Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next

Pull networking updates from David Miller:

 1) More iov_iter conversion work from Al Viro.

    [ The "crypto: switch af_alg_make_sg() to iov_iter" commit was
      wrong, and this pull actually adds an extra commit on top of the
      branch I'm pulling to fix that up, so that the pre-merge state is
      ok.   - Linus ]

 2) Various optimizations to the ipv4 forwarding information base trie
    lookup implementation.  From Alexander Duyck.

 3) Remove sock_iocb altogether, from CHristoph Hellwig.

 4) Allow congestion control algorithm selection via routing metrics.
    From Daniel Borkmann.

 5) Make ipv4 uncached route list per-cpu, from Eric Dumazet.

 6) Handle rfs hash collisions more gracefully, also from Eric Dumazet.

 7) Add xmit_more support to r8169, e1000, and e1000e drivers.  From
    Florian Westphal.

 8) Transparent Ethernet Bridging support for GRO, from Jesse Gross.

 9) Add BPF packet actions to packet scheduler, from Jiri Pirko.

10) Add support for uniqu flow IDs to openvswitch, from Joe Stringer.

11) New NetCP ethernet driver, from Muralidharan Karicheri and Wingman
    Kwok.

12) More sanely handle out-of-window dupacks, which can result in
    serious ACK storms.  From Neal Cardwell.

13) Various rhashtable bug fixes and enhancements, from Herbert Xu,
    Patrick McHardy, and Thomas Graf.

14) Support xmit_more in be2net, from Sathya Perla.

15) Group Policy extensions for vxlan, from Thomas Graf.

16) Remove Checksum Offload support for vxlan, from Tom Herbert.

17) Like ipv4, support lockless transmit over ipv6 UDP sockets.  From
    Vlad Yasevich.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1494+1 commits)
  crypto: fix af_alg_make_sg() conversion to iov_iter
  ipv4: Namespecify TCP PMTU mechanism
  i40e: Fix for stats init function call in Rx setup
  tcp: don't include Fast Open option in SYN-ACK on pure SYN-data
  openvswitch: Only set TUNNEL_VXLAN_OPT if VXLAN-GBP metadata is set
  ipv6: Make __ipv6_select_ident static
  ipv6: Fix fragment id assignment on LE arches.
  bridge: Fix inability to add non-vlan fdb entry
  net: Mellanox: Delete unnecessary checks before the function call "vunmap"
  cxgb4: Add support in cxgb4 to get expansion rom version via ethtool
  ethtool: rename reserved1 memeber in ethtool_drvinfo for expansion ROM version
  net: dsa: Remove redundant phy_attach()
  IB/mlx4: Reset flow support for IB kernel ULPs
  IB/mlx4: Always use the correct port for mirrored multicast attachments
  net/bonding: Fix potential bad memory access during bonding events
  tipc: remove tipc_snprintf
  tipc: nl compat add noop and remove legacy nl framework
  tipc: convert legacy nl stats show to nl compat
  tipc: convert legacy nl net id get to nl compat
  tipc: convert legacy nl net id set to nl compat
  ...

4 files changed, 152 insertions, 46 deletions

diff --git a/drivers/net/wireless/iwlwifi/pcie/drv.c b/drivers/net/wireless/iwlwifi/pcie/drv.c
index d5aadb00dd9e..dbd6bcf52205 100644
--- a/drivers/net/wireless/iwlwifi/pcie/drv.c
+++ b/drivers/net/wireless/iwlwifi/pcie/drv.c
@@ -415,6 +415,8 @@ static const struct pci_device_id iwl_hw_card_ids[] = {
 	{IWL_PCI_DEVICE(0x24F3, 0x0010, iwl8260_2ac_cfg)},
 	{IWL_PCI_DEVICE(0x24F3, 0x0004, iwl8260_2n_cfg)},
 	{IWL_PCI_DEVICE(0x24F4, 0x0030, iwl8260_2ac_cfg)},
+	{IWL_PCI_DEVICE(0x24F5, 0x0010, iwl4165_2ac_cfg)},
+	{IWL_PCI_DEVICE(0x24F6, 0x0030, iwl4165_2ac_cfg)},
 #endif /* CONFIG_IWLMVM */
 
 	{0}
diff --git a/drivers/net/wireless/iwlwifi/pcie/internal.h b/drivers/net/wireless/iwlwifi/pcie/internal.h
index 1aea6b66c594..cae0eb8835ce 100644
--- a/drivers/net/wireless/iwlwifi/pcie/internal.h
+++ b/drivers/net/wireless/iwlwifi/pcie/internal.h
@@ -216,6 +216,7 @@ struct iwl_pcie_txq_scratch_buf {
  * @need_update: indicates need to update read/write index
  * @active: stores if queue is active
  * @ampdu: true if this queue is an ampdu queue for an specific RA/TID
+ * @wd_timeout: queue watchdog timeout (jiffies) - per queue
  *
  * A Tx queue consists of circular buffer of BDs (a.k.a. TFDs, transmit frame
  * descriptors) and required locking structures.
@@ -232,6 +233,7 @@ struct iwl_txq {
 	bool need_update;
 	u8 active;
 	bool ampdu;
+	unsigned long wd_timeout;
 };
 
 static inline dma_addr_t
@@ -259,7 +261,6 @@ iwl_pcie_get_scratchbuf_dma(struct iwl_txq *txq, int idx)
  * @bc_table_dword: true if the BC table expects DWORD (as opposed to bytes)
  * @scd_set_active: should the transport configure the SCD for HCMD queue
  * @rx_page_order: page order for receive buffer size
- * @wd_timeout: queue watchdog timeout (jiffies)
  * @reg_lock: protect hw register access
  * @cmd_in_flight: true when we have a host command in flight
  * @fw_mon_phys: physical address of the buffer for the firmware monitor
@@ -302,6 +303,7 @@ struct iwl_trans_pcie {
 
 	u8 cmd_queue;
 	u8 cmd_fifo;
+	unsigned int cmd_q_wdg_timeout;
 	u8 n_no_reclaim_cmds;
 	u8 no_reclaim_cmds[MAX_NO_RECLAIM_CMDS];
 
@@ -312,12 +314,14 @@ struct iwl_trans_pcie {
 
 	const char *const *command_names;
 
-	/* queue watchdog */
-	unsigned long wd_timeout;
-
 	/*protect hw register */
 	spinlock_t reg_lock;
 	bool cmd_in_flight;
+	bool ref_cmd_in_flight;
+
+	/* protect ref counter */
+	spinlock_t ref_lock;
+	u32 ref_count;
 
 	dma_addr_t fw_mon_phys;
 	struct page *fw_mon_page;
@@ -368,7 +372,8 @@ void iwl_pcie_tx_start(struct iwl_trans *trans, u32 scd_base_addr);
 int iwl_pcie_tx_stop(struct iwl_trans *trans);
 void iwl_pcie_tx_free(struct iwl_trans *trans);
 void iwl_trans_pcie_txq_enable(struct iwl_trans *trans, int queue, u16 ssn,
-			       const struct iwl_trans_txq_scd_cfg *cfg);
+			       const struct iwl_trans_txq_scd_cfg *cfg,
+			       unsigned int wdg_timeout);
 void iwl_trans_pcie_txq_disable(struct iwl_trans *trans, int queue,
 				bool configure_scd);
 int iwl_trans_pcie_tx(struct iwl_trans *trans, struct sk_buff *skb,
@@ -381,6 +386,9 @@ void iwl_trans_pcie_reclaim(struct iwl_trans *trans, int txq_id, int ssn,
 			    struct sk_buff_head *skbs);
 void iwl_trans_pcie_tx_reset(struct iwl_trans *trans);
 
+void iwl_trans_pcie_ref(struct iwl_trans *trans);
+void iwl_trans_pcie_unref(struct iwl_trans *trans);
+
 static inline u16 iwl_pcie_tfd_tb_get_len(struct iwl_tfd *tfd, u8 idx)
 {
 	struct iwl_tfd_tb *tb = &tfd->tbs[idx];
diff --git a/drivers/net/wireless/iwlwifi/pcie/trans.c b/drivers/net/wireless/iwlwifi/pcie/trans.c
index 523fe0c88dcb..69935aa5a1b3 100644
--- a/drivers/net/wireless/iwlwifi/pcie/trans.c
+++ b/drivers/net/wireless/iwlwifi/pcie/trans.c
@@ -75,6 +75,7 @@
 #include "iwl-trans.h"
 #include "iwl-csr.h"
 #include "iwl-prph.h"
+#include "iwl-scd.h"
 #include "iwl-agn-hw.h"
 #include "iwl-fw-error-dump.h"
 #include "internal.h"
@@ -443,10 +444,25 @@ static int iwl_pcie_apm_stop_master(struct iwl_trans *trans)
 	return ret;
 }
 
-static void iwl_pcie_apm_stop(struct iwl_trans *trans)
+static void iwl_pcie_apm_stop(struct iwl_trans *trans, bool op_mode_leave)
 {
 	IWL_DEBUG_INFO(trans, "Stop card, put in low power state\n");
 
+	if (op_mode_leave) {
+		if (!test_bit(STATUS_DEVICE_ENABLED, &trans->status))
+			iwl_pcie_apm_init(trans);
+
+		/* inform ME that we are leaving */
+		if (trans->cfg->device_family == IWL_DEVICE_FAMILY_7000)
+			iwl_set_bits_prph(trans, APMG_PCIDEV_STT_REG,
+					  APMG_PCIDEV_STT_VAL_WAKE_ME);
+		else if (trans->cfg->device_family == IWL_DEVICE_FAMILY_8000)
+			iwl_set_bit(trans, CSR_HW_IF_CONFIG_REG,
+				    CSR_HW_IF_CONFIG_REG_PREPARE |
+				    CSR_HW_IF_CONFIG_REG_ENABLE_PME);
+		mdelay(5);
+	}
+
 	clear_bit(STATUS_DEVICE_ENABLED, &trans->status);
 
 	/* Stop device's DMA activity */
@@ -707,6 +723,11 @@ static int iwl_pcie_load_cpu_sections_8000b(struct iwl_trans *trans,
 
 	*first_ucode_section = last_read_idx;
 
+	if (cpu == 1)
+		iwl_write_direct32(trans, FH_UCODE_LOAD_STATUS, 0xFFFF);
+	else
+		iwl_write_direct32(trans, FH_UCODE_LOAD_STATUS, 0xFFFFFFFF);
+
 	return 0;
 }
 
@@ -893,8 +914,8 @@ static int iwl_pcie_load_given_ucode_8000b(struct iwl_trans *trans,
 	if (ret)
 		return ret;
 
-	/* Notify FW loading is done */
-	iwl_write_direct32(trans, FH_UCODE_LOAD_STATUS, 0xFFFFFFFF);
+	if (trans->dbg_dest_tlv)
+		iwl_pcie_apply_destination(trans);
 
 	/* wait for image verification to complete  */
 	ret = iwl_poll_prph_bit(trans, LMPM_SECURE_BOOT_CPU1_STATUS_ADDR_B0,
@@ -916,6 +937,7 @@ static int iwl_pcie_load_given_ucode_8000b(struct iwl_trans *trans,
 static int iwl_trans_pcie_start_fw(struct iwl_trans *trans,
 				   const struct fw_img *fw, bool run_in_rfkill)
 {
+	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
 	int ret;
 	bool hw_rfkill;
 
@@ -945,6 +967,9 @@ static int iwl_trans_pcie_start_fw(struct iwl_trans *trans,
 		return ret;
 	}
 
+	/* init ref_count to 1 (should be cleared when ucode is loaded) */
+	trans_pcie->ref_count = 1;
+
 	/* make sure rfkill handshake bits are cleared */
 	iwl_write32(trans, CSR_UCODE_DRV_GP1_CLR, CSR_UCODE_SW_BIT_RFKILL);
 	iwl_write32(trans, CSR_UCODE_DRV_GP1_CLR,
@@ -960,7 +985,7 @@ static int iwl_trans_pcie_start_fw(struct iwl_trans *trans,
 
 	/* Load the given image to the HW */
 	if ((trans->cfg->device_family == IWL_DEVICE_FAMILY_8000) &&
-	    (CSR_HW_REV_STEP(trans->hw_rev) == SILICON_B_STEP))
+	    (CSR_HW_REV_STEP(trans->hw_rev) != SILICON_A_STEP))
 		return iwl_pcie_load_given_ucode_8000b(trans, fw);
 	else
 		return iwl_pcie_load_given_ucode(trans, fw);
@@ -1010,7 +1035,7 @@ static void iwl_trans_pcie_stop_device(struct iwl_trans *trans)
 		      CSR_GP_CNTRL_REG_FLAG_MAC_ACCESS_REQ);
 
 	/* Stop the device, and put it in low power state */
-	iwl_pcie_apm_stop(trans);
+	iwl_pcie_apm_stop(trans, false);
 
 	/* stop and reset the on-board processor */
 	iwl_write32(trans, CSR_RESET, CSR_RESET_REG_FLAG_SW_RESET);
@@ -1192,7 +1217,7 @@ static void iwl_trans_pcie_op_mode_leave(struct iwl_trans *trans)
 	iwl_disable_interrupts(trans);
 	spin_unlock(&trans_pcie->irq_lock);
 
-	iwl_pcie_apm_stop(trans);
+	iwl_pcie_apm_stop(trans, true);
 
 	spin_lock(&trans_pcie->irq_lock);
 	iwl_disable_interrupts(trans);
@@ -1244,6 +1269,7 @@ static void iwl_trans_pcie_configure(struct iwl_trans *trans,
 
 	trans_pcie->cmd_queue = trans_cfg->cmd_queue;
 	trans_pcie->cmd_fifo = trans_cfg->cmd_fifo;
+	trans_pcie->cmd_q_wdg_timeout = trans_cfg->cmd_q_wdg_timeout;
 	if (WARN_ON(trans_cfg->n_no_reclaim_cmds > MAX_NO_RECLAIM_CMDS))
 		trans_pcie->n_no_reclaim_cmds = 0;
 	else
@@ -1258,9 +1284,6 @@ static void iwl_trans_pcie_configure(struct iwl_trans *trans,
 	else
 		trans_pcie->rx_page_order = get_order(4 * 1024);
 
-	trans_pcie->wd_timeout =
-		msecs_to_jiffies(trans_cfg->queue_watchdog_timeout);
-
 	trans_pcie->command_names = trans_cfg->command_names;
 	trans_pcie->bc_table_dword = trans_cfg->bc_table_dword;
 	trans_pcie->scd_set_active = trans_cfg->scd_set_active;
@@ -1540,6 +1563,38 @@ static void iwl_trans_pcie_set_bits_mask(struct iwl_trans *trans, u32 reg,
 	spin_unlock_irqrestore(&trans_pcie->reg_lock, flags);
 }
 
+void iwl_trans_pcie_ref(struct iwl_trans *trans)
+{
+	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
+	unsigned long flags;
+
+	if (iwlwifi_mod_params.d0i3_disable)
+		return;
+
+	spin_lock_irqsave(&trans_pcie->ref_lock, flags);
+	IWL_DEBUG_RPM(trans, "ref_counter: %d\n", trans_pcie->ref_count);
+	trans_pcie->ref_count++;
+	spin_unlock_irqrestore(&trans_pcie->ref_lock, flags);
+}
+
+void iwl_trans_pcie_unref(struct iwl_trans *trans)
+{
+	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
+	unsigned long flags;
+
+	if (iwlwifi_mod_params.d0i3_disable)
+		return;
+
+	spin_lock_irqsave(&trans_pcie->ref_lock, flags);
+	IWL_DEBUG_RPM(trans, "ref_counter: %d\n", trans_pcie->ref_count);
+	if (WARN_ON_ONCE(trans_pcie->ref_count == 0)) {
+		spin_unlock_irqrestore(&trans_pcie->ref_lock, flags);
+		return;
+	}
+	trans_pcie->ref_count--;
+	spin_unlock_irqrestore(&trans_pcie->ref_lock, flags);
+}
+
 static const char *get_csr_string(int cmd)
 {
 #define IWL_CMD(x) case x: return #x
@@ -2264,6 +2319,9 @@ static const struct iwl_trans_ops trans_ops_pcie = {
 	.release_nic_access = iwl_trans_pcie_release_nic_access,
 	.set_bits_mask = iwl_trans_pcie_set_bits_mask,
 
+	.ref = iwl_trans_pcie_ref,
+	.unref = iwl_trans_pcie_unref,
+
 	.dump_data = iwl_trans_pcie_dump_data,
 };
 
@@ -2291,6 +2349,7 @@ struct iwl_trans *iwl_trans_pcie_alloc(struct pci_dev *pdev,
 	trans_pcie->trans = trans;
 	spin_lock_init(&trans_pcie->irq_lock);
 	spin_lock_init(&trans_pcie->reg_lock);
+	spin_lock_init(&trans_pcie->ref_lock);
 	init_waitqueue_head(&trans_pcie->ucode_write_waitq);
 
 	err = pci_enable_device(pdev);
@@ -2404,6 +2463,7 @@ struct iwl_trans *iwl_trans_pcie_alloc(struct pci_dev *pdev,
 	}
 
 	trans_pcie->inta_mask = CSR_INI_SET_MASK;
+	trans->d0i3_mode = IWL_D0I3_MODE_ON_SUSPEND;
 
 	return trans;
 
diff --git a/drivers/net/wireless/iwlwifi/pcie/tx.c b/drivers/net/wireless/iwlwifi/pcie/tx.c
index 8a6c7a084aa1..af0bce736358 100644
--- a/drivers/net/wireless/iwlwifi/pcie/tx.c
+++ b/drivers/net/wireless/iwlwifi/pcie/tx.c
@@ -147,7 +147,6 @@ static void iwl_pcie_free_dma_ptr(struct iwl_trans *trans,
 static void iwl_pcie_txq_stuck_timer(unsigned long data)
 {
 	struct iwl_txq *txq = (void *)data;
-	struct iwl_queue *q = &txq->q;
 	struct iwl_trans_pcie *trans_pcie = txq->trans_pcie;
 	struct iwl_trans *trans = iwl_trans_pcie_get_trans(trans_pcie);
 	u32 scd_sram_addr = trans_pcie->scd_base_addr +
@@ -164,7 +163,7 @@ static void iwl_pcie_txq_stuck_timer(unsigned long data)
 	spin_unlock(&txq->lock);
 
 	IWL_ERR(trans, "Queue %d stuck for %u ms.\n", txq->q.id,
-		jiffies_to_msecs(trans_pcie->wd_timeout));
+		jiffies_to_msecs(txq->wd_timeout));
 	IWL_ERR(trans, "Current SW read_ptr %d write_ptr %d\n",
 		txq->q.read_ptr, txq->q.write_ptr);
 
@@ -198,11 +197,6 @@ static void iwl_pcie_txq_stuck_timer(unsigned long data)
 			iwl_read_prph(trans, SCD_QUEUE_WRPTR(i)));
 	}
 
-	for (i = q->read_ptr; i != q->write_ptr;
-	     i = iwl_queue_inc_wrap(i))
-		IWL_ERR(trans, "scratch %d = 0x%08x\n", i,
-			le32_to_cpu(txq->scratchbufs[i].scratch));
-
 	iwl_force_nmi(trans);
 }
 
@@ -680,7 +674,8 @@ void iwl_pcie_tx_start(struct iwl_trans *trans, u32 scd_base_addr)
 		iwl_write_prph(trans, SCD_CHAINEXT_EN, 0);
 
 	iwl_trans_ac_txq_enable(trans, trans_pcie->cmd_queue,
-				trans_pcie->cmd_fifo);
+				trans_pcie->cmd_fifo,
+				trans_pcie->cmd_q_wdg_timeout);
 
 	/* Activate all Tx DMA/FIFO channels */
 	iwl_scd_activate_fifos(trans);
@@ -722,7 +717,12 @@ void iwl_trans_pcie_tx_reset(struct iwl_trans *trans)
 	iwl_write_direct32(trans, FH_KW_MEM_ADDR_REG,
 			   trans_pcie->kw.dma >> 4);
 
-	iwl_pcie_tx_start(trans, trans_pcie->scd_base_addr);
+	/*
+	 * Send 0 as the scd_base_addr since the device may have be reset
+	 * while we were in WoWLAN in which case SCD_SRAM_BASE_ADDR will
+	 * contain garbage.
+	 */
+	iwl_pcie_tx_start(trans, 0);
 }
 
 /*
@@ -898,6 +898,10 @@ int iwl_pcie_tx_init(struct iwl_trans *trans)
 		}
 	}
 
+	if (trans->cfg->base_params->num_of_queues > 20)
+		iwl_set_bits_prph(trans, SCD_GP_CTRL,
+				  SCD_GP_CTRL_ENABLE_31_QUEUES);
+
 	return 0;
 error:
 	/*Upon error, free only if we allocated something */
@@ -906,10 +910,9 @@ error:
 	return ret;
 }
 
-static inline void iwl_pcie_txq_progress(struct iwl_trans_pcie *trans_pcie,
-					   struct iwl_txq *txq)
+static inline void iwl_pcie_txq_progress(struct iwl_txq *txq)
 {
-	if (!trans_pcie->wd_timeout)
+	if (!txq->wd_timeout)
 		return;
 
 	/*
@@ -919,7 +922,7 @@ static inline void iwl_pcie_txq_progress(struct iwl_trans_pcie *trans_pcie,
 	if (txq->q.read_ptr == txq->q.write_ptr)
 		del_timer(&txq->stuck_timer);
 	else
-		mod_timer(&txq->stuck_timer, jiffies + trans_pcie->wd_timeout);
+		mod_timer(&txq->stuck_timer, jiffies + txq->wd_timeout);
 }
 
 /* Frees buffers until index _not_ inclusive */
@@ -981,21 +984,35 @@ void iwl_trans_pcie_reclaim(struct iwl_trans *trans, int txq_id, int ssn,
 		iwl_pcie_txq_free_tfd(trans, txq);
 	}
 
-	iwl_pcie_txq_progress(trans_pcie, txq);
+	iwl_pcie_txq_progress(txq);
 
 	if (iwl_queue_space(&txq->q) > txq->q.low_mark)
 		iwl_wake_queue(trans, txq);
+
+	if (q->read_ptr == q->write_ptr) {
+		IWL_DEBUG_RPM(trans, "Q %d - last tx reclaimed\n", q->id);
+		iwl_trans_pcie_unref(trans);
+	}
+
 out:
 	spin_unlock_bh(&txq->lock);
 }
 
-static int iwl_pcie_set_cmd_in_flight(struct iwl_trans *trans)
+static int iwl_pcie_set_cmd_in_flight(struct iwl_trans *trans,
+				      const struct iwl_host_cmd *cmd)
 {
 	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
 	int ret;
 
 	lockdep_assert_held(&trans_pcie->reg_lock);
 
+	if (!(cmd->flags & CMD_SEND_IN_IDLE) &&
+	    !trans_pcie->ref_cmd_in_flight) {
+		trans_pcie->ref_cmd_in_flight = true;
+		IWL_DEBUG_RPM(trans, "set ref_cmd_in_flight - ref\n");
+		iwl_trans_pcie_ref(trans);
+	}
+
 	if (trans_pcie->cmd_in_flight)
 		return 0;
 
@@ -1036,6 +1053,12 @@ static int iwl_pcie_clear_cmd_in_flight(struct iwl_trans *trans)
 
 	lockdep_assert_held(&trans_pcie->reg_lock);
 
+	if (trans_pcie->ref_cmd_in_flight) {
+		trans_pcie->ref_cmd_in_flight = false;
+		IWL_DEBUG_RPM(trans, "clear ref_cmd_in_flight - unref\n");
+		iwl_trans_pcie_unref(trans);
+	}
+
 	if (WARN_ON(!trans_pcie->cmd_in_flight))
 		return 0;
 
@@ -1089,7 +1112,7 @@ static void iwl_pcie_cmdq_reclaim(struct iwl_trans *trans, int txq_id, int idx)
 		spin_unlock_irqrestore(&trans_pcie->reg_lock, flags);
 	}
 
-	iwl_pcie_txq_progress(trans_pcie, txq);
+	iwl_pcie_txq_progress(txq);
 }
 
 static int iwl_pcie_txq_set_ratid_map(struct iwl_trans *trans, u16 ra_tid,
@@ -1122,14 +1145,18 @@ static int iwl_pcie_txq_set_ratid_map(struct iwl_trans *trans, u16 ra_tid,
 #define BUILD_RAxTID(sta_id, tid)	(((sta_id) << 4) + (tid))
 
 void iwl_trans_pcie_txq_enable(struct iwl_trans *trans, int txq_id, u16 ssn,
-			       const struct iwl_trans_txq_scd_cfg *cfg)
+			       const struct iwl_trans_txq_scd_cfg *cfg,
+			       unsigned int wdg_timeout)
 {
 	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
+	struct iwl_txq *txq = &trans_pcie->txq[txq_id];
 	int fifo = -1;
 
 	if (test_and_set_bit(txq_id, trans_pcie->queue_used))
 		WARN_ONCE(1, "queue %d already used - expect issues", txq_id);
 
+	txq->wd_timeout = msecs_to_jiffies(wdg_timeout);
+
 	if (cfg) {
 		fifo = cfg->fifo;
 
@@ -1153,7 +1180,7 @@ void iwl_trans_pcie_txq_enable(struct iwl_trans *trans, int txq_id, u16 ssn,
 
 			/* enable aggregations for the queue */
 			iwl_scd_txq_enable_agg(trans, txq_id);
-			trans_pcie->txq[txq_id].ampdu = true;
+			txq->ampdu = true;
 		} else {
 			/*
 			 * disable aggregations for the queue, this will also
@@ -1162,20 +1189,20 @@ void iwl_trans_pcie_txq_enable(struct iwl_trans *trans, int txq_id, u16 ssn,
 			 */
 			iwl_scd_txq_disable_agg(trans, txq_id);
 
-			ssn = trans_pcie->txq[txq_id].q.read_ptr;
+			ssn = txq->q.read_ptr;
 		}
 	}
 
 	/* Place first TFD at index corresponding to start sequence number.
 	 * Assumes that ssn_idx is valid (!= 0xFFF) */
-	trans_pcie->txq[txq_id].q.read_ptr = (ssn & 0xff);
-	trans_pcie->txq[txq_id].q.write_ptr = (ssn & 0xff);
+	txq->q.read_ptr = (ssn & 0xff);
+	txq->q.write_ptr = (ssn & 0xff);
+	iwl_write_direct32(trans, HBUS_TARG_WRPTR,
+			   (ssn & 0xff) | (txq_id << 8));
 
 	if (cfg) {
 		u8 frame_limit = cfg->frame_limit;
 
-		iwl_write_direct32(trans, HBUS_TARG_WRPTR,
-				   (ssn & 0xff) | (txq_id << 8));
 		iwl_write_prph(trans, SCD_QUEUE_RDPTR(txq_id), ssn);
 
 		/* Set up Tx window size and frame limit for this queue */
@@ -1200,11 +1227,17 @@ void iwl_trans_pcie_txq_enable(struct iwl_trans *trans, int txq_id, u16 ssn,
 		if (txq_id == trans_pcie->cmd_queue &&
 		    trans_pcie->scd_set_active)
 			iwl_scd_enable_set_active(trans, BIT(txq_id));
+
+		IWL_DEBUG_TX_QUEUES(trans,
+				    "Activate queue %d on FIFO %d WrPtr: %d\n",
+				    txq_id, fifo, ssn & 0xff);
+	} else {
+		IWL_DEBUG_TX_QUEUES(trans,
+				    "Activate queue %d WrPtr: %d\n",
+				    txq_id, ssn & 0xff);
 	}
 
-	trans_pcie->txq[txq_id].active = true;
-	IWL_DEBUG_TX_QUEUES(trans, "Activate queue %d on FIFO %d WrPtr: %d\n",
-			    txq_id, fifo, ssn & 0xff);
+	txq->active = true;
 }
 
 void iwl_trans_pcie_txq_disable(struct iwl_trans *trans, int txq_id,
@@ -1469,11 +1502,11 @@ static int iwl_pcie_enqueue_hcmd(struct iwl_trans *trans,
 	trace_iwlwifi_dev_hcmd(trans->dev, cmd, cmd_size, &out_cmd->hdr);
 
 	/* start timer if queue currently empty */
-	if (q->read_ptr == q->write_ptr && trans_pcie->wd_timeout)
-		mod_timer(&txq->stuck_timer, jiffies + trans_pcie->wd_timeout);
+	if (q->read_ptr == q->write_ptr && txq->wd_timeout)
+		mod_timer(&txq->stuck_timer, jiffies + txq->wd_timeout);
 
 	spin_lock_irqsave(&trans_pcie->reg_lock, flags);
-	ret = iwl_pcie_set_cmd_in_flight(trans);
+	ret = iwl_pcie_set_cmd_in_flight(trans, cmd);
 	if (ret < 0) {
 		idx = ret;
 		spin_unlock_irqrestore(&trans_pcie->reg_lock, flags);
@@ -1819,9 +1852,12 @@ int iwl_trans_pcie_tx(struct iwl_trans *trans, struct sk_buff *skb,
 	wait_write_ptr = ieee80211_has_morefrags(fc);
 
 	/* start timer if queue currently empty */
-	if (txq->need_update && q->read_ptr == q->write_ptr &&
-	    trans_pcie->wd_timeout)
-		mod_timer(&txq->stuck_timer, jiffies + trans_pcie->wd_timeout);
+	if (q->read_ptr == q->write_ptr) {
+		if (txq->wd_timeout)
+			mod_timer(&txq->stuck_timer, jiffies + txq->wd_timeout);
+		IWL_DEBUG_RPM(trans, "Q: %d first tx - take ref\n", q->id);
+		iwl_trans_pcie_ref(trans);
+	}
 
 	/* Tell device the write index *just past* this latest filled TFD */
 	q->write_ptr = iwl_queue_inc_wrap(q->write_ptr);