -9,7 +9,7 @@
  * The device works as an standard CDC device, it has 2 interfaces, the first
  * one is for firmware access and the second is the serial one.
  * The protocol is very simply, there are two posibilities reading or writing.
- * When writting the first urb must have a Header that starts with 0x20 0x29 the
+ * When writing the first urb must have a Header that starts with 0x20 0x29 the
  * next two bytes must say how much data will be sended.
  * When reading the process is almost equal except that the header starts with
  * 0x00 0x20.
@@ -18,7 +18,7 @@
  * buffer: The First and Second byte is used for a Header, the Third and Fourth
  * tells the  device the amount of information the package holds.
  * Packages are 60 bytes long Header Stuff.
- * When writting to the device the first two bytes of the header are 0x20 0x29
+ * When writing to the device the first two bytes of the header are 0x20 0x29
  * When reading the bytes are 0x00 0x20, or 0x00 0x10, there is an strange
  * situation, when too much data arrives to the device because it sends the data
  * but with out the header. I will use a simply hack to override this situation,
diff --git a/drivers/usb/serial/io_edgeport.c b/drivers/usb/serial/io_edgeport.c
index 18f74ac76565..4807f960150b 100644
--- a/drivers/usb/serial/io_edgeport.c
+++ b/drivers/usb/serial/io_edgeport.c
@@ -2465,7 +2465,7 @@ static int send_cmd_write_uart_register (struct edgeport_port *edge_port, __u8 r
 	    ((edge_serial->is_epic) &&
 	     (!edge_serial->epic_descriptor.Supports.IOSPWriteMCR) &&
 	     (regNum == MCR))) {
-		dbg("SendCmdWriteUartReg - Not writting to MCR Register");
+		dbg("SendCmdWriteUartReg - Not writing to MCR Register");
 		return 0;
 	}
 
@@ -2473,7 +2473,7 @@ static int send_cmd_write_uart_register (struct edgeport_port *edge_port, __u8 r
 	    ((edge_serial->is_epic) &&
 	     (!edge_serial->epic_descriptor.Supports.IOSPWriteLCR) &&
 	     (regNum == LCR))) {
-		dbg ("SendCmdWriteUartReg - Not writting to LCR Register");
+		dbg ("SendCmdWriteUartReg - Not writing to LCR Register");
 		return 0;
 	}
 
diff --git a/drivers/video/Kconfig b/drivers/video/Kconfig
index 1132ba5ff391..9a256d2ff9dc 100644
--- a/drivers/video/Kconfig
+++ b/drivers/video/Kconfig
@@ -1348,6 +1348,20 @@ config FB_VOODOO1
 	  Please read the <file:Documentation/fb/README-sstfb.txt> for supported
 	  options and other important info  support.
 
+config FB_VT8623
+	tristate "VIA VT8623 support"
+	depends on FB && PCI
+	select FB_CFB_FILLRECT
+	select FB_CFB_COPYAREA
+	select FB_CFB_IMAGEBLIT
+	select FB_TILEBLITTING
+	select FB_SVGALIB
+	select VGASTATE
+	select FONT_8x16 if FRAMEBUFFER_CONSOLE
+	---help---
+	  Driver for CastleRock integrated graphics core in the
+	  VIA VT8623 [Apollo CLE266] chipset.
+
 config FB_CYBLA
 	tristate "Cyberblade/i1 support"
 	depends on FB && PCI && X86_32 && !64BIT
@@ -1401,6 +1415,20 @@ config FB_TRIDENT_ACCEL
 	This will compile the Trident frame buffer device with
 	acceleration functions.
 
+config FB_ARK
+	tristate "ARK 2000PV support"
+	depends on FB && PCI
+	select FB_CFB_FILLRECT
+	select FB_CFB_COPYAREA
+	select FB_CFB_IMAGEBLIT
+	select FB_TILEBLITTING
+	select FB_SVGALIB
+	select VGASTATE
+	select FONT_8x16 if FRAMEBUFFER_CONSOLE
+	---help---
+	  Driver for PCI graphics boards with ARK 2000PV chip
+	  and ICS 5342 RAMDAC.
+
 config FB_PM3
 	tristate "Permedia3 support"
 	depends on FB && PCI && BROKEN
diff --git a/drivers/video/Makefile b/drivers/video/Makefile
index a916c204274f..0b70567458fb 100644
--- a/drivers/video/Makefile
+++ b/drivers/video/Makefile
@@ -54,10 +54,12 @@ obj-$(CONFIG_FB_VALKYRIE)         += valkyriefb.o
 obj-$(CONFIG_FB_CT65550)          += chipsfb.o
 obj-$(CONFIG_FB_IMSTT)            += imsttfb.o
 obj-$(CONFIG_FB_FM2)              += fm2fb.o
+obj-$(CONFIG_FB_VT8623)           += vt8623fb.o
 obj-$(CONFIG_FB_CYBLA)            += cyblafb.o
 obj-$(CONFIG_FB_TRIDENT)          += tridentfb.o
 obj-$(CONFIG_FB_LE80578)          += vermilion/
 obj-$(CONFIG_FB_S3)               += s3fb.o
+obj-$(CONFIG_FB_ARK)              += arkfb.o
 obj-$(CONFIG_FB_STI)              += stifb.o
 obj-$(CONFIG_FB_FFB)              += ffb.o sbuslib.o
 obj-$(CONFIG_FB_CG6)              += cg6.o sbuslib.o
diff --git a/drivers/video/arkfb.c b/drivers/video/arkfb.c
new file mode 100644
index 000000000000..ba6fede5c466
--- /dev/null
+++ b/drivers/video/arkfb.c
@@ -0,0 +1,1200 @@
+/*
+ *  linux/drivers/video/arkfb.c -- Frame buffer device driver for ARK 2000PV
+ *  with ICS 5342 dac (it is easy to add support for different dacs).
+ *
+ *  Copyright (c) 2007 Ondrej Zajicek <santiago@crfreenet.org>
+ *
+ *  This file is subject to the terms and conditions of the GNU General Public
+ *  License.  See the file COPYING in the main directory of this archive for
+ *  more details.
+ *
+ *  Code is based on s3fb
+ */
+
+#include <linux/version.h>
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/string.h>
+#include <linux/mm.h>
+#include <linux/tty.h>
+#include <linux/slab.h>
+#include <linux/delay.h>
+#include <linux/fb.h>
+#include <linux/svga.h>
+#include <linux/init.h>
+#include <linux/pci.h>
+#include <linux/console.h> /* Why should fb driver call console functions? because acquire_console_sem() */
+#include <video/vga.h>
+
+#ifdef CONFIG_MTRR
+#include <asm/mtrr.h>
+#endif
+
+struct arkfb_info {
+	int mclk_freq;
+	int mtrr_reg;
+
+	struct dac_info *dac;
+	struct vgastate state;
+	struct mutex open_lock;
+	unsigned int ref_count;
+	u32 pseudo_palette[16];
+};
+
+
+/* ------------------------------------------------------------------------- */
+
+
+static const struct svga_fb_format arkfb_formats[] = {
+	{ 0,  {0, 6, 0},  {0, 6, 0},  {0, 6, 0}, {0, 0, 0}, 0,
+		FB_TYPE_TEXT, FB_AUX_TEXT_SVGA_STEP4,	FB_VISUAL_PSEUDOCOLOR, 8, 8},
+	{ 4,  {0, 6, 0},  {0, 6, 0},  {0, 6, 0}, {0, 0, 0}, 0,
+		FB_TYPE_PACKED_PIXELS, 0,		FB_VISUAL_PSEUDOCOLOR, 8, 16},
+	{ 4,  {0, 6, 0},  {0, 6, 0},  {0, 6, 0}, {0, 0, 0}, 1,
+		FB_TYPE_INTERLEAVED_PLANES, 1,		FB_VISUAL_PSEUDOCOLOR, 8, 16},
+	{ 8,  {0, 6, 0},  {0, 6, 0},  {0, 6, 0}, {0, 0, 0}, 0,
+		FB_TYPE_PACKED_PIXELS, 0,		FB_VISUAL_PSEUDOCOLOR, 8, 8},
+	{16,  {10, 5, 0}, {5, 5, 0},  {0, 5, 0}, {0, 0, 0}, 0,
+		FB_TYPE_PACKED_PIXELS, 0,		FB_VISUAL_TRUECOLOR, 4, 4},
+	{16,  {11, 5, 0}, {5, 6, 0},  {0, 5, 0}, {0, 0, 0}, 0,
+		FB_TYPE_PACKED_PIXELS, 0,		FB_VISUAL_TRUECOLOR, 4, 4},
+	{24,  {16, 8, 0}, {8, 8, 0},  {0, 8, 0}, {0, 0, 0}, 0,
+		FB_TYPE_PACKED_PIXELS, 0,		FB_VISUAL_TRUECOLOR, 8, 8},
+	{32,  {16, 8, 0}, {8, 8, 0},  {0, 8, 0}, {0, 0, 0}, 0,
+		FB_TYPE_PACKED_PIXELS, 0,		FB_VISUAL_TRUECOLOR, 2, 2},
+	SVGA_FORMAT_END
+};
+
+
+/* CRT timing register sets */
+
+static const struct vga_regset ark_h_total_regs[]        = {{0x00, 0, 7}, {0x41, 7, 7}, VGA_REGSET_END};
+static const struct vga_regset ark_h_display_regs[]      = {{0x01, 0, 7}, {0x41, 6, 6}, VGA_REGSET_END};
+static const struct vga_regset ark_h_blank_start_regs[]  = {{0x02, 0, 7}, {0x41, 5, 5}, VGA_REGSET_END};
+static const struct vga_regset ark_h_blank_end_regs[]    = {{0x03, 0, 4}, {0x05, 7, 7	}, VGA_REGSET_END};
+static const struct vga_regset ark_h_sync_start_regs[]   = {{0x04, 0, 7}, {0x41, 4, 4}, VGA_REGSET_END};
+static const struct vga_regset ark_h_sync_end_regs[]     = {{0x05, 0, 4}, VGA_REGSET_END};
+
+static const struct vga_regset ark_v_total_regs[]        = {{0x06, 0, 7}, {0x07, 0, 0}, {0x07, 5, 5}, {0x40, 7, 7}, VGA_REGSET_END};
+static const struct vga_regset ark_v_display_regs[]      = {{0x12, 0, 7}, {0x07, 1, 1}, {0x07, 6, 6}, {0x40, 6, 6}, VGA_REGSET_END};
+static const struct vga_regset ark_v_blank_start_regs[]  = {{0x15, 0, 7}, {0x07, 3, 3}, {0x09, 5, 5}, {0x40, 5, 5}, VGA_REGSET_END};
+// const struct vga_regset ark_v_blank_end_regs[]    = {{0x16, 0, 6}, VGA_REGSET_END};
+static const struct vga_regset ark_v_blank_end_regs[]    = {{0x16, 0, 7}, VGA_REGSET_END};
+static const struct vga_regset ark_v_sync_start_regs[]   = {{0x10, 0, 7}, {0x07, 2, 2}, {0x07, 7, 7}, {0x40, 4, 4}, VGA_REGSET_END};
+static const struct vga_regset ark_v_sync_end_regs[]     = {{0x11, 0, 3}, VGA_REGSET_END};
+
+static const struct vga_regset ark_line_compare_regs[]   = {{0x18, 0, 7}, {0x07, 4, 4}, {0x09, 6, 6}, VGA_REGSET_END};
+static const struct vga_regset ark_start_address_regs[]  = {{0x0d, 0, 7}, {0x0c, 0, 7}, {0x40, 0, 2}, VGA_REGSET_END};
+static const struct vga_regset ark_offset_regs[]         = {{0x13, 0, 7}, {0x41, 3, 3}, VGA_REGSET_END};
+
+static const struct svga_timing_regs ark_timing_regs     = {
+	ark_h_total_regs, ark_h_display_regs, ark_h_blank_start_regs,
+	ark_h_blank_end_regs, ark_h_sync_start_regs, ark_h_sync_end_regs,
+	ark_v_total_regs, ark_v_display_regs, ark_v_blank_start_regs,
+	ark_v_blank_end_regs, ark_v_sync_start_regs, ark_v_sync_end_regs,
+};
+
+
+/* ------------------------------------------------------------------------- */
+
+
+/* Module parameters */
+
+static char *mode = "640x480-8@60";
+
+#ifdef CONFIG_MTRR
+static int mtrr = 1;
+#endif
+
+MODULE_AUTHOR("(c) 2007 Ondrej Zajicek <santiago@crfreenet.org>");
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("fbdev driver for ARK 2000PV");
+
+module_param(mode, charp, 0444);
+MODULE_PARM_DESC(mode, "Default video mode ('640x480-8@60', etc)");
+
+#ifdef CONFIG_MTRR
+module_param(mtrr, int, 0444);
+MODULE_PARM_DESC(mtrr, "Enable write-combining with MTRR (1=enable, 0=disable, default=1)");
+#endif
+
+static int threshold = 4;
+
+module_param(threshold, int, 0644);
+MODULE_PARM_DESC(threshold, "FIFO threshold");
+
+
+/* ------------------------------------------------------------------------- */
+
+
+static void arkfb_settile(struct fb_info *info, struct fb_tilemap *map)
+{
+	const u8 *font = map->data;
+	u8 __iomem *fb = (u8 __iomem *)info->screen_base;
+	int i, c;
+
+	if ((map->width != 8) || (map->height != 16) ||
+	    (map->depth != 1) || (map->length != 256)) {
+	    	printk(KERN_ERR "fb%d: unsupported font parameters: width %d, "
+		       "height %d, depth %d, length %d\n", info->node,
+		       map->width, map->height, map->depth, map->length);
+		return;
+	}
+
+	fb += 2;
+	for (c = 0; c < map->length; c++) {
+		for (i = 0; i < map->height; i++) {
+			fb_writeb(font[i], &fb[i * 4]);
+			fb_writeb(font[i], &fb[i * 4 + (128 * 8)]);
+		}
+		fb += 128;
+
+		if ((c % 8) == 7)
+			fb += 128*8;
+
+		font += map->height;
+	}
+}
+
+static struct fb_tile_ops arkfb_tile_ops = {
+	.fb_settile	= arkfb_settile,
+	.fb_tilecopy	= svga_tilecopy,
+	.fb_tilefill    = svga_tilefill,
+	.fb_tileblit    = svga_tileblit,
+	.fb_tilecursor  = svga_tilecursor,
+	.fb_get_tilemax = svga_get_tilemax,
+};
+
+
+/* ------------------------------------------------------------------------- */
+
+
+/* image data is MSB-first, fb structure is MSB-first too */
+static inline u32 expand_color(u32 c)
+{
+	return ((c & 1) | ((c & 2) << 7) | ((c & 4) << 14) | ((c & 8) << 21)) * 0xFF;
+}
+
+/* arkfb_iplan_imageblit silently assumes that almost everything is 8-pixel aligned */
+static void arkfb_iplan_imageblit(struct fb_info *info, const struct fb_image *image)
+{
+	u32 fg = expand_color(image->fg_color);
+	u32 bg = expand_color(image->bg_color);
+	const u8 *src1, *src;
+	u8 __iomem *dst1;
+	u32 __iomem *dst;
+	u32 val;
+	int x, y;
+
+	src1 = image->data;
+	dst1 = info->screen_base + (image->dy * info->fix.line_length)
+		 + ((image->dx / 8) * 4);
+
+	for (y = 0; y < image->height; y++) {
+		src = src1;
+		dst = (u32 __iomem *) dst1;
+		for (x = 0; x < image->width; x += 8) {
+			val = *(src++) * 0x01010101;
+			val = (val & fg) | (~val & bg);
+			fb_writel(val, dst++);
+		}
+		src1 += image->width / 8;
+		dst1 += info->fix.line_length;
+	}
+
+}
+
+/* arkfb_iplan_fillrect silently assumes that almost everything is 8-pixel aligned */
+static void arkfb_iplan_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
+{
+	u32 fg = expand_color(rect->color);
+	u8 __iomem *dst1;
+	u32 __iomem *dst;
+	int x, y;
+
+	dst1 = info->screen_base + (rect->dy * info->fix.line_length)
+		 + ((rect->dx / 8) * 4);
+
+	for (y = 0; y < rect->height; y++) {
+		dst = (u32 __iomem *) dst1;
+		for (x = 0; x < rect->width; x += 8) {
+			fb_writel(fg, dst++);
+		}
+		dst1 += info->fix.line_length;
+	}
+
+}
+
+
+/* image data is MSB-first, fb structure is high-nibble-in-low-byte-first */
+static inline u32 expand_pixel(u32 c)
+{
+	return (((c &  1) << 24) | ((c &  2) << 27) | ((c &  4) << 14) | ((c &   8) << 17) |
+		((c & 16) <<  4) | ((c & 32) <<  7) | ((c & 64) >>  6) | ((c & 128) >>  3)) * 0xF;
+}
+
+/* arkfb_cfb4_imageblit silently assumes that almost everything is 8-pixel aligned */
+static void arkfb_cfb4_imageblit(struct fb_info *info, const struct fb_image *image)
+{
+	u32 fg = image->fg_color * 0x11111111;
+	u32 bg = image->bg_color * 0x11111111;
+	const u8 *src1, *src;
+	u8 __iomem *dst1;
+	u32 __iomem *dst;
+	u32 val;
+	int x, y;
+
+	src1 = image->data;
+	dst1 = info->screen_base + (image->dy * info->fix.line_length)
+		 + ((image->dx / 8) * 4);
+
+	for (y = 0; y < image->height; y++) {
+		src = src1;
+		dst = (u32 __iomem *) dst1;
+		for (x = 0; x < image->width; x += 8) {
+			val = expand_pixel(*(src++));
+			val = (val & fg) | (~val & bg);
+			fb_writel(val, dst++);
+		}
+		src1 += image->width / 8;
+		dst1 += info->fix.line_length;
+	}
+
+}
+
+static void arkfb_imageblit(struct fb_info *info, const struct fb_image *image)
+{
+	if ((info->var.bits_per_pixel == 4) && (image->depth == 1)
+	    && ((image->width % 8) == 0) && ((image->dx % 8) == 0)) {
+		if (info->fix.type == FB_TYPE_INTERLEAVED_PLANES)
+			arkfb_iplan_imageblit(info, image);
+		else
+			arkfb_cfb4_imageblit(info, image);
+	} else
+		cfb_imageblit(info, image);
+}
+
+static void arkfb_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
+{
+	if ((info->var.bits_per_pixel == 4)
+	    && ((rect->width % 8) == 0) && ((rect->dx % 8) == 0)
+	    && (info->fix.type == FB_TYPE_INTERLEAVED_PLANES))
+		arkfb_iplan_fillrect(info, rect);
+	 else
+		cfb_fillrect(info, rect);
+}
+
+
+/* ------------------------------------------------------------------------- */
+
+
+enum
+{
+	DAC_PSEUDO8_8,
+	DAC_RGB1555_8,
+	DAC_RGB0565_8,
+	DAC_RGB0888_8,
+	DAC_RGB8888_8,
+	DAC_PSEUDO8_16,
+	DAC_RGB1555_16,
+	DAC_RGB0565_16,
+	DAC_RGB0888_16,
+	DAC_RGB8888_16,
+	DAC_MAX
+};
+
+struct dac_ops {
+	int (*dac_get_mode)(struct dac_info *info);
+	int (*dac_set_mode)(struct dac_info *info, int mode);
+	int (*dac_get_freq)(struct dac_info *info, int channel);
+	int (*dac_set_freq)(struct dac_info *info, int channel, u32 freq);
+	void (*dac_release)(struct dac_info *info);
+};
+
+typedef void (*dac_read_regs_t)(void *data, u8 *code, int count);
+typedef void (*dac_write_regs_t)(void *data, u8 *code, int count);
+
+struct dac_info
+{
+	struct dac_ops *dacops;
+	dac_read_regs_t dac_read_regs;
+	dac_write_regs_t dac_write_regs;
+	void *data;
+};
+
+
+static inline u8 dac_read_reg(struct dac_info *info, u8 reg)
+{
+	u8 code[2] = {reg, 0};
+	info->dac_read_regs(info->data, code, 1);
+	return code[1];
+}
+
+static inline void dac_read_regs(struct dac_info *info, u8 *code, int count)
+{
+	info->dac_read_regs(info->data, code, count);
+}
+
+static inline void dac_write_reg(struct dac_info *info, u8 reg, u8 val)
+{
+	u8 code[2] = {reg, val};
+	info->dac_write_regs(info->data, code, 1);
+}
+
+static inline void dac_write_regs(struct dac_info *info, u8 *code, int count)
+{
+	info->dac_write_regs(info->data, code, count);
+}
+
+static inline int dac_set_mode(struct dac_info *info, int mode)
+{
+	return info->dacops->dac_set_mode(info, mode);
+}
+
+static inline int dac_set_freq(struct dac_info *info, int channel, u32 freq)
+{
+	return info->dacops->dac_set_freq(info, channel, freq);
+}
+
+static inline void dac_release(struct dac_info *info)
+{
+	info->dacops->dac_release(info);
+}
+
+
+/* ------------------------------------------------------------------------- */
+
+
+/* ICS5342 DAC */
+
+struct ics5342_info
+{
+	struct dac_info dac;
+	u8 mode;
+};
+
+#define DAC_PAR(info) ((struct ics5342_info *) info)
+
+/* LSB is set to distinguish unused slots */
+static const u8 ics5342_mode_table[DAC_MAX] = {
+	[DAC_PSEUDO8_8]  = 0x01, [DAC_RGB1555_8]  = 0x21, [DAC_RGB0565_8]  = 0x61,
+	[DAC_RGB0888_8]  = 0x41, [DAC_PSEUDO8_16] = 0x11, [DAC_RGB1555_16] = 0x31,
+	[DAC_RGB0565_16] = 0x51, [DAC_RGB0888_16] = 0x91, [DAC_RGB8888_16] = 0x71
+};
+
+static int ics5342_set_mode(struct dac_info *info, int mode)
+{
+	u8 code;
+
+	if (mode >= DAC_MAX)
+		return -EINVAL;
+
+	code = ics5342_mode_table[mode];
+
+	if (! code)
+		return -EINVAL;
+
+	dac_write_reg(info, 6, code & 0xF0);
+	DAC_PAR(info)->mode = mode;
+
+	return 0;
+}
+
+static const struct svga_pll ics5342_pll = {3, 129, 3, 33, 0, 3,
+	60000, 250000, 14318};
+
+/* pd4 - allow only posdivider 4 (r=2) */
+static const struct svga_pll ics5342_pll_pd4 = {3, 129, 3, 33, 2, 2,
+	60000, 335000, 14318};
+
+/* 270 MHz should be upper bound for VCO clock according to specs,
+   but that is too restrictive in pd4 case */
+
+static int ics5342_set_freq(struct dac_info *info, int channel, u32 freq)
+{
+	u16 m, n, r;
+
+	/* only postdivider 4 (r=2) is valid in mode DAC_PSEUDO8_16 */
+	int rv = svga_compute_pll((DAC_PAR(info)->mode == DAC_PSEUDO8_16)
+				  ? &ics5342_pll_pd4 : &ics5342_pll,
+				  freq, &m, &n, &r, 0);
+
+	if (rv < 0) {
+		return -EINVAL;
+	} else {
+		u8 code[6] = {4, 3, 5, m-2, 5, (n-2) | (r << 5)};
+		dac_write_regs(info, code, 3);
+		return 0;
+	}
+}
+
+static void ics5342_release(struct dac_info *info)
+{
+	ics5342_set_mode(info, DAC_PSEUDO8_8);
+	kfree(info);
+}
+
+static struct dac_ops ics5342_ops = {
+	.dac_set_mode	= ics5342_set_mode,
+	.dac_set_freq	= ics5342_set_freq,
+	.dac_release	= ics5342_release
+};
+
+
+static struct dac_info * ics5342_init(dac_read_regs_t drr, dac_write_regs_t dwr, void *data)
+{
+	struct dac_info *info = kzalloc(sizeof(struct ics5342_info), GFP_KERNEL);
+
+	if (! info)
+		return NULL;
+
+	info->dacops = &ics5342_ops;
+	info->dac_read_regs = drr;
+	info->dac_write_regs = dwr;
+	info->data = data;
+	DAC_PAR(info)->mode = DAC_PSEUDO8_8; /* estimation */
+	return info;
+}
+
+
+/* ------------------------------------------------------------------------- */
+
+
+static unsigned short dac_regs[4] = {0x3c8, 0x3c9, 0x3c6, 0x3c7};
+
+static void ark_dac_read_regs(void *data, u8 *code, int count)
+{
+	u8 regval = vga_rseq(NULL, 0x1C);
+
+	while (count != 0)
+	{
+		vga_wseq(NULL, 0x1C, regval | (code[0] & 4) ? 0x80 : 0);
+		code[1] = vga_r(NULL, dac_regs[code[0] & 3]);
+		count--;
+		code += 2;
+	}
+
+	vga_wseq(NULL, 0x1C, regval);
+}
+
+static void ark_dac_write_regs(void *data, u8 *code, int count)
+{
+	u8 regval = vga_rseq(NULL, 0x1C);
+
+	while (count != 0)
+	{
+		vga_wseq(NULL, 0x1C, regval | (code[0] & 4) ? 0x80 : 0);
+		vga_w(NULL, dac_regs[code[0] & 3], code[1]);
+		count--;
+		code += 2;
+	}
+
+	vga_wseq(NULL, 0x1C, regval);
+}
+
+
+static void ark_set_pixclock(struct fb_info *info, u32 pixclock)
+{
+	struct arkfb_info *par = info->par;
+	u8 regval;
+
+	int rv = dac_set_freq(par->dac, 0, 1000000000 / pixclock);
+	if (rv < 0) {
+		printk(KERN_ERR "fb%d: cannot set requested pixclock, keeping old value\n", info->node);
+		return;
+	}
+
+	/* Set VGA misc register  */
+	regval = vga_r(NULL, VGA_MIS_R);
+	vga_w(NULL, VGA_MIS_W, regval | VGA_MIS_ENB_PLL_LOAD);
+}
+
+
+/* Open framebuffer */
+
+static int arkfb_open(struct fb_info *info, int user)
+{
+	struct arkfb_info *par = info->par;
+
+	mutex_lock(&(par->open_lock));
+	if (par->ref_count == 0) {
+		memset(&(par->state), 0, sizeof(struct vgastate));
+		par->state.flags = VGA_SAVE_MODE | VGA_SAVE_FONTS | VGA_SAVE_CMAP;
+		par->state.num_crtc = 0x60;
+		par->state.num_seq = 0x30;
+		save_vga(&(par->state));
+	}
+
+	par->ref_count++;
+	mutex_unlock(&(par->open_lock));
+
+	return 0;
+}
+
+/* Close framebuffer */
+
+static int arkfb_release(struct fb_info *info, int user)
+{
+	struct arkfb_info *par = info->par;
+
+	mutex_lock(&(par->open_lock));
+	if (par->ref_count == 0) {
+		mutex_unlock(&(par->open_lock));
+		return -EINVAL;
+	}
+
+	if (par->ref_count == 1) {
+		restore_vga(&(par->state));
+		dac_set_mode(par->dac, DAC_PSEUDO8_8);
+	}
+
+	par->ref_count--;
+	mutex_unlock(&(par->open_lock));
+
+	return 0;
+}
+
+/* Validate passed in var */
+
+static int arkfb_check_var(struct fb_var_screeninfo *var, struct fb_info *info)
+{
+	int rv, mem, step;
+
+	/* Find appropriate format */
+	rv = svga_match_format (arkfb_formats, var, NULL);
+	if (rv < 0)
+	{
+		printk(KERN_ERR "fb%d: unsupported mode requested\n", info->node);
+		return rv;
+	}
+
+	/* Do not allow to have real resoulution larger than virtual */
+	if (var->xres > var->xres_virtual)
+		var->xres_virtual = var->xres;
+
+	if (var->yres > var->yres_virtual)
+		var->yres_virtual = var->yres;
+
+	/* Round up xres_virtual to have proper alignment of lines */
+	step = arkfb_formats[rv].xresstep - 1;
+	var->xres_virtual = (var->xres_virtual+step) & ~step;
+
+
+	/* Check whether have enough memory */
+	mem = ((var->bits_per_pixel * var->xres_virtual) >> 3) * var->yres_virtual;
+	if (mem > info->screen_size)
+	{
+		printk(KERN_ERR "fb%d: not enough framebuffer memory (%d kB requested , %d kB available)\n", info->node, mem >> 10, (unsigned int) (info->screen_size >> 10));
+		return -EINVAL;
+	}
+
+	rv = svga_check_timings (&ark_timing_regs, var, info->node);
+	if (rv < 0)
+	{
+		printk(KERN_ERR "fb%d: invalid timings requested\n", info->node);
+		return rv;
+	}
+
+	/* Interlaced mode is broken */
+	if (var->vmode & FB_VMODE_INTERLACED)
+		return -EINVAL;
+
+	return 0;
+}
+
+/* Set video mode from par */
+
+static int arkfb_set_par(struct fb_info *info)
+{
+	struct arkfb_info *par = info->par;
+	u32 value, mode, hmul, hdiv, offset_value, screen_size;
+	u32 bpp = info->var.bits_per_pixel;
+	u8 regval;
+
+	if (bpp != 0) {
+		info->fix.ypanstep = 1;
+		info->fix.line_length = (info->var.xres_virtual * bpp) / 8;
+
+		info->flags &= ~FBINFO_MISC_TILEBLITTING;
+		info->tileops = NULL;
+
+		/* in 4bpp supports 8p wide tiles only, any tiles otherwise */
+		info->pixmap.blit_x = (bpp == 4) ? (1 << (8 - 1)) : (~(u32)0);
+		info->pixmap.blit_y = ~(u32)0;
+
+		offset_value = (info->var.xres_virtual * bpp) / 64;
+		screen_size = info->var.yres_virtual * info->fix.line_length;
+	} else {
+		info->fix.ypanstep = 16;
+		info->fix.line_length = 0;
+
+		info->flags |= FBINFO_MISC_TILEBLITTING;
+		info->tileops = &arkfb_tile_ops;
+
+		/* supports 8x16 tiles only */
+		info->pixmap.blit_x = 1 << (8 - 1);
+		info->pixmap.blit_y = 1 << (16 - 1);
+
+		offset_value = info->var.xres_virtual / 16;
+		screen_size = (info->var.xres_virtual * info->var.yres_virtual) / 64;
+	}
+
+	info->var.xoffset = 0;
+	info->var.yoffset = 0;
+	info->var.activate = FB_ACTIVATE_NOW;
+
+	/* Unlock registers */
+	svga_wcrt_mask(0x11, 0x00, 0x80);
+
+	/* Blank screen and turn off sync */
+	svga_wseq_mask(0x01, 0x20, 0x20);
+	svga_wcrt_mask(0x17, 0x00, 0x80);
+
+	/* Set default values */
+	svga_set_default_gfx_regs();
+	svga_set_default_atc_regs();
+	svga_set_default_seq_regs();
+	svga_set_default_crt_regs();
+	svga_wcrt_multi(ark_line_compare_regs, 0xFFFFFFFF);
+	svga_wcrt_multi(ark_start_address_regs, 0);
+
+	/* ARK specific initialization */
+	svga_wseq_mask(0x10, 0x1F, 0x1F); /* enable linear framebuffer and full memory access */
+	svga_wseq_mask(0x12, 0x03, 0x03); /* 4 MB linear framebuffer size */
+
+	vga_wseq(NULL, 0x13, info->fix.smem_start >> 16);
+	vga_wseq(NULL, 0x14, info->fix.smem_start >> 24);
+	vga_wseq(NULL, 0x15, 0);
+	vga_wseq(NULL, 0x16, 0);
+
+	/* Set the FIFO threshold register */
+	/* It is fascinating way to store 5-bit value in 8-bit register */
+	regval = 0x10 | ((threshold & 0x0E) >> 1) | (threshold & 0x01) << 7 | (threshold & 0x10) << 1;
+	vga_wseq(NULL, 0x18, regval);
+
+	/* Set the offset register */
+	pr_debug("fb%d: offset register       : %d\n", info->node, offset_value);
+	svga_wcrt_multi(ark_offset_regs, offset_value);
+
+	/* fix for hi-res textmode */
+	svga_wcrt_mask(0x40, 0x08, 0x08);
+
+	if (info->var.vmode & FB_VMODE_DOUBLE)
+		svga_wcrt_mask(0x09, 0x80, 0x80);
+	else
+		svga_wcrt_mask(0x09, 0x00, 0x80);
+
+	if (info->var.vmode & FB_VMODE_INTERLACED)
+		svga_wcrt_mask(0x44, 0x04, 0x04);
+	else
+		svga_wcrt_mask(0x44, 0x00, 0x04);
+
+	hmul = 1;
+	hdiv = 1;
+	mode = svga_match_format(arkfb_formats, &(info->var), &(info->fix));
+
+	/* Set mode-specific register values */
+	switch (mode) {
+	case 0:
+		pr_debug("fb%d: text mode\n", info->node);
+		svga_set_textmode_vga_regs();
+
+		vga_wseq(NULL, 0x11, 0x10); /* basic VGA mode */
+		svga_wcrt_mask(0x46, 0x00, 0x04); /* 8bit pixel path */
+		dac_set_mode(par->dac, DAC_PSEUDO8_8);
+
+		break;
+	case 1:
+		pr_debug("fb%d: 4 bit pseudocolor\n", info->node);
+		vga_wgfx(NULL, VGA_GFX_MODE, 0x40);
+
+		vga_wseq(NULL, 0x11, 0x10); /* basic VGA mode */
+		svga_wcrt_mask(0x46, 0x00, 0x04); /* 8bit pixel path */
+		dac_set_mode(par->dac, DAC_PSEUDO8_8);
+		break;
+	case 2:
+		pr_debug("fb%d: 4 bit pseudocolor, planar\n", info->node);
+
+		vga_wseq(NULL, 0x11, 0x10); /* basic VGA mode */
+		svga_wcrt_mask(0x46, 0x00, 0x04); /* 8bit pixel path */
+		dac_set_mode(par->dac, DAC_PSEUDO8_8);
+		break;
+	case 3:
+		pr_debug("fb%d: 8 bit pseudocolor\n", info->node);
+
+		vga_wseq(NULL, 0x11, 0x16); /* 8bpp accel mode */
+
+		if (info->var.pixclock > 20000) {
+			pr_debug("fb%d: not using multiplex\n", info->node);
+			svga_wcrt_mask(0x46, 0x00, 0x04); /* 8bit pixel path */
+			dac_set_mode(par->dac, DAC_PSEUDO8_8);
+		} else {
+			pr_debug("fb%d: using multiplex\n", info->node);
+			svga_wcrt_mask(0x46, 0x04, 0x04); /* 16bit pixel path */
+			dac_set_mode(par->dac, DAC_PSEUDO8_16);
+			hdiv = 2;
+		}
+		break;
+	case 4:
+		pr_debug("fb%d: 5/5/5 truecolor\n", info->node);
+
+		vga_wseq(NULL, 0x11, 0x1A); /* 16bpp accel mode */
+		svga_wcrt_mask(0x46, 0x04, 0x04); /* 16bit pixel path */
+		dac_set_mode(par->dac, DAC_RGB1555_16);
+		break;
+	case 5:
+		pr_debug("fb%d: 5/6/5 truecolor\n", info->node);
+
+		vga_wseq(NULL, 0x11, 0x1A); /* 16bpp accel mode */
+		svga_wcrt_mask(0x46, 0x04, 0x04); /* 16bit pixel path */
+		dac_set_mode(par->dac, DAC_RGB0565_16);
+		break;
+	case 6:
+		pr_debug("fb%d: 8/8/8 truecolor\n", info->node);
+
+		vga_wseq(NULL, 0x11, 0x16); /* 8bpp accel mode ??? */
+		svga_wcrt_mask(0x46, 0x04, 0x04); /* 16bit pixel path */
+		dac_set_mode(par->dac, DAC_RGB0888_16);
+		hmul = 3;
+		hdiv = 2;
+		break;
+	case 7:
+		pr_debug("fb%d: 8/8/8/8 truecolor\n", info->node);
+
+		vga_wseq(NULL, 0x11, 0x1E); /* 32bpp accel mode */
+		svga_wcrt_mask(0x46, 0x04, 0x04); /* 16bit pixel path */
+		dac_set_mode(par->dac, DAC_RGB8888_16);
+		hmul = 2;
+		break;
+	default:
+		printk(KERN_ERR "fb%d: unsupported mode - bug\n", info->node);
+		return -EINVAL;
+	}
+
+	ark_set_pixclock(info, (hdiv * info->var.pixclock) / hmul);
+	svga_set_timings(&ark_timing_regs, &(info->var), hmul, hdiv,
+			 (info->var.vmode & FB_VMODE_DOUBLE)     ? 2 : 1,
+			 (info->var.vmode & FB_VMODE_INTERLACED) ? 2 : 1,
+			  hmul, info->node);
+
+	/* Set interlaced mode start/end register */
+	value = info->var.xres + info->var.left_margin + info->var.right_margin + info->var.hsync_len;
+	value = ((value * hmul / hdiv) / 8) - 5;
+	vga_wcrt(NULL, 0x42, (value + 1) / 2);
+
+	memset_io(info->screen_base, 0x00, screen_size);
+	/* Device and screen back on */
+	svga_wcrt_mask(0x17, 0x80, 0x80);
+	svga_wseq_mask(0x01, 0x00, 0x20);
+
+	return 0;
+}
+
+/* Set a colour register */
+
+static int arkfb_setcolreg(u_int regno, u_int red, u_int green, u_int blue,
+				u_int transp, struct fb_info *fb)
+{
+	switch (fb->var.bits_per_pixel) {
+	case 0:
+	case 4:
+		if (regno >= 16)
+			return -EINVAL;
+
+		if ((fb->var.bits_per_pixel == 4) &&
+		    (fb->var.nonstd == 0)) {
+			outb(0xF0, VGA_PEL_MSK);
+			outb(regno*16, VGA_PEL_IW);
+		} else {
+			outb(0x0F, VGA_PEL_MSK);
+			outb(regno, VGA_PEL_IW);
+		}
+		outb(red >> 10, VGA_PEL_D);
+		outb(green >> 10, VGA_PEL_D);
+		outb(blue >> 10, VGA_PEL_D);
+		break;
+	case 8:
+		if (regno >= 256)
+			return -EINVAL;
+
+		outb(0xFF, VGA_PEL_MSK);
+		outb(regno, VGA_PEL_IW);
+		outb(red >> 10, VGA_PEL_D);
+		outb(green >> 10, VGA_PEL_D);
+		outb(blue >> 10, VGA_PEL_D);
+		break;
+	case 16:
+		if (regno >= 16)
+			return 0;
+
+		if (fb->var.green.length == 5)
+			((u32*)fb->pseudo_palette)[regno] = ((red & 0xF800) >> 1) |
+				((green & 0xF800) >> 6) | ((blue & 0xF800) >> 11);
+		else if (fb->var.green.length == 6)
+			((u32*)fb->pseudo_palette)[regno] = (red & 0xF800) |
+				((green & 0xFC00) >> 5) | ((blue & 0xF800) >> 11);
+		else
+			return -EINVAL;
+		break;
+	case 24:
+	case 32:
+		if (regno >= 16)
+			return 0;
+
+		((u32*)fb->pseudo_palette)[regno] = ((red & 0xFF00) << 8) |
+			(green & 0xFF00) | ((blue & 0xFF00) >> 8);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+/* Set the display blanking state */
+
+static int arkfb_blank(int blank_mode, struct fb_info *info)
+{
+	switch (blank_mode) {
+	case FB_BLANK_UNBLANK:
+		pr_debug("fb%d: unblank\n", info->node);
+		svga_wseq_mask(0x01, 0x00, 0x20);
+		svga_wcrt_mask(0x17, 0x80, 0x80);
+		break;
+	case FB_BLANK_NORMAL:
+		pr_debug("fb%d: blank\n", info->node);
+		svga_wseq_mask(0x01, 0x20, 0x20);
+		svga_wcrt_mask(0x17, 0x80, 0x80);
+		break;
+	case FB_BLANK_POWERDOWN:
+	case FB_BLANK_HSYNC_SUSPEND:
+	case FB_BLANK_VSYNC_SUSPEND:
+		pr_debug("fb%d: sync down\n", info->node);
+		svga_wseq_mask(0x01, 0x20, 0x20);
+		svga_wcrt_mask(0x17, 0x00, 0x80);
+		break;
+	}
+	return 0;
+}
+
+
+/* Pan the display */
+
+static int arkfb_pan_display(struct fb_var_screeninfo *var, struct fb_info *info)
+{
+	unsigned int offset;
+
+	/* Calculate the offset */
+	if (var->bits_per_pixel == 0) {
+		offset = (var->yoffset / 16) * (var->xres_virtual / 2) + (var->xoffset / 2);
+		offset = offset >> 2;
+	} else {
+		offset = (var->yoffset * info->fix.line_length) +
+			 (var->xoffset * var->bits_per_pixel / 8);
+		offset = offset >> ((var->bits_per_pixel == 4) ? 2 : 3);
+	}
+
+	/* Set the offset */
+	svga_wcrt_multi(ark_start_address_regs, offset);
+
+	return 0;
+}
+
+
+/* ------------------------------------------------------------------------- */
+
+
+/* Frame buffer operations */
+
+static struct fb_ops arkfb_ops = {
+	.owner		= THIS_MODULE,
+	.fb_open	= arkfb_open,
+	.fb_release	= arkfb_release,
+	.fb_check_var	= arkfb_check_var,
+	.fb_set_par	= arkfb_set_par,
+	.fb_setcolreg	= arkfb_setcolreg,
+	.fb_blank	= arkfb_blank,
+	.fb_pan_display	= arkfb_pan_display,
+	.fb_fillrect	= arkfb_fillrect,
+	.fb_copyarea	= cfb_copyarea,
+	.fb_imageblit	= arkfb_imageblit,
+	.fb_get_caps    = svga_get_caps,
+};
+
+
+/* ------------------------------------------------------------------------- */
+
+
+/* PCI probe */
+static int __devinit ark_pci_probe(struct pci_dev *dev, const struct pci_device_id *id)
+{
+	struct fb_info *info;
+	struct arkfb_info *par;
+	int rc;
+	u8 regval;
+
+	/* Ignore secondary VGA device because there is no VGA arbitration */
+	if (! svga_primary_device(dev)) {
+		dev_info(&(dev->dev), "ignoring secondary device\n");
+		return -ENODEV;
+	}
+
+	/* Allocate and fill driver data structure */
+	info = framebuffer_alloc(sizeof(struct arkfb_info), NULL);
+	if (! info) {
+		dev_err(&(dev->dev), "cannot allocate memory\n");
+		return -ENOMEM;
+	}
+
+	par = info->par;
+	mutex_init(&par->open_lock);
+
+	info->flags = FBINFO_PARTIAL_PAN_OK | FBINFO_HWACCEL_YPAN;
+	info->fbops = &arkfb_ops;
+
+	/* Prepare PCI device */
+	rc = pci_enable_device(dev);
+	if (rc < 0) {
+		dev_err(&(dev->dev), "cannot enable PCI device\n");
+		goto err_enable_device;
+	}
+
+	rc = pci_request_regions(dev, "arkfb");
+	if (rc < 0) {
+		dev_err(&(dev->dev), "cannot reserve framebuffer region\n");
+		goto err_request_regions;
+	}
+
+	par->dac = ics5342_init(ark_dac_read_regs, ark_dac_write_regs, info);
+	if (! par->dac) {
+		rc = -ENOMEM;
+		dev_err(&(dev->dev), "RAMDAC initialization failed\n");
+		goto err_dac;
+	}
+
+	info->fix.smem_start = pci_resource_start(dev, 0);
+	info->fix.smem_len = pci_resource_len(dev, 0);
+
+	/* Map physical IO memory address into kernel space */
+	info->screen_base = pci_iomap(dev, 0, 0);
+	if (! info->screen_base) {
+		rc = -ENOMEM;
+		dev_err(&(dev->dev), "iomap for framebuffer failed\n");
+		goto err_iomap;
+	}
+
+	/* FIXME get memsize */
+	regval = vga_rseq(NULL, 0x10);
+	info->screen_size = (1 << (regval >> 6)) << 20;
+	info->fix.smem_len = info->screen_size;
+
+	strcpy(info->fix.id, "ARK 2000PV");
+	info->fix.mmio_start = 0;
+	info->fix.mmio_len = 0;
+	info->fix.type = FB_TYPE_PACKED_PIXELS;
+	info->fix.visual = FB_VISUAL_PSEUDOCOLOR;
+	info->fix.ypanstep = 0;
+	info->fix.accel = FB_ACCEL_NONE;
+	info->pseudo_palette = (void*) (par->pseudo_palette);
+
+	/* Prepare startup mode */
+	rc = fb_find_mode(&(info->var), info, mode, NULL, 0, NULL, 8);
+	if (! ((rc == 1) || (rc == 2))) {
+		rc = -EINVAL;
+		dev_err(&(dev->dev), "mode %s not found\n", mode);
+		goto err_find_mode;
+	}
+
+	rc = fb_alloc_cmap(&info->cmap, 256, 0);
+	if (rc < 0) {
+		dev_err(&(dev->dev), "cannot allocate colormap\n");
+		goto err_alloc_cmap;
+	}
+
+	rc = register_framebuffer(info);
+	if (rc < 0) {
+		dev_err(&(dev->dev), "cannot register framebugger\n");
+		goto err_reg_fb;
+	}
+
+	printk(KERN_INFO "fb%d: %s on %s, %d MB RAM\n", info->node, info->fix.id,
+		 pci_name(dev), info->fix.smem_len >> 20);
+
+	/* Record a reference to the driver data */
+	pci_set_drvdata(dev, info);
+
+#ifdef CONFIG_MTRR
+	if (mtrr) {
+		par->mtrr_reg = -1;
+		par->mtrr_reg = mtrr_add(info->fix.smem_start, info->fix.smem_len, MTRR_TYPE_WRCOMB, 1);
+	}
+#endif
+
+	return 0;
+
+	/* Error handling */
+err_reg_fb:
+	fb_dealloc_cmap(&info->cmap);
+err_alloc_cmap:
+err_find_mode:
+	pci_iounmap(dev, info->screen_base);
+err_iomap:
+	dac_release(par->dac);
+err_dac:
+	pci_release_regions(dev);
+err_request_regions:
+/*	pci_disable_device(dev); */
+err_enable_device:
+	framebuffer_release(info);
+	return rc;
+}
+
+/* PCI remove */
+
+static void __devexit ark_pci_remove(struct pci_dev *dev)
+{
+	struct fb_info *info = pci_get_drvdata(dev);
+	struct arkfb_info *par = info->par;
+
+	if (info) {
+#ifdef CONFIG_MTRR
+		if (par->mtrr_reg >= 0) {
+			mtrr_del(par->mtrr_reg, 0, 0);
+			par->mtrr_reg = -1;
+		}
+#endif
+
+		dac_release(par->dac);
+		unregister_framebuffer(info);
+		fb_dealloc_cmap(&info->cmap);
+
+		pci_iounmap(dev, info->screen_base);
+		pci_release_regions(dev);
+/*		pci_disable_device(dev); */
+
+		pci_set_drvdata(dev, NULL);
+		framebuffer_release(info);
+	}
+}
+
+
+#ifdef CONFIG_PM
+/* PCI suspend */
+
+static int ark_pci_suspend (struct pci_dev* dev, pm_message_t state)
+{
+	struct fb_info *info = pci_get_drvdata(dev);
+	struct arkfb_info *par = info->par;
+
+	dev_info(&(dev->dev), "suspend\n");
+
+	acquire_console_sem();
+	mutex_lock(&(par->open_lock));
+
+	if ((state.event == PM_EVENT_FREEZE) || (par->ref_count == 0)) {
+		mutex_unlock(&(par->open_lock));
+		release_console_sem();
+		return 0;
+	}
+
+	fb_set_suspend(info, 1);
+
+	pci_save_state(dev);
+	pci_disable_device(dev);
+	pci_set_power_state(dev, pci_choose_state(dev, state));
+
+	mutex_unlock(&(par->open_lock));
+	release_console_sem();
+
+	return 0;
+}
+
+
+/* PCI resume */
+
+static int ark_pci_resume (struct pci_dev* dev)
+{
+	struct fb_info *info = pci_get_drvdata(dev);
+	struct arkfb_info *par = info->par;
+
+	dev_info(&(dev->dev), "resume\n");
+
+	acquire_console_sem();
+	mutex_lock(&(par->open_lock));
+
+	if (par->ref_count == 0) {
+		mutex_unlock(&(par->open_lock));
+		release_console_sem();
+		return 0;
+	}
+
+	pci_set_power_state(dev, PCI_D0);
+	pci_restore_state(dev);
+
+	if (pci_enable_device(dev))
+		goto fail;
+
+	pci_set_master(dev);
+
+	arkfb_set_par(info);
+	fb_set_suspend(info, 0);
+
+	mutex_unlock(&(par->open_lock));
+fail:
+	release_console_sem();
+	return 0;
+}
+#else
+#define ark_pci_suspend NULL
+#define ark_pci_resume NULL
+#endif /* CONFIG_PM */
+
+/* List of boards that we are trying to support */
+
+static struct pci_device_id ark_devices[] __devinitdata = {
+	{PCI_DEVICE(0xEDD8, 0xA099)},
+	{0, 0, 0, 0, 0, 0, 0}
+};
+
+
+MODULE_DEVICE_TABLE(pci, ark_devices);
+
+static struct pci_driver arkfb_pci_driver = {
+	.name		= "arkfb",
+	.id_table	= ark_devices,
+	.probe		= ark_pci_probe,
+	.remove		= __devexit_p(ark_pci_remove),
+	.suspend	= ark_pci_suspend,
+	.resume		= ark_pci_resume,
+};
+
+/* Cleanup */
+
+static void __exit arkfb_cleanup(void)
+{
+	pr_debug("arkfb: cleaning up\n");
+	pci_unregister_driver(&arkfb_pci_driver);
+}
+
+/* Driver Initialisation */
+
+static int __init arkfb_init(void)
+{
+
+#ifndef MODULE
+	char *option = NULL;
+
+	if (fb_get_options("arkfb", &option))
+		return -ENODEV;
+
+	if (option && *option)
+		mode = option;
+#endif
+
+	pr_debug("arkfb: initializing\n");
+	return pci_register_driver(&arkfb_pci_driver);
+}
+
+module_init(arkfb_init);
+module_exit(arkfb_cleanup);
diff --git a/drivers/video/console/softcursor.c b/drivers/video/console/softcursor.c
index f577bd80e020..03cfb7ac5733 100644
--- a/drivers/video/console/softcursor.c
+++ b/drivers/video/console/softcursor.c
@@ -1,5 +1,5 @@
 /*
- * linux/drivers/video/softcursor.c
+ * linux/drivers/video/console/softcursor.c
  *
  * Generic software cursor for frame buffer devices
  *
diff --git a/drivers/video/fbmem.c b/drivers/video/fbmem.c
index 08d4e11d9121..38c2e2558f5e 100644
--- a/drivers/video/fbmem.c
+++ b/drivers/video/fbmem.c
@@ -1236,6 +1236,10 @@ fb_mmap(struct file *file, struct vm_area_struct * vma)
 	pgprot_val(vma->vm_page_prot) |= _PAGE_NO_CACHE;
 #elif defined(__arm__) || defined(__sh__) || defined(__m32r__)
 	vma->vm_page_prot = pgprot_writecombine(vma->vm_page_prot);
+#elif defined(__avr32__)
+	vma->vm_page_prot = __pgprot((pgprot_val(vma->vm_page_prot)
+				      & ~_PAGE_CACHABLE)
+				     | (_PAGE_BUFFER | _PAGE_DIRTY));
 #elif defined(__ia64__)
 	if (efi_range_is_wc(vma->vm_start, vma->vm_end - vma->vm_start))
 		vma->vm_page_prot = pgprot_writecombine(vma->vm_page_prot);
diff --git a/drivers/video/i810/i810_main.c b/drivers/video/i810/i810_main.c
index 7e760197cf29..1a7d7789d877 100644
--- a/drivers/video/i810/i810_main.c
+++ b/drivers/video/i810/i810_main.c
@@ -1717,7 +1717,7 @@ static int __devinit i810_alloc_agp_mem(struct fb_info *info)
  * @info: pointer to device specific info structure
  *
  * DESCRIPTION:
- * Sets the the user monitor's horizontal and vertical
+ * Sets the user monitor's horizontal and vertical
  * frequency limits
  */
 static void __devinit i810_init_monspecs(struct fb_info *info)
diff --git a/drivers/video/matrox/matroxfb_Ti3026.c b/drivers/video/matrox/matroxfb_Ti3026.c
index a5690a5f29d5..9445cdb759b1 100644
--- a/drivers/video/matrox/matroxfb_Ti3026.c
+++ b/drivers/video/matrox/matroxfb_Ti3026.c
@@ -72,7 +72,7 @@
  *     (c) 1998 Gerd Knorr <kraxel@cs.tu-berlin.de>
  *
  * (following author is not in any relation with this code, but his ideas
- *  were used when writting this driver)
+ *  were used when writing this driver)
  *
  *		 FreeVBE/AF (Matrox), "Shawn Hargreaves" <shawn@talula.demon.co.uk>
  *
diff --git a/drivers/video/matrox/matroxfb_accel.c b/drivers/video/matrox/matroxfb_accel.c
index a5c825d99466..c57aaadf410c 100644
--- a/drivers/video/matrox/matroxfb_accel.c
+++ b/drivers/video/matrox/matroxfb_accel.c
@@ -70,7 +70,7 @@
  *     (c) 1998 Gerd Knorr <kraxel@cs.tu-berlin.de>
  *
  * (following author is not in any relation with this code, but his ideas
- *  were used when writting this driver)
+ *  were used when writing this driver)
  *
  *		 FreeVBE/AF (Matrox), "Shawn Hargreaves" <shawn@talula.demon.co.uk>
  *
diff --git a/drivers/video/matrox/matroxfb_base.c b/drivers/video/matrox/matroxfb_base.c
index cb2aa402ddfd..c8559a756b75 100644
--- a/drivers/video/matrox/matroxfb_base.c
+++ b/drivers/video/matrox/matroxfb_base.c
@@ -93,7 +93,7 @@
  *     (c) 1998 Gerd Knorr <kraxel@cs.tu-berlin.de>
  *
  * (following author is not in any relation with this code, but his ideas
- *  were used when writting this driver)
+ *  were used when writing this driver)
  *
  *		 FreeVBE/AF (Matrox), "Shawn Hargreaves" <shawn@talula.demon.co.uk>
  *
diff --git a/drivers/video/matrox/matroxfb_misc.c b/drivers/video/matrox/matroxfb_misc.c
index 18886b629cb1..5948e54b9ef9 100644
--- a/drivers/video/matrox/matroxfb_misc.c
+++ b/drivers/video/matrox/matroxfb_misc.c
@@ -78,7 +78,7 @@
  *     (c) 1998 Gerd Knorr <kraxel@cs.tu-berlin.de>
  *
  * (following author is not in any relation with this code, but his ideas
- *  were used when writting this driver)
+ *  were used when writing this driver)
  *
  *		 FreeVBE/AF (Matrox), "Shawn Hargreaves" <shawn@talula.demon.co.uk>
  *
diff --git a/drivers/video/nvidia/nv_hw.c b/drivers/video/nvidia/nv_hw.c
index f297c7b14a41..c627955aa124 100644
--- a/drivers/video/nvidia/nv_hw.c
+++ b/drivers/video/nvidia/nv_hw.c
@@ -149,8 +149,7 @@ static void nvGetClocks(struct nvidia_par *par, unsigned int *MClk,
 		pll = NV_RD32(par->PMC, 0x4024);
 		M = pll & 0xFF;
 		N = (pll >> 8) & 0xFF;
-		if (((par->Chipset & 0xfff0) == 0x0290) ||
-				((par->Chipset & 0xfff0) == 0x0390)) {
+		if (((par->Chipset & 0xfff0) == 0x0290) || ((par->Chipset & 0xfff0) == 0x0390) || ((par->Chipset & 0xfff0) == 0x02E0)) {
 			MB = 1;
 			NB = 1;
 		} else {
@@ -963,6 +962,7 @@ void NVLoadStateExt(struct nvidia_par *par, RIVA_HW_STATE * state)
 
 		if (((par->Chipset & 0xfff0) == 0x0090) ||
 		    ((par->Chipset & 0xfff0) == 0x01D0) ||
+		    ((par->Chipset & 0xfff0) == 0x02E0) ||
 		    ((par->Chipset & 0xfff0) == 0x0290))
 			regions = 15;
 		for(i = 0; i < regions; i++) {
@@ -1275,6 +1275,7 @@ void NVLoadStateExt(struct nvidia_par *par, RIVA_HW_STATE * state)
 						0x00100000);
 					break;
 				case 0x0090:
+				case 0x02E0:
 				case 0x0290:
 					NV_WR32(par->PRAMDAC, 0x0608,
 						NV_RD32(par->PRAMDAC, 0x0608) |
@@ -1352,6 +1353,7 @@ void NVLoadStateExt(struct nvidia_par *par, RIVA_HW_STATE * state)
 			} else {
 				if (((par->Chipset & 0xfff0) == 0x0090) ||
 				    ((par->Chipset & 0xfff0) == 0x01D0) ||
+				    ((par->Chipset & 0xfff0) == 0x02E0) ||
 				    ((par->Chipset & 0xfff0) == 0x0290)) {
 					for (i = 0; i < 60; i++) {
 						NV_WR32(par->PGRAPH,
@@ -1403,6 +1405,7 @@ void NVLoadStateExt(struct nvidia_par *par, RIVA_HW_STATE * state)
 				} else {
 					if ((par->Chipset & 0xfff0) == 0x0090 ||
 					    (par->Chipset & 0xfff0) == 0x01D0 ||
+					    (par->Chipset & 0xfff0) == 0x02E0 ||
 					    (par->Chipset & 0xfff0) == 0x0290) {
 						NV_WR32(par->PGRAPH, 0x0DF0,
 							NV_RD32(par->PFB, 0x0200));
diff --git a/drivers/video/nvidia/nvidia.c b/drivers/video/nvidia/nvidia.c
index 7c36b5fe582e..f85edf084da3 100644
--- a/drivers/video/nvidia/nvidia.c
+++ b/drivers/video/nvidia/nvidia.c
@@ -1243,6 +1243,7 @@ static u32 __devinit nvidia_get_arch(struct fb_info *info)
 	case 0x0140:		/* GeForce 6600 */
 	case 0x0160:		/* GeForce 6200 */
 	case 0x01D0:		/* GeForce 7200, 7300, 7400 */
+	case 0x02E0:		/* GeForce 7300 GT */
 	case 0x0090:		/* GeForce 7800 */
 	case 0x0210:		/* GeForce 6800 */
 	case 0x0220:		/* GeForce 6200 */
diff --git a/drivers/video/s3fb.c b/drivers/video/s3fb.c
index 756fafb41d78..d11735895a01 100644
--- a/drivers/video/s3fb.c
+++ b/drivers/video/s3fb.c
@@ -796,23 +796,6 @@ static int s3fb_pan_display(struct fb_var_screeninfo *var, struct fb_info *info)
 	return 0;
 }
 
-/* Get capabilities of accelerator based on the mode */
-
-static void s3fb_get_caps(struct fb_info *info, struct fb_blit_caps *caps,
-			  struct fb_var_screeninfo *var)
-{
-	if (var->bits_per_pixel == 0) {
-		/* can only support 256 8x16 bitmap */
-		caps->x = 1 << (8 - 1);
-		caps->y = 1 << (16 - 1);
-		caps->len = 256;
-	} else {
-		caps->x = ~(u32)0;
-		caps->y = ~(u32)0;
-		caps->len = ~(u32)0;
-	}
-}
-
 /* ------------------------------------------------------------------------- */
 
 /* Frame buffer operations */
@@ -829,7 +812,7 @@ static struct fb_ops s3fb_ops = {
 	.fb_fillrect	= s3fb_fillrect,
 	.fb_copyarea	= cfb_copyarea,
 	.fb_imageblit	= s3fb_imageblit,
-	.fb_get_caps    = s3fb_get_caps,
+	.fb_get_caps    = svga_get_caps,
 };
 
 /* ------------------------------------------------------------------------- */
diff --git a/drivers/video/skeletonfb.c b/drivers/video/skeletonfb.c
index 842b5cd054c6..836a612af977 100644
--- a/drivers/video/skeletonfb.c
+++ b/drivers/video/skeletonfb.c
@@ -14,7 +14,7 @@
  *  of it. 
  *
  *  First the roles of struct fb_info and struct display have changed. Struct
- *  display will go away. The way the the new framebuffer console code will
+ *  display will go away. The way the new framebuffer console code will
  *  work is that it will act to translate data about the tty/console in 
  *  struct vc_data to data in a device independent way in struct fb_info. Then
  *  various functions in struct fb_ops will be called to store the device 
diff --git a/drivers/video/svgalib.c b/drivers/video/svgalib.c
index 079cdc911e48..25df928d37d8 100644
--- a/drivers/video/svgalib.c
+++ b/drivers/video/svgalib.c
@@ -347,6 +347,23 @@ int svga_get_tilemax(struct fb_info *info)
 	return 256;
 }
 
+/* Get capabilities of accelerator based on the mode */
+
+void svga_get_caps(struct fb_info *info, struct fb_blit_caps *caps,
+		   struct fb_var_screeninfo *var)
+{
+	if (var->bits_per_pixel == 0) {
+		/* can only support 256 8x16 bitmap */
+		caps->x = 1 << (8 - 1);
+		caps->y = 1 << (16 - 1);
+		caps->len = 256;
+	} else {
+		caps->x = (var->bits_per_pixel == 4) ? 1 << (8 - 1) : ~(u32)0;
+		caps->y = ~(u32)0;
+		caps->len = ~(u32)0;
+	}
+}
+EXPORT_SYMBOL(svga_get_caps);
 
 /* ------------------------------------------------------------------------- */
 
diff --git a/drivers/video/vt8623fb.c b/drivers/video/vt8623fb.c
new file mode 100644
index 000000000000..5e9755e464a1
--- /dev/null
+++ b/drivers/video/vt8623fb.c
@@ -0,0 +1,927 @@
+/*
+ * linux/drivers/video/vt8623fb.c - fbdev driver for
+ * integrated graphic core in VIA VT8623 [CLE266] chipset
+ *
+ * Copyright (c) 2006-2007 Ondrej Zajicek <santiago@crfreenet.org>
+ *
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file COPYING in the main directory of this archive for
+ * more details.
+ *
+ * Code is based on s3fb, some parts are from David Boucher's viafb
+ * (http://davesdomain.org.uk/viafb/)
+ */
+
+#include <linux/version.h>
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/string.h>
+#include <linux/mm.h>
+#include <linux/tty.h>
+#include <linux/slab.h>
+#include <linux/delay.h>
+#include <linux/fb.h>
+#include <linux/svga.h>
+#include <linux/init.h>
+#include <linux/pci.h>
+#include <linux/console.h> /* Why should fb driver call console functions? because acquire_console_sem() */
+#include <video/vga.h>
+
+#ifdef CONFIG_MTRR
+#include <asm/mtrr.h>
+#endif
+
+struct vt8623fb_info {
+	char __iomem *mmio_base;
+	int mtrr_reg;
+	struct vgastate state;
+	struct mutex open_lock;
+	unsigned int ref_count;
+	u32 pseudo_palette[16];
+};
+
+
+
+/* ------------------------------------------------------------------------- */
+
+static const struct svga_fb_format vt8623fb_formats[] = {
+	{ 0,  {0, 6, 0},  {0, 6, 0},  {0, 6, 0}, {0, 0, 0}, 0,
+		FB_TYPE_TEXT, FB_AUX_TEXT_SVGA_STEP8,	FB_VISUAL_PSEUDOCOLOR, 16, 16},
+	{ 4,  {0, 6, 0},  {0, 6, 0},  {0, 6, 0}, {0, 0, 0}, 0,
+		FB_TYPE_PACKED_PIXELS, 0,		FB_VISUAL_PSEUDOCOLOR, 16, 16},
+	{ 4,  {0, 6, 0},  {0, 6, 0},  {0, 6, 0}, {0, 0, 0}, 1,
+		FB_TYPE_INTERLEAVED_PLANES, 1,		FB_VISUAL_PSEUDOCOLOR, 16, 16},
+	{ 8,  {0, 6, 0},  {0, 6, 0},  {0, 6, 0}, {0, 0, 0}, 0,
+		FB_TYPE_PACKED_PIXELS, 0,		FB_VISUAL_PSEUDOCOLOR, 8, 8},
+/*	{16,  {10, 5, 0}, {5, 5, 0},  {0, 5, 0}, {0, 0, 0}, 0,
+		FB_TYPE_PACKED_PIXELS, 0,		FB_VISUAL_TRUECOLOR, 4, 4},	*/
+	{16,  {11, 5, 0}, {5, 6, 0},  {0, 5, 0}, {0, 0, 0}, 0,
+		FB_TYPE_PACKED_PIXELS, 0,		FB_VISUAL_TRUECOLOR, 4, 4},
+	{32,  {16, 8, 0}, {8, 8, 0},  {0, 8, 0}, {0, 0, 0}, 0,
+		FB_TYPE_PACKED_PIXELS, 0,		FB_VISUAL_TRUECOLOR, 2, 2},
+	SVGA_FORMAT_END
+};
+
+static const struct svga_pll vt8623_pll = {2, 127, 2, 7, 0, 3,
+	60000, 300000, 14318};
+
+/* CRT timing register sets */
+
+struct vga_regset vt8623_h_total_regs[]       = {{0x00, 0, 7}, {0x36, 3, 3}, VGA_REGSET_END};
+struct vga_regset vt8623_h_display_regs[]     = {{0x01, 0, 7}, VGA_REGSET_END};
+struct vga_regset vt8623_h_blank_start_regs[] = {{0x02, 0, 7}, VGA_REGSET_END};
+struct vga_regset vt8623_h_blank_end_regs[]   = {{0x03, 0, 4}, {0x05, 7, 7}, {0x33, 5, 5}, VGA_REGSET_END};
+struct vga_regset vt8623_h_sync_start_regs[]  = {{0x04, 0, 7}, {0x33, 4, 4}, VGA_REGSET_END};
+struct vga_regset vt8623_h_sync_end_regs[]    = {{0x05, 0, 4}, VGA_REGSET_END};
+
+struct vga_regset vt8623_v_total_regs[]       = {{0x06, 0, 7}, {0x07, 0, 0}, {0x07, 5, 5}, {0x35, 0, 0}, VGA_REGSET_END};
+struct vga_regset vt8623_v_display_regs[]     = {{0x12, 0, 7}, {0x07, 1, 1}, {0x07, 6, 6}, {0x35, 2, 2}, VGA_REGSET_END};
+struct vga_regset vt8623_v_blank_start_regs[] = {{0x15, 0, 7}, {0x07, 3, 3}, {0x09, 5, 5}, {0x35, 3, 3}, VGA_REGSET_END};
+struct vga_regset vt8623_v_blank_end_regs[]   = {{0x16, 0, 7}, VGA_REGSET_END};
+struct vga_regset vt8623_v_sync_start_regs[]  = {{0x10, 0, 7}, {0x07, 2, 2}, {0x07, 7, 7}, {0x35, 1, 1}, VGA_REGSET_END};
+struct vga_regset vt8623_v_sync_end_regs[]    = {{0x11, 0, 3}, VGA_REGSET_END};
+
+struct vga_regset vt8623_offset_regs[]        = {{0x13, 0, 7}, {0x35, 5, 7}, VGA_REGSET_END};
+struct vga_regset vt8623_line_compare_regs[]  = {{0x18, 0, 7}, {0x07, 4, 4}, {0x09, 6, 6}, {0x33, 0, 2}, {0x35, 4, 4}, VGA_REGSET_END};
+struct vga_regset vt8623_fetch_count_regs[]   = {{0x1C, 0, 7}, {0x1D, 0, 1}, VGA_REGSET_END};
+struct vga_regset vt8623_start_address_regs[] = {{0x0d, 0, 7}, {0x0c, 0, 7}, {0x34, 0, 7}, {0x48, 0, 1}, VGA_REGSET_END};
+
+struct svga_timing_regs vt8623_timing_regs     = {
+	vt8623_h_total_regs, vt8623_h_display_regs, vt8623_h_blank_start_regs,
+	vt8623_h_blank_end_regs, vt8623_h_sync_start_regs, vt8623_h_sync_end_regs,
+	vt8623_v_total_regs, vt8623_v_display_regs, vt8623_v_blank_start_regs,
+	vt8623_v_blank_end_regs, vt8623_v_sync_start_regs, vt8623_v_sync_end_regs,
+};
+
+
+/* ------------------------------------------------------------------------- */
+
+
+/* Module parameters */
+
+static char *mode = "640x480-8@60";
+
+#ifdef CONFIG_MTRR
+static int mtrr = 1;
+#endif
+
+MODULE_AUTHOR("(c) 2006 Ondrej Zajicek <santiago@crfreenet.org>");
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("fbdev driver for integrated graphics core in VIA VT8623 [CLE266]");
+
+module_param(mode, charp, 0644);
+MODULE_PARM_DESC(mode, "Default video mode ('640x480-8@60', etc)");
+
+#ifdef CONFIG_MTRR
+module_param(mtrr, int, 0444);
+MODULE_PARM_DESC(mtrr, "Enable write-combining with MTRR (1=enable, 0=disable, default=1)");
+#endif
+
+
+/* ------------------------------------------------------------------------- */
+
+
+static struct fb_tile_ops vt8623fb_tile_ops = {
+	.fb_settile	= svga_settile,
+	.fb_tilecopy	= svga_tilecopy,
+	.fb_tilefill    = svga_tilefill,
+	.fb_tileblit    = svga_tileblit,
+	.fb_tilecursor  = svga_tilecursor,
+	.fb_get_tilemax = svga_get_tilemax,
+};
+
+
+/* ------------------------------------------------------------------------- */
+
+
+/* image data is MSB-first, fb structure is MSB-first too */
+static inline u32 expand_color(u32 c)
+{
+	return ((c & 1) | ((c & 2) << 7) | ((c & 4) << 14) | ((c & 8) << 21)) * 0xFF;
+}
+
+/* vt8623fb_iplan_imageblit silently assumes that almost everything is 8-pixel aligned */
+static void vt8623fb_iplan_imageblit(struct fb_info *info, const struct fb_image *image)
+{
+	u32 fg = expand_color(image->fg_color);
+	u32 bg = expand_color(image->bg_color);
+	const u8 *src1, *src;
+	u8 __iomem *dst1;
+	u32 __iomem *dst;
+	u32 val;
+	int x, y;
+
+	src1 = image->data;
+	dst1 = info->screen_base + (image->dy * info->fix.line_length)
+		 + ((image->dx / 8) * 4);
+
+	for (y = 0; y < image->height; y++) {
+		src = src1;
+		dst = (u32 __iomem *) dst1;
+		for (x = 0; x < image->width; x += 8) {
+			val = *(src++) * 0x01010101;
+			val = (val & fg) | (~val & bg);
+			fb_writel(val, dst++);
+		}
+		src1 += image->width / 8;
+		dst1 += info->fix.line_length;
+	}
+}
+
+/* vt8623fb_iplan_fillrect silently assumes that almost everything is 8-pixel aligned */
+static void vt8623fb_iplan_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
+{
+	u32 fg = expand_color(rect->color);
+	u8 __iomem *dst1;
+	u32 __iomem *dst;
+	int x, y;
+
+	dst1 = info->screen_base + (rect->dy * info->fix.line_length)
+		 + ((rect->dx / 8) * 4);
+
+	for (y = 0; y < rect->height; y++) {
+		dst = (u32 __iomem *) dst1;
+		for (x = 0; x < rect->width; x += 8) {
+			fb_writel(fg, dst++);
+		}
+		dst1 += info->fix.line_length;
+	}
+}
+
+
+/* image data is MSB-first, fb structure is high-nibble-in-low-byte-first */
+static inline u32 expand_pixel(u32 c)
+{
+	return (((c &  1) << 24) | ((c &  2) << 27) | ((c &  4) << 14) | ((c &   8) << 17) |
+		((c & 16) <<  4) | ((c & 32) <<  7) | ((c & 64) >>  6) | ((c & 128) >>  3)) * 0xF;
+}
+
+/* vt8623fb_cfb4_imageblit silently assumes that almost everything is 8-pixel aligned */
+static void vt8623fb_cfb4_imageblit(struct fb_info *info, const struct fb_image *image)
+{
+	u32 fg = image->fg_color * 0x11111111;
+	u32 bg = image->bg_color * 0x11111111;
+	const u8 *src1, *src;
+	u8 __iomem *dst1;
+	u32 __iomem *dst;
+	u32 val;
+	int x, y;
+
+	src1 = image->data;
+	dst1 = info->screen_base + (image->dy * info->fix.line_length)
+		 + ((image->dx / 8) * 4);
+
+	for (y = 0; y < image->height; y++) {
+		src = src1;
+		dst = (u32 __iomem *) dst1;
+		for (x = 0; x < image->width; x += 8) {
+			val = expand_pixel(*(src++));
+			val = (val & fg) | (~val & bg);
+			fb_writel(val, dst++);
+		}
+		src1 += image->width / 8;
+		dst1 += info->fix.line_length;
+	}
+}
+
+static void vt8623fb_imageblit(struct fb_info *info, const struct fb_image *image)
+{
+	if ((info->var.bits_per_pixel == 4) && (image->depth == 1)
+	    && ((image->width % 8) == 0) && ((image->dx % 8) == 0)) {
+		if (info->fix.type == FB_TYPE_INTERLEAVED_PLANES)
+			vt8623fb_iplan_imageblit(info, image);
+		else
+			vt8623fb_cfb4_imageblit(info, image);
+	} else
+		cfb_imageblit(info, image);
+}
+
+static void vt8623fb_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
+{
+	if ((info->var.bits_per_pixel == 4)
+	    && ((rect->width % 8) == 0) && ((rect->dx % 8) == 0)
+	    && (info->fix.type == FB_TYPE_INTERLEAVED_PLANES))
+		vt8623fb_iplan_fillrect(info, rect);
+	 else
+		cfb_fillrect(info, rect);
+}
+
+
+/* ------------------------------------------------------------------------- */
+
+
+static void vt8623_set_pixclock(struct fb_info *info, u32 pixclock)
+{
+	u16 m, n, r;
+	u8 regval;
+	int rv;
+
+	rv = svga_compute_pll(&vt8623_pll, 1000000000 / pixclock, &m, &n, &r, info->node);
+	if (rv < 0) {
+		printk(KERN_ERR "fb%d: cannot set requested pixclock, keeping old value\n", info->node);
+		return;
+	}
+
+	/* Set VGA misc register  */
+	regval = vga_r(NULL, VGA_MIS_R);
+	vga_w(NULL, VGA_MIS_W, regval | VGA_MIS_ENB_PLL_LOAD);
+
+	/* Set clock registers */
+	vga_wseq(NULL, 0x46, (n  | (r << 6)));
+	vga_wseq(NULL, 0x47, m);
+
+	udelay(1000);
+
+	/* PLL reset */
+	svga_wseq_mask(0x40, 0x02, 0x02);
+	svga_wseq_mask(0x40, 0x00, 0x02);
+}
+
+
+static int vt8623fb_open(struct fb_info *info, int user)
+{
+	struct vt8623fb_info *par = info->par;
+
+	mutex_lock(&(par->open_lock));
+	if (par->ref_count == 0) {
+		memset(&(par->state), 0, sizeof(struct vgastate));
+		par->state.flags = VGA_SAVE_MODE | VGA_SAVE_FONTS | VGA_SAVE_CMAP;
+		par->state.num_crtc = 0xA2;
+		par->state.num_seq = 0x50;
+		save_vga(&(par->state));
+	}
+
+	par->ref_count++;
+	mutex_unlock(&(par->open_lock));
+
+	return 0;
+}
+
+static int vt8623fb_release(struct fb_info *info, int user)
+{
+	struct vt8623fb_info *par = info->par;
+
+	mutex_lock(&(par->open_lock));
+	if (par->ref_count == 0) {
+		mutex_unlock(&(par->open_lock));
+		return -EINVAL;
+	}
+
+	if (par->ref_count == 1)
+		restore_vga(&(par->state));
+
+	par->ref_count--;
+	mutex_unlock(&(par->open_lock));
+
+	return 0;
+}
+
+static int vt8623fb_check_var(struct fb_var_screeninfo *var, struct fb_info *info)
+{
+	int rv, mem, step;
+
+	/* Find appropriate format */
+	rv = svga_match_format (vt8623fb_formats, var, NULL);
+	if (rv < 0)
+	{
+		printk(KERN_ERR "fb%d: unsupported mode requested\n", info->node);
+		return rv;
+	}
+
+	/* Do not allow to have real resoulution larger than virtual */
+	if (var->xres > var->xres_virtual)
+		var->xres_virtual = var->xres;
+
+	if (var->yres > var->yres_virtual)
+		var->yres_virtual = var->yres;
+
+	/* Round up xres_virtual to have proper alignment of lines */
+	step = vt8623fb_formats[rv].xresstep - 1;
+	var->xres_virtual = (var->xres_virtual+step) & ~step;
+
+	/* Check whether have enough memory */
+	mem = ((var->bits_per_pixel * var->xres_virtual) >> 3) * var->yres_virtual;
+	if (mem > info->screen_size)
+	{
+		printk(KERN_ERR "fb%d: not enough framebuffer memory (%d kB requested , %d kB available)\n", info->node, mem >> 10, (unsigned int) (info->screen_size >> 10));
+		return -EINVAL;
+	}
+
+	/* Text mode is limited to 256 kB of memory */
+	if ((var->bits_per_pixel == 0) && (mem > (256*1024)))
+	{
+		printk(KERN_ERR "fb%d: text framebuffer size too large (%d kB requested, 256 kB possible)\n", info->node, mem >> 10);
+		return -EINVAL;
+	}
+
+	rv = svga_check_timings (&vt8623_timing_regs, var, info->node);
+	if (rv < 0)
+	{
+		printk(KERN_ERR "fb%d: invalid timings requested\n", info->node);
+		return rv;
+	}
+
+	/* Interlaced mode not supported */
+	if (var->vmode & FB_VMODE_INTERLACED)
+		return -EINVAL;
+
+	return 0;
+}
+
+
+static int vt8623fb_set_par(struct fb_info *info)
+{
+	u32 mode, offset_value, fetch_value, screen_size;
+	u32 bpp = info->var.bits_per_pixel;
+
+	if (bpp != 0) {
+		info->fix.ypanstep = 1;
+		info->fix.line_length = (info->var.xres_virtual * bpp) / 8;
+
+		info->flags &= ~FBINFO_MISC_TILEBLITTING;
+		info->tileops = NULL;
+
+		/* in 4bpp supports 8p wide tiles only, any tiles otherwise */
+		info->pixmap.blit_x = (bpp == 4) ? (1 << (8 - 1)) : (~(u32)0);
+		info->pixmap.blit_y = ~(u32)0;
+
+		offset_value = (info->var.xres_virtual * bpp) / 64;
+		fetch_value  = ((info->var.xres * bpp) / 128) + 4;
+
+		if (bpp == 4)
+			fetch_value  = (info->var.xres / 8) + 8; /* + 0 is OK */
+
+		screen_size  = info->var.yres_virtual * info->fix.line_length;
+	} else {
+		info->fix.ypanstep = 16;
+		info->fix.line_length = 0;
+
+		info->flags |= FBINFO_MISC_TILEBLITTING;
+		info->tileops = &vt8623fb_tile_ops;
+
+		/* supports 8x16 tiles only */
+		info->pixmap.blit_x = 1 << (8 - 1);
+		info->pixmap.blit_y = 1 << (16 - 1);
+
+		offset_value = info->var.xres_virtual / 16;
+		fetch_value  = (info->var.xres / 8) + 8;
+		screen_size  = (info->var.xres_virtual * info->var.yres_virtual) / 64;
+	}
+
+	info->var.xoffset = 0;
+	info->var.yoffset = 0;
+	info->var.activate = FB_ACTIVATE_NOW;
+
+	/* Unlock registers */
+	svga_wseq_mask(0x10, 0x01, 0x01);
+	svga_wcrt_mask(0x11, 0x00, 0x80);
+	svga_wcrt_mask(0x47, 0x00, 0x01);
+
+	/* Device, screen and sync off */
+	svga_wseq_mask(0x01, 0x20, 0x20);
+	svga_wcrt_mask(0x36, 0x30, 0x30);
+	svga_wcrt_mask(0x17, 0x00, 0x80);
+
+	/* Set default values */
+	svga_set_default_gfx_regs();
+	svga_set_default_atc_regs();
+	svga_set_default_seq_regs();
+	svga_set_default_crt_regs();
+	svga_wcrt_multi(vt8623_line_compare_regs, 0xFFFFFFFF);
+	svga_wcrt_multi(vt8623_start_address_regs, 0);
+
+	svga_wcrt_multi(vt8623_offset_regs, offset_value);
+	svga_wseq_multi(vt8623_fetch_count_regs, fetch_value);
+
+	if (info->var.vmode & FB_VMODE_DOUBLE)
+		svga_wcrt_mask(0x09, 0x80, 0x80);
+	else
+		svga_wcrt_mask(0x09, 0x00, 0x80);
+
+	svga_wseq_mask(0x1E, 0xF0, 0xF0); // DI/DVP bus
+	svga_wseq_mask(0x2A, 0x0F, 0x0F); // DI/DVP bus
+	svga_wseq_mask(0x16, 0x08, 0xBF); // FIFO read treshold
+	vga_wseq(NULL, 0x17, 0x1F);       // FIFO depth
+	vga_wseq(NULL, 0x18, 0x4E);
+	svga_wseq_mask(0x1A, 0x08, 0x08); // enable MMIO ?
+
+	vga_wcrt(NULL, 0x32, 0x00);
+	vga_wcrt(NULL, 0x34, 0x00);
+	vga_wcrt(NULL, 0x6A, 0x80);
+	vga_wcrt(NULL, 0x6A, 0xC0);
+
+	vga_wgfx(NULL, 0x20, 0x00);
+	vga_wgfx(NULL, 0x21, 0x00);
+	vga_wgfx(NULL, 0x22, 0x00);
+
+	/* Set SR15 according to number of bits per pixel */
+	mode = svga_match_format(vt8623fb_formats, &(info->var), &(info->fix));
+	switch (mode) {
+	case 0:
+		pr_debug("fb%d: text mode\n", info->node);
+		svga_set_textmode_vga_regs();
+		svga_wseq_mask(0x15, 0x00, 0xFE);
+		svga_wcrt_mask(0x11, 0x60, 0x70);
+		break;
+	case 1:
+		pr_debug("fb%d: 4 bit pseudocolor\n", info->node);
+		vga_wgfx(NULL, VGA_GFX_MODE, 0x40);
+		svga_wseq_mask(0x15, 0x20, 0xFE);
+		svga_wcrt_mask(0x11, 0x00, 0x70);
+		break;
+	case 2:
+		pr_debug("fb%d: 4 bit pseudocolor, planar\n", info->node);
+		svga_wseq_mask(0x15, 0x00, 0xFE);
+		svga_wcrt_mask(0x11, 0x00, 0x70);
+		break;
+	case 3:
+		pr_debug("fb%d: 8 bit pseudocolor\n", info->node);
+		svga_wseq_mask(0x15, 0x22, 0xFE);
+		break;
+	case 4:
+		pr_debug("fb%d: 5/6/5 truecolor\n", info->node);
+		svga_wseq_mask(0x15, 0xB6, 0xFE);
+		break;
+	case 5:
+		pr_debug("fb%d: 8/8/8 truecolor\n", info->node);
+		svga_wseq_mask(0x15, 0xAE, 0xFE);
+		break;
+	default:
+		printk(KERN_ERR "vt8623fb: unsupported mode - bug\n");
+		return (-EINVAL);
+	}
+
+	vt8623_set_pixclock(info, info->var.pixclock);
+	svga_set_timings(&vt8623_timing_regs, &(info->var), 1, 1,
+			 (info->var.vmode & FB_VMODE_DOUBLE) ? 2 : 1, 1,
+			 1, info->node);
+
+	memset_io(info->screen_base, 0x00, screen_size);
+
+	/* Device and screen back on */
+	svga_wcrt_mask(0x17, 0x80, 0x80);
+	svga_wcrt_mask(0x36, 0x00, 0x30);
+	svga_wseq_mask(0x01, 0x00, 0x20);
+
+	return 0;
+}
+
+
+static int vt8623fb_setcolreg(u_int regno, u_int red, u_int green, u_int blue,
+				u_int transp, struct fb_info *fb)
+{
+	switch (fb->var.bits_per_pixel) {
+	case 0:
+	case 4:
+		if (regno >= 16)
+			return -EINVAL;
+
+		outb(0x0F, VGA_PEL_MSK);
+		outb(regno, VGA_PEL_IW);
+		outb(red >> 10, VGA_PEL_D);
+		outb(green >> 10, VGA_PEL_D);
+		outb(blue >> 10, VGA_PEL_D);
+		break;
+	case 8:
+		if (regno >= 256)
+			return -EINVAL;
+
+		outb(0xFF, VGA_PEL_MSK);
+		outb(regno, VGA_PEL_IW);
+		outb(red >> 10, VGA_PEL_D);
+		outb(green >> 10, VGA_PEL_D);
+		outb(blue >> 10, VGA_PEL_D);
+		break;
+	case 16:
+		if (regno >= 16)
+			return 0;
+
+		if (fb->var.green.length == 5)
+			((u32*)fb->pseudo_palette)[regno] = ((red & 0xF800) >> 1) |
+				((green & 0xF800) >> 6) | ((blue & 0xF800) >> 11);
+		else if (fb->var.green.length == 6)
+			((u32*)fb->pseudo_palette)[regno] = (red & 0xF800) |
+				((green & 0xFC00) >> 5) | ((blue & 0xF800) >> 11);
+		else
+			return -EINVAL;
+		break;
+	case 24:
+	case 32:
+		if (regno >= 16)
+			return 0;
+
+		/* ((transp & 0xFF00) << 16) */
+		((u32*)fb->pseudo_palette)[regno] = ((red & 0xFF00) << 8) |
+			(green & 0xFF00) | ((blue & 0xFF00) >> 8);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+
+static int vt8623fb_blank(int blank_mode, struct fb_info *info)
+{
+	switch (blank_mode) {
+	case FB_BLANK_UNBLANK:
+		pr_debug("fb%d: unblank\n", info->node);
+		svga_wcrt_mask(0x36, 0x00, 0x30);
+		svga_wseq_mask(0x01, 0x00, 0x20);
+		break;
+	case FB_BLANK_NORMAL:
+		pr_debug("fb%d: blank\n", info->node);
+		svga_wcrt_mask(0x36, 0x00, 0x30);
+		svga_wseq_mask(0x01, 0x20, 0x20);
+		break;
+	case FB_BLANK_HSYNC_SUSPEND:
+		pr_debug("fb%d: DPMS standby (hsync off)\n", info->node);
+		svga_wcrt_mask(0x36, 0x10, 0x30);
+		svga_wseq_mask(0x01, 0x20, 0x20);
+		break;
+	case FB_BLANK_VSYNC_SUSPEND:
+		pr_debug("fb%d: DPMS suspend (vsync off)\n", info->node);
+		svga_wcrt_mask(0x36, 0x20, 0x30);
+		svga_wseq_mask(0x01, 0x20, 0x20);
+		break;
+	case FB_BLANK_POWERDOWN:
+		pr_debug("fb%d: DPMS off (no sync)\n", info->node);
+		svga_wcrt_mask(0x36, 0x30, 0x30);
+		svga_wseq_mask(0x01, 0x20, 0x20);
+		break;
+	}
+
+	return 0;
+}
+
+
+static int vt8623fb_pan_display(struct fb_var_screeninfo *var, struct fb_info *info)
+{
+	unsigned int offset;
+
+	/* Calculate the offset */
+	if (var->bits_per_pixel == 0) {
+		offset = (var->yoffset / 16) * var->xres_virtual + var->xoffset;
+		offset = offset >> 3;
+	} else {
+		offset = (var->yoffset * info->fix.line_length) +
+			 (var->xoffset * var->bits_per_pixel / 8);
+		offset = offset >> ((var->bits_per_pixel == 4) ? 2 : 1);
+	}
+
+	/* Set the offset */
+	svga_wcrt_multi(vt8623_start_address_regs, offset);
+
+	return 0;
+}
+
+
+/* ------------------------------------------------------------------------- */
+
+
+/* Frame buffer operations */
+
+static struct fb_ops vt8623fb_ops = {
+	.owner		= THIS_MODULE,
+	.fb_open	= vt8623fb_open,
+	.fb_release	= vt8623fb_release,
+	.fb_check_var	= vt8623fb_check_var,
+	.fb_set_par	= vt8623fb_set_par,
+	.fb_setcolreg	= vt8623fb_setcolreg,
+	.fb_blank	= vt8623fb_blank,
+	.fb_pan_display	= vt8623fb_pan_display,
+	.fb_fillrect	= vt8623fb_fillrect,
+	.fb_copyarea	= cfb_copyarea,
+	.fb_imageblit	= vt8623fb_imageblit,
+	.fb_get_caps    = svga_get_caps,
+};
+
+
+/* PCI probe */
+
+static int __devinit vt8623_pci_probe(struct pci_dev *dev, const struct pci_device_id *id)
+{
+	struct fb_info *info;
+	struct vt8623fb_info *par;
+	unsigned int memsize1, memsize2;
+	int rc;
+
+	/* Ignore secondary VGA device because there is no VGA arbitration */
+	if (! svga_primary_device(dev)) {
+		dev_info(&(dev->dev), "ignoring secondary device\n");
+		return -ENODEV;
+	}
+
+	/* Allocate and fill driver data structure */
+	info = framebuffer_alloc(sizeof(struct vt8623fb_info), NULL);
+	if (! info) {
+		dev_err(&(dev->dev), "cannot allocate memory\n");
+		return -ENOMEM;
+	}
+
+	par = info->par;
+	mutex_init(&par->open_lock);
+
+	info->flags = FBINFO_PARTIAL_PAN_OK | FBINFO_HWACCEL_YPAN;
+	info->fbops = &vt8623fb_ops;
+
+	/* Prepare PCI device */
+
+	rc = pci_enable_device(dev);
+	if (rc < 0) {
+		dev_err(&(dev->dev), "cannot enable PCI device\n");
+		goto err_enable_device;
+	}
+
+	rc = pci_request_regions(dev, "vt8623fb");
+	if (rc < 0) {
+		dev_err(&(dev->dev), "cannot reserve framebuffer region\n");
+		goto err_request_regions;
+	}
+
+	info->fix.smem_start = pci_resource_start(dev, 0);
+	info->fix.smem_len = pci_resource_len(dev, 0);
+	info->fix.mmio_start = pci_resource_start(dev, 1);
+	info->fix.mmio_len = pci_resource_len(dev, 1);
+
+	/* Map physical IO memory address into kernel space */
+	info->screen_base = pci_iomap(dev, 0, 0);
+	if (! info->screen_base) {
+		rc = -ENOMEM;
+		dev_err(&(dev->dev), "iomap for framebuffer failed\n");
+		goto err_iomap_1;
+	}
+
+	par->mmio_base = pci_iomap(dev, 1, 0);
+	if (! par->mmio_base) {
+		rc = -ENOMEM;
+		dev_err(&(dev->dev), "iomap for MMIO failed\n");
+		goto err_iomap_2;
+	}
+
+	/* Find how many physical memory there is on card */
+	memsize1 = (vga_rseq(NULL, 0x34) + 1) >> 1;
+	memsize2 = vga_rseq(NULL, 0x39) << 2;
+
+	if ((16 <= memsize1) && (memsize1 <= 64) && (memsize1 == memsize2))
+		info->screen_size = memsize1 << 20;
+	else {
+		dev_err(&(dev->dev), "memory size detection failed (%x %x), suppose 16 MB\n", memsize1, memsize2);
+		info->screen_size = 16 << 20;
+	}
+
+	info->fix.smem_len = info->screen_size;
+	strcpy(info->fix.id, "VIA VT8623");
+	info->fix.type = FB_TYPE_PACKED_PIXELS;
+	info->fix.visual = FB_VISUAL_PSEUDOCOLOR;
+	info->fix.ypanstep = 0;
+	info->fix.accel = FB_ACCEL_NONE;
+	info->pseudo_palette = (void*)par->pseudo_palette;
+
+	/* Prepare startup mode */
+
+	rc = fb_find_mode(&(info->var), info, mode, NULL, 0, NULL, 8);
+	if (! ((rc == 1) || (rc == 2))) {
+		rc = -EINVAL;
+		dev_err(&(dev->dev), "mode %s not found\n", mode);
+		goto err_find_mode;
+	}
+
+	rc = fb_alloc_cmap(&info->cmap, 256, 0);
+	if (rc < 0) {
+		dev_err(&(dev->dev), "cannot allocate colormap\n");
+		goto err_alloc_cmap;
+	}
+
+	rc = register_framebuffer(info);
+	if (rc < 0) {
+		dev_err(&(dev->dev), "cannot register framebugger\n");
+		goto err_reg_fb;
+	}
+
+	printk(KERN_INFO "fb%d: %s on %s, %d MB RAM\n", info->node, info->fix.id,
+		 pci_name(dev), info->fix.smem_len >> 20);
+
+	/* Record a reference to the driver data */
+	pci_set_drvdata(dev, info);
+
+#ifdef CONFIG_MTRR
+	if (mtrr) {
+		par->mtrr_reg = -1;
+		par->mtrr_reg = mtrr_add(info->fix.smem_start, info->fix.smem_len, MTRR_TYPE_WRCOMB, 1);
+	}
+#endif
+
+	return 0;
+
+	/* Error handling */
+err_reg_fb:
+	fb_dealloc_cmap(&info->cmap);
+err_alloc_cmap:
+err_find_mode:
+	pci_iounmap(dev, par->mmio_base);
+err_iomap_2:
+	pci_iounmap(dev, info->screen_base);
+err_iomap_1:
+	pci_release_regions(dev);
+err_request_regions:
+/*	pci_disable_device(dev); */
+err_enable_device:
+	framebuffer_release(info);
+	return rc;
+}
+
+/* PCI remove */
+
+static void __devexit vt8623_pci_remove(struct pci_dev *dev)
+{
+	struct fb_info *info = pci_get_drvdata(dev);
+	struct vt8623fb_info *par = info->par;
+
+	if (info) {
+#ifdef CONFIG_MTRR
+		if (par->mtrr_reg >= 0) {
+			mtrr_del(par->mtrr_reg, 0, 0);
+			par->mtrr_reg = -1;
+		}
+#endif
+
+		unregister_framebuffer(info);
+		fb_dealloc_cmap(&info->cmap);
+
+		pci_iounmap(dev, info->screen_base);
+		pci_iounmap(dev, par->mmio_base);
+		pci_release_regions(dev);
+/*		pci_disable_device(dev); */
+
+		pci_set_drvdata(dev, NULL);
+		framebuffer_release(info);
+	}
+}
+
+
+#ifdef CONFIG_PM
+/* PCI suspend */
+
+static int vt8623_pci_suspend(struct pci_dev* dev, pm_message_t state)
+{
+	struct fb_info *info = pci_get_drvdata(dev);
+	struct vt8623fb_info *par = info->par;
+
+	dev_info(&(dev->dev), "suspend\n");
+
+	acquire_console_sem();
+	mutex_lock(&(par->open_lock));
+
+	if ((state.event == PM_EVENT_FREEZE) || (par->ref_count == 0)) {
+		mutex_unlock(&(par->open_lock));
+		release_console_sem();
+		return 0;
+	}
+
+	fb_set_suspend(info, 1);
+
+	pci_save_state(dev);
+	pci_disable_device(dev);
+	pci_set_power_state(dev, pci_choose_state(dev, state));
+
+	mutex_unlock(&(par->open_lock));
+	release_console_sem();
+
+	return 0;
+}
+
+
+/* PCI resume */
+
+static int vt8623_pci_resume(struct pci_dev* dev)
+{
+	struct fb_info *info = pci_get_drvdata(dev);
+	struct vt8623fb_info *par = info->par;
+
+	dev_info(&(dev->dev), "resume\n");
+
+	acquire_console_sem();
+	mutex_lock(&(par->open_lock));
+
+	if (par->ref_count == 0) {
+		mutex_unlock(&(par->open_lock));
+		release_console_sem();
+		return 0;
+	}
+
+	pci_set_power_state(dev, PCI_D0);
+	pci_restore_state(dev);
+
+	if (pci_enable_device(dev))
+		goto fail;
+
+	pci_set_master(dev);
+
+	vt8623fb_set_par(info);
+	fb_set_suspend(info, 0);
+
+	mutex_unlock(&(par->open_lock));
+fail:
+	release_console_sem();
+
+	return 0;
+}
+#else
+#define vt8623_pci_suspend NULL
+#define vt8623_pci_resume NULL
+#endif /* CONFIG_PM */
+
+/* List of boards that we are trying to support */
+
+static struct pci_device_id vt8623_devices[] __devinitdata = {
+	{PCI_DEVICE(PCI_VENDOR_ID_VIA, 0x3122)},
+	{0, 0, 0, 0, 0, 0, 0}
+};
+
+MODULE_DEVICE_TABLE(pci, vt8623_devices);
+
+static struct pci_driver vt8623fb_pci_driver = {
+	.name		= "vt8623fb",
+	.id_table	= vt8623_devices,
+	.probe		= vt8623_pci_probe,
+	.remove		= __devexit_p(vt8623_pci_remove),
+	.suspend	= vt8623_pci_suspend,
+	.resume		= vt8623_pci_resume,
+};
+
+/* Cleanup */
+
+static void __exit vt8623fb_cleanup(void)
+{
+	pr_debug("vt8623fb: cleaning up\n");
+	pci_unregister_driver(&vt8623fb_pci_driver);
+}
+
+/* Driver Initialisation */
+
+int __init vt8623fb_init(void)
+{
+
+#ifndef MODULE
+	char *option = NULL;
+
+	if (fb_get_options("vt8623fb", &option))
+		return -ENODEV;
+
+	if (option && *option)
+		mode = option;
+#endif
+
+	pr_debug("vt8623fb: initializing\n");
+	return pci_register_driver(&vt8623fb_pci_driver);
+}
+
+/* ------------------------------------------------------------------------- */
+
+/* Modularization */
+
+module_init(vt8623fb_init);
+module_exit(vt8623fb_cleanup);