From 62fdfeaf8b1f487060b6e160e7b5cd90287607c9 Mon Sep 17 00:00:00 2001
From: Eric Anholt <eric@anholt.net>
Date: Fri, 21 May 2010 13:26:39 -0700
Subject: drm/i915: Move ringbuffer-related code to intel_ringbuffer.c.

This is preparation for supporting multiple ringbuffers on Ironlake.
The non-copy-and-paste changes are:
- de-staticing functions
- I915_GEM_GPU_DOMAINS moving to i915_drv.h to be used by both files.
- i915_gem_add_request had only half its implementation
  copy-and-pasted out of the middle of it.
---
 drivers/gpu/drm/i915/i915_drv.h | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

(limited to 'drivers/gpu/drm/i915/i915_drv.h')

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 7f797ef1ab39..114653aa9ae2 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -31,6 +31,7 @@
 #define _I915_DRV_H_
 
 #include "i915_reg.h"
+#include "i915_drm.h"
 #include "intel_bios.h"
 #include <linux/io-mapping.h>
 
@@ -55,6 +56,8 @@ enum plane {
 
 #define I915_NUM_PIPE	2
 
+#define I915_GEM_GPU_DOMAINS	(~(I915_GEM_DOMAIN_CPU | I915_GEM_DOMAIN_GTT))
+
 /* Interface history:
  *
  * 1.1: Original.
@@ -849,6 +852,9 @@ extern u32 gm45_get_vblank_counter(struct drm_device *dev, int crtc);
 extern int i915_vblank_swap(struct drm_device *dev, void *data,
 			    struct drm_file *file_priv);
 extern void i915_enable_irq(drm_i915_private_t *dev_priv, u32 mask);
+extern void i915_disable_irq(drm_i915_private_t *dev_priv, u32 mask);
+void ironlake_enable_graphics_irq(drm_i915_private_t *dev_priv, u32 mask);
+void ironlake_disable_graphics_irq(drm_i915_private_t *dev_priv, u32 mask);
 
 void
 i915_enable_pipestat(drm_i915_private_t *dev_priv, int pipe, u32 mask);
@@ -956,6 +962,8 @@ void i915_gem_object_flush_write_domain(struct drm_gem_object *obj);
 
 void i915_gem_shrinker_init(void);
 void i915_gem_shrinker_exit(void);
+int i915_gem_init_pipe_control(struct drm_device *dev);
+void i915_gem_cleanup_pipe_control(struct drm_device *dev);
 
 /* i915_gem_tiling.c */
 void i915_gem_detect_bit_6_swizzle(struct drm_device *dev);
@@ -1006,6 +1014,16 @@ static inline void ironlake_opregion_gse_intr(struct drm_device *dev) { return;
 static inline void opregion_enable_asle(struct drm_device *dev) { return; }
 #endif
 
+/* intel_ringbuffer.c */
+extern void i915_gem_flush(struct drm_device *dev,
+			   uint32_t invalidate_domains,
+			   uint32_t flush_domains);
+extern int i915_dispatch_gem_execbuffer(struct drm_device *dev,
+					struct drm_i915_gem_execbuffer2 *exec,
+					struct drm_clip_rect *cliprects,
+					uint64_t exec_offset);
+extern uint32_t i915_ring_add_request(struct drm_device *dev);
+
 /* modesetting */
 extern void intel_modeset_init(struct drm_device *dev);
 extern void intel_modeset_cleanup(struct drm_device *dev);
-- 
cgit v1.2.3


From d3301d86b4bf2bcf649982ae464211d8bcf9575a Mon Sep 17 00:00:00 2001
From: Eric Anholt <eric@anholt.net>
Date: Fri, 21 May 2010 13:55:54 -0700
Subject: drm/i915: Rename dev_priv->ring to dev_priv->render_ring.

With the advent of the BSD ring, be clear about which ring this is.
The docs are pretty consistent with calling this the Render engine at
this point.
---
 drivers/gpu/drm/i915/i915_debugfs.c     |  8 +++----
 drivers/gpu/drm/i915/i915_dma.c         | 38 ++++++++++++++++-----------------
 drivers/gpu/drm/i915/i915_drv.c         |  2 +-
 drivers/gpu/drm/i915/i915_drv.h         | 20 ++++++++---------
 drivers/gpu/drm/i915/i915_gem.c         |  2 +-
 drivers/gpu/drm/i915/i915_irq.c         | 12 +++++------
 drivers/gpu/drm/i915/intel_ringbuffer.c | 28 ++++++++++++------------
 7 files changed, 55 insertions(+), 55 deletions(-)

(limited to 'drivers/gpu/drm/i915/i915_drv.h')

diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index 322070c0c631..488175c70c7d 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -317,14 +317,14 @@ static int i915_ringbuffer_data(struct seq_file *m, void *data)
 	u8 *virt;
 	uint32_t *ptr, off;
 
-	if (!dev_priv->ring.ring_obj) {
+	if (!dev_priv->render_ring.ring_obj) {
 		seq_printf(m, "No ringbuffer setup\n");
 		return 0;
 	}
 
-	virt = dev_priv->ring.virtual_start;
+	virt = dev_priv->render_ring.virtual_start;
 
-	for (off = 0; off < dev_priv->ring.Size; off += 4) {
+	for (off = 0; off < dev_priv->render_ring.Size; off += 4) {
 		ptr = (uint32_t *)(virt + off);
 		seq_printf(m, "%08x :  %08x\n", off, *ptr);
 	}
@@ -344,7 +344,7 @@ static int i915_ringbuffer_info(struct seq_file *m, void *data)
 
 	seq_printf(m, "RingHead :  %08x\n", head);
 	seq_printf(m, "RingTail :  %08x\n", tail);
-	seq_printf(m, "RingSize :  %08lx\n", dev_priv->ring.Size);
+	seq_printf(m, "RingSize :  %08lx\n", dev_priv->render_ring.Size);
 	seq_printf(m, "Acthd :     %08x\n", I915_READ(IS_I965G(dev) ? ACTHD_I965 : ACTHD));
 
 	return 0;
diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index a657e3315950..6de7eace4319 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -95,7 +95,7 @@ void i915_kernel_lost_context(struct drm_device * dev)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	struct drm_i915_master_private *master_priv;
-	drm_i915_ring_buffer_t *ring = &(dev_priv->ring);
+	drm_i915_ring_buffer_t *ring = &(dev_priv->render_ring);
 
 	/*
 	 * We should never lose context on the ring with modesetting
@@ -128,11 +128,11 @@ static int i915_dma_cleanup(struct drm_device * dev)
 	if (dev->irq_enabled)
 		drm_irq_uninstall(dev);
 
-	if (dev_priv->ring.virtual_start) {
-		drm_core_ioremapfree(&dev_priv->ring.map, dev);
-		dev_priv->ring.virtual_start = NULL;
-		dev_priv->ring.map.handle = NULL;
-		dev_priv->ring.map.size = 0;
+	if (dev_priv->render_ring.virtual_start) {
+		drm_core_ioremapfree(&dev_priv->render_ring.map, dev);
+		dev_priv->render_ring.virtual_start = NULL;
+		dev_priv->render_ring.map.handle = NULL;
+		dev_priv->render_ring.map.size = 0;
 	}
 
 	/* Clear the HWS virtual address at teardown */
@@ -156,24 +156,24 @@ static int i915_initialize(struct drm_device * dev, drm_i915_init_t * init)
 	}
 
 	if (init->ring_size != 0) {
-		if (dev_priv->ring.ring_obj != NULL) {
+		if (dev_priv->render_ring.ring_obj != NULL) {
 			i915_dma_cleanup(dev);
 			DRM_ERROR("Client tried to initialize ringbuffer in "
 				  "GEM mode\n");
 			return -EINVAL;
 		}
 
-		dev_priv->ring.Size = init->ring_size;
+		dev_priv->render_ring.Size = init->ring_size;
 
-		dev_priv->ring.map.offset = init->ring_start;
-		dev_priv->ring.map.size = init->ring_size;
-		dev_priv->ring.map.type = 0;
-		dev_priv->ring.map.flags = 0;
-		dev_priv->ring.map.mtrr = 0;
+		dev_priv->render_ring.map.offset = init->ring_start;
+		dev_priv->render_ring.map.size = init->ring_size;
+		dev_priv->render_ring.map.type = 0;
+		dev_priv->render_ring.map.flags = 0;
+		dev_priv->render_ring.map.mtrr = 0;
 
-		drm_core_ioremap_wc(&dev_priv->ring.map, dev);
+		drm_core_ioremap_wc(&dev_priv->render_ring.map, dev);
 
-		if (dev_priv->ring.map.handle == NULL) {
+		if (dev_priv->render_ring.map.handle == NULL) {
 			i915_dma_cleanup(dev);
 			DRM_ERROR("can not ioremap virtual address for"
 				  " ring buffer\n");
@@ -181,7 +181,7 @@ static int i915_initialize(struct drm_device * dev, drm_i915_init_t * init)
 		}
 	}
 
-	dev_priv->ring.virtual_start = dev_priv->ring.map.handle;
+	dev_priv->render_ring.virtual_start = dev_priv->render_ring.map.handle;
 
 	dev_priv->cpp = init->cpp;
 	dev_priv->back_offset = init->back_offset;
@@ -203,7 +203,7 @@ static int i915_dma_resume(struct drm_device * dev)
 
 	DRM_DEBUG_DRIVER("%s\n", __func__);
 
-	if (dev_priv->ring.map.handle == NULL) {
+	if (dev_priv->render_ring.map.handle == NULL) {
 		DRM_ERROR("can not ioremap virtual address for"
 			  " ring buffer\n");
 		return -ENOMEM;
@@ -332,7 +332,7 @@ static int i915_emit_cmds(struct drm_device * dev, int *buffer, int dwords)
 	int i;
 	RING_LOCALS;
 
-	if ((dwords+1) * sizeof(int) >= dev_priv->ring.Size - 8)
+	if ((dwords+1) * sizeof(int) >= dev_priv->render_ring.Size - 8)
 		return -EINVAL;
 
 	BEGIN_LP_RING((dwords+1)&~1);
@@ -563,7 +563,7 @@ static int i915_quiescent(struct drm_device * dev)
 	drm_i915_private_t *dev_priv = dev->dev_private;
 
 	i915_kernel_lost_context(dev);
-	return i915_wait_ring(dev, dev_priv->ring.Size - 8, __func__);
+	return i915_wait_ring(dev, dev_priv->render_ring.Size - 8, __func__);
 }
 
 static int i915_flush_ioctl(struct drm_device *dev, void *data,
diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index 5c51e45ab68d..a1814f65fdb4 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -389,7 +389,7 @@ int i965_reset(struct drm_device *dev, u8 flags)
 	 */
 	if (drm_core_check_feature(dev, DRIVER_MODESET) ||
 	    !dev_priv->mm.suspended) {
-		drm_i915_ring_buffer_t *ring = &dev_priv->ring;
+		drm_i915_ring_buffer_t *ring = &dev_priv->render_ring;
 		struct drm_gem_object *obj = ring->ring_obj;
 		struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
 		dev_priv->mm.suspended = 0;
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 114653aa9ae2..a39440cf1dee 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -244,7 +244,7 @@ typedef struct drm_i915_private {
 	void __iomem *regs;
 
 	struct pci_dev *bridge_dev;
-	drm_i915_ring_buffer_t ring;
+	drm_i915_ring_buffer_t render_ring;
 
 	drm_dma_handle_t *status_page_dmah;
 	void *hw_status_page;
@@ -1044,7 +1044,7 @@ extern int intel_trans_dp_port_sel (struct drm_crtc *crtc);
  * has access to the ring.
  */
 #define RING_LOCK_TEST_WITH_RETURN(dev, file_priv) do {			\
-	if (((drm_i915_private_t *)dev->dev_private)->ring.ring_obj == NULL) \
+	if (((drm_i915_private_t *)dev->dev_private)->render_ring.ring_obj == NULL) \
 		LOCK_TEST_WITH_RETURN(dev, file_priv);			\
 } while (0)
 
@@ -1066,15 +1066,15 @@ extern int intel_trans_dp_port_sel (struct drm_crtc *crtc);
 	int bytes__ = 4*(n);						\
 	if (I915_VERBOSE) DRM_DEBUG("BEGIN_LP_RING(%d)\n", (n));	\
 	/* a wrap must occur between instructions so pad beforehand */	\
-	if (unlikely (dev_priv->ring.tail + bytes__ > dev_priv->ring.Size)) \
+	if (unlikely (dev_priv->render_ring.tail + bytes__ > dev_priv->render_ring.Size)) \
 		i915_wrap_ring(dev);					\
-	if (unlikely (dev_priv->ring.space < bytes__))			\
+	if (unlikely (dev_priv->render_ring.space < bytes__))			\
 		i915_wait_ring(dev, bytes__, __func__);			\
 	ring_virt__ = (unsigned int *)					\
-	        (dev_priv->ring.virtual_start + dev_priv->ring.tail);	\
-	dev_priv->ring.tail += bytes__;					\
-	dev_priv->ring.tail &= dev_priv->ring.Size - 1;			\
-	dev_priv->ring.space -= bytes__;				\
+	        (dev_priv->render_ring.virtual_start + dev_priv->render_ring.tail);	\
+	dev_priv->render_ring.tail += bytes__;					\
+	dev_priv->render_ring.tail &= dev_priv->render_ring.Size - 1;			\
+	dev_priv->render_ring.space -= bytes__;				\
 } while (0)
 
 #define OUT_RING(n) do {						\
@@ -1084,8 +1084,8 @@ extern int intel_trans_dp_port_sel (struct drm_crtc *crtc);
 
 #define ADVANCE_LP_RING() do {						\
 	if (I915_VERBOSE)						\
-		DRM_DEBUG("ADVANCE_LP_RING %x\n", dev_priv->ring.tail);	\
-	I915_WRITE(PRB0_TAIL, dev_priv->ring.tail);			\
+		DRM_DEBUG("ADVANCE_LP_RING %x\n", dev_priv->render_ring.tail);	\
+	I915_WRITE(PRB0_TAIL, dev_priv->render_ring.tail);			\
 } while(0)
 
 /**
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 4f2f5f8cdca2..95dbe5628a25 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -4378,7 +4378,7 @@ i915_gem_idle(struct drm_device *dev)
 
 	mutex_lock(&dev->struct_mutex);
 
-	if (dev_priv->mm.suspended || dev_priv->ring.ring_obj == NULL) {
+	if (dev_priv->mm.suspended || dev_priv->render_ring.ring_obj == NULL) {
 		mutex_unlock(&dev->struct_mutex);
 		return 0;
 	}
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index 896184bfeb12..dd91c97de968 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -536,17 +536,17 @@ i915_ringbuffer_last_batch(struct drm_device *dev)
 	 */
 	bbaddr = 0;
 	head = I915_READ(PRB0_HEAD) & HEAD_ADDR;
-	ring = (u32 *)(dev_priv->ring.virtual_start + head);
+	ring = (u32 *)(dev_priv->render_ring.virtual_start + head);
 
-	while (--ring >= (u32 *)dev_priv->ring.virtual_start) {
+	while (--ring >= (u32 *)dev_priv->render_ring.virtual_start) {
 		bbaddr = i915_get_bbaddr(dev, ring);
 		if (bbaddr)
 			break;
 	}
 
 	if (bbaddr == 0) {
-		ring = (u32 *)(dev_priv->ring.virtual_start + dev_priv->ring.Size);
-		while (--ring >= (u32 *)dev_priv->ring.virtual_start) {
+		ring = (u32 *)(dev_priv->render_ring.virtual_start + dev_priv->render_ring.Size);
+		while (--ring >= (u32 *)dev_priv->render_ring.virtual_start) {
 			bbaddr = i915_get_bbaddr(dev, ring);
 			if (bbaddr)
 				break;
@@ -639,7 +639,7 @@ static void i915_capture_error_state(struct drm_device *dev)
 	error->batchbuffer[1] = i915_error_object_create(dev, batchbuffer[1]);
 
 	/* Record the ringbuffer */
-	error->ringbuffer = i915_error_object_create(dev, dev_priv->ring.ring_obj);
+	error->ringbuffer = i915_error_object_create(dev, dev_priv->render_ring.ring_obj);
 
 	/* Record buffers on the active list. */
 	error->active_bo = NULL;
@@ -1056,7 +1056,7 @@ int i915_irq_emit(struct drm_device *dev, void *data,
 	drm_i915_irq_emit_t *emit = data;
 	int result;
 
-	if (!dev_priv || !dev_priv->ring.virtual_start) {
+	if (!dev_priv || !dev_priv->render_ring.virtual_start) {
 		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index 13a796fafaee..06058ddb4eed 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -368,7 +368,7 @@ i915_gem_init_ringbuffer(struct drm_device *dev)
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	struct drm_gem_object *obj;
 	struct drm_i915_gem_object *obj_priv;
-	drm_i915_ring_buffer_t *ring = &dev_priv->ring;
+	drm_i915_ring_buffer_t *ring = &dev_priv->render_ring;
 	int ret;
 	u32 head;
 
@@ -403,7 +403,7 @@ i915_gem_init_ringbuffer(struct drm_device *dev)
 	drm_core_ioremap_wc(&ring->map, dev);
 	if (ring->map.handle == NULL) {
 		DRM_ERROR("Failed to map ringbuffer.\n");
-		memset(&dev_priv->ring, 0, sizeof(dev_priv->ring));
+		memset(&dev_priv->render_ring, 0, sizeof(dev_priv->render_ring));
 		i915_gem_object_unpin(obj);
 		drm_gem_object_unreference(obj);
 		i915_gem_cleanup_hws(dev);
@@ -481,15 +481,15 @@ i915_gem_cleanup_ringbuffer(struct drm_device *dev)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 
-	if (dev_priv->ring.ring_obj == NULL)
+	if (dev_priv->render_ring.ring_obj == NULL)
 		return;
 
-	drm_core_ioremapfree(&dev_priv->ring.map, dev);
+	drm_core_ioremapfree(&dev_priv->render_ring.map, dev);
 
-	i915_gem_object_unpin(dev_priv->ring.ring_obj);
-	drm_gem_object_unreference(dev_priv->ring.ring_obj);
-	dev_priv->ring.ring_obj = NULL;
-	memset(&dev_priv->ring, 0, sizeof(dev_priv->ring));
+	i915_gem_object_unpin(dev_priv->render_ring.ring_obj);
+	drm_gem_object_unreference(dev_priv->render_ring.ring_obj);
+	dev_priv->render_ring.ring_obj = NULL;
+	memset(&dev_priv->render_ring, 0, sizeof(dev_priv->render_ring));
 
 	i915_gem_cleanup_hws(dev);
 }
@@ -503,21 +503,21 @@ int i915_wrap_ring(struct drm_device *dev)
 	volatile unsigned int *virt;
 	int rem;
 
-	rem = dev_priv->ring.Size - dev_priv->ring.tail;
-	if (dev_priv->ring.space < rem) {
+	rem = dev_priv->render_ring.Size - dev_priv->render_ring.tail;
+	if (dev_priv->render_ring.space < rem) {
 		int ret = i915_wait_ring(dev, rem, __func__);
 		if (ret)
 			return ret;
 	}
-	dev_priv->ring.space -= rem;
+	dev_priv->render_ring.space -= rem;
 
 	virt = (unsigned int *)
-		(dev_priv->ring.virtual_start + dev_priv->ring.tail);
+		(dev_priv->render_ring.virtual_start + dev_priv->render_ring.tail);
 	rem /= 4;
 	while (rem--)
 		*virt++ = MI_NOOP;
 
-	dev_priv->ring.tail = 0;
+	dev_priv->render_ring.tail = 0;
 
 	return 0;
 }
@@ -525,7 +525,7 @@ int i915_wrap_ring(struct drm_device *dev)
 int i915_wait_ring(struct drm_device * dev, int n, const char *caller)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
-	drm_i915_ring_buffer_t *ring = &(dev_priv->ring);
+	drm_i915_ring_buffer_t *ring = &(dev_priv->render_ring);
 	u32 acthd_reg = IS_I965G(dev) ? ACTHD_I965 : ACTHD;
 	u32 last_acthd = I915_READ(acthd_reg);
 	u32 acthd;
-- 
cgit v1.2.3


From 8187a2b70e34c727a06617441f74f202b6fefaf9 Mon Sep 17 00:00:00 2001
From: Zou Nan hai <nanhai.zou@intel.com>
Date: Fri, 21 May 2010 09:08:55 +0800
Subject: drm/i915: introduce intel_ring_buffer structure (V2)

Introduces a more complete intel_ring_buffer structure with callbacks
for setup and management of a particular ringbuffer, and converts the
render ring buffer consumers to use it.

Signed-off-by: Zou Nan hai <nanhai.zou@intel.com>
Signed-off-by: Xiang Hai hao <haihao.xiang@intel.com>
[anholt: Fixed up whitespace fail and rebased against prep patches]
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_debugfs.c     |   6 +-
 drivers/gpu/drm/i915/i915_dma.c         |  58 ++--
 drivers/gpu/drm/i915/i915_drv.c         |  29 +-
 drivers/gpu/drm/i915/i915_drv.h         |  80 ++---
 drivers/gpu/drm/i915/i915_gem.c         |  76 ++++-
 drivers/gpu/drm/i915/i915_irq.c         |  15 +-
 drivers/gpu/drm/i915/intel_display.c    |   1 -
 drivers/gpu/drm/i915/intel_overlay.c    |   8 -
 drivers/gpu/drm/i915/intel_ringbuffer.c | 582 +++++++++++++++++++-------------
 drivers/gpu/drm/i915/intel_ringbuffer.h | 124 +++++++
 include/drm/i915_drm.h                  |   4 +-
 11 files changed, 606 insertions(+), 377 deletions(-)
 create mode 100644 drivers/gpu/drm/i915/intel_ringbuffer.h

(limited to 'drivers/gpu/drm/i915/i915_drv.h')

diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index 488175c70c7d..4fddf094deb2 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -317,14 +317,14 @@ static int i915_ringbuffer_data(struct seq_file *m, void *data)
 	u8 *virt;
 	uint32_t *ptr, off;
 
-	if (!dev_priv->render_ring.ring_obj) {
+	if (!dev_priv->render_ring.gem_object) {
 		seq_printf(m, "No ringbuffer setup\n");
 		return 0;
 	}
 
 	virt = dev_priv->render_ring.virtual_start;
 
-	for (off = 0; off < dev_priv->render_ring.Size; off += 4) {
+	for (off = 0; off < dev_priv->render_ring.size; off += 4) {
 		ptr = (uint32_t *)(virt + off);
 		seq_printf(m, "%08x :  %08x\n", off, *ptr);
 	}
@@ -344,7 +344,7 @@ static int i915_ringbuffer_info(struct seq_file *m, void *data)
 
 	seq_printf(m, "RingHead :  %08x\n", head);
 	seq_printf(m, "RingTail :  %08x\n", tail);
-	seq_printf(m, "RingSize :  %08lx\n", dev_priv->render_ring.Size);
+	seq_printf(m, "RingSize :  %08lx\n", dev_priv->render_ring.size);
 	seq_printf(m, "Acthd :     %08x\n", I915_READ(IS_I965G(dev) ? ACTHD_I965 : ACTHD));
 
 	return 0;
diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 6de7eace4319..2541428b2fe5 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -40,7 +40,6 @@
 #include <linux/vga_switcheroo.h>
 #include <linux/slab.h>
 
-
 /**
  * Sets up the hardware status page for devices that need a physical address
  * in the register.
@@ -56,10 +55,11 @@ static int i915_init_phys_hws(struct drm_device *dev)
 		DRM_ERROR("Can not allocate hardware status page\n");
 		return -ENOMEM;
 	}
-	dev_priv->hw_status_page = dev_priv->status_page_dmah->vaddr;
+	dev_priv->render_ring.status_page.page_addr
+		= dev_priv->status_page_dmah->vaddr;
 	dev_priv->dma_status_page = dev_priv->status_page_dmah->busaddr;
 
-	memset(dev_priv->hw_status_page, 0, PAGE_SIZE);
+	memset(dev_priv->render_ring.status_page.page_addr, 0, PAGE_SIZE);
 
 	if (IS_I965G(dev))
 		dev_priv->dma_status_page |= (dev_priv->dma_status_page >> 28) &
@@ -95,7 +95,7 @@ void i915_kernel_lost_context(struct drm_device * dev)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	struct drm_i915_master_private *master_priv;
-	drm_i915_ring_buffer_t *ring = &(dev_priv->render_ring);
+	struct intel_ring_buffer *ring = &dev_priv->render_ring;
 
 	/*
 	 * We should never lose context on the ring with modesetting
@@ -108,7 +108,7 @@ void i915_kernel_lost_context(struct drm_device * dev)
 	ring->tail = I915_READ(PRB0_TAIL) & TAIL_ADDR;
 	ring->space = ring->head - (ring->tail + 8);
 	if (ring->space < 0)
-		ring->space += ring->Size;
+		ring->space += ring->size;
 
 	if (!dev->primary->master)
 		return;
@@ -128,12 +128,7 @@ static int i915_dma_cleanup(struct drm_device * dev)
 	if (dev->irq_enabled)
 		drm_irq_uninstall(dev);
 
-	if (dev_priv->render_ring.virtual_start) {
-		drm_core_ioremapfree(&dev_priv->render_ring.map, dev);
-		dev_priv->render_ring.virtual_start = NULL;
-		dev_priv->render_ring.map.handle = NULL;
-		dev_priv->render_ring.map.size = 0;
-	}
+	intel_cleanup_ring_buffer(dev, &dev_priv->render_ring);
 
 	/* Clear the HWS virtual address at teardown */
 	if (I915_NEED_GFX_HWS(dev))
@@ -156,14 +151,14 @@ static int i915_initialize(struct drm_device * dev, drm_i915_init_t * init)
 	}
 
 	if (init->ring_size != 0) {
-		if (dev_priv->render_ring.ring_obj != NULL) {
+		if (dev_priv->render_ring.gem_object != NULL) {
 			i915_dma_cleanup(dev);
 			DRM_ERROR("Client tried to initialize ringbuffer in "
 				  "GEM mode\n");
 			return -EINVAL;
 		}
 
-		dev_priv->render_ring.Size = init->ring_size;
+		dev_priv->render_ring.size = init->ring_size;
 
 		dev_priv->render_ring.map.offset = init->ring_start;
 		dev_priv->render_ring.map.size = init->ring_size;
@@ -201,26 +196,29 @@ static int i915_dma_resume(struct drm_device * dev)
 {
 	drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private;
 
+	struct intel_ring_buffer *ring;
 	DRM_DEBUG_DRIVER("%s\n", __func__);
 
-	if (dev_priv->render_ring.map.handle == NULL) {
+	ring = &dev_priv->render_ring;
+
+	if (ring->map.handle == NULL) {
 		DRM_ERROR("can not ioremap virtual address for"
 			  " ring buffer\n");
 		return -ENOMEM;
 	}
 
 	/* Program Hardware Status Page */
-	if (!dev_priv->hw_status_page) {
+	if (!ring->status_page.page_addr) {
 		DRM_ERROR("Can not find hardware status page\n");
 		return -EINVAL;
 	}
 	DRM_DEBUG_DRIVER("hw status page @ %p\n",
-				dev_priv->hw_status_page);
-
-	if (dev_priv->status_gfx_addr != 0)
-		I915_WRITE(HWS_PGA, dev_priv->status_gfx_addr);
+				ring->status_page.page_addr);
+	if (ring->status_page.gfx_addr != 0)
+		ring->setup_status_page(dev, ring);
 	else
 		I915_WRITE(HWS_PGA, dev_priv->dma_status_page);
+
 	DRM_DEBUG_DRIVER("Enabled hardware status page\n");
 
 	return 0;
@@ -330,9 +328,8 @@ static int i915_emit_cmds(struct drm_device * dev, int *buffer, int dwords)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	int i;
-	RING_LOCALS;
 
-	if ((dwords+1) * sizeof(int) >= dev_priv->render_ring.Size - 8)
+	if ((dwords+1) * sizeof(int) >= dev_priv->render_ring.size - 8)
 		return -EINVAL;
 
 	BEGIN_LP_RING((dwords+1)&~1);
@@ -365,9 +362,7 @@ i915_emit_box(struct drm_device *dev,
 	      struct drm_clip_rect *boxes,
 	      int i, int DR1, int DR4)
 {
-	drm_i915_private_t *dev_priv = dev->dev_private;
 	struct drm_clip_rect box = boxes[i];
-	RING_LOCALS;
 
 	if (box.y2 <= box.y1 || box.x2 <= box.x1 || box.y2 <= 0 || box.x2 <= 0) {
 		DRM_ERROR("Bad box %d,%d..%d,%d\n",
@@ -404,7 +399,6 @@ static void i915_emit_breadcrumb(struct drm_device *dev)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	struct drm_i915_master_private *master_priv = dev->primary->master->driver_priv;
-	RING_LOCALS;
 
 	dev_priv->counter++;
 	if (dev_priv->counter > 0x7FFFFFFFUL)
@@ -458,10 +452,8 @@ static int i915_dispatch_batchbuffer(struct drm_device * dev,
 				     drm_i915_batchbuffer_t * batch,
 				     struct drm_clip_rect *cliprects)
 {
-	drm_i915_private_t *dev_priv = dev->dev_private;
 	int nbox = batch->num_cliprects;
 	int i = 0, count;
-	RING_LOCALS;
 
 	if ((batch->start | batch->used) & 0x7) {
 		DRM_ERROR("alignment");
@@ -510,7 +502,6 @@ static int i915_dispatch_flip(struct drm_device * dev)
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	struct drm_i915_master_private *master_priv =
 		dev->primary->master->driver_priv;
-	RING_LOCALS;
 
 	if (!master_priv->sarea_priv)
 		return -EINVAL;
@@ -563,7 +554,8 @@ static int i915_quiescent(struct drm_device * dev)
 	drm_i915_private_t *dev_priv = dev->dev_private;
 
 	i915_kernel_lost_context(dev);
-	return i915_wait_ring(dev, dev_priv->render_ring.Size - 8, __func__);
+	return intel_wait_ring_buffer(dev, &dev_priv->render_ring,
+				      dev_priv->render_ring.size - 8);
 }
 
 static int i915_flush_ioctl(struct drm_device *dev, void *data,
@@ -805,6 +797,7 @@ static int i915_set_status_page(struct drm_device *dev, void *data,
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	drm_i915_hws_addr_t *hws = data;
+	struct intel_ring_buffer *ring = &dev_priv->render_ring;
 
 	if (!I915_NEED_GFX_HWS(dev))
 		return -EINVAL;
@@ -821,7 +814,7 @@ static int i915_set_status_page(struct drm_device *dev, void *data,
 
 	DRM_DEBUG_DRIVER("set status page addr 0x%08x\n", (u32)hws->addr);
 
-	dev_priv->status_gfx_addr = hws->addr & (0x1ffff<<12);
+	ring->status_page.gfx_addr = hws->addr & (0x1ffff<<12);
 
 	dev_priv->hws_map.offset = dev->agp->base + hws->addr;
 	dev_priv->hws_map.size = 4*1024;
@@ -837,10 +830,10 @@ static int i915_set_status_page(struct drm_device *dev, void *data,
 				" G33 hw status page\n");
 		return -ENOMEM;
 	}
-	dev_priv->hw_status_page = dev_priv->hws_map.handle;
+	ring->status_page.page_addr = dev_priv->hws_map.handle;
+	memset(ring->status_page.page_addr, 0, PAGE_SIZE);
+	I915_WRITE(HWS_PGA, ring->status_page.gfx_addr);
 
-	memset(dev_priv->hw_status_page, 0, PAGE_SIZE);
-	I915_WRITE(HWS_PGA, dev_priv->status_gfx_addr);
 	DRM_DEBUG_DRIVER("load hws HWS_PGA with gfx mem 0x%x\n",
 				dev_priv->status_gfx_addr);
 	DRM_DEBUG_DRIVER("load hws at %p\n",
@@ -1639,7 +1632,6 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)
 
 	spin_lock_init(&dev_priv->user_irq_lock);
 	spin_lock_init(&dev_priv->error_lock);
-	dev_priv->user_irq_refcount = 0;
 	dev_priv->trace_irq_seqno = 0;
 
 	ret = drm_vblank_init(dev, I915_NUM_PIPE);
diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index a1814f65fdb4..c57c54f403da 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -388,33 +388,10 @@ int i965_reset(struct drm_device *dev, u8 flags)
 	 * switched away).
 	 */
 	if (drm_core_check_feature(dev, DRIVER_MODESET) ||
-	    !dev_priv->mm.suspended) {
-		drm_i915_ring_buffer_t *ring = &dev_priv->render_ring;
-		struct drm_gem_object *obj = ring->ring_obj;
-		struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
+			!dev_priv->mm.suspended) {
+		struct intel_ring_buffer *ring = &dev_priv->render_ring;
 		dev_priv->mm.suspended = 0;
-
-		/* Stop the ring if it's running. */
-		I915_WRITE(PRB0_CTL, 0);
-		I915_WRITE(PRB0_TAIL, 0);
-		I915_WRITE(PRB0_HEAD, 0);
-
-		/* Initialize the ring. */
-		I915_WRITE(PRB0_START, obj_priv->gtt_offset);
-		I915_WRITE(PRB0_CTL,
-			   ((obj->size - 4096) & RING_NR_PAGES) |
-			   RING_NO_REPORT |
-			   RING_VALID);
-		if (!drm_core_check_feature(dev, DRIVER_MODESET))
-			i915_kernel_lost_context(dev);
-		else {
-			ring->head = I915_READ(PRB0_HEAD) & HEAD_ADDR;
-			ring->tail = I915_READ(PRB0_TAIL) & TAIL_ADDR;
-			ring->space = ring->head - (ring->tail + 8);
-			if (ring->space < 0)
-				ring->space += ring->Size;
-		}
-
+		ring->init(dev, ring);
 		mutex_unlock(&dev->struct_mutex);
 		drm_irq_uninstall(dev);
 		drm_irq_install(dev);
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index a39440cf1dee..6bb7933d49dc 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -31,8 +31,8 @@
 #define _I915_DRV_H_
 
 #include "i915_reg.h"
-#include "i915_drm.h"
 #include "intel_bios.h"
+#include "intel_ringbuffer.h"
 #include <linux/io-mapping.h>
 
 /* General customization:
@@ -92,16 +92,6 @@ struct drm_i915_gem_phys_object {
 	struct drm_gem_object *cur_obj;
 };
 
-typedef struct _drm_i915_ring_buffer {
-	unsigned long Size;
-	u8 *virtual_start;
-	int head;
-	int tail;
-	int space;
-	drm_local_map_t map;
-	struct drm_gem_object *ring_obj;
-} drm_i915_ring_buffer_t;
-
 struct mem_block {
 	struct mem_block *next;
 	struct mem_block *prev;
@@ -244,7 +234,7 @@ typedef struct drm_i915_private {
 	void __iomem *regs;
 
 	struct pci_dev *bridge_dev;
-	drm_i915_ring_buffer_t render_ring;
+	struct intel_ring_buffer render_ring;
 
 	drm_dma_handle_t *status_page_dmah;
 	void *hw_status_page;
@@ -270,8 +260,6 @@ typedef struct drm_i915_private {
 	atomic_t irq_received;
 	/** Protects user_irq_refcount and irq_mask_reg */
 	spinlock_t user_irq_lock;
-	/** Refcount for i915_user_irq_get() versus i915_user_irq_put(). */
-	int user_irq_refcount;
 	u32 trace_irq_seqno;
 	/** Cached value of IMR to avoid reads in updating the bitfield */
 	u32 irq_mask_reg;
@@ -832,9 +820,7 @@ extern int i915_irq_emit(struct drm_device *dev, void *data,
 			 struct drm_file *file_priv);
 extern int i915_irq_wait(struct drm_device *dev, void *data,
 			 struct drm_file *file_priv);
-void i915_user_irq_get(struct drm_device *dev);
 void i915_trace_irq_get(struct drm_device *dev, u32 seqno);
-void i915_user_irq_put(struct drm_device *dev);
 extern void i915_enable_interrupt (struct drm_device *dev);
 
 extern irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS);
@@ -853,8 +839,10 @@ extern int i915_vblank_swap(struct drm_device *dev, void *data,
 			    struct drm_file *file_priv);
 extern void i915_enable_irq(drm_i915_private_t *dev_priv, u32 mask);
 extern void i915_disable_irq(drm_i915_private_t *dev_priv, u32 mask);
-void ironlake_enable_graphics_irq(drm_i915_private_t *dev_priv, u32 mask);
-void ironlake_disable_graphics_irq(drm_i915_private_t *dev_priv, u32 mask);
+extern void ironlake_enable_graphics_irq(drm_i915_private_t *dev_priv,
+		u32 mask);
+extern void ironlake_disable_graphics_irq(drm_i915_private_t *dev_priv,
+		u32 mask);
 
 void
 i915_enable_pipestat(drm_i915_private_t *dev_priv, int pipe, u32 mask);
@@ -962,8 +950,6 @@ void i915_gem_object_flush_write_domain(struct drm_gem_object *obj);
 
 void i915_gem_shrinker_init(void);
 void i915_gem_shrinker_exit(void);
-int i915_gem_init_pipe_control(struct drm_device *dev);
-void i915_gem_cleanup_pipe_control(struct drm_device *dev);
 
 /* i915_gem_tiling.c */
 void i915_gem_detect_bit_6_swizzle(struct drm_device *dev);
@@ -1014,16 +1000,6 @@ static inline void ironlake_opregion_gse_intr(struct drm_device *dev) { return;
 static inline void opregion_enable_asle(struct drm_device *dev) { return; }
 #endif
 
-/* intel_ringbuffer.c */
-extern void i915_gem_flush(struct drm_device *dev,
-			   uint32_t invalidate_domains,
-			   uint32_t flush_domains);
-extern int i915_dispatch_gem_execbuffer(struct drm_device *dev,
-					struct drm_i915_gem_execbuffer2 *exec,
-					struct drm_clip_rect *cliprects,
-					uint64_t exec_offset);
-extern uint32_t i915_ring_add_request(struct drm_device *dev);
-
 /* modesetting */
 extern void intel_modeset_init(struct drm_device *dev);
 extern void intel_modeset_cleanup(struct drm_device *dev);
@@ -1044,7 +1020,8 @@ extern int intel_trans_dp_port_sel (struct drm_crtc *crtc);
  * has access to the ring.
  */
 #define RING_LOCK_TEST_WITH_RETURN(dev, file_priv) do {			\
-	if (((drm_i915_private_t *)dev->dev_private)->render_ring.ring_obj == NULL) \
+	if (((drm_i915_private_t *)dev->dev_private)->render_ring.gem_object \
+			== NULL)					\
 		LOCK_TEST_WITH_RETURN(dev, file_priv);			\
 } while (0)
 
@@ -1060,32 +1037,27 @@ extern int intel_trans_dp_port_sel (struct drm_crtc *crtc);
 
 #define I915_VERBOSE 0
 
-#define RING_LOCALS	volatile unsigned int *ring_virt__;
-
-#define BEGIN_LP_RING(n) do {						\
-	int bytes__ = 4*(n);						\
-	if (I915_VERBOSE) DRM_DEBUG("BEGIN_LP_RING(%d)\n", (n));	\
-	/* a wrap must occur between instructions so pad beforehand */	\
-	if (unlikely (dev_priv->render_ring.tail + bytes__ > dev_priv->render_ring.Size)) \
-		i915_wrap_ring(dev);					\
-	if (unlikely (dev_priv->render_ring.space < bytes__))			\
-		i915_wait_ring(dev, bytes__, __func__);			\
-	ring_virt__ = (unsigned int *)					\
-	        (dev_priv->render_ring.virtual_start + dev_priv->render_ring.tail);	\
-	dev_priv->render_ring.tail += bytes__;					\
-	dev_priv->render_ring.tail &= dev_priv->render_ring.Size - 1;			\
-	dev_priv->render_ring.space -= bytes__;				\
+#define BEGIN_LP_RING(n)  do { \
+	drm_i915_private_t *dev_priv = dev->dev_private;                \
+	if (I915_VERBOSE)						\
+		DRM_DEBUG("   BEGIN_LP_RING %x\n", (int)(n));		\
+	intel_ring_begin(dev, &dev_priv->render_ring, 4*(n));		\
 } while (0)
 
-#define OUT_RING(n) do {						\
-	if (I915_VERBOSE) DRM_DEBUG("   OUT_RING %x\n", (int)(n));	\
-	*ring_virt__++ = (n);						\
+
+#define OUT_RING(x) do {						\
+	drm_i915_private_t *dev_priv = dev->dev_private;		\
+	if (I915_VERBOSE)						\
+		DRM_DEBUG("   OUT_RING %x\n", (int)(x));		\
+	intel_ring_emit(dev, &dev_priv->render_ring, x);		\
 } while (0)
 
 #define ADVANCE_LP_RING() do {						\
+	drm_i915_private_t *dev_priv = dev->dev_private;                \
 	if (I915_VERBOSE)						\
-		DRM_DEBUG("ADVANCE_LP_RING %x\n", dev_priv->render_ring.tail);	\
-	I915_WRITE(PRB0_TAIL, dev_priv->render_ring.tail);			\
+		DRM_DEBUG("ADVANCE_LP_RING %x\n",			\
+				dev_priv->render_ring.tail);		\
+	intel_ring_advance(dev, &dev_priv->render_ring);		\
 } while(0)
 
 /**
@@ -1103,14 +1075,12 @@ extern int intel_trans_dp_port_sel (struct drm_crtc *crtc);
  *
  * The area from dword 0x20 to 0x3ff is available for driver usage.
  */
-#define READ_HWSP(dev_priv, reg)  (((volatile u32*)(dev_priv->hw_status_page))[reg])
+#define READ_HWSP(dev_priv, reg)  (((volatile u32 *)\
+			(dev_priv->render_ring.status_page.page_addr))[reg])
 #define READ_BREADCRUMB(dev_priv) READ_HWSP(dev_priv, I915_BREADCRUMB_INDEX)
 #define I915_GEM_HWS_INDEX		0x20
 #define I915_BREADCRUMB_INDEX		0x21
 
-extern int i915_wrap_ring(struct drm_device * dev);
-extern int i915_wait_ring(struct drm_device * dev, int n, const char *caller);
-
 #define INTEL_INFO(dev)	(((struct drm_i915_private *) (dev)->dev_private)->info)
 
 #define IS_I830(dev)		((dev)->pci_device == 0x3577)
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 95dbe5628a25..58b6e814fae1 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -1590,6 +1590,7 @@ i915_gem_process_flushing_list(struct drm_device *dev,
 		}
 	}
 }
+
 uint32_t
 i915_add_request(struct drm_device *dev, struct drm_file *file_priv,
 		 uint32_t flush_domains)
@@ -1607,7 +1608,8 @@ i915_add_request(struct drm_device *dev, struct drm_file *file_priv,
 	if (request == NULL)
 		return 0;
 
-	seqno = i915_ring_add_request(dev);
+	seqno = dev_priv->render_ring.add_request(dev, &dev_priv->render_ring,
+						  file_priv, flush_domains);
 
 	DRM_DEBUG_DRIVER("%d\n", seqno);
 
@@ -1645,10 +1647,8 @@ i915_add_request(struct drm_device *dev, struct drm_file *file_priv,
 static uint32_t
 i915_retire_commands(struct drm_device *dev)
 {
-	drm_i915_private_t *dev_priv = dev->dev_private;
 	uint32_t cmd = MI_FLUSH | MI_NO_WRITE_FLUSH;
 	uint32_t flush_domains = 0;
-	RING_LOCALS;
 
 	/* The sampler always gets flushed on i965 (sigh) */
 	if (IS_I965G(dev))
@@ -1746,7 +1746,9 @@ i915_gem_retire_requests(struct drm_device *dev)
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	uint32_t seqno;
 
-	if (!dev_priv->hw_status_page || list_empty(&dev_priv->mm.request_list))
+	struct intel_ring_buffer *ring = &(dev_priv->render_ring);
+	if (!ring->status_page.page_addr
+			|| list_empty(&dev_priv->mm.request_list))
 		return;
 
 	seqno = i915_get_gem_seqno(dev);
@@ -1773,7 +1775,8 @@ i915_gem_retire_requests(struct drm_device *dev)
 
 	if (unlikely (dev_priv->trace_irq_seqno &&
 		      i915_seqno_passed(dev_priv->trace_irq_seqno, seqno))) {
-		i915_user_irq_put(dev);
+
+		ring->user_irq_put(dev, ring);
 		dev_priv->trace_irq_seqno = 0;
 	}
 }
@@ -1803,6 +1806,7 @@ i915_do_wait_request(struct drm_device *dev, uint32_t seqno, int interruptible)
 	u32 ier;
 	int ret = 0;
 
+	struct intel_ring_buffer *ring = &dev_priv->render_ring;
 	BUG_ON(seqno == 0);
 
 	if (atomic_read(&dev_priv->mm.wedged))
@@ -1823,7 +1827,7 @@ i915_do_wait_request(struct drm_device *dev, uint32_t seqno, int interruptible)
 		trace_i915_gem_request_wait_begin(dev, seqno);
 
 		dev_priv->mm.waiting_gem_seqno = seqno;
-		i915_user_irq_get(dev);
+		ring->user_irq_get(dev, ring);
 		if (interruptible)
 			ret = wait_event_interruptible(dev_priv->irq_queue,
 				i915_seqno_passed(i915_get_gem_seqno(dev), seqno) ||
@@ -1833,7 +1837,7 @@ i915_do_wait_request(struct drm_device *dev, uint32_t seqno, int interruptible)
 				i915_seqno_passed(i915_get_gem_seqno(dev), seqno) ||
 				atomic_read(&dev_priv->mm.wedged));
 
-		i915_user_irq_put(dev);
+		ring->user_irq_put(dev, ring);
 		dev_priv->mm.waiting_gem_seqno = 0;
 
 		trace_i915_gem_request_wait_end(dev, seqno);
@@ -1867,6 +1871,19 @@ i915_wait_request(struct drm_device *dev, uint32_t seqno)
 }
 
 
+static void
+i915_gem_flush(struct drm_device *dev,
+	       uint32_t invalidate_domains,
+	       uint32_t flush_domains)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	if (flush_domains & I915_GEM_DOMAIN_CPU)
+		drm_agp_chipset_flush(dev);
+	dev_priv->render_ring.flush(dev, &dev_priv->render_ring,
+			invalidate_domains,
+			flush_domains);
+}
+
 /**
  * Ensures that all rendering to the object has completed and the object is
  * safe to unbind from the GTT or access from the CPU.
@@ -3820,7 +3837,11 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,
 #endif
 
 	/* Exec the batchbuffer */
-	ret = i915_dispatch_gem_execbuffer(dev, args, cliprects, exec_offset);
+	ret = dev_priv->render_ring.dispatch_gem_execbuffer(dev,
+							    &dev_priv->render_ring,
+							    args,
+							    cliprects,
+							    exec_offset);
 	if (ret) {
 		DRM_ERROR("dispatch failed %d\n", ret);
 		goto err;
@@ -4378,7 +4399,8 @@ i915_gem_idle(struct drm_device *dev)
 
 	mutex_lock(&dev->struct_mutex);
 
-	if (dev_priv->mm.suspended || dev_priv->render_ring.ring_obj == NULL) {
+	if (dev_priv->mm.suspended ||
+			dev_priv->render_ring.gem_object == NULL) {
 		mutex_unlock(&dev->struct_mutex);
 		return 0;
 	}
@@ -4420,7 +4442,7 @@ i915_gem_idle(struct drm_device *dev)
  * 965+ support PIPE_CONTROL commands, which provide finer grained control
  * over cache flushing.
  */
-int
+static int
 i915_gem_init_pipe_control(struct drm_device *dev)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
@@ -4459,7 +4481,8 @@ err:
 	return ret;
 }
 
-void
+
+static void
 i915_gem_cleanup_pipe_control(struct drm_device *dev)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
@@ -4476,6 +4499,37 @@ i915_gem_cleanup_pipe_control(struct drm_device *dev)
 	dev_priv->seqno_page = NULL;
 }
 
+int
+i915_gem_init_ringbuffer(struct drm_device *dev)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	int ret;
+	dev_priv->render_ring = render_ring;
+	if (!I915_NEED_GFX_HWS(dev)) {
+		dev_priv->render_ring.status_page.page_addr
+			= dev_priv->status_page_dmah->vaddr;
+		memset(dev_priv->render_ring.status_page.page_addr,
+				0, PAGE_SIZE);
+	}
+	if (HAS_PIPE_CONTROL(dev)) {
+		ret = i915_gem_init_pipe_control(dev);
+		if (ret)
+			return ret;
+	}
+	ret = intel_init_ring_buffer(dev, &dev_priv->render_ring);
+	return ret;
+}
+
+void
+i915_gem_cleanup_ringbuffer(struct drm_device *dev)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+
+	intel_cleanup_ring_buffer(dev, &dev_priv->render_ring);
+	if (HAS_PIPE_CONTROL(dev))
+		i915_gem_cleanup_pipe_control(dev);
+}
+
 int
 i915_gem_entervt_ioctl(struct drm_device *dev, void *data,
 		       struct drm_file *file_priv)
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index dd91c97de968..e07c643c8365 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -545,7 +545,8 @@ i915_ringbuffer_last_batch(struct drm_device *dev)
 	}
 
 	if (bbaddr == 0) {
-		ring = (u32 *)(dev_priv->render_ring.virtual_start + dev_priv->render_ring.Size);
+		ring = (u32 *)(dev_priv->render_ring.virtual_start
+				+ dev_priv->render_ring.size);
 		while (--ring >= (u32 *)dev_priv->render_ring.virtual_start) {
 			bbaddr = i915_get_bbaddr(dev, ring);
 			if (bbaddr)
@@ -639,7 +640,8 @@ static void i915_capture_error_state(struct drm_device *dev)
 	error->batchbuffer[1] = i915_error_object_create(dev, batchbuffer[1]);
 
 	/* Record the ringbuffer */
-	error->ringbuffer = i915_error_object_create(dev, dev_priv->render_ring.ring_obj);
+	error->ringbuffer = i915_error_object_create(dev,
+			dev_priv->render_ring.gem_object);
 
 	/* Record buffers on the active list. */
 	error->active_bo = NULL;
@@ -984,7 +986,6 @@ static int i915_emit_irq(struct drm_device * dev)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	struct drm_i915_master_private *master_priv = dev->primary->master->driver_priv;
-	RING_LOCALS;
 
 	i915_kernel_lost_context(dev);
 
@@ -1009,9 +1010,10 @@ static int i915_emit_irq(struct drm_device * dev)
 void i915_trace_irq_get(struct drm_device *dev, u32 seqno)
 {
 	drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private;
+	struct intel_ring_buffer *render_ring = &dev_priv->render_ring;
 
 	if (dev_priv->trace_irq_seqno == 0)
-		i915_user_irq_get(dev);
+		render_ring->user_irq_get(dev, render_ring);
 
 	dev_priv->trace_irq_seqno = seqno;
 }
@@ -1021,6 +1023,7 @@ static int i915_wait_irq(struct drm_device * dev, int irq_nr)
 	drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private;
 	struct drm_i915_master_private *master_priv = dev->primary->master->driver_priv;
 	int ret = 0;
+	struct intel_ring_buffer *render_ring = &dev_priv->render_ring;
 
 	DRM_DEBUG_DRIVER("irq_nr=%d breadcrumb=%d\n", irq_nr,
 		  READ_BREADCRUMB(dev_priv));
@@ -1034,10 +1037,10 @@ static int i915_wait_irq(struct drm_device * dev, int irq_nr)
 	if (master_priv->sarea_priv)
 		master_priv->sarea_priv->perf_boxes |= I915_BOX_WAIT;
 
-	i915_user_irq_get(dev);
+	render_ring->user_irq_get(dev, render_ring);
 	DRM_WAIT_ON(ret, dev_priv->irq_queue, 3 * DRM_HZ,
 		    READ_BREADCRUMB(dev_priv) >= irq_nr);
-	i915_user_irq_put(dev);
+	render_ring->user_irq_put(dev, render_ring);
 
 	if (ret == -EBUSY) {
 		DRM_ERROR("EBUSY -- rec: %d emitted: %d\n",
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index f469a84cacfd..b867f3c78408 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -4629,7 +4629,6 @@ static int intel_crtc_page_flip(struct drm_crtc *crtc,
 	unsigned long flags;
 	int pipesrc_reg = (intel_crtc->pipe == 0) ? PIPEASRC : PIPEBSRC;
 	int ret, pipesrc;
-	RING_LOCALS;
 
 	work = kzalloc(sizeof *work, GFP_KERNEL);
 	if (work == NULL)
diff --git a/drivers/gpu/drm/i915/intel_overlay.c b/drivers/gpu/drm/i915/intel_overlay.c
index b0e17b06eb6e..93da83782e5e 100644
--- a/drivers/gpu/drm/i915/intel_overlay.c
+++ b/drivers/gpu/drm/i915/intel_overlay.c
@@ -211,9 +211,7 @@ static void intel_overlay_unmap_regs_atomic(struct intel_overlay *overlay)
 static int intel_overlay_on(struct intel_overlay *overlay)
 {
 	struct drm_device *dev = overlay->dev;
-        drm_i915_private_t *dev_priv = dev->dev_private;
 	int ret;
-	RING_LOCALS;
 
 	BUG_ON(overlay->active);
 
@@ -248,7 +246,6 @@ static void intel_overlay_continue(struct intel_overlay *overlay,
         drm_i915_private_t *dev_priv = dev->dev_private;
 	u32 flip_addr = overlay->flip_addr;
 	u32 tmp;
-	RING_LOCALS;
 
 	BUG_ON(!overlay->active);
 
@@ -274,7 +271,6 @@ static int intel_overlay_wait_flip(struct intel_overlay *overlay)
         drm_i915_private_t *dev_priv = dev->dev_private;
 	int ret;
 	u32 tmp;
-	RING_LOCALS;
 
 	if (overlay->last_flip_req != 0) {
 		ret = i915_do_wait_request(dev, overlay->last_flip_req, 1);
@@ -314,9 +310,7 @@ static int intel_overlay_off(struct intel_overlay *overlay)
 {
 	u32 flip_addr = overlay->flip_addr;
 	struct drm_device *dev = overlay->dev;
-        drm_i915_private_t *dev_priv = dev->dev_private;
 	int ret;
-	RING_LOCALS;
 
 	BUG_ON(!overlay->active);
 
@@ -390,11 +384,9 @@ int intel_overlay_recover_from_interrupt(struct intel_overlay *overlay,
 					 int interruptible)
 {
 	struct drm_device *dev = overlay->dev;
-        drm_i915_private_t *dev_priv = dev->dev_private;
 	struct drm_gem_object *obj;
 	u32 flip_addr;
 	int ret;
-	RING_LOCALS;
 
 	if (overlay->hw_wedged == HW_WEDGED)
 		return -EIO;
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index 06058ddb4eed..5715c4d8cce9 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -29,30 +29,24 @@
 
 #include "drmP.h"
 #include "drm.h"
-#include "i915_drm.h"
 #include "i915_drv.h"
+#include "i915_drm.h"
 #include "i915_trace.h"
-#include "intel_drv.h"
 
-void
-i915_gem_flush(struct drm_device *dev,
-	       uint32_t invalidate_domains,
-	       uint32_t flush_domains)
+static void
+render_ring_flush(struct drm_device *dev,
+		struct intel_ring_buffer *ring,
+		u32	invalidate_domains,
+		u32	flush_domains)
 {
-	drm_i915_private_t *dev_priv = dev->dev_private;
-	uint32_t cmd;
-	RING_LOCALS;
-
 #if WATCH_EXEC
 	DRM_INFO("%s: invalidate %08x flush %08x\n", __func__,
 		  invalidate_domains, flush_domains);
 #endif
-	trace_i915_gem_request_flush(dev, dev_priv->mm.next_gem_seqno,
+	u32 cmd;
+	trace_i915_gem_request_flush(dev, ring->next_seqno,
 				     invalidate_domains, flush_domains);
 
-	if (flush_domains & I915_GEM_DOMAIN_CPU)
-		drm_agp_chipset_flush(dev);
-
 	if ((invalidate_domains | flush_domains) & I915_GEM_GPU_DOMAINS) {
 		/*
 		 * read/write caches:
@@ -100,19 +94,130 @@ i915_gem_flush(struct drm_device *dev,
 #if WATCH_EXEC
 		DRM_INFO("%s: queue flush %08x to ring\n", __func__, cmd);
 #endif
-		BEGIN_LP_RING(2);
-		OUT_RING(cmd);
-		OUT_RING(MI_NOOP);
-		ADVANCE_LP_RING();
+		intel_ring_begin(dev, ring, 8);
+		intel_ring_emit(dev, ring, cmd);
+		intel_ring_emit(dev, ring, MI_NOOP);
+		intel_ring_advance(dev, ring);
 	}
+}
+
+static unsigned int render_ring_get_head(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	return I915_READ(PRB0_HEAD) & HEAD_ADDR;
+}
 
+static unsigned int render_ring_get_tail(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	return I915_READ(PRB0_TAIL) & TAIL_ADDR;
 }
+
+static unsigned int render_ring_get_active_head(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	u32 acthd_reg = IS_I965G(dev) ? ACTHD_I965 : ACTHD;
+
+	return I915_READ(acthd_reg);
+}
+
+static void render_ring_advance_ring(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	I915_WRITE(PRB0_TAIL, ring->tail);
+}
+
+static int init_ring_common(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	u32 head;
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	struct drm_i915_gem_object *obj_priv;
+	obj_priv = to_intel_bo(ring->gem_object);
+
+	/* Stop the ring if it's running. */
+	I915_WRITE(ring->regs.ctl, 0);
+	I915_WRITE(ring->regs.head, 0);
+	I915_WRITE(ring->regs.tail, 0);
+
+	/* Initialize the ring. */
+	I915_WRITE(ring->regs.start, obj_priv->gtt_offset);
+	head = ring->get_head(dev, ring);
+
+	/* G45 ring initialization fails to reset head to zero */
+	if (head != 0) {
+		DRM_ERROR("%s head not reset to zero "
+				"ctl %08x head %08x tail %08x start %08x\n",
+				ring->name,
+				I915_READ(ring->regs.ctl),
+				I915_READ(ring->regs.head),
+				I915_READ(ring->regs.tail),
+				I915_READ(ring->regs.start));
+
+		I915_WRITE(ring->regs.head, 0);
+
+		DRM_ERROR("%s head forced to zero "
+				"ctl %08x head %08x tail %08x start %08x\n",
+				ring->name,
+				I915_READ(ring->regs.ctl),
+				I915_READ(ring->regs.head),
+				I915_READ(ring->regs.tail),
+				I915_READ(ring->regs.start));
+	}
+
+	I915_WRITE(ring->regs.ctl,
+			((ring->gem_object->size - PAGE_SIZE) & RING_NR_PAGES)
+			| RING_NO_REPORT | RING_VALID);
+
+	head = I915_READ(ring->regs.head) & HEAD_ADDR;
+	/* If the head is still not zero, the ring is dead */
+	if (head != 0) {
+		DRM_ERROR("%s initialization failed "
+				"ctl %08x head %08x tail %08x start %08x\n",
+				ring->name,
+				I915_READ(ring->regs.ctl),
+				I915_READ(ring->regs.head),
+				I915_READ(ring->regs.tail),
+				I915_READ(ring->regs.start));
+		return -EIO;
+	}
+
+	if (!drm_core_check_feature(dev, DRIVER_MODESET))
+		i915_kernel_lost_context(dev);
+	else {
+		ring->head = ring->get_head(dev, ring);
+		ring->tail = ring->get_tail(dev, ring);
+		ring->space = ring->head - (ring->tail + 8);
+		if (ring->space < 0)
+			ring->space += ring->size;
+	}
+	return 0;
+}
+
+static int init_render_ring(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	int ret = init_ring_common(dev, ring);
+	if (IS_I9XX(dev) && !IS_GEN3(dev)) {
+		I915_WRITE(MI_MODE,
+				(VS_TIMER_DISPATCH) << 16 | VS_TIMER_DISPATCH);
+	}
+	return ret;
+}
+
 #define PIPE_CONTROL_FLUSH(addr)					\
+do {									\
 	OUT_RING(GFX_OP_PIPE_CONTROL | PIPE_CONTROL_QW_WRITE |		\
 		 PIPE_CONTROL_DEPTH_STALL);				\
 	OUT_RING(addr | PIPE_CONTROL_GLOBAL_GTT);			\
 	OUT_RING(0);							\
 	OUT_RING(0);							\
+} while (0)
 
 /**
  * Creates a new sequence number, emitting a write of it to the status page
@@ -122,21 +227,15 @@ i915_gem_flush(struct drm_device *dev,
  *
  * Returned sequence numbers are nonzero on success.
  */
-uint32_t
-i915_ring_add_request(struct drm_device *dev)
+static u32
+render_ring_add_request(struct drm_device *dev,
+		struct intel_ring_buffer *ring,
+		struct drm_file *file_priv,
+		u32 flush_domains)
 {
+	u32 seqno;
 	drm_i915_private_t *dev_priv = dev->dev_private;
-	uint32_t seqno;
-	RING_LOCALS;
-
-	/* Grab the seqno we're going to make this request be, and bump the
-	 * next (skipping 0 so it can be the reserved no-seqno value).
-	 */
-	seqno = dev_priv->mm.next_gem_seqno;
-	dev_priv->mm.next_gem_seqno++;
-	if (dev_priv->mm.next_gem_seqno == 0)
-		dev_priv->mm.next_gem_seqno++;
-
+	seqno = intel_ring_get_seqno(dev, ring);
 	if (HAS_PIPE_CONTROL(dev)) {
 		u32 scratch_addr = dev_priv->seqno_gfx_addr + 128;
 
@@ -181,13 +280,26 @@ i915_ring_add_request(struct drm_device *dev)
 	return seqno;
 }
 
-void i915_user_irq_get(struct drm_device *dev)
+static u32
+render_ring_get_gem_seqno(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private;
+	if (HAS_PIPE_CONTROL(dev))
+		return ((volatile u32 *)(dev_priv->seqno_page))[0];
+	else
+		return intel_read_status_page(ring, I915_GEM_HWS_INDEX);
+}
+
+static void
+render_ring_get_user_irq(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
 {
 	drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private;
 	unsigned long irqflags;
 
 	spin_lock_irqsave(&dev_priv->user_irq_lock, irqflags);
-	if (dev->irq_enabled && (++dev_priv->user_irq_refcount == 1)) {
+	if (dev->irq_enabled && (++ring->user_irq_refcount == 1)) {
 		if (HAS_PCH_SPLIT(dev))
 			ironlake_enable_graphics_irq(dev_priv, GT_PIPE_NOTIFY);
 		else
@@ -196,14 +308,16 @@ void i915_user_irq_get(struct drm_device *dev)
 	spin_unlock_irqrestore(&dev_priv->user_irq_lock, irqflags);
 }
 
-void i915_user_irq_put(struct drm_device *dev)
+static void
+render_ring_put_user_irq(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
 {
 	drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private;
 	unsigned long irqflags;
 
 	spin_lock_irqsave(&dev_priv->user_irq_lock, irqflags);
-	BUG_ON(dev->irq_enabled && dev_priv->user_irq_refcount <= 0);
-	if (dev->irq_enabled && (--dev_priv->user_irq_refcount == 0)) {
+	BUG_ON(dev->irq_enabled && ring->user_irq_refcount <= 0);
+	if (dev->irq_enabled && (--ring->user_irq_refcount == 0)) {
 		if (HAS_PCH_SPLIT(dev))
 			ironlake_disable_graphics_irq(dev_priv, GT_PIPE_NOTIFY);
 		else
@@ -212,20 +326,31 @@ void i915_user_irq_put(struct drm_device *dev)
 	spin_unlock_irqrestore(&dev_priv->user_irq_lock, irqflags);
 }
 
-/** Dispatch a batchbuffer to the ring
- */
-int
-i915_dispatch_gem_execbuffer(struct drm_device *dev,
-			      struct drm_i915_gem_execbuffer2 *exec,
-			      struct drm_clip_rect *cliprects,
-			      uint64_t exec_offset)
+static void render_setup_status_page(struct drm_device *dev,
+	struct	intel_ring_buffer *ring)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	if (IS_GEN6(dev)) {
+		I915_WRITE(HWS_PGA_GEN6, ring->status_page.gfx_addr);
+		I915_READ(HWS_PGA_GEN6); /* posting read */
+	} else {
+		I915_WRITE(HWS_PGA, ring->status_page.gfx_addr);
+		I915_READ(HWS_PGA); /* posting read */
+	}
+
+}
+
+static int
+render_ring_dispatch_gem_execbuffer(struct drm_device *dev,
+		struct intel_ring_buffer *ring,
+		struct drm_i915_gem_execbuffer2 *exec,
+		struct drm_clip_rect *cliprects,
+		uint64_t exec_offset)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	int nbox = exec->num_cliprects;
 	int i = 0, count;
 	uint32_t exec_start, exec_len;
-	RING_LOCALS;
-
 	exec_start = (uint32_t) exec_offset + exec->batch_start_offset;
 	exec_len = (uint32_t) exec->batch_len;
 
@@ -242,74 +367,61 @@ i915_dispatch_gem_execbuffer(struct drm_device *dev,
 		}
 
 		if (IS_I830(dev) || IS_845G(dev)) {
-			BEGIN_LP_RING(4);
-			OUT_RING(MI_BATCH_BUFFER);
-			OUT_RING(exec_start | MI_BATCH_NON_SECURE);
-			OUT_RING(exec_start + exec_len - 4);
-			OUT_RING(0);
-			ADVANCE_LP_RING();
+			intel_ring_begin(dev, ring, 4);
+			intel_ring_emit(dev, ring, MI_BATCH_BUFFER);
+			intel_ring_emit(dev, ring,
+					exec_start | MI_BATCH_NON_SECURE);
+			intel_ring_emit(dev, ring, exec_start + exec_len - 4);
+			intel_ring_emit(dev, ring, 0);
 		} else {
-			BEGIN_LP_RING(2);
+			intel_ring_begin(dev, ring, 4);
 			if (IS_I965G(dev)) {
-				OUT_RING(MI_BATCH_BUFFER_START |
-					 (2 << 6) |
-					 MI_BATCH_NON_SECURE_I965);
-				OUT_RING(exec_start);
+				intel_ring_emit(dev, ring,
+						MI_BATCH_BUFFER_START | (2 << 6)
+						| MI_BATCH_NON_SECURE_I965);
+				intel_ring_emit(dev, ring, exec_start);
 			} else {
-				OUT_RING(MI_BATCH_BUFFER_START |
-					 (2 << 6));
-				OUT_RING(exec_start | MI_BATCH_NON_SECURE);
+				intel_ring_emit(dev, ring, MI_BATCH_BUFFER_START
+						| (2 << 6));
+				intel_ring_emit(dev, ring, exec_start |
+						MI_BATCH_NON_SECURE);
 			}
-			ADVANCE_LP_RING();
 		}
+		intel_ring_advance(dev, ring);
 	}
 
 	/* XXX breadcrumb */
 	return 0;
 }
 
-static void
-i915_gem_cleanup_hws(struct drm_device *dev)
+static void cleanup_status_page(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	struct drm_gem_object *obj;
 	struct drm_i915_gem_object *obj_priv;
 
-	if (dev_priv->hws_obj == NULL)
+	obj = ring->status_page.obj;
+	if (obj == NULL)
 		return;
-
-	obj = dev_priv->hws_obj;
 	obj_priv = to_intel_bo(obj);
 
 	kunmap(obj_priv->pages[0]);
 	i915_gem_object_unpin(obj);
 	drm_gem_object_unreference(obj);
-	dev_priv->hws_obj = NULL;
+	ring->status_page.obj = NULL;
 
 	memset(&dev_priv->hws_map, 0, sizeof(dev_priv->hws_map));
-	dev_priv->hw_status_page = NULL;
-
-	if (HAS_PIPE_CONTROL(dev))
-		i915_gem_cleanup_pipe_control(dev);
-
-	/* Write high address into HWS_PGA when disabling. */
-	I915_WRITE(HWS_PGA, 0x1ffff000);
 }
 
-static int
-i915_gem_init_hws(struct drm_device *dev)
+static int init_status_page(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	struct drm_gem_object *obj;
 	struct drm_i915_gem_object *obj_priv;
 	int ret;
 
-	/* If we need a physical address for the status page, it's already
-	 * initialized at driver load time.
-	 */
-	if (!I915_NEED_GFX_HWS(dev))
-		return 0;
-
 	obj = i915_gem_alloc_object(dev, 4096);
 	if (obj == NULL) {
 		DRM_ERROR("Failed to allocate status page\n");
@@ -321,36 +433,21 @@ i915_gem_init_hws(struct drm_device *dev)
 
 	ret = i915_gem_object_pin(obj, 4096);
 	if (ret != 0) {
-		drm_gem_object_unreference(obj);
 		goto err_unref;
 	}
 
-	dev_priv->status_gfx_addr = obj_priv->gtt_offset;
-
-	dev_priv->hw_status_page = kmap(obj_priv->pages[0]);
-	if (dev_priv->hw_status_page == NULL) {
-		DRM_ERROR("Failed to map status page.\n");
+	ring->status_page.gfx_addr = obj_priv->gtt_offset;
+	ring->status_page.page_addr = kmap(obj_priv->pages[0]);
+	if (ring->status_page.page_addr == NULL) {
 		memset(&dev_priv->hws_map, 0, sizeof(dev_priv->hws_map));
-		ret = -EINVAL;
 		goto err_unpin;
 	}
+	ring->status_page.obj = obj;
+	memset(ring->status_page.page_addr, 0, PAGE_SIZE);
 
-	if (HAS_PIPE_CONTROL(dev)) {
-		ret = i915_gem_init_pipe_control(dev);
-		if (ret)
-			goto err_unpin;
-	}
-
-	dev_priv->hws_obj = obj;
-	memset(dev_priv->hw_status_page, 0, PAGE_SIZE);
-	if (IS_GEN6(dev)) {
-		I915_WRITE(HWS_PGA_GEN6, dev_priv->status_gfx_addr);
-		I915_READ(HWS_PGA_GEN6); /* posting read */
-	} else {
-		I915_WRITE(HWS_PGA, dev_priv->status_gfx_addr);
-		I915_READ(HWS_PGA); /* posting read */
-	}
-	DRM_DEBUG_DRIVER("hws offset: 0x%08x\n", dev_priv->status_gfx_addr);
+	ring->setup_status_page(dev, ring);
+	DRM_DEBUG_DRIVER("%s hws offset: 0x%08x\n",
+			ring->name, ring->status_page.gfx_addr);
 
 	return 0;
 
@@ -359,43 +456,42 @@ err_unpin:
 err_unref:
 	drm_gem_object_unreference(obj);
 err:
-	return 0;
+	return ret;
 }
 
-int
-i915_gem_init_ringbuffer(struct drm_device *dev)
+
+int intel_init_ring_buffer(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
 {
-	drm_i915_private_t *dev_priv = dev->dev_private;
-	struct drm_gem_object *obj;
-	struct drm_i915_gem_object *obj_priv;
-	drm_i915_ring_buffer_t *ring = &dev_priv->render_ring;
 	int ret;
-	u32 head;
+	struct drm_i915_gem_object *obj_priv;
+	struct drm_gem_object *obj;
+	ring->dev = dev;
 
-	ret = i915_gem_init_hws(dev);
-	if (ret != 0)
-		return ret;
+	if (I915_NEED_GFX_HWS(dev)) {
+		ret = init_status_page(dev, ring);
+		if (ret)
+			return ret;
+	}
 
-	obj = i915_gem_alloc_object(dev, 128 * 1024);
+	obj = i915_gem_alloc_object(dev, ring->size);
 	if (obj == NULL) {
 		DRM_ERROR("Failed to allocate ringbuffer\n");
-		i915_gem_cleanup_hws(dev);
-		return -ENOMEM;
+		ret = -ENOMEM;
+		goto cleanup;
 	}
-	obj_priv = to_intel_bo(obj);
 
-	ret = i915_gem_object_pin(obj, 4096);
+	ring->gem_object = obj;
+
+	ret = i915_gem_object_pin(obj, ring->alignment);
 	if (ret != 0) {
 		drm_gem_object_unreference(obj);
-		i915_gem_cleanup_hws(dev);
-		return ret;
+		goto cleanup;
 	}
 
-	/* Set up the kernel mapping for the ring. */
-	ring->Size = obj->size;
-
+	obj_priv = to_intel_bo(obj);
+	ring->map.size = ring->size;
 	ring->map.offset = dev->agp->base + obj_priv->gtt_offset;
-	ring->map.size = obj->size;
 	ring->map.type = 0;
 	ring->map.flags = 0;
 	ring->map.mtrr = 0;
@@ -403,143 +499,85 @@ i915_gem_init_ringbuffer(struct drm_device *dev)
 	drm_core_ioremap_wc(&ring->map, dev);
 	if (ring->map.handle == NULL) {
 		DRM_ERROR("Failed to map ringbuffer.\n");
-		memset(&dev_priv->render_ring, 0, sizeof(dev_priv->render_ring));
 		i915_gem_object_unpin(obj);
 		drm_gem_object_unreference(obj);
-		i915_gem_cleanup_hws(dev);
-		return -EINVAL;
-	}
-	ring->ring_obj = obj;
-	ring->virtual_start = ring->map.handle;
-
-	/* Stop the ring if it's running. */
-	I915_WRITE(PRB0_CTL, 0);
-	I915_WRITE(PRB0_TAIL, 0);
-	I915_WRITE(PRB0_HEAD, 0);
-
-	/* Initialize the ring. */
-	I915_WRITE(PRB0_START, obj_priv->gtt_offset);
-	head = I915_READ(PRB0_HEAD) & HEAD_ADDR;
-
-	/* G45 ring initialization fails to reset head to zero */
-	if (head != 0) {
-		DRM_ERROR("Ring head not reset to zero "
-			  "ctl %08x head %08x tail %08x start %08x\n",
-			  I915_READ(PRB0_CTL),
-			  I915_READ(PRB0_HEAD),
-			  I915_READ(PRB0_TAIL),
-			  I915_READ(PRB0_START));
-		I915_WRITE(PRB0_HEAD, 0);
-
-		DRM_ERROR("Ring head forced to zero "
-			  "ctl %08x head %08x tail %08x start %08x\n",
-			  I915_READ(PRB0_CTL),
-			  I915_READ(PRB0_HEAD),
-			  I915_READ(PRB0_TAIL),
-			  I915_READ(PRB0_START));
+		ret = -EINVAL;
+		goto cleanup;
 	}
 
-	I915_WRITE(PRB0_CTL,
-		   ((obj->size - 4096) & RING_NR_PAGES) |
-		   RING_NO_REPORT |
-		   RING_VALID);
-
-	head = I915_READ(PRB0_HEAD) & HEAD_ADDR;
-
-	/* If the head is still not zero, the ring is dead */
-	if (head != 0) {
-		DRM_ERROR("Ring initialization failed "
-			  "ctl %08x head %08x tail %08x start %08x\n",
-			  I915_READ(PRB0_CTL),
-			  I915_READ(PRB0_HEAD),
-			  I915_READ(PRB0_TAIL),
-			  I915_READ(PRB0_START));
-		return -EIO;
+	ring->virtual_start = ring->map.handle;
+	ret = ring->init(dev, ring);
+	if (ret != 0) {
+		intel_cleanup_ring_buffer(dev, ring);
+		return ret;
 	}
 
-	/* Update our cache of the ring state */
 	if (!drm_core_check_feature(dev, DRIVER_MODESET))
 		i915_kernel_lost_context(dev);
 	else {
-		ring->head = I915_READ(PRB0_HEAD) & HEAD_ADDR;
-		ring->tail = I915_READ(PRB0_TAIL) & TAIL_ADDR;
+		ring->head = ring->get_head(dev, ring);
+		ring->tail = ring->get_tail(dev, ring);
 		ring->space = ring->head - (ring->tail + 8);
 		if (ring->space < 0)
-			ring->space += ring->Size;
+			ring->space += ring->size;
 	}
-
-	if (IS_I9XX(dev) && !IS_GEN3(dev)) {
-		I915_WRITE(MI_MODE,
-			   (VS_TIMER_DISPATCH) << 16 | VS_TIMER_DISPATCH);
-	}
-
-	return 0;
+	INIT_LIST_HEAD(&ring->active_list);
+	INIT_LIST_HEAD(&ring->request_list);
+	return ret;
+cleanup:
+	cleanup_status_page(dev, ring);
+	return ret;
 }
 
-void
-i915_gem_cleanup_ringbuffer(struct drm_device *dev)
+void intel_cleanup_ring_buffer(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
 {
-	drm_i915_private_t *dev_priv = dev->dev_private;
-
-	if (dev_priv->render_ring.ring_obj == NULL)
+	if (ring->gem_object == NULL)
 		return;
 
-	drm_core_ioremapfree(&dev_priv->render_ring.map, dev);
-
-	i915_gem_object_unpin(dev_priv->render_ring.ring_obj);
-	drm_gem_object_unreference(dev_priv->render_ring.ring_obj);
-	dev_priv->render_ring.ring_obj = NULL;
-	memset(&dev_priv->render_ring, 0, sizeof(dev_priv->render_ring));
+	drm_core_ioremapfree(&ring->map, dev);
 
-	i915_gem_cleanup_hws(dev);
+	i915_gem_object_unpin(ring->gem_object);
+	drm_gem_object_unreference(ring->gem_object);
+	ring->gem_object = NULL;
+	cleanup_status_page(dev, ring);
 }
 
-/* As a ringbuffer is only allowed to wrap between instructions, fill
- * the tail with NOOPs.
- */
-int i915_wrap_ring(struct drm_device *dev)
+int intel_wrap_ring_buffer(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
 {
-	drm_i915_private_t *dev_priv = dev->dev_private;
-	volatile unsigned int *virt;
+	unsigned int *virt;
 	int rem;
+	rem = ring->size - ring->tail;
 
-	rem = dev_priv->render_ring.Size - dev_priv->render_ring.tail;
-	if (dev_priv->render_ring.space < rem) {
-		int ret = i915_wait_ring(dev, rem, __func__);
+	if (ring->space < rem) {
+		int ret = intel_wait_ring_buffer(dev, ring, rem);
 		if (ret)
 			return ret;
 	}
-	dev_priv->render_ring.space -= rem;
 
-	virt = (unsigned int *)
-		(dev_priv->render_ring.virtual_start + dev_priv->render_ring.tail);
+	virt = (unsigned int *)(ring->virtual_start + ring->tail);
 	rem /= 4;
 	while (rem--)
 		*virt++ = MI_NOOP;
 
-	dev_priv->render_ring.tail = 0;
+	ring->tail = 0;
 
 	return 0;
 }
 
-int i915_wait_ring(struct drm_device * dev, int n, const char *caller)
+int intel_wait_ring_buffer(struct drm_device *dev,
+		struct intel_ring_buffer *ring, int n)
 {
-	drm_i915_private_t *dev_priv = dev->dev_private;
-	drm_i915_ring_buffer_t *ring = &(dev_priv->render_ring);
-	u32 acthd_reg = IS_I965G(dev) ? ACTHD_I965 : ACTHD;
-	u32 last_acthd = I915_READ(acthd_reg);
-	u32 acthd;
-	u32 last_head = I915_READ(PRB0_HEAD) & HEAD_ADDR;
-	int i;
+	unsigned long end;
 
 	trace_i915_ring_wait_begin (dev);
-
-	for (i = 0; i < 100000; i++) {
-		ring->head = I915_READ(PRB0_HEAD) & HEAD_ADDR;
-		acthd = I915_READ(acthd_reg);
+	end = jiffies + 3 * HZ;
+	do {
+		ring->head = ring->get_head(dev, ring);
 		ring->space = ring->head - (ring->tail + 8);
 		if (ring->space < 0)
-			ring->space += ring->Size;
+			ring->space += ring->size;
 		if (ring->space >= n) {
 			trace_i915_ring_wait_end (dev);
 			return 0;
@@ -550,19 +588,97 @@ int i915_wait_ring(struct drm_device * dev, int n, const char *caller)
 			if (master_priv->sarea_priv)
 				master_priv->sarea_priv->perf_boxes |= I915_BOX_WAIT;
 		}
+		yield();
+	} while (!time_after(jiffies, end));
+	trace_i915_ring_wait_end (dev);
+	return -EBUSY;
+}
 
+void intel_ring_begin(struct drm_device *dev,
+		struct intel_ring_buffer *ring, int n)
+{
+	if (unlikely(ring->tail + n > ring->size))
+		intel_wrap_ring_buffer(dev, ring);
+	if (unlikely(ring->space < n))
+		intel_wait_ring_buffer(dev, ring, n);
+}
 
-		if (ring->head != last_head)
-			i = 0;
-		if (acthd != last_acthd)
-			i = 0;
+void intel_ring_emit(struct drm_device *dev,
+		struct intel_ring_buffer *ring, unsigned int data)
+{
+	unsigned int *virt = ring->virtual_start + ring->tail;
+	*virt = data;
+	ring->tail += 4;
+	ring->tail &= ring->size - 1;
+	ring->space -= 4;
+}
 
-		last_head = ring->head;
-		last_acthd = acthd;
-		msleep_interruptible(10);
+void intel_ring_advance(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	ring->advance_ring(dev, ring);
+}
 
-	}
+void intel_fill_struct(struct drm_device *dev,
+		struct intel_ring_buffer *ring,
+		void *data,
+		unsigned int len)
+{
+	unsigned int *virt = ring->virtual_start + ring->tail;
+	BUG_ON((len&~(4-1)) != 0);
+	intel_ring_begin(dev, ring, len);
+	memcpy(virt, data, len);
+	ring->tail += len;
+	ring->tail &= ring->size - 1;
+	ring->space -= len;
+	intel_ring_advance(dev, ring);
+}
 
-	trace_i915_ring_wait_end (dev);
-	return -EBUSY;
+u32 intel_ring_get_seqno(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	u32 seqno;
+	seqno = ring->next_seqno;
+
+	/* reserve 0 for non-seqno */
+	if (++ring->next_seqno == 0)
+		ring->next_seqno = 1;
+	return seqno;
 }
+
+struct intel_ring_buffer render_ring = {
+	.name			= "render ring",
+	.regs                   = {
+		.ctl = PRB0_CTL,
+		.head = PRB0_HEAD,
+		.tail = PRB0_TAIL,
+		.start = PRB0_START
+	},
+	.ring_flag		= I915_EXEC_RENDER,
+	.size			= 32 * PAGE_SIZE,
+	.alignment		= PAGE_SIZE,
+	.virtual_start		= NULL,
+	.dev			= NULL,
+	.gem_object		= NULL,
+	.head			= 0,
+	.tail			= 0,
+	.space			= 0,
+	.next_seqno		= 1,
+	.user_irq_refcount	= 0,
+	.irq_gem_seqno		= 0,
+	.waiting_gem_seqno	= 0,
+	.setup_status_page	= render_setup_status_page,
+	.init			= init_render_ring,
+	.get_head		= render_ring_get_head,
+	.get_tail		= render_ring_get_tail,
+	.get_active_head	= render_ring_get_active_head,
+	.advance_ring		= render_ring_advance_ring,
+	.flush			= render_ring_flush,
+	.add_request		= render_ring_add_request,
+	.get_gem_seqno		= render_ring_get_gem_seqno,
+	.user_irq_get		= render_ring_get_user_irq,
+	.user_irq_put		= render_ring_put_user_irq,
+	.dispatch_gem_execbuffer = render_ring_dispatch_gem_execbuffer,
+	.status_page		= {NULL, 0, NULL},
+	.map			= {0,}
+};
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
new file mode 100644
index 000000000000..d5568d3766de
--- /dev/null
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -0,0 +1,124 @@
+#ifndef _INTEL_RINGBUFFER_H_
+#define _INTEL_RINGBUFFER_H_
+
+struct  intel_hw_status_page {
+	void		*page_addr;
+	unsigned int	gfx_addr;
+	struct		drm_gem_object *obj;
+};
+
+struct drm_i915_gem_execbuffer2;
+struct  intel_ring_buffer {
+	const char	*name;
+	struct		ring_regs {
+			u32 ctl;
+			u32 head;
+			u32 tail;
+			u32 start;
+	} regs;
+	unsigned int	ring_flag;
+	unsigned long	size;
+	unsigned int	alignment;
+	void		*virtual_start;
+	struct		drm_device *dev;
+	struct		drm_gem_object *gem_object;
+
+	unsigned int	head;
+	unsigned int	tail;
+	unsigned int	space;
+	u32		next_seqno;
+	struct intel_hw_status_page status_page;
+
+	u32		irq_gem_seqno;		/* last seq seem at irq time */
+	u32		waiting_gem_seqno;
+	int		user_irq_refcount;
+	void		(*user_irq_get)(struct drm_device *dev,
+			struct intel_ring_buffer *ring);
+	void		(*user_irq_put)(struct drm_device *dev,
+			struct intel_ring_buffer *ring);
+	void		(*setup_status_page)(struct drm_device *dev,
+			struct	intel_ring_buffer *ring);
+
+	int		(*init)(struct drm_device *dev,
+			struct intel_ring_buffer *ring);
+
+	unsigned int	(*get_head)(struct drm_device *dev,
+			struct intel_ring_buffer *ring);
+	unsigned int	(*get_tail)(struct drm_device *dev,
+			struct intel_ring_buffer *ring);
+	unsigned int	(*get_active_head)(struct drm_device *dev,
+			struct intel_ring_buffer *ring);
+	void		(*advance_ring)(struct drm_device *dev,
+			struct intel_ring_buffer *ring);
+	void		(*flush)(struct drm_device *dev,
+			struct intel_ring_buffer *ring,
+			u32	invalidate_domains,
+			u32	flush_domains);
+	u32		(*add_request)(struct drm_device *dev,
+			struct intel_ring_buffer *ring,
+			struct drm_file *file_priv,
+			u32 flush_domains);
+	u32		(*get_gem_seqno)(struct drm_device *dev,
+			struct intel_ring_buffer *ring);
+	int		(*dispatch_gem_execbuffer)(struct drm_device *dev,
+			struct intel_ring_buffer *ring,
+			struct drm_i915_gem_execbuffer2 *exec,
+			struct drm_clip_rect *cliprects,
+			uint64_t exec_offset);
+
+	/**
+	 * List of objects currently involved in rendering from the
+	 * ringbuffer.
+	 *
+	 * Includes buffers having the contents of their GPU caches
+	 * flushed, not necessarily primitives.  last_rendering_seqno
+	 * represents when the rendering involved will be completed.
+	 *
+	 * A reference is held on the buffer while on this list.
+	 */
+	struct list_head active_list;
+
+	/**
+	 * List of breadcrumbs associated with GPU requests currently
+	 * outstanding.
+	 */
+	struct list_head request_list;
+
+	wait_queue_head_t irq_queue;
+	drm_local_map_t map;
+};
+
+static inline u32
+intel_read_status_page(struct intel_ring_buffer *ring,
+		int reg)
+{
+	u32 *regs = ring->status_page.page_addr;
+	return regs[reg];
+}
+
+int intel_init_ring_buffer(struct drm_device *dev,
+		struct intel_ring_buffer *ring);
+void intel_cleanup_ring_buffer(struct drm_device *dev,
+		struct intel_ring_buffer *ring);
+int intel_wait_ring_buffer(struct drm_device *dev,
+		struct intel_ring_buffer *ring, int n);
+int intel_wrap_ring_buffer(struct drm_device *dev,
+		struct intel_ring_buffer *ring);
+void intel_ring_begin(struct drm_device *dev,
+		struct intel_ring_buffer *ring, int n);
+void intel_ring_emit(struct drm_device *dev,
+		struct intel_ring_buffer *ring, u32 data);
+void intel_fill_struct(struct drm_device *dev,
+		struct intel_ring_buffer *ring,
+		void *data,
+		unsigned int len);
+void intel_ring_advance(struct drm_device *dev,
+		struct intel_ring_buffer *ring);
+
+u32 intel_ring_get_seqno(struct drm_device *dev,
+		struct intel_ring_buffer *ring);
+
+extern struct intel_ring_buffer render_ring;
+extern struct intel_ring_buffer bsd_ring;
+
+#endif /* _INTEL_RINGBUFFER_H_ */
diff --git a/include/drm/i915_drm.h b/include/drm/i915_drm.h
index b64a8d7cdf6d..e9168704cabe 100644
--- a/include/drm/i915_drm.h
+++ b/include/drm/i915_drm.h
@@ -616,7 +616,9 @@ struct drm_i915_gem_execbuffer2 {
 	__u32 num_cliprects;
 	/** This is a struct drm_clip_rect *cliprects */
 	__u64 cliprects_ptr;
-	__u64 flags; /* currently unused */
+#define I915_EXEC_RENDER                 (1<<0)
+#define I915_EXEC_BSD                    (1<<1)
+	__u64 flags;
 	__u64 rsvd1;
 	__u64 rsvd2;
 };
-- 
cgit v1.2.3


From 852835f343146a82a528c3b712b373661d4fa17a Mon Sep 17 00:00:00 2001
From: Zou Nan hai <nanhai.zou@intel.com>
Date: Fri, 21 May 2010 09:08:56 +0800
Subject: drm/i915: convert some gem structures to per-ring V2

The active list and request list move into the ringbuffer structure,
so each can track its active objects in the order they are in that
ring.  The flushing list does not, as it doesn't matter which ring
caused data to end up in the render cache.  Objects gain a pointer to
the ring they are active on (if any).

Signed-off-by: Zou Nan hai <nanhai.zou@intel.com>
Signed-off-by: Xiang Hai hao <haihao.xiang@intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_debugfs.c  |  12 +-
 drivers/gpu/drm/i915/i915_dma.c      |   8 +-
 drivers/gpu/drm/i915/i915_drv.c      |   2 +-
 drivers/gpu/drm/i915/i915_drv.h      |  39 +++----
 drivers/gpu/drm/i915/i915_gem.c      | 210 ++++++++++++++++++++---------------
 drivers/gpu/drm/i915/i915_irq.c      |  45 +++++---
 drivers/gpu/drm/i915/intel_overlay.c |  44 +++++---
 7 files changed, 207 insertions(+), 153 deletions(-)

(limited to 'drivers/gpu/drm/i915/i915_drv.h')

diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index 4fddf094deb2..c864858d5064 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -77,7 +77,7 @@ static int i915_gem_object_list_info(struct seq_file *m, void *data)
 	case ACTIVE_LIST:
 		seq_printf(m, "Active:\n");
 		lock = &dev_priv->mm.active_list_lock;
-		head = &dev_priv->mm.active_list;
+		head = &dev_priv->render_ring.active_list;
 		break;
 	case INACTIVE_LIST:
 		seq_printf(m, "Inactive:\n");
@@ -129,7 +129,8 @@ static int i915_gem_request_info(struct seq_file *m, void *data)
 	struct drm_i915_gem_request *gem_request;
 
 	seq_printf(m, "Request:\n");
-	list_for_each_entry(gem_request, &dev_priv->mm.request_list, list) {
+	list_for_each_entry(gem_request, &dev_priv->render_ring.request_list,
+			list) {
 		seq_printf(m, "    %d @ %d\n",
 			   gem_request->seqno,
 			   (int) (jiffies - gem_request->emitted_jiffies));
@@ -145,7 +146,7 @@ static int i915_gem_seqno_info(struct seq_file *m, void *data)
 
 	if (dev_priv->hw_status_page != NULL) {
 		seq_printf(m, "Current sequence: %d\n",
-			   i915_get_gem_seqno(dev));
+			   i915_get_gem_seqno(dev,  &dev_priv->render_ring));
 	} else {
 		seq_printf(m, "Current sequence: hws uninitialized\n");
 	}
@@ -197,7 +198,7 @@ static int i915_interrupt_info(struct seq_file *m, void *data)
 		   atomic_read(&dev_priv->irq_received));
 	if (dev_priv->hw_status_page != NULL) {
 		seq_printf(m, "Current sequence:    %d\n",
-			   i915_get_gem_seqno(dev));
+			   i915_get_gem_seqno(dev,  &dev_priv->render_ring));
 	} else {
 		seq_printf(m, "Current sequence:    hws uninitialized\n");
 	}
@@ -287,7 +288,8 @@ static int i915_batchbuffer_info(struct seq_file *m, void *data)
 
 	spin_lock(&dev_priv->mm.active_list_lock);
 
-	list_for_each_entry(obj_priv, &dev_priv->mm.active_list, list) {
+	list_for_each_entry(obj_priv, &dev_priv->render_ring.active_list,
+			list) {
 		obj = &obj_priv->base;
 		if (obj->read_domains & I915_GEM_DOMAIN_COMMAND) {
 		    ret = i915_gem_object_get_pages(obj, 0);
diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 2541428b2fe5..f485880300ce 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -82,7 +82,8 @@ static void i915_free_hws(struct drm_device *dev)
 		dev_priv->status_page_dmah = NULL;
 	}
 
-	if (dev_priv->status_gfx_addr) {
+	if (dev_priv->render_ring.status_page.gfx_addr) {
+		dev_priv->render_ring.status_page.gfx_addr = 0;
 		dev_priv->status_gfx_addr = 0;
 		drm_core_ioremapfree(&dev_priv->hws_map, dev);
 	}
@@ -835,9 +836,9 @@ static int i915_set_status_page(struct drm_device *dev, void *data,
 	I915_WRITE(HWS_PGA, ring->status_page.gfx_addr);
 
 	DRM_DEBUG_DRIVER("load hws HWS_PGA with gfx mem 0x%x\n",
-				dev_priv->status_gfx_addr);
+			dev_priv->status_gfx_addr);
 	DRM_DEBUG_DRIVER("load hws at %p\n",
-				dev_priv->hw_status_page);
+			dev_priv->hw_status_page);
 	return 0;
 }
 
@@ -1510,7 +1511,6 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)
 	resource_size_t base, size;
 	int ret = 0, mmio_bar;
 	uint32_t agp_size, prealloc_size, prealloc_start;
-
 	/* i915 has 4 more counters */
 	dev->counters += 4;
 	dev->types[6] = _DRM_STAT_IRQ;
diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index c57c54f403da..d40f62d36453 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -340,7 +340,7 @@ int i965_reset(struct drm_device *dev, u8 flags)
 	/*
 	 * Clear request list
 	 */
-	i915_gem_retire_requests(dev);
+	i915_gem_retire_requests(dev, &dev_priv->render_ring);
 
 	if (need_display)
 		i915_save_display(dev);
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 6bb7933d49dc..3f35989ba74c 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -505,18 +505,7 @@ typedef struct drm_i915_private {
 		 */
 		struct list_head shrink_list;
 
-		/**
-		 * List of objects currently involved in rendering from the
-		 * ringbuffer.
-		 *
-		 * Includes buffers having the contents of their GPU caches
-		 * flushed, not necessarily primitives.  last_rendering_seqno
-		 * represents when the rendering involved will be completed.
-		 *
-		 * A reference is held on the buffer while on this list.
-		 */
 		spinlock_t active_list_lock;
-		struct list_head active_list;
 
 		/**
 		 * List of objects which are not in the ringbuffer but which
@@ -553,12 +542,6 @@ typedef struct drm_i915_private {
 		/** LRU list of objects with fence regs on them. */
 		struct list_head fence_list;
 
-		/**
-		 * List of breadcrumbs associated with GPU requests currently
-		 * outstanding.
-		 */
-		struct list_head request_list;
-
 		/**
 		 * We leave the user IRQ off as much as possible,
 		 * but this means that requests will finish and never
@@ -683,6 +666,9 @@ struct drm_i915_gem_object {
 	 */
 	uint32_t gtt_offset;
 
+	/* Which ring is refering to is this object */
+	struct intel_ring_buffer *ring;
+
 	/**
 	 * Fake offset for use by mmap(2)
 	 */
@@ -756,6 +742,9 @@ struct drm_i915_gem_object {
  * an emission time with seqnos for tracking how far ahead of the GPU we are.
  */
 struct drm_i915_gem_request {
+	/** On Which ring this request was generated */
+	struct intel_ring_buffer *ring;
+
 	/** GEM sequence number associated with this request. */
 	uint32_t seqno;
 
@@ -916,11 +905,13 @@ void i915_gem_object_unpin(struct drm_gem_object *obj);
 int i915_gem_object_unbind(struct drm_gem_object *obj);
 void i915_gem_release_mmap(struct drm_gem_object *obj);
 void i915_gem_lastclose(struct drm_device *dev);
-uint32_t i915_get_gem_seqno(struct drm_device *dev);
+uint32_t i915_get_gem_seqno(struct drm_device *dev,
+		struct intel_ring_buffer *ring);
 bool i915_seqno_passed(uint32_t seq1, uint32_t seq2);
 int i915_gem_object_get_fence_reg(struct drm_gem_object *obj);
 int i915_gem_object_put_fence_reg(struct drm_gem_object *obj);
-void i915_gem_retire_requests(struct drm_device *dev);
+void i915_gem_retire_requests(struct drm_device *dev,
+		 struct intel_ring_buffer *ring);
 void i915_gem_retire_work_handler(struct work_struct *work);
 void i915_gem_clflush_object(struct drm_gem_object *obj);
 int i915_gem_object_set_domain(struct drm_gem_object *obj,
@@ -931,9 +922,13 @@ void i915_gem_cleanup_ringbuffer(struct drm_device *dev);
 int i915_gem_do_init(struct drm_device *dev, unsigned long start,
 		     unsigned long end);
 int i915_gem_idle(struct drm_device *dev);
-uint32_t i915_add_request(struct drm_device *dev, struct drm_file *file_priv,
-			  uint32_t flush_domains);
-int i915_do_wait_request(struct drm_device *dev, uint32_t seqno, int interruptible);
+uint32_t i915_add_request(struct drm_device *dev,
+		struct drm_file *file_priv,
+		uint32_t flush_domains,
+		struct intel_ring_buffer *ring);
+int i915_do_wait_request(struct drm_device *dev,
+		uint32_t seqno, int interruptible,
+		struct intel_ring_buffer *ring);
 int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
 int i915_gem_object_set_to_gtt_domain(struct drm_gem_object *obj,
 				      int write);
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 58b6e814fae1..af664ba923c5 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -1482,11 +1482,14 @@ i915_gem_object_put_pages(struct drm_gem_object *obj)
 }
 
 static void
-i915_gem_object_move_to_active(struct drm_gem_object *obj, uint32_t seqno)
+i915_gem_object_move_to_active(struct drm_gem_object *obj, uint32_t seqno,
+			       struct intel_ring_buffer *ring)
 {
 	struct drm_device *dev = obj->dev;
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
+	BUG_ON(ring == NULL);
+	obj_priv->ring = ring;
 
 	/* Add a reference if we're newly entering the active list. */
 	if (!obj_priv->active) {
@@ -1495,8 +1498,7 @@ i915_gem_object_move_to_active(struct drm_gem_object *obj, uint32_t seqno)
 	}
 	/* Move from whatever list we were on to the tail of execution. */
 	spin_lock(&dev_priv->mm.active_list_lock);
-	list_move_tail(&obj_priv->list,
-		       &dev_priv->mm.active_list);
+	list_move_tail(&obj_priv->list, &ring->active_list);
 	spin_unlock(&dev_priv->mm.active_list_lock);
 	obj_priv->last_rendering_seqno = seqno;
 }
@@ -1549,6 +1551,7 @@ i915_gem_object_move_to_inactive(struct drm_gem_object *obj)
 	BUG_ON(!list_empty(&obj_priv->gpu_write_list));
 
 	obj_priv->last_rendering_seqno = 0;
+	obj_priv->ring = NULL;
 	if (obj_priv->active) {
 		obj_priv->active = 0;
 		drm_gem_object_unreference(obj);
@@ -1558,7 +1561,8 @@ i915_gem_object_move_to_inactive(struct drm_gem_object *obj)
 
 static void
 i915_gem_process_flushing_list(struct drm_device *dev,
-			       uint32_t flush_domains, uint32_t seqno)
+			       uint32_t flush_domains, uint32_t seqno,
+			       struct intel_ring_buffer *ring)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	struct drm_i915_gem_object *obj_priv, *next;
@@ -1569,12 +1573,13 @@ i915_gem_process_flushing_list(struct drm_device *dev,
 		struct drm_gem_object *obj = &obj_priv->base;
 
 		if ((obj->write_domain & flush_domains) ==
-		    obj->write_domain) {
+		    obj->write_domain &&
+		    obj_priv->ring->ring_flag == ring->ring_flag) {
 			uint32_t old_write_domain = obj->write_domain;
 
 			obj->write_domain = 0;
 			list_del_init(&obj_priv->gpu_write_list);
-			i915_gem_object_move_to_active(obj, seqno);
+			i915_gem_object_move_to_active(obj, seqno, ring);
 
 			/* update the fence lru list */
 			if (obj_priv->fence_reg != I915_FENCE_REG_NONE) {
@@ -1593,7 +1598,7 @@ i915_gem_process_flushing_list(struct drm_device *dev,
 
 uint32_t
 i915_add_request(struct drm_device *dev, struct drm_file *file_priv,
-		 uint32_t flush_domains)
+		 uint32_t flush_domains, struct intel_ring_buffer *ring)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	struct drm_i915_file_private *i915_file_priv = NULL;
@@ -1608,15 +1613,14 @@ i915_add_request(struct drm_device *dev, struct drm_file *file_priv,
 	if (request == NULL)
 		return 0;
 
-	seqno = dev_priv->render_ring.add_request(dev, &dev_priv->render_ring,
-						  file_priv, flush_domains);
-
-	DRM_DEBUG_DRIVER("%d\n", seqno);
+	seqno = ring->add_request(dev, ring, file_priv, flush_domains);
 
 	request->seqno = seqno;
+	request->ring = ring;
 	request->emitted_jiffies = jiffies;
-	was_empty = list_empty(&dev_priv->mm.request_list);
-	list_add_tail(&request->list, &dev_priv->mm.request_list);
+	was_empty = list_empty(&ring->request_list);
+	list_add_tail(&request->list, &ring->request_list);
+
 	if (i915_file_priv) {
 		list_add_tail(&request->client_list,
 			      &i915_file_priv->mm.request_list);
@@ -1628,7 +1632,7 @@ i915_add_request(struct drm_device *dev, struct drm_file *file_priv,
 	 * domain we're flushing with our flush.
 	 */
 	if (flush_domains != 0) 
-		i915_gem_process_flushing_list(dev, flush_domains, seqno);
+		i915_gem_process_flushing_list(dev, flush_domains, seqno, ring);
 
 	if (!dev_priv->mm.suspended) {
 		mod_timer(&dev_priv->hangcheck_timer, jiffies + DRM_I915_HANGCHECK_PERIOD);
@@ -1645,18 +1649,16 @@ i915_add_request(struct drm_device *dev, struct drm_file *file_priv,
  * before signalling the CPU
  */
 static uint32_t
-i915_retire_commands(struct drm_device *dev)
+i915_retire_commands(struct drm_device *dev, struct intel_ring_buffer *ring)
 {
-	uint32_t cmd = MI_FLUSH | MI_NO_WRITE_FLUSH;
 	uint32_t flush_domains = 0;
 
 	/* The sampler always gets flushed on i965 (sigh) */
 	if (IS_I965G(dev))
 		flush_domains |= I915_GEM_DOMAIN_SAMPLER;
-	BEGIN_LP_RING(2);
-	OUT_RING(cmd);
-	OUT_RING(0); /* noop */
-	ADVANCE_LP_RING();
+
+	ring->flush(dev, ring,
+			I915_GEM_DOMAIN_COMMAND, flush_domains);
 	return flush_domains;
 }
 
@@ -1676,11 +1678,11 @@ i915_gem_retire_request(struct drm_device *dev,
 	 * by the ringbuffer to the flushing/inactive lists as appropriate.
 	 */
 	spin_lock(&dev_priv->mm.active_list_lock);
-	while (!list_empty(&dev_priv->mm.active_list)) {
+	while (!list_empty(&request->ring->active_list)) {
 		struct drm_gem_object *obj;
 		struct drm_i915_gem_object *obj_priv;
 
-		obj_priv = list_first_entry(&dev_priv->mm.active_list,
+		obj_priv = list_first_entry(&request->ring->active_list,
 					    struct drm_i915_gem_object,
 					    list);
 		obj = &obj_priv->base;
@@ -1727,37 +1729,33 @@ i915_seqno_passed(uint32_t seq1, uint32_t seq2)
 }
 
 uint32_t
-i915_get_gem_seqno(struct drm_device *dev)
+i915_get_gem_seqno(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
 {
-	drm_i915_private_t *dev_priv = dev->dev_private;
-
-	if (HAS_PIPE_CONTROL(dev))
-		return ((volatile u32 *)(dev_priv->seqno_page))[0];
-	else
-		return READ_HWSP(dev_priv, I915_GEM_HWS_INDEX);
+	return ring->get_gem_seqno(dev, ring);
 }
 
 /**
  * This function clears the request list as sequence numbers are passed.
  */
 void
-i915_gem_retire_requests(struct drm_device *dev)
+i915_gem_retire_requests(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	uint32_t seqno;
 
-	struct intel_ring_buffer *ring = &(dev_priv->render_ring);
 	if (!ring->status_page.page_addr
-			|| list_empty(&dev_priv->mm.request_list))
+			|| list_empty(&ring->request_list))
 		return;
 
-	seqno = i915_get_gem_seqno(dev);
+	seqno = i915_get_gem_seqno(dev, ring);
 
-	while (!list_empty(&dev_priv->mm.request_list)) {
+	while (!list_empty(&ring->request_list)) {
 		struct drm_i915_gem_request *request;
 		uint32_t retiring_seqno;
 
-		request = list_first_entry(&dev_priv->mm.request_list,
+		request = list_first_entry(&ring->request_list,
 					   struct drm_i915_gem_request,
 					   list);
 		retiring_seqno = request->seqno;
@@ -1792,27 +1790,28 @@ i915_gem_retire_work_handler(struct work_struct *work)
 	dev = dev_priv->dev;
 
 	mutex_lock(&dev->struct_mutex);
-	i915_gem_retire_requests(dev);
+	i915_gem_retire_requests(dev, &dev_priv->render_ring);
+
 	if (!dev_priv->mm.suspended &&
-	    !list_empty(&dev_priv->mm.request_list))
+			(!list_empty(&dev_priv->render_ring.request_list)))
 		queue_delayed_work(dev_priv->wq, &dev_priv->mm.retire_work, HZ);
 	mutex_unlock(&dev->struct_mutex);
 }
 
 int
-i915_do_wait_request(struct drm_device *dev, uint32_t seqno, int interruptible)
+i915_do_wait_request(struct drm_device *dev, uint32_t seqno,
+		int interruptible, struct intel_ring_buffer *ring)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	u32 ier;
 	int ret = 0;
 
-	struct intel_ring_buffer *ring = &dev_priv->render_ring;
 	BUG_ON(seqno == 0);
 
 	if (atomic_read(&dev_priv->mm.wedged))
 		return -EIO;
 
-	if (!i915_seqno_passed(i915_get_gem_seqno(dev), seqno)) {
+	if (!i915_seqno_passed(ring->get_gem_seqno(dev, ring), seqno)) {
 		if (HAS_PCH_SPLIT(dev))
 			ier = I915_READ(DEIER) | I915_READ(GTIER);
 		else
@@ -1826,19 +1825,21 @@ i915_do_wait_request(struct drm_device *dev, uint32_t seqno, int interruptible)
 
 		trace_i915_gem_request_wait_begin(dev, seqno);
 
-		dev_priv->mm.waiting_gem_seqno = seqno;
+		ring->waiting_gem_seqno = seqno;
 		ring->user_irq_get(dev, ring);
 		if (interruptible)
-			ret = wait_event_interruptible(dev_priv->irq_queue,
-				i915_seqno_passed(i915_get_gem_seqno(dev), seqno) ||
-				atomic_read(&dev_priv->mm.wedged));
+			ret = wait_event_interruptible(ring->irq_queue,
+				i915_seqno_passed(
+					ring->get_gem_seqno(dev, ring), seqno)
+				|| atomic_read(&dev_priv->mm.wedged));
 		else
-			wait_event(dev_priv->irq_queue,
-				i915_seqno_passed(i915_get_gem_seqno(dev), seqno) ||
-				atomic_read(&dev_priv->mm.wedged));
+			wait_event(ring->irq_queue,
+				i915_seqno_passed(
+					ring->get_gem_seqno(dev, ring), seqno)
+				|| atomic_read(&dev_priv->mm.wedged));
 
 		ring->user_irq_put(dev, ring);
-		dev_priv->mm.waiting_gem_seqno = 0;
+		ring->waiting_gem_seqno = 0;
 
 		trace_i915_gem_request_wait_end(dev, seqno);
 	}
@@ -1847,7 +1848,7 @@ i915_do_wait_request(struct drm_device *dev, uint32_t seqno, int interruptible)
 
 	if (ret && ret != -ERESTARTSYS)
 		DRM_ERROR("%s returns %d (awaiting %d at %d)\n",
-			  __func__, ret, seqno, i915_get_gem_seqno(dev));
+			  __func__, ret, seqno, ring->get_gem_seqno(dev, ring));
 
 	/* Directly dispatch request retiring.  While we have the work queue
 	 * to handle this, the waiter on a request often wants an associated
@@ -1855,7 +1856,7 @@ i915_do_wait_request(struct drm_device *dev, uint32_t seqno, int interruptible)
 	 * a separate wait queue to handle that.
 	 */
 	if (ret == 0)
-		i915_gem_retire_requests(dev);
+		i915_gem_retire_requests(dev, ring);
 
 	return ret;
 }
@@ -1865,12 +1866,12 @@ i915_do_wait_request(struct drm_device *dev, uint32_t seqno, int interruptible)
  * request and object lists appropriately for that event.
  */
 static int
-i915_wait_request(struct drm_device *dev, uint32_t seqno)
+i915_wait_request(struct drm_device *dev, uint32_t seqno,
+		struct intel_ring_buffer *ring)
 {
-	return i915_do_wait_request(dev, seqno, 1);
+	return i915_do_wait_request(dev, seqno, 1, ring);
 }
 
-
 static void
 i915_gem_flush(struct drm_device *dev,
 	       uint32_t invalidate_domains,
@@ -1884,6 +1885,19 @@ i915_gem_flush(struct drm_device *dev,
 			flush_domains);
 }
 
+static void
+i915_gem_flush_ring(struct drm_device *dev,
+	       uint32_t invalidate_domains,
+	       uint32_t flush_domains,
+	       struct intel_ring_buffer *ring)
+{
+	if (flush_domains & I915_GEM_DOMAIN_CPU)
+		drm_agp_chipset_flush(dev);
+	ring->flush(dev, ring,
+			invalidate_domains,
+			flush_domains);
+}
+
 /**
  * Ensures that all rendering to the object has completed and the object is
  * safe to unbind from the GTT or access from the CPU.
@@ -1908,7 +1922,8 @@ i915_gem_object_wait_rendering(struct drm_gem_object *obj)
 		DRM_INFO("%s: object %p wait for seqno %08x\n",
 			  __func__, obj, obj_priv->last_rendering_seqno);
 #endif
-		ret = i915_wait_request(dev, obj_priv->last_rendering_seqno);
+		ret = i915_wait_request(dev,
+				obj_priv->last_rendering_seqno, obj_priv->ring);
 		if (ret != 0)
 			return ret;
 	}
@@ -2025,10 +2040,11 @@ i915_gpu_idle(struct drm_device *dev)
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	bool lists_empty;
 	uint32_t seqno;
+	int ret;
 
 	spin_lock(&dev_priv->mm.active_list_lock);
 	lists_empty = list_empty(&dev_priv->mm.flushing_list) &&
-		      list_empty(&dev_priv->mm.active_list);
+		      list_empty(&dev_priv->render_ring.active_list);
 	spin_unlock(&dev_priv->mm.active_list_lock);
 
 	if (lists_empty)
@@ -2036,11 +2052,13 @@ i915_gpu_idle(struct drm_device *dev)
 
 	/* Flush everything onto the inactive list. */
 	i915_gem_flush(dev, I915_GEM_GPU_DOMAINS, I915_GEM_GPU_DOMAINS);
-	seqno = i915_add_request(dev, NULL, I915_GEM_GPU_DOMAINS);
+	seqno = i915_add_request(dev, NULL, I915_GEM_GPU_DOMAINS,
+			&dev_priv->render_ring);
 	if (seqno == 0)
 		return -ENOMEM;
+	ret = i915_wait_request(dev, seqno, &dev_priv->render_ring);
 
-	return i915_wait_request(dev, seqno);
+	return ret;
 }
 
 static int
@@ -2053,7 +2071,7 @@ i915_gem_evict_everything(struct drm_device *dev)
 	spin_lock(&dev_priv->mm.active_list_lock);
 	lists_empty = (list_empty(&dev_priv->mm.inactive_list) &&
 		       list_empty(&dev_priv->mm.flushing_list) &&
-		       list_empty(&dev_priv->mm.active_list));
+		       list_empty(&dev_priv->render_ring.active_list));
 	spin_unlock(&dev_priv->mm.active_list_lock);
 
 	if (lists_empty)
@@ -2073,7 +2091,7 @@ i915_gem_evict_everything(struct drm_device *dev)
 	spin_lock(&dev_priv->mm.active_list_lock);
 	lists_empty = (list_empty(&dev_priv->mm.inactive_list) &&
 		       list_empty(&dev_priv->mm.flushing_list) &&
-		       list_empty(&dev_priv->mm.active_list));
+		       list_empty(&dev_priv->render_ring.active_list));
 	spin_unlock(&dev_priv->mm.active_list_lock);
 	BUG_ON(!lists_empty);
 
@@ -2087,8 +2105,9 @@ i915_gem_evict_something(struct drm_device *dev, int min_size)
 	struct drm_gem_object *obj;
 	int ret;
 
+	struct intel_ring_buffer *render_ring = &dev_priv->render_ring;
 	for (;;) {
-		i915_gem_retire_requests(dev);
+		i915_gem_retire_requests(dev, render_ring);
 
 		/* If there's an inactive buffer available now, grab it
 		 * and be done.
@@ -2112,14 +2131,15 @@ i915_gem_evict_something(struct drm_device *dev, int min_size)
 		 * things, wait for the next to finish and hopefully leave us
 		 * a buffer to evict.
 		 */
-		if (!list_empty(&dev_priv->mm.request_list)) {
+		if (!list_empty(&render_ring->request_list)) {
 			struct drm_i915_gem_request *request;
 
-			request = list_first_entry(&dev_priv->mm.request_list,
+			request = list_first_entry(&render_ring->request_list,
 						   struct drm_i915_gem_request,
 						   list);
 
-			ret = i915_wait_request(dev, request->seqno);
+			ret = i915_wait_request(dev,
+					request->seqno, request->ring);
 			if (ret)
 				return ret;
 
@@ -2146,10 +2166,13 @@ i915_gem_evict_something(struct drm_device *dev, int min_size)
 			if (obj != NULL) {
 				uint32_t seqno;
 
-				i915_gem_flush(dev,
+				i915_gem_flush_ring(dev,
+					       obj->write_domain,
 					       obj->write_domain,
-					       obj->write_domain);
-				seqno = i915_add_request(dev, NULL, obj->write_domain);
+					       obj_priv->ring);
+				seqno = i915_add_request(dev, NULL,
+						obj->write_domain,
+						obj_priv->ring);
 				if (seqno == 0)
 					return -ENOMEM;
 				continue;
@@ -2685,6 +2708,7 @@ i915_gem_object_flush_gpu_write_domain(struct drm_gem_object *obj)
 {
 	struct drm_device *dev = obj->dev;
 	uint32_t old_write_domain;
+	struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
 
 	if ((obj->write_domain & I915_GEM_GPU_DOMAINS) == 0)
 		return;
@@ -2692,7 +2716,7 @@ i915_gem_object_flush_gpu_write_domain(struct drm_gem_object *obj)
 	/* Queue the GPU write cache flushing we need. */
 	old_write_domain = obj->write_domain;
 	i915_gem_flush(dev, 0, obj->write_domain);
-	(void) i915_add_request(dev, NULL, obj->write_domain);
+	(void) i915_add_request(dev, NULL, obj->write_domain, obj_priv->ring);
 	BUG_ON(obj->write_domain);
 
 	trace_i915_gem_object_change_domain(obj,
@@ -2832,7 +2856,10 @@ i915_gem_object_set_to_display_plane(struct drm_gem_object *obj)
 		DRM_INFO("%s: object %p wait for seqno %08x\n",
 			  __func__, obj, obj_priv->last_rendering_seqno);
 #endif
-		ret = i915_do_wait_request(dev, obj_priv->last_rendering_seqno, 0);
+		ret = i915_do_wait_request(dev,
+				obj_priv->last_rendering_seqno,
+				0,
+				obj_priv->ring);
 		if (ret != 0)
 			return ret;
 	}
@@ -3451,7 +3478,7 @@ i915_gem_ring_throttle(struct drm_device *dev, struct drm_file *file_priv)
 		if (time_after_eq(request->emitted_jiffies, recent_enough))
 			break;
 
-		ret = i915_wait_request(dev, request->seqno);
+		ret = i915_wait_request(dev, request->seqno, request->ring);
 		if (ret != 0)
 			break;
 	}
@@ -3608,6 +3635,8 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,
 	uint32_t seqno, flush_domains, reloc_index;
 	int pin_tries, flips;
 
+	struct intel_ring_buffer *ring = NULL;
+
 #if WATCH_EXEC
 	DRM_INFO("buffers_ptr %d buffer_count %d len %08x\n",
 		  (int) args->buffers_ptr, args->buffer_count, args->batch_len);
@@ -3665,6 +3694,8 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,
 		goto pre_mutex_err;
 	}
 
+	ring = &dev_priv->render_ring;
+
 	/* Look up object handles */
 	flips = 0;
 	for (i = 0; i < args->buffer_count; i++) {
@@ -3798,9 +3829,12 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,
 		i915_gem_flush(dev,
 			       dev->invalidate_domains,
 			       dev->flush_domains);
-		if (dev->flush_domains & I915_GEM_GPU_DOMAINS)
+		if (dev->flush_domains & I915_GEM_GPU_DOMAINS) {
 			(void)i915_add_request(dev, file_priv,
-					       dev->flush_domains);
+					dev->flush_domains,
+					&dev_priv->render_ring);
+
+		}
 	}
 
 	for (i = 0; i < args->buffer_count; i++) {
@@ -3837,11 +3871,8 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,
 #endif
 
 	/* Exec the batchbuffer */
-	ret = dev_priv->render_ring.dispatch_gem_execbuffer(dev,
-							    &dev_priv->render_ring,
-							    args,
-							    cliprects,
-							    exec_offset);
+	ret = ring->dispatch_gem_execbuffer(dev, ring, args,
+			cliprects, exec_offset);
 	if (ret) {
 		DRM_ERROR("dispatch failed %d\n", ret);
 		goto err;
@@ -3851,7 +3882,7 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,
 	 * Ensure that the commands in the batch buffer are
 	 * finished before the interrupt fires
 	 */
-	flush_domains = i915_retire_commands(dev);
+	flush_domains = i915_retire_commands(dev, ring);
 
 	i915_verify_inactive(dev, __FILE__, __LINE__);
 
@@ -3862,12 +3893,13 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,
 	 * *some* interrupts representing completion of buffers that we can
 	 * wait on when trying to clear up gtt space).
 	 */
-	seqno = i915_add_request(dev, file_priv, flush_domains);
+	seqno = i915_add_request(dev, file_priv, flush_domains, ring);
 	BUG_ON(seqno == 0);
 	for (i = 0; i < args->buffer_count; i++) {
 		struct drm_gem_object *obj = object_list[i];
+		obj_priv = to_intel_bo(obj);
 
-		i915_gem_object_move_to_active(obj, seqno);
+		i915_gem_object_move_to_active(obj, seqno, ring);
 #if WATCH_LRU
 		DRM_INFO("%s: move to exec list %p\n", __func__, obj);
 #endif
@@ -3979,7 +4011,7 @@ i915_gem_execbuffer(struct drm_device *dev, void *data,
 	exec2.DR4 = args->DR4;
 	exec2.num_cliprects = args->num_cliprects;
 	exec2.cliprects_ptr = args->cliprects_ptr;
-	exec2.flags = 0;
+	exec2.flags = I915_EXEC_RENDER;
 
 	ret = i915_gem_do_execbuffer(dev, data, file_priv, &exec2, exec2_list);
 	if (!ret) {
@@ -4218,6 +4250,7 @@ i915_gem_busy_ioctl(struct drm_device *dev, void *data,
 	struct drm_i915_gem_busy *args = data;
 	struct drm_gem_object *obj;
 	struct drm_i915_gem_object *obj_priv;
+	drm_i915_private_t *dev_priv = dev->dev_private;
 
 	obj = drm_gem_object_lookup(dev, file_priv, args->handle);
 	if (obj == NULL) {
@@ -4232,7 +4265,7 @@ i915_gem_busy_ioctl(struct drm_device *dev, void *data,
 	 * actually unmasked, and our working set ends up being larger than
 	 * required.
 	 */
-	i915_gem_retire_requests(dev);
+	i915_gem_retire_requests(dev, &dev_priv->render_ring);
 
 	obj_priv = to_intel_bo(obj);
 	/* Don't count being on the flushing list against the object being
@@ -4555,12 +4588,12 @@ i915_gem_entervt_ioctl(struct drm_device *dev, void *data,
 	}
 
 	spin_lock(&dev_priv->mm.active_list_lock);
-	BUG_ON(!list_empty(&dev_priv->mm.active_list));
+	BUG_ON(!list_empty(&dev_priv->render_ring.active_list));
 	spin_unlock(&dev_priv->mm.active_list_lock);
 
 	BUG_ON(!list_empty(&dev_priv->mm.flushing_list));
 	BUG_ON(!list_empty(&dev_priv->mm.inactive_list));
-	BUG_ON(!list_empty(&dev_priv->mm.request_list));
+	BUG_ON(!list_empty(&dev_priv->render_ring.request_list));
 	mutex_unlock(&dev->struct_mutex);
 
 	drm_irq_install(dev);
@@ -4599,18 +4632,16 @@ i915_gem_load(struct drm_device *dev)
 	drm_i915_private_t *dev_priv = dev->dev_private;
 
 	spin_lock_init(&dev_priv->mm.active_list_lock);
-	INIT_LIST_HEAD(&dev_priv->mm.active_list);
 	INIT_LIST_HEAD(&dev_priv->mm.flushing_list);
 	INIT_LIST_HEAD(&dev_priv->mm.gpu_write_list);
 	INIT_LIST_HEAD(&dev_priv->mm.inactive_list);
-	INIT_LIST_HEAD(&dev_priv->mm.request_list);
 	INIT_LIST_HEAD(&dev_priv->mm.fence_list);
+	INIT_LIST_HEAD(&dev_priv->render_ring.active_list);
+	INIT_LIST_HEAD(&dev_priv->render_ring.request_list);
 	for (i = 0; i < 16; i++)
 		INIT_LIST_HEAD(&dev_priv->fence_regs[i].lru_list);
 	INIT_DELAYED_WORK(&dev_priv->mm.retire_work,
 			  i915_gem_retire_work_handler);
-	dev_priv->mm.next_gem_seqno = 1;
-
 	spin_lock(&shrink_list_lock);
 	list_add(&dev_priv->mm.shrink_list, &shrink_list);
 	spin_unlock(&shrink_list_lock);
@@ -4842,7 +4873,7 @@ i915_gpu_is_active(struct drm_device *dev)
 
 	spin_lock(&dev_priv->mm.active_list_lock);
 	lists_empty = list_empty(&dev_priv->mm.flushing_list) &&
-		      list_empty(&dev_priv->mm.active_list);
+		      list_empty(&dev_priv->render_ring.active_list);
 	spin_unlock(&dev_priv->mm.active_list_lock);
 
 	return !lists_empty;
@@ -4887,8 +4918,7 @@ rescan:
 			continue;
 
 		spin_unlock(&shrink_list_lock);
-
-		i915_gem_retire_requests(dev);
+		i915_gem_retire_requests(dev, &dev_priv->render_ring);
 
 		list_for_each_entry_safe(obj_priv, next_obj,
 					 &dev_priv->mm.inactive_list,
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index e07c643c8365..8a667f1db75a 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -331,6 +331,7 @@ irqreturn_t ironlake_irq_handler(struct drm_device *dev)
 	int ret = IRQ_NONE;
 	u32 de_iir, gt_iir, de_ier, pch_iir;
 	struct drm_i915_master_private *master_priv;
+	struct intel_ring_buffer *render_ring = &dev_priv->render_ring;
 
 	/* disable master interrupt before clearing iir  */
 	de_ier = I915_READ(DEIER);
@@ -354,10 +355,10 @@ irqreturn_t ironlake_irq_handler(struct drm_device *dev)
 	}
 
 	if (gt_iir & GT_PIPE_NOTIFY) {
-		u32 seqno = i915_get_gem_seqno(dev);
-		dev_priv->mm.irq_gem_seqno = seqno;
+		u32 seqno = render_ring->get_gem_seqno(dev, render_ring);
+		render_ring->irq_gem_seqno = seqno;
 		trace_i915_gem_request_complete(dev, seqno);
-		DRM_WAKEUP(&dev_priv->irq_queue);
+		DRM_WAKEUP(&dev_priv->render_ring.irq_queue);
 		dev_priv->hangcheck_count = 0;
 		mod_timer(&dev_priv->hangcheck_timer, jiffies + DRM_I915_HANGCHECK_PERIOD);
 	}
@@ -588,7 +589,7 @@ static void i915_capture_error_state(struct drm_device *dev)
 		return;
 	}
 
-	error->seqno = i915_get_gem_seqno(dev);
+	error->seqno = i915_get_gem_seqno(dev, &dev_priv->render_ring);
 	error->eir = I915_READ(EIR);
 	error->pgtbl_er = I915_READ(PGTBL_ER);
 	error->pipeastat = I915_READ(PIPEASTAT);
@@ -616,7 +617,9 @@ static void i915_capture_error_state(struct drm_device *dev)
 	batchbuffer[0] = NULL;
 	batchbuffer[1] = NULL;
 	count = 0;
-	list_for_each_entry(obj_priv, &dev_priv->mm.active_list, list) {
+	list_for_each_entry(obj_priv,
+			&dev_priv->render_ring.active_list, list) {
+
 		struct drm_gem_object *obj = &obj_priv->base;
 
 		if (batchbuffer[0] == NULL &&
@@ -653,7 +656,8 @@ static void i915_capture_error_state(struct drm_device *dev)
 
 	if (error->active_bo) {
 		int i = 0;
-		list_for_each_entry(obj_priv, &dev_priv->mm.active_list, list) {
+		list_for_each_entry(obj_priv,
+				&dev_priv->render_ring.active_list, list) {
 			struct drm_gem_object *obj = &obj_priv->base;
 
 			error->active_bo[i].size = obj->size;
@@ -831,7 +835,7 @@ static void i915_handle_error(struct drm_device *dev, bool wedged)
 		/*
 		 * Wakeup waiting processes so they don't hang
 		 */
-		DRM_WAKEUP(&dev_priv->irq_queue);
+		DRM_WAKEUP(&dev_priv->render_ring.irq_queue);
 	}
 
 	queue_work(dev_priv->wq, &dev_priv->error_work);
@@ -850,6 +854,7 @@ irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS)
 	unsigned long irqflags;
 	int irq_received;
 	int ret = IRQ_NONE;
+	struct intel_ring_buffer *render_ring = &dev_priv->render_ring;
 
 	atomic_inc(&dev_priv->irq_received);
 
@@ -930,10 +935,11 @@ irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS)
 		}
 
 		if (iir & I915_USER_INTERRUPT) {
-			u32 seqno = i915_get_gem_seqno(dev);
-			dev_priv->mm.irq_gem_seqno = seqno;
+			u32 seqno =
+				render_ring->get_gem_seqno(dev, render_ring);
+			render_ring->irq_gem_seqno = seqno;
 			trace_i915_gem_request_complete(dev, seqno);
-			DRM_WAKEUP(&dev_priv->irq_queue);
+			DRM_WAKEUP(&dev_priv->render_ring.irq_queue);
 			dev_priv->hangcheck_count = 0;
 			mod_timer(&dev_priv->hangcheck_timer, jiffies + DRM_I915_HANGCHECK_PERIOD);
 		}
@@ -1038,7 +1044,7 @@ static int i915_wait_irq(struct drm_device * dev, int irq_nr)
 		master_priv->sarea_priv->perf_boxes |= I915_BOX_WAIT;
 
 	render_ring->user_irq_get(dev, render_ring);
-	DRM_WAIT_ON(ret, dev_priv->irq_queue, 3 * DRM_HZ,
+	DRM_WAIT_ON(ret, dev_priv->render_ring.irq_queue, 3 * DRM_HZ,
 		    READ_BREADCRUMB(dev_priv) >= irq_nr);
 	render_ring->user_irq_put(dev, render_ring);
 
@@ -1205,9 +1211,12 @@ int i915_vblank_swap(struct drm_device *dev, void *data,
 	return -EINVAL;
 }
 
-struct drm_i915_gem_request *i915_get_tail_request(struct drm_device *dev) {
+struct drm_i915_gem_request *
+i915_get_tail_request(struct drm_device *dev)
+{
 	drm_i915_private_t *dev_priv = dev->dev_private;
-	return list_entry(dev_priv->mm.request_list.prev, struct drm_i915_gem_request, list);
+	return list_entry(dev_priv->render_ring.request_list.prev,
+			struct drm_i915_gem_request, list);
 }
 
 /**
@@ -1232,8 +1241,10 @@ void i915_hangcheck_elapsed(unsigned long data)
 		acthd = I915_READ(ACTHD_I965);
 
 	/* If all work is done then ACTHD clearly hasn't advanced. */
-	if (list_empty(&dev_priv->mm.request_list) ||
-		       i915_seqno_passed(i915_get_gem_seqno(dev), i915_get_tail_request(dev)->seqno)) {
+	if (list_empty(&dev_priv->render_ring.request_list) ||
+		i915_seqno_passed(i915_get_gem_seqno(dev,
+				&dev_priv->render_ring),
+			i915_get_tail_request(dev)->seqno)) {
 		dev_priv->hangcheck_count = 0;
 		return;
 	}
@@ -1300,7 +1311,7 @@ static int ironlake_irq_postinstall(struct drm_device *dev)
 	(void) I915_READ(DEIER);
 
 	/* user interrupt should be enabled, but masked initial */
-	dev_priv->gt_irq_mask_reg = 0xffffffff;
+	dev_priv->gt_irq_mask_reg = ~render_mask;
 	dev_priv->gt_irq_enable_reg = render_mask;
 
 	I915_WRITE(GTIIR, I915_READ(GTIIR));
@@ -1363,7 +1374,7 @@ int i915_driver_irq_postinstall(struct drm_device *dev)
 	u32 enable_mask = I915_INTERRUPT_ENABLE_FIX | I915_INTERRUPT_ENABLE_VAR;
 	u32 error_mask;
 
-	DRM_INIT_WAITQUEUE(&dev_priv->irq_queue);
+	DRM_INIT_WAITQUEUE(&dev_priv->render_ring.irq_queue);
 
 	dev_priv->vblank_pipe = DRM_I915_VBLANK_PIPE_A | DRM_I915_VBLANK_PIPE_B;
 
diff --git a/drivers/gpu/drm/i915/intel_overlay.c b/drivers/gpu/drm/i915/intel_overlay.c
index 93da83782e5e..d7ad5139d17c 100644
--- a/drivers/gpu/drm/i915/intel_overlay.c
+++ b/drivers/gpu/drm/i915/intel_overlay.c
@@ -212,6 +212,7 @@ static int intel_overlay_on(struct intel_overlay *overlay)
 {
 	struct drm_device *dev = overlay->dev;
 	int ret;
+	drm_i915_private_t *dev_priv = dev->dev_private;
 
 	BUG_ON(overlay->active);
 
@@ -225,11 +226,13 @@ static int intel_overlay_on(struct intel_overlay *overlay)
 	OUT_RING(MI_NOOP);
 	ADVANCE_LP_RING();
 
-	overlay->last_flip_req = i915_add_request(dev, NULL, 0);
+	overlay->last_flip_req =
+		i915_add_request(dev, NULL, 0, &dev_priv->render_ring);
 	if (overlay->last_flip_req == 0)
 		return -ENOMEM;
 
-	ret = i915_do_wait_request(dev, overlay->last_flip_req, 1);
+	ret = i915_do_wait_request(dev,
+			overlay->last_flip_req, 1, &dev_priv->render_ring);
 	if (ret != 0)
 		return ret;
 
@@ -262,7 +265,8 @@ static void intel_overlay_continue(struct intel_overlay *overlay,
 	OUT_RING(flip_addr);
         ADVANCE_LP_RING();
 
-	overlay->last_flip_req = i915_add_request(dev, NULL, 0);
+	overlay->last_flip_req =
+		i915_add_request(dev, NULL, 0, &dev_priv->render_ring);
 }
 
 static int intel_overlay_wait_flip(struct intel_overlay *overlay)
@@ -273,7 +277,8 @@ static int intel_overlay_wait_flip(struct intel_overlay *overlay)
 	u32 tmp;
 
 	if (overlay->last_flip_req != 0) {
-		ret = i915_do_wait_request(dev, overlay->last_flip_req, 1);
+		ret = i915_do_wait_request(dev, overlay->last_flip_req,
+				1, &dev_priv->render_ring);
 		if (ret == 0) {
 			overlay->last_flip_req = 0;
 
@@ -292,11 +297,13 @@ static int intel_overlay_wait_flip(struct intel_overlay *overlay)
         OUT_RING(MI_NOOP);
         ADVANCE_LP_RING();
 
-	overlay->last_flip_req = i915_add_request(dev, NULL, 0);
+	overlay->last_flip_req =
+		i915_add_request(dev, NULL, 0, &dev_priv->render_ring);
 	if (overlay->last_flip_req == 0)
 		return -ENOMEM;
 
-	ret = i915_do_wait_request(dev, overlay->last_flip_req, 1);
+	ret = i915_do_wait_request(dev, overlay->last_flip_req,
+			1, &dev_priv->render_ring);
 	if (ret != 0)
 		return ret;
 
@@ -310,6 +317,7 @@ static int intel_overlay_off(struct intel_overlay *overlay)
 {
 	u32 flip_addr = overlay->flip_addr;
 	struct drm_device *dev = overlay->dev;
+	drm_i915_private_t *dev_priv = dev->dev_private;
 	int ret;
 
 	BUG_ON(!overlay->active);
@@ -330,11 +338,13 @@ static int intel_overlay_off(struct intel_overlay *overlay)
         OUT_RING(MI_NOOP);
         ADVANCE_LP_RING();
 
-	overlay->last_flip_req = i915_add_request(dev, NULL, 0);
+	overlay->last_flip_req =
+		i915_add_request(dev, NULL, 0, &dev_priv->render_ring);
 	if (overlay->last_flip_req == 0)
 		return -ENOMEM;
 
-	ret = i915_do_wait_request(dev, overlay->last_flip_req, 1);
+	ret = i915_do_wait_request(dev, overlay->last_flip_req,
+			1, &dev_priv->render_ring);
 	if (ret != 0)
 		return ret;
 
@@ -348,11 +358,13 @@ static int intel_overlay_off(struct intel_overlay *overlay)
         OUT_RING(MI_NOOP);
 	ADVANCE_LP_RING();
 
-	overlay->last_flip_req = i915_add_request(dev, NULL, 0);
+	overlay->last_flip_req =
+		i915_add_request(dev, NULL, 0, &dev_priv->render_ring);
 	if (overlay->last_flip_req == 0)
 		return -ENOMEM;
 
-	ret = i915_do_wait_request(dev, overlay->last_flip_req, 1);
+	ret = i915_do_wait_request(dev, overlay->last_flip_req,
+			1, &dev_priv->render_ring);
 	if (ret != 0)
 		return ret;
 
@@ -385,6 +397,7 @@ int intel_overlay_recover_from_interrupt(struct intel_overlay *overlay,
 {
 	struct drm_device *dev = overlay->dev;
 	struct drm_gem_object *obj;
+	drm_i915_private_t *dev_priv = dev->dev_private;
 	u32 flip_addr;
 	int ret;
 
@@ -392,12 +405,14 @@ int intel_overlay_recover_from_interrupt(struct intel_overlay *overlay,
 		return -EIO;
 
 	if (overlay->last_flip_req == 0) {
-		overlay->last_flip_req = i915_add_request(dev, NULL, 0);
+		overlay->last_flip_req =
+			i915_add_request(dev, NULL, 0, &dev_priv->render_ring);
 		if (overlay->last_flip_req == 0)
 			return -ENOMEM;
 	}
 
-	ret = i915_do_wait_request(dev, overlay->last_flip_req, interruptible);
+	ret = i915_do_wait_request(dev, overlay->last_flip_req,
+			interruptible, &dev_priv->render_ring);
 	if (ret != 0)
 		return ret;
 
@@ -421,12 +436,13 @@ int intel_overlay_recover_from_interrupt(struct intel_overlay *overlay,
 			OUT_RING(MI_NOOP);
 			ADVANCE_LP_RING();
 
-			overlay->last_flip_req = i915_add_request(dev, NULL, 0);
+			overlay->last_flip_req = i915_add_request(dev, NULL,
+					0, &dev_priv->render_ring);
 			if (overlay->last_flip_req == 0)
 				return -ENOMEM;
 
 			ret = i915_do_wait_request(dev, overlay->last_flip_req,
-					interruptible);
+					interruptible, &dev_priv->render_ring);
 			if (ret != 0)
 				return ret;
 
-- 
cgit v1.2.3


From d1b851fc0d105caa6b6e3e7c92d2987dfb52cbe0 Mon Sep 17 00:00:00 2001
From: Zou Nan hai <nanhai.zou@intel.com>
Date: Fri, 21 May 2010 09:08:57 +0800
Subject: drm/i915: implement BSD ring buffer V2

The BSD (bit stream decoder) ring is used for accessing the BSD engine
which decodes video bitstream for H.264 and VC1 on G45+.  It is
asynchronous with the render ring and has access to separate parts of
the GPU from it, though the render cache is coherent between the two.

Signed-off-by: Zou Nan hai <nanhai.zou@intel.com>
Signed-off-by: Xiang Hai hao <haihao.xiang@intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_dma.c         |   2 +
 drivers/gpu/drm/i915/i915_drv.h         |   2 +
 drivers/gpu/drm/i915/i915_gem.c         | 107 +++++++++++++++++++---
 drivers/gpu/drm/i915/i915_irq.c         |  13 ++-
 drivers/gpu/drm/i915/i915_reg.h         |  14 +++
 drivers/gpu/drm/i915/intel_ringbuffer.c | 153 ++++++++++++++++++++++++++++++++
 6 files changed, 276 insertions(+), 15 deletions(-)

(limited to 'drivers/gpu/drm/i915/i915_drv.h')

diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index f485880300ce..1dbed700800e 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -130,6 +130,8 @@ static int i915_dma_cleanup(struct drm_device * dev)
 		drm_irq_uninstall(dev);
 
 	intel_cleanup_ring_buffer(dev, &dev_priv->render_ring);
+	if (HAS_BSD(dev))
+		intel_cleanup_ring_buffer(dev, &dev_priv->bsd_ring);
 
 	/* Clear the HWS virtual address at teardown */
 	if (I915_NEED_GFX_HWS(dev))
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 3f35989ba74c..6bc0fc080f2b 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -235,6 +235,7 @@ typedef struct drm_i915_private {
 
 	struct pci_dev *bridge_dev;
 	struct intel_ring_buffer render_ring;
+	struct intel_ring_buffer bsd_ring;
 
 	drm_dma_handle_t *status_page_dmah;
 	void *hw_status_page;
@@ -1121,6 +1122,7 @@ extern int intel_trans_dp_port_sel (struct drm_crtc *crtc);
 			 (dev)->pci_device == 0x2A42 ||		\
 			 (dev)->pci_device == 0x2E42)
 
+#define HAS_BSD(dev)            (IS_IRONLAKE(dev) || IS_G4X(dev))
 #define I915_NEED_GFX_HWS(dev)	(INTEL_INFO(dev)->need_gfx_hws)
 
 /* With the 945 and later, Y tiling got adjusted so that it was 32 128-byte
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index af664ba923c5..c51495f15718 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -1730,7 +1730,7 @@ i915_seqno_passed(uint32_t seq1, uint32_t seq2)
 
 uint32_t
 i915_get_gem_seqno(struct drm_device *dev,
-		struct intel_ring_buffer *ring)
+		   struct intel_ring_buffer *ring)
 {
 	return ring->get_gem_seqno(dev, ring);
 }
@@ -1792,8 +1792,13 @@ i915_gem_retire_work_handler(struct work_struct *work)
 	mutex_lock(&dev->struct_mutex);
 	i915_gem_retire_requests(dev, &dev_priv->render_ring);
 
+	if (HAS_BSD(dev))
+		i915_gem_retire_requests(dev, &dev_priv->bsd_ring);
+
 	if (!dev_priv->mm.suspended &&
-			(!list_empty(&dev_priv->render_ring.request_list)))
+		(!list_empty(&dev_priv->render_ring.request_list) ||
+			(HAS_BSD(dev) &&
+			 !list_empty(&dev_priv->bsd_ring.request_list))))
 		queue_delayed_work(dev_priv->wq, &dev_priv->mm.retire_work, HZ);
 	mutex_unlock(&dev->struct_mutex);
 }
@@ -1883,6 +1888,11 @@ i915_gem_flush(struct drm_device *dev,
 	dev_priv->render_ring.flush(dev, &dev_priv->render_ring,
 			invalidate_domains,
 			flush_domains);
+
+	if (HAS_BSD(dev))
+		dev_priv->bsd_ring.flush(dev, &dev_priv->bsd_ring,
+				invalidate_domains,
+				flush_domains);
 }
 
 static void
@@ -2039,12 +2049,14 @@ i915_gpu_idle(struct drm_device *dev)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	bool lists_empty;
-	uint32_t seqno;
+	uint32_t seqno1, seqno2;
 	int ret;
 
 	spin_lock(&dev_priv->mm.active_list_lock);
-	lists_empty = list_empty(&dev_priv->mm.flushing_list) &&
-		      list_empty(&dev_priv->render_ring.active_list);
+	lists_empty = (list_empty(&dev_priv->mm.flushing_list) &&
+		       list_empty(&dev_priv->render_ring.active_list) &&
+		       (!HAS_BSD(dev) ||
+			list_empty(&dev_priv->bsd_ring.active_list)));
 	spin_unlock(&dev_priv->mm.active_list_lock);
 
 	if (lists_empty)
@@ -2052,11 +2064,23 @@ i915_gpu_idle(struct drm_device *dev)
 
 	/* Flush everything onto the inactive list. */
 	i915_gem_flush(dev, I915_GEM_GPU_DOMAINS, I915_GEM_GPU_DOMAINS);
-	seqno = i915_add_request(dev, NULL, I915_GEM_GPU_DOMAINS,
+	seqno1 = i915_add_request(dev, NULL, I915_GEM_GPU_DOMAINS,
 			&dev_priv->render_ring);
-	if (seqno == 0)
+	if (seqno1 == 0)
 		return -ENOMEM;
-	ret = i915_wait_request(dev, seqno, &dev_priv->render_ring);
+	ret = i915_wait_request(dev, seqno1, &dev_priv->render_ring);
+
+	if (HAS_BSD(dev)) {
+		seqno2 = i915_add_request(dev, NULL, I915_GEM_GPU_DOMAINS,
+				&dev_priv->bsd_ring);
+		if (seqno2 == 0)
+			return -ENOMEM;
+
+		ret = i915_wait_request(dev, seqno2, &dev_priv->bsd_ring);
+		if (ret)
+			return ret;
+	}
+
 
 	return ret;
 }
@@ -2071,7 +2095,9 @@ i915_gem_evict_everything(struct drm_device *dev)
 	spin_lock(&dev_priv->mm.active_list_lock);
 	lists_empty = (list_empty(&dev_priv->mm.inactive_list) &&
 		       list_empty(&dev_priv->mm.flushing_list) &&
-		       list_empty(&dev_priv->render_ring.active_list));
+		       list_empty(&dev_priv->render_ring.active_list) &&
+		       (!HAS_BSD(dev)
+			|| list_empty(&dev_priv->bsd_ring.active_list)));
 	spin_unlock(&dev_priv->mm.active_list_lock);
 
 	if (lists_empty)
@@ -2091,7 +2117,9 @@ i915_gem_evict_everything(struct drm_device *dev)
 	spin_lock(&dev_priv->mm.active_list_lock);
 	lists_empty = (list_empty(&dev_priv->mm.inactive_list) &&
 		       list_empty(&dev_priv->mm.flushing_list) &&
-		       list_empty(&dev_priv->render_ring.active_list));
+		       list_empty(&dev_priv->render_ring.active_list) &&
+		       (!HAS_BSD(dev)
+			|| list_empty(&dev_priv->bsd_ring.active_list)));
 	spin_unlock(&dev_priv->mm.active_list_lock);
 	BUG_ON(!lists_empty);
 
@@ -2106,9 +2134,13 @@ i915_gem_evict_something(struct drm_device *dev, int min_size)
 	int ret;
 
 	struct intel_ring_buffer *render_ring = &dev_priv->render_ring;
+	struct intel_ring_buffer *bsd_ring = &dev_priv->bsd_ring;
 	for (;;) {
 		i915_gem_retire_requests(dev, render_ring);
 
+		if (HAS_BSD(dev))
+			i915_gem_retire_requests(dev, bsd_ring);
+
 		/* If there's an inactive buffer available now, grab it
 		 * and be done.
 		 */
@@ -2146,6 +2178,21 @@ i915_gem_evict_something(struct drm_device *dev, int min_size)
 			continue;
 		}
 
+		if (HAS_BSD(dev) && !list_empty(&bsd_ring->request_list)) {
+			struct drm_i915_gem_request *request;
+
+			request = list_first_entry(&bsd_ring->request_list,
+						   struct drm_i915_gem_request,
+						   list);
+
+			ret = i915_wait_request(dev,
+					request->seqno, request->ring);
+			if (ret)
+				return ret;
+
+			continue;
+		}
+
 		/* If we didn't have anything on the request list but there
 		 * are buffers awaiting a flush, emit one and try again.
 		 * When we wait on it, those buffers waiting for that flush
@@ -3641,6 +3688,16 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,
 	DRM_INFO("buffers_ptr %d buffer_count %d len %08x\n",
 		  (int) args->buffers_ptr, args->buffer_count, args->batch_len);
 #endif
+	if (args->flags & I915_EXEC_BSD) {
+		if (!HAS_BSD(dev)) {
+			DRM_ERROR("execbuf with wrong flag\n");
+			return -EINVAL;
+		}
+		ring = &dev_priv->bsd_ring;
+	} else {
+		ring = &dev_priv->render_ring;
+	}
+
 
 	if (args->buffer_count < 1) {
 		DRM_ERROR("execbuf with %d buffers\n", args->buffer_count);
@@ -3694,8 +3751,6 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,
 		goto pre_mutex_err;
 	}
 
-	ring = &dev_priv->render_ring;
-
 	/* Look up object handles */
 	flips = 0;
 	for (i = 0; i < args->buffer_count; i++) {
@@ -3834,6 +3889,10 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,
 					dev->flush_domains,
 					&dev_priv->render_ring);
 
+			if (HAS_BSD(dev))
+				(void)i915_add_request(dev, file_priv,
+						dev->flush_domains,
+						&dev_priv->bsd_ring);
 		}
 	}
 
@@ -4267,6 +4326,9 @@ i915_gem_busy_ioctl(struct drm_device *dev, void *data,
 	 */
 	i915_gem_retire_requests(dev, &dev_priv->render_ring);
 
+	if (HAS_BSD(dev))
+		i915_gem_retire_requests(dev, &dev_priv->bsd_ring);
+
 	obj_priv = to_intel_bo(obj);
 	/* Don't count being on the flushing list against the object being
 	 * done.  Otherwise, a buffer left on the flushing list but not getting
@@ -4433,7 +4495,9 @@ i915_gem_idle(struct drm_device *dev)
 	mutex_lock(&dev->struct_mutex);
 
 	if (dev_priv->mm.suspended ||
-			dev_priv->render_ring.gem_object == NULL) {
+			(dev_priv->render_ring.gem_object == NULL) ||
+			(HAS_BSD(dev) &&
+			 dev_priv->bsd_ring.gem_object == NULL)) {
 		mutex_unlock(&dev->struct_mutex);
 		return 0;
 	}
@@ -4550,6 +4614,10 @@ i915_gem_init_ringbuffer(struct drm_device *dev)
 			return ret;
 	}
 	ret = intel_init_ring_buffer(dev, &dev_priv->render_ring);
+	if (!ret && HAS_BSD(dev)) {
+		dev_priv->bsd_ring = bsd_ring;
+		ret = intel_init_ring_buffer(dev, &dev_priv->bsd_ring);
+	}
 	return ret;
 }
 
@@ -4559,6 +4627,8 @@ i915_gem_cleanup_ringbuffer(struct drm_device *dev)
 	drm_i915_private_t *dev_priv = dev->dev_private;
 
 	intel_cleanup_ring_buffer(dev, &dev_priv->render_ring);
+	if (HAS_BSD(dev))
+		intel_cleanup_ring_buffer(dev, &dev_priv->bsd_ring);
 	if (HAS_PIPE_CONTROL(dev))
 		i915_gem_cleanup_pipe_control(dev);
 }
@@ -4589,11 +4659,13 @@ i915_gem_entervt_ioctl(struct drm_device *dev, void *data,
 
 	spin_lock(&dev_priv->mm.active_list_lock);
 	BUG_ON(!list_empty(&dev_priv->render_ring.active_list));
+	BUG_ON(HAS_BSD(dev) && !list_empty(&dev_priv->bsd_ring.active_list));
 	spin_unlock(&dev_priv->mm.active_list_lock);
 
 	BUG_ON(!list_empty(&dev_priv->mm.flushing_list));
 	BUG_ON(!list_empty(&dev_priv->mm.inactive_list));
 	BUG_ON(!list_empty(&dev_priv->render_ring.request_list));
+	BUG_ON(HAS_BSD(dev) && !list_empty(&dev_priv->bsd_ring.request_list));
 	mutex_unlock(&dev->struct_mutex);
 
 	drm_irq_install(dev);
@@ -4638,6 +4710,10 @@ i915_gem_load(struct drm_device *dev)
 	INIT_LIST_HEAD(&dev_priv->mm.fence_list);
 	INIT_LIST_HEAD(&dev_priv->render_ring.active_list);
 	INIT_LIST_HEAD(&dev_priv->render_ring.request_list);
+	if (HAS_BSD(dev)) {
+		INIT_LIST_HEAD(&dev_priv->bsd_ring.active_list);
+		INIT_LIST_HEAD(&dev_priv->bsd_ring.request_list);
+	}
 	for (i = 0; i < 16; i++)
 		INIT_LIST_HEAD(&dev_priv->fence_regs[i].lru_list);
 	INIT_DELAYED_WORK(&dev_priv->mm.retire_work,
@@ -4874,6 +4950,8 @@ i915_gpu_is_active(struct drm_device *dev)
 	spin_lock(&dev_priv->mm.active_list_lock);
 	lists_empty = list_empty(&dev_priv->mm.flushing_list) &&
 		      list_empty(&dev_priv->render_ring.active_list);
+	if (HAS_BSD(dev))
+		lists_empty &= list_empty(&dev_priv->bsd_ring.active_list);
 	spin_unlock(&dev_priv->mm.active_list_lock);
 
 	return !lists_empty;
@@ -4920,6 +4998,9 @@ rescan:
 		spin_unlock(&shrink_list_lock);
 		i915_gem_retire_requests(dev, &dev_priv->render_ring);
 
+		if (HAS_BSD(dev))
+			i915_gem_retire_requests(dev, &dev_priv->bsd_ring);
+
 		list_for_each_entry_safe(obj_priv, next_obj,
 					 &dev_priv->mm.inactive_list,
 					 list) {
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index 8a667f1db75a..0a3a5806a12e 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -53,7 +53,7 @@
 	 I915_RENDER_COMMAND_PARSER_ERROR_INTERRUPT)
 
 /** Interrupts that we mask and unmask at runtime. */
-#define I915_INTERRUPT_ENABLE_VAR (I915_USER_INTERRUPT)
+#define I915_INTERRUPT_ENABLE_VAR (I915_USER_INTERRUPT | I915_BSD_USER_INTERRUPT)
 
 #define I915_PIPE_VBLANK_STATUS	(PIPE_START_VBLANK_INTERRUPT_STATUS |\
 				 PIPE_VBLANK_INTERRUPT_STATUS)
@@ -362,6 +362,9 @@ irqreturn_t ironlake_irq_handler(struct drm_device *dev)
 		dev_priv->hangcheck_count = 0;
 		mod_timer(&dev_priv->hangcheck_timer, jiffies + DRM_I915_HANGCHECK_PERIOD);
 	}
+	if (gt_iir & GT_BSD_USER_INTERRUPT)
+		DRM_WAKEUP(&dev_priv->bsd_ring.irq_queue);
+
 
 	if (de_iir & DE_GSE)
 		ironlake_opregion_gse_intr(dev);
@@ -944,6 +947,9 @@ irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS)
 			mod_timer(&dev_priv->hangcheck_timer, jiffies + DRM_I915_HANGCHECK_PERIOD);
 		}
 
+		if (HAS_BSD(dev) && (iir & I915_BSD_USER_INTERRUPT))
+			DRM_WAKEUP(&dev_priv->bsd_ring.irq_queue);
+
 		if (iir & I915_DISPLAY_PLANE_A_FLIP_PENDING_INTERRUPT)
 			intel_prepare_page_flip(dev, 0);
 
@@ -1297,7 +1303,7 @@ static int ironlake_irq_postinstall(struct drm_device *dev)
 	/* enable kind of interrupts always enabled */
 	u32 display_mask = DE_MASTER_IRQ_CONTROL | DE_GSE | DE_PCH_EVENT |
 			   DE_PLANEA_FLIP_DONE | DE_PLANEB_FLIP_DONE;
-	u32 render_mask = GT_PIPE_NOTIFY;
+	u32 render_mask = GT_PIPE_NOTIFY | GT_BSD_USER_INTERRUPT;
 	u32 hotplug_mask = SDE_CRT_HOTPLUG | SDE_PORTB_HOTPLUG |
 			   SDE_PORTC_HOTPLUG | SDE_PORTD_HOTPLUG;
 
@@ -1376,6 +1382,9 @@ int i915_driver_irq_postinstall(struct drm_device *dev)
 
 	DRM_INIT_WAITQUEUE(&dev_priv->render_ring.irq_queue);
 
+	if (HAS_BSD(dev))
+		DRM_INIT_WAITQUEUE(&dev_priv->bsd_ring.irq_queue);
+
 	dev_priv->vblank_pipe = DRM_I915_VBLANK_PIPE_A | DRM_I915_VBLANK_PIPE_B;
 
 	if (HAS_PCH_SPLIT(dev))
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index f3e39cc46f0d..784cf3c914e9 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -334,6 +334,7 @@
 #define   I915_DEBUG_INTERRUPT				(1<<2)
 #define   I915_USER_INTERRUPT				(1<<1)
 #define   I915_ASLE_INTERRUPT				(1<<0)
+#define   I915_BSD_USER_INTERRUPT                      (1<<25)
 #define EIR		0x020b0
 #define EMR		0x020b4
 #define ESR		0x020b8
@@ -368,6 +369,17 @@
 #define BB_ADDR		0x02140 /* 8 bytes */
 #define GFX_FLSH_CNTL	0x02170 /* 915+ only */
 
+/*
+ * BSD (bit stream decoder instruction and interrupt control register defines
+ * (G4X and Ironlake only)
+ */
+
+#define BSD_RING_TAIL          0x04030
+#define BSD_RING_HEAD          0x04034
+#define BSD_RING_START         0x04038
+#define BSD_RING_CTL           0x0403c
+#define BSD_RING_ACTHD         0x04074
+#define BSD_HWS_PGA            0x04080
 
 /*
  * Framebuffer compression (915+ only)
@@ -2355,6 +2367,8 @@
 #define GT_PIPE_NOTIFY		(1 << 4)
 #define GT_SYNC_STATUS          (1 << 2)
 #define GT_USER_INTERRUPT       (1 << 0)
+#define GT_BSD_USER_INTERRUPT   (1 << 5)
+
 
 #define GTISR   0x44010
 #define GTIMR   0x44014
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index 5715c4d8cce9..f6b84fe8099a 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -340,6 +340,119 @@ static void render_setup_status_page(struct drm_device *dev,
 
 }
 
+void
+bsd_ring_flush(struct drm_device *dev,
+		struct intel_ring_buffer *ring,
+		u32     invalidate_domains,
+		u32     flush_domains)
+{
+	intel_ring_begin(dev, ring, 8);
+	intel_ring_emit(dev, ring, MI_FLUSH);
+	intel_ring_emit(dev, ring, MI_NOOP);
+	intel_ring_advance(dev, ring);
+}
+
+static inline unsigned int bsd_ring_get_head(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	return I915_READ(BSD_RING_HEAD) & HEAD_ADDR;
+}
+
+static inline unsigned int bsd_ring_get_tail(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	return I915_READ(BSD_RING_TAIL) & TAIL_ADDR;
+}
+
+static inline unsigned int bsd_ring_get_active_head(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	return I915_READ(BSD_RING_ACTHD);
+}
+
+static inline void bsd_ring_advance_ring(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	I915_WRITE(BSD_RING_TAIL, ring->tail);
+}
+
+static int init_bsd_ring(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	return init_ring_common(dev, ring);
+}
+
+static u32
+bsd_ring_add_request(struct drm_device *dev,
+		struct intel_ring_buffer *ring,
+		struct drm_file *file_priv,
+		u32 flush_domains)
+{
+	u32 seqno;
+	seqno = intel_ring_get_seqno(dev, ring);
+	intel_ring_begin(dev, ring, 4);
+	intel_ring_emit(dev, ring, MI_STORE_DWORD_INDEX);
+	intel_ring_emit(dev, ring,
+			I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT);
+	intel_ring_emit(dev, ring, seqno);
+	intel_ring_emit(dev, ring, MI_USER_INTERRUPT);
+	intel_ring_advance(dev, ring);
+
+	DRM_DEBUG_DRIVER("%s %d\n", ring->name, seqno);
+
+	return seqno;
+}
+
+static void bsd_setup_status_page(struct drm_device *dev,
+		struct  intel_ring_buffer *ring)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	I915_WRITE(BSD_HWS_PGA, ring->status_page.gfx_addr);
+	I915_READ(BSD_HWS_PGA);
+}
+
+static void
+bsd_ring_get_user_irq(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	/* do nothing */
+}
+static void
+bsd_ring_put_user_irq(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	/* do nothing */
+}
+
+static u32
+bsd_ring_get_gem_seqno(struct drm_device *dev,
+		struct intel_ring_buffer *ring)
+{
+	return intel_read_status_page(ring, I915_GEM_HWS_INDEX);
+}
+
+static int
+bsd_ring_dispatch_gem_execbuffer(struct drm_device *dev,
+		struct intel_ring_buffer *ring,
+		struct drm_i915_gem_execbuffer2 *exec,
+		struct drm_clip_rect *cliprects,
+		uint64_t exec_offset)
+{
+	uint32_t exec_start;
+	exec_start = (uint32_t) exec_offset + exec->batch_start_offset;
+	intel_ring_begin(dev, ring, 2);
+	intel_ring_emit(dev, ring, MI_BATCH_BUFFER_START |
+			(2 << 6) | MI_BATCH_NON_SECURE_I965);
+	intel_ring_emit(dev, ring, exec_start);
+	intel_ring_advance(dev, ring);
+	return 0;
+}
+
+
 static int
 render_ring_dispatch_gem_execbuffer(struct drm_device *dev,
 		struct intel_ring_buffer *ring,
@@ -588,6 +701,7 @@ int intel_wait_ring_buffer(struct drm_device *dev,
 			if (master_priv->sarea_priv)
 				master_priv->sarea_priv->perf_boxes |= I915_BOX_WAIT;
 		}
+
 		yield();
 	} while (!time_after(jiffies, end));
 	trace_i915_ring_wait_end (dev);
@@ -682,3 +796,42 @@ struct intel_ring_buffer render_ring = {
 	.status_page		= {NULL, 0, NULL},
 	.map			= {0,}
 };
+
+/* ring buffer for bit-stream decoder */
+
+struct intel_ring_buffer bsd_ring = {
+	.name                   = "bsd ring",
+	.regs			= {
+		.ctl = BSD_RING_CTL,
+		.head = BSD_RING_HEAD,
+		.tail = BSD_RING_TAIL,
+		.start = BSD_RING_START
+	},
+	.ring_flag		= I915_EXEC_BSD,
+	.size			= 32 * PAGE_SIZE,
+	.alignment		= PAGE_SIZE,
+	.virtual_start		= NULL,
+	.dev			= NULL,
+	.gem_object		= NULL,
+	.head			= 0,
+	.tail			= 0,
+	.space			= 0,
+	.next_seqno		= 1,
+	.user_irq_refcount	= 0,
+	.irq_gem_seqno		= 0,
+	.waiting_gem_seqno	= 0,
+	.setup_status_page	= bsd_setup_status_page,
+	.init			= init_bsd_ring,
+	.get_head		= bsd_ring_get_head,
+	.get_tail		= bsd_ring_get_tail,
+	.get_active_head	= bsd_ring_get_active_head,
+	.advance_ring		= bsd_ring_advance_ring,
+	.flush			= bsd_ring_flush,
+	.add_request		= bsd_ring_add_request,
+	.get_gem_seqno		= bsd_ring_get_gem_seqno,
+	.user_irq_get		= bsd_ring_get_user_irq,
+	.user_irq_put		= bsd_ring_put_user_irq,
+	.dispatch_gem_execbuffer = bsd_ring_dispatch_gem_execbuffer,
+	.status_page		= {NULL, 0, NULL},
+	.map			= {0,}
+};
-- 
cgit v1.2.3


From 7648fa99eb77a2e1a90b7beaa420e07d819b9c11 Mon Sep 17 00:00:00 2001
From: Jesse Barnes <jbarnes@virtuousgeek.org>
Date: Thu, 20 May 2010 14:28:11 -0700
Subject: drm/i915: add power monitoring support

Add power monitoring support to the i915 driver for use by the IPS
driver.  Export the available power info to the IPS driver through a few
new inter-driver hooks.  When used together, the IPS driver and this
patch can significantly increase graphics performance on Ironlake class
chips.

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
[anholt: Fixed 32-bit compile.  stupid obfuscating div_u64()]
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_debugfs.c  |  56 +++-
 drivers/gpu/drm/i915/i915_dma.c      | 533 ++++++++++++++++++++++++++++++++++-
 drivers/gpu/drm/i915/i915_drv.h      |  20 +-
 drivers/gpu/drm/i915/i915_irq.c      |  28 +-
 drivers/gpu/drm/i915/i915_reg.h      |  41 +++
 drivers/gpu/drm/i915/intel_display.c | 147 ++++++++--
 6 files changed, 772 insertions(+), 53 deletions(-)

(limited to 'drivers/gpu/drm/i915/i915_drv.h')

diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index c864858d5064..eb11e1e049cb 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -491,11 +491,14 @@ static int i915_cur_delayinfo(struct seq_file *m, void *unused)
 	struct drm_device *dev = node->minor->dev;
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	u16 rgvswctl = I915_READ16(MEMSWCTL);
+	u16 rgvstat = I915_READ16(MEMSTAT_ILK);
 
-	seq_printf(m, "Last command: 0x%01x\n", (rgvswctl >> 13) & 0x3);
-	seq_printf(m, "Command status: %d\n", (rgvswctl >> 12) & 1);
-	seq_printf(m, "P%d DELAY 0x%02x\n", (rgvswctl >> 8) & 0xf,
-		   rgvswctl & 0x3f);
+	seq_printf(m, "Requested P-state: %d\n", (rgvswctl >> 8) & 0xf);
+	seq_printf(m, "Requested VID: %d\n", rgvswctl & 0x3f);
+	seq_printf(m, "Current VID: %d\n", (rgvstat & MEMSTAT_VID_MASK) >>
+		   MEMSTAT_VID_SHIFT);
+	seq_printf(m, "Current P-state: %d\n",
+		   (rgvstat & MEMSTAT_PSTATE_MASK) >> MEMSTAT_PSTATE_SHIFT);
 
 	return 0;
 }
@@ -510,7 +513,8 @@ static int i915_delayfreq_table(struct seq_file *m, void *unused)
 
 	for (i = 0; i < 16; i++) {
 		delayfreq = I915_READ(PXVFREQ_BASE + i * 4);
-		seq_printf(m, "P%02dVIDFREQ: 0x%08x\n", i, delayfreq);
+		seq_printf(m, "P%02dVIDFREQ: 0x%08x (VID: %d)\n", i, delayfreq,
+			   (delayfreq & PXVFREQ_PX_MASK) >> PXVFREQ_PX_SHIFT);
 	}
 
 	return 0;
@@ -543,6 +547,8 @@ static int i915_drpc_info(struct seq_file *m, void *unused)
 	struct drm_device *dev = node->minor->dev;
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	u32 rgvmodectl = I915_READ(MEMMODECTL);
+	u32 rstdbyctl = I915_READ(MCHBAR_RENDER_STANDBY);
+	u16 crstandvid = I915_READ16(CRSTANDVID);
 
 	seq_printf(m, "HD boost: %s\n", (rgvmodectl & MEMMODE_BOOST_EN) ?
 		   "yes" : "no");
@@ -557,9 +563,13 @@ static int i915_drpc_info(struct seq_file *m, void *unused)
 		   rgvmodectl & MEMMODE_RCLK_GATE ? "yes" : "no");
 	seq_printf(m, "Starting frequency: P%d\n",
 		   (rgvmodectl & MEMMODE_FSTART_MASK) >> MEMMODE_FSTART_SHIFT);
-	seq_printf(m, "Max frequency: P%d\n",
+	seq_printf(m, "Max P-state: P%d\n",
 		   (rgvmodectl & MEMMODE_FMAX_MASK) >> MEMMODE_FMAX_SHIFT);
-	seq_printf(m, "Min frequency: P%d\n", (rgvmodectl & MEMMODE_FMIN_MASK));
+	seq_printf(m, "Min P-state: P%d\n", (rgvmodectl & MEMMODE_FMIN_MASK));
+	seq_printf(m, "RS1 VID: %d\n", (crstandvid & 0x3f));
+	seq_printf(m, "RS2 VID: %d\n", ((crstandvid >> 8) & 0x3f));
+	seq_printf(m, "Render standby enabled: %s\n",
+		   (rstdbyctl & RCX_SW_EXIT) ? "no" : "yes");
 
 	return 0;
 }
@@ -623,6 +633,36 @@ static int i915_sr_status(struct seq_file *m, void *unused)
 	return 0;
 }
 
+static int i915_emon_status(struct seq_file *m, void *unused)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	unsigned long temp, chipset, gfx;
+
+	temp = i915_mch_val(dev_priv);
+	chipset = i915_chipset_val(dev_priv);
+	gfx = i915_gfx_val(dev_priv);
+
+	seq_printf(m, "GMCH temp: %ld\n", temp);
+	seq_printf(m, "Chipset power: %ld\n", chipset);
+	seq_printf(m, "GFX power: %ld\n", gfx);
+	seq_printf(m, "Total power: %ld\n", chipset + gfx);
+
+	return 0;
+}
+
+static int i915_gfxec(struct seq_file *m, void *unused)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	drm_i915_private_t *dev_priv = dev->dev_private;
+
+	seq_printf(m, "GFXEC: %ld\n", (unsigned long)I915_READ(0x112f4));
+
+	return 0;
+}
+
 static int
 i915_wedged_open(struct inode *inode,
 		 struct file *filp)
@@ -745,6 +785,8 @@ static struct drm_info_list i915_debugfs_list[] = {
 	{"i915_delayfreq_table", i915_delayfreq_table, 0},
 	{"i915_inttoext_table", i915_inttoext_table, 0},
 	{"i915_drpc_info", i915_drpc_info, 0},
+	{"i915_emon_status", i915_emon_status, 0},
+	{"i915_gfxec", i915_gfxec, 0},
 	{"i915_fbc_status", i915_fbc_status, 0},
 	{"i915_sr_status", i915_sr_status, 0},
 };
diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 1dbed700800e..12e92f2cc3a7 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -1458,14 +1458,11 @@ void i915_master_destroy(struct drm_device *dev, struct drm_master *master)
 	master->driver_priv = NULL;
 }
 
-static void i915_get_mem_freq(struct drm_device *dev)
+static void i915_pineview_get_mem_freq(struct drm_device *dev)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	u32 tmp;
 
-	if (!IS_PINEVIEW(dev))
-		return;
-
 	tmp = I915_READ(CLKCFG);
 
 	switch (tmp & CLKCFG_FSB_MASK) {
@@ -1496,6 +1493,519 @@ static void i915_get_mem_freq(struct drm_device *dev)
 	}
 }
 
+static void i915_ironlake_get_mem_freq(struct drm_device *dev)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	u16 ddrpll, csipll;
+
+	ddrpll = I915_READ16(DDRMPLL1);
+	csipll = I915_READ16(CSIPLL0);
+
+	switch (ddrpll & 0xff) {
+	case 0xc:
+		dev_priv->mem_freq = 800;
+		break;
+	case 0x10:
+		dev_priv->mem_freq = 1066;
+		break;
+	case 0x14:
+		dev_priv->mem_freq = 1333;
+		break;
+	case 0x18:
+		dev_priv->mem_freq = 1600;
+		break;
+	default:
+		DRM_DEBUG_DRIVER("unknown memory frequency 0x%02x\n",
+				 ddrpll & 0xff);
+		dev_priv->mem_freq = 0;
+		break;
+	}
+
+	dev_priv->r_t = dev_priv->mem_freq;
+
+	switch (csipll & 0x3ff) {
+	case 0x00c:
+		dev_priv->fsb_freq = 3200;
+		break;
+	case 0x00e:
+		dev_priv->fsb_freq = 3733;
+		break;
+	case 0x010:
+		dev_priv->fsb_freq = 4266;
+		break;
+	case 0x012:
+		dev_priv->fsb_freq = 4800;
+		break;
+	case 0x014:
+		dev_priv->fsb_freq = 5333;
+		break;
+	case 0x016:
+		dev_priv->fsb_freq = 5866;
+		break;
+	case 0x018:
+		dev_priv->fsb_freq = 6400;
+		break;
+	default:
+		DRM_DEBUG_DRIVER("unknown fsb frequency 0x%04x\n",
+				 csipll & 0x3ff);
+		dev_priv->fsb_freq = 0;
+		break;
+	}
+
+	if (dev_priv->fsb_freq == 3200) {
+		dev_priv->c_m = 0;
+	} else if (dev_priv->fsb_freq > 3200 && dev_priv->fsb_freq <= 4800) {
+		dev_priv->c_m = 1;
+	} else {
+		dev_priv->c_m = 2;
+	}
+}
+
+struct v_table {
+	u8 vid;
+	unsigned long vd; /* in .1 mil */
+	unsigned long vm; /* in .1 mil */
+	u8 pvid;
+};
+
+static struct v_table v_table[] = {
+	{ 0, 16125, 15000, 0x7f, },
+	{ 1, 16000, 14875, 0x7e, },
+	{ 2, 15875, 14750, 0x7d, },
+	{ 3, 15750, 14625, 0x7c, },
+	{ 4, 15625, 14500, 0x7b, },
+	{ 5, 15500, 14375, 0x7a, },
+	{ 6, 15375, 14250, 0x79, },
+	{ 7, 15250, 14125, 0x78, },
+	{ 8, 15125, 14000, 0x77, },
+	{ 9, 15000, 13875, 0x76, },
+	{ 10, 14875, 13750, 0x75, },
+	{ 11, 14750, 13625, 0x74, },
+	{ 12, 14625, 13500, 0x73, },
+	{ 13, 14500, 13375, 0x72, },
+	{ 14, 14375, 13250, 0x71, },
+	{ 15, 14250, 13125, 0x70, },
+	{ 16, 14125, 13000, 0x6f, },
+	{ 17, 14000, 12875, 0x6e, },
+	{ 18, 13875, 12750, 0x6d, },
+	{ 19, 13750, 12625, 0x6c, },
+	{ 20, 13625, 12500, 0x6b, },
+	{ 21, 13500, 12375, 0x6a, },
+	{ 22, 13375, 12250, 0x69, },
+	{ 23, 13250, 12125, 0x68, },
+	{ 24, 13125, 12000, 0x67, },
+	{ 25, 13000, 11875, 0x66, },
+	{ 26, 12875, 11750, 0x65, },
+	{ 27, 12750, 11625, 0x64, },
+	{ 28, 12625, 11500, 0x63, },
+	{ 29, 12500, 11375, 0x62, },
+	{ 30, 12375, 11250, 0x61, },
+	{ 31, 12250, 11125, 0x60, },
+	{ 32, 12125, 11000, 0x5f, },
+	{ 33, 12000, 10875, 0x5e, },
+	{ 34, 11875, 10750, 0x5d, },
+	{ 35, 11750, 10625, 0x5c, },
+	{ 36, 11625, 10500, 0x5b, },
+	{ 37, 11500, 10375, 0x5a, },
+	{ 38, 11375, 10250, 0x59, },
+	{ 39, 11250, 10125, 0x58, },
+	{ 40, 11125, 10000, 0x57, },
+	{ 41, 11000, 9875, 0x56, },
+	{ 42, 10875, 9750, 0x55, },
+	{ 43, 10750, 9625, 0x54, },
+	{ 44, 10625, 9500, 0x53, },
+	{ 45, 10500, 9375, 0x52, },
+	{ 46, 10375, 9250, 0x51, },
+	{ 47, 10250, 9125, 0x50, },
+	{ 48, 10125, 9000, 0x4f, },
+	{ 49, 10000, 8875, 0x4e, },
+	{ 50, 9875, 8750, 0x4d, },
+	{ 51, 9750, 8625, 0x4c, },
+	{ 52, 9625, 8500, 0x4b, },
+	{ 53, 9500, 8375, 0x4a, },
+	{ 54, 9375, 8250, 0x49, },
+	{ 55, 9250, 8125, 0x48, },
+	{ 56, 9125, 8000, 0x47, },
+	{ 57, 9000, 7875, 0x46, },
+	{ 58, 8875, 7750, 0x45, },
+	{ 59, 8750, 7625, 0x44, },
+	{ 60, 8625, 7500, 0x43, },
+	{ 61, 8500, 7375, 0x42, },
+	{ 62, 8375, 7250, 0x41, },
+	{ 63, 8250, 7125, 0x40, },
+	{ 64, 8125, 7000, 0x3f, },
+	{ 65, 8000, 6875, 0x3e, },
+	{ 66, 7875, 6750, 0x3d, },
+	{ 67, 7750, 6625, 0x3c, },
+	{ 68, 7625, 6500, 0x3b, },
+	{ 69, 7500, 6375, 0x3a, },
+	{ 70, 7375, 6250, 0x39, },
+	{ 71, 7250, 6125, 0x38, },
+	{ 72, 7125, 6000, 0x37, },
+	{ 73, 7000, 5875, 0x36, },
+	{ 74, 6875, 5750, 0x35, },
+	{ 75, 6750, 5625, 0x34, },
+	{ 76, 6625, 5500, 0x33, },
+	{ 77, 6500, 5375, 0x32, },
+	{ 78, 6375, 5250, 0x31, },
+	{ 79, 6250, 5125, 0x30, },
+	{ 80, 6125, 5000, 0x2f, },
+	{ 81, 6000, 4875, 0x2e, },
+	{ 82, 5875, 4750, 0x2d, },
+	{ 83, 5750, 4625, 0x2c, },
+	{ 84, 5625, 4500, 0x2b, },
+	{ 85, 5500, 4375, 0x2a, },
+	{ 86, 5375, 4250, 0x29, },
+	{ 87, 5250, 4125, 0x28, },
+	{ 88, 5125, 4000, 0x27, },
+	{ 89, 5000, 3875, 0x26, },
+	{ 90, 4875, 3750, 0x25, },
+	{ 91, 4750, 3625, 0x24, },
+	{ 92, 4625, 3500, 0x23, },
+	{ 93, 4500, 3375, 0x22, },
+	{ 94, 4375, 3250, 0x21, },
+	{ 95, 4250, 3125, 0x20, },
+	{ 96, 4125, 3000, 0x1f, },
+	{ 97, 4125, 3000, 0x1e, },
+	{ 98, 4125, 3000, 0x1d, },
+	{ 99, 4125, 3000, 0x1c, },
+	{ 100, 4125, 3000, 0x1b, },
+	{ 101, 4125, 3000, 0x1a, },
+	{ 102, 4125, 3000, 0x19, },
+	{ 103, 4125, 3000, 0x18, },
+	{ 104, 4125, 3000, 0x17, },
+	{ 105, 4125, 3000, 0x16, },
+	{ 106, 4125, 3000, 0x15, },
+	{ 107, 4125, 3000, 0x14, },
+	{ 108, 4125, 3000, 0x13, },
+	{ 109, 4125, 3000, 0x12, },
+	{ 110, 4125, 3000, 0x11, },
+	{ 111, 4125, 3000, 0x10, },
+	{ 112, 4125, 3000, 0x0f, },
+	{ 113, 4125, 3000, 0x0e, },
+	{ 114, 4125, 3000, 0x0d, },
+	{ 115, 4125, 3000, 0x0c, },
+	{ 116, 4125, 3000, 0x0b, },
+	{ 117, 4125, 3000, 0x0a, },
+	{ 118, 4125, 3000, 0x09, },
+	{ 119, 4125, 3000, 0x08, },
+	{ 120, 1125, 0, 0x07, },
+	{ 121, 1000, 0, 0x06, },
+	{ 122, 875, 0, 0x05, },
+	{ 123, 750, 0, 0x04, },
+	{ 124, 625, 0, 0x03, },
+	{ 125, 500, 0, 0x02, },
+	{ 126, 375, 0, 0x01, },
+	{ 127, 0, 0, 0x00, },
+};
+
+struct cparams {
+	int i;
+	int t;
+	int m;
+	int c;
+};
+
+static struct cparams cparams[] = {
+	{ 1, 1333, 301, 28664 },
+	{ 1, 1066, 294, 24460 },
+	{ 1, 800, 294, 25192 },
+	{ 0, 1333, 276, 27605 },
+	{ 0, 1066, 276, 27605 },
+	{ 0, 800, 231, 23784 },
+};
+
+unsigned long i915_chipset_val(struct drm_i915_private *dev_priv)
+{
+	u64 total_count, diff, ret;
+	u32 count1, count2, count3, m = 0, c = 0;
+	unsigned long now = jiffies_to_msecs(jiffies), diff1;
+	int i;
+
+	diff1 = now - dev_priv->last_time1;
+
+	count1 = I915_READ(DMIEC);
+	count2 = I915_READ(DDREC);
+	count3 = I915_READ(CSIEC);
+
+	total_count = count1 + count2 + count3;
+
+	/* FIXME: handle per-counter overflow */
+	if (total_count < dev_priv->last_count1) {
+		diff = ~0UL - dev_priv->last_count1;
+		diff += total_count;
+	} else {
+		diff = total_count - dev_priv->last_count1;
+	}
+
+	for (i = 0; i < ARRAY_SIZE(cparams); i++) {
+		if (cparams[i].i == dev_priv->c_m &&
+		    cparams[i].t == dev_priv->r_t) {
+			m = cparams[i].m;
+			c = cparams[i].c;
+			break;
+		}
+	}
+
+	div_u64(diff, diff1);
+	ret = ((m * diff) + c);
+	div_u64(ret, 10);
+
+	dev_priv->last_count1 = total_count;
+	dev_priv->last_time1 = now;
+
+	return ret;
+}
+
+unsigned long i915_mch_val(struct drm_i915_private *dev_priv)
+{
+	unsigned long m, x, b;
+	u32 tsfs;
+
+	tsfs = I915_READ(TSFS);
+
+	m = ((tsfs & TSFS_SLOPE_MASK) >> TSFS_SLOPE_SHIFT);
+	x = I915_READ8(TR1);
+
+	b = tsfs & TSFS_INTR_MASK;
+
+	return ((m * x) / 127) - b;
+}
+
+static unsigned long pvid_to_extvid(struct drm_i915_private *dev_priv, u8 pxvid)
+{
+	unsigned long val = 0;
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(v_table); i++) {
+		if (v_table[i].pvid == pxvid) {
+			if (IS_MOBILE(dev_priv->dev))
+				val = v_table[i].vm;
+			else
+				val = v_table[i].vd;
+		}
+	}
+
+	return val;
+}
+
+void i915_update_gfx_val(struct drm_i915_private *dev_priv)
+{
+	struct timespec now, diff1;
+	u64 diff;
+	unsigned long diffms;
+	u32 count;
+
+	getrawmonotonic(&now);
+	diff1 = timespec_sub(now, dev_priv->last_time2);
+
+	/* Don't divide by 0 */
+	diffms = diff1.tv_sec * 1000 + diff1.tv_nsec / 1000000;
+	if (!diffms)
+		return;
+
+	count = I915_READ(GFXEC);
+
+	if (count < dev_priv->last_count2) {
+		diff = ~0UL - dev_priv->last_count2;
+		diff += count;
+	} else {
+		diff = count - dev_priv->last_count2;
+	}
+
+	dev_priv->last_count2 = count;
+	dev_priv->last_time2 = now;
+
+	/* More magic constants... */
+	diff = diff * 1181;
+	div_u64(diff, diffms * 10);
+	dev_priv->gfx_power = diff;
+}
+
+unsigned long i915_gfx_val(struct drm_i915_private *dev_priv)
+{
+	unsigned long t, corr, state1, corr2, state2;
+	u32 pxvid, ext_v;
+
+	pxvid = I915_READ(PXVFREQ_BASE + (dev_priv->cur_delay * 4));
+	pxvid = (pxvid >> 24) & 0x7f;
+	ext_v = pvid_to_extvid(dev_priv, pxvid);
+
+	state1 = ext_v;
+
+	t = i915_mch_val(dev_priv);
+
+	/* Revel in the empirically derived constants */
+
+	/* Correction factor in 1/100000 units */
+	if (t > 80)
+		corr = ((t * 2349) + 135940);
+	else if (t >= 50)
+		corr = ((t * 964) + 29317);
+	else /* < 50 */
+		corr = ((t * 301) + 1004);
+
+	corr = corr * ((150142 * state1) / 10000 - 78642);
+	corr /= 100000;
+	corr2 = (corr * dev_priv->corr);
+
+	state2 = (corr2 * state1) / 10000;
+	state2 /= 100; /* convert to mW */
+
+	i915_update_gfx_val(dev_priv);
+
+	return dev_priv->gfx_power + state2;
+}
+
+/* Global for IPS driver to get at the current i915 device */
+static struct drm_i915_private *i915_mch_dev;
+/*
+ * Lock protecting IPS related data structures
+ *   - i915_mch_dev
+ *   - dev_priv->max_delay
+ *   - dev_priv->min_delay
+ *   - dev_priv->fmax
+ *   - dev_priv->gpu_busy
+ */
+DEFINE_SPINLOCK(mchdev_lock);
+
+/**
+ * i915_read_mch_val - return value for IPS use
+ *
+ * Calculate and return a value for the IPS driver to use when deciding whether
+ * we have thermal and power headroom to increase CPU or GPU power budget.
+ */
+unsigned long i915_read_mch_val(void)
+{
+  	struct drm_i915_private *dev_priv;
+	unsigned long chipset_val, graphics_val, ret = 0;
+
+  	spin_lock(&mchdev_lock);
+	if (!i915_mch_dev)
+		goto out_unlock;
+	dev_priv = i915_mch_dev;
+
+	chipset_val = i915_chipset_val(dev_priv);
+	graphics_val = i915_gfx_val(dev_priv);
+
+	ret = chipset_val + graphics_val;
+
+out_unlock:
+  	spin_unlock(&mchdev_lock);
+
+  	return ret;
+}
+EXPORT_SYMBOL_GPL(i915_read_mch_val);
+
+/**
+ * i915_gpu_raise - raise GPU frequency limit
+ *
+ * Raise the limit; IPS indicates we have thermal headroom.
+ */
+bool i915_gpu_raise(void)
+{
+  	struct drm_i915_private *dev_priv;
+	bool ret = true;
+
+  	spin_lock(&mchdev_lock);
+	if (!i915_mch_dev) {
+		ret = false;
+		goto out_unlock;
+	}
+	dev_priv = i915_mch_dev;
+
+	if (dev_priv->max_delay > dev_priv->fmax)
+		dev_priv->max_delay--;
+
+out_unlock:
+  	spin_unlock(&mchdev_lock);
+
+  	return ret;
+}
+EXPORT_SYMBOL_GPL(i915_gpu_raise);
+
+/**
+ * i915_gpu_lower - lower GPU frequency limit
+ *
+ * IPS indicates we're close to a thermal limit, so throttle back the GPU
+ * frequency maximum.
+ */
+bool i915_gpu_lower(void)
+{
+  	struct drm_i915_private *dev_priv;
+	bool ret = true;
+
+  	spin_lock(&mchdev_lock);
+	if (!i915_mch_dev) {
+		ret = false;
+		goto out_unlock;
+	}
+	dev_priv = i915_mch_dev;
+
+	if (dev_priv->max_delay < dev_priv->min_delay)
+		dev_priv->max_delay++;
+
+out_unlock:
+  	spin_unlock(&mchdev_lock);
+
+  	return ret;
+}
+EXPORT_SYMBOL_GPL(i915_gpu_lower);
+
+/**
+ * i915_gpu_busy - indicate GPU business to IPS
+ *
+ * Tell the IPS driver whether or not the GPU is busy.
+ */
+bool i915_gpu_busy(void)
+{
+  	struct drm_i915_private *dev_priv;
+	bool ret = false;
+
+  	spin_lock(&mchdev_lock);
+	if (!i915_mch_dev)
+		goto out_unlock;
+	dev_priv = i915_mch_dev;
+
+	ret = dev_priv->busy;
+
+out_unlock:
+  	spin_unlock(&mchdev_lock);
+
+  	return ret;
+}
+EXPORT_SYMBOL_GPL(i915_gpu_busy);
+
+/**
+ * i915_gpu_turbo_disable - disable graphics turbo
+ *
+ * Disable graphics turbo by resetting the max frequency and setting the
+ * current frequency to the default.
+ */
+bool i915_gpu_turbo_disable(void)
+{
+  	struct drm_i915_private *dev_priv;
+	bool ret = true;
+
+  	spin_lock(&mchdev_lock);
+	if (!i915_mch_dev) {
+		ret = false;
+		goto out_unlock;
+	}
+	dev_priv = i915_mch_dev;
+
+	dev_priv->max_delay = dev_priv->fstart;
+
+	if (!ironlake_set_drps(dev_priv->dev, dev_priv->fstart))
+		ret = false;
+
+out_unlock:
+  	spin_unlock(&mchdev_lock);
+
+  	return ret;
+}
+EXPORT_SYMBOL_GPL(i915_gpu_turbo_disable);
+
 /**
  * i915_driver_load - setup chip and create an initial config
  * @dev: DRM device
@@ -1616,7 +2126,10 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)
 			goto out_workqueue_free;
 	}
 
-	i915_get_mem_freq(dev);
+	if (IS_PINEVIEW(dev))
+		i915_pineview_get_mem_freq(dev);
+	else if (IS_IRONLAKE(dev))
+		i915_ironlake_get_mem_freq(dev);
 
 	/* On the 945G/GM, the chipset reports the MSI capability on the
 	 * integrated graphics even though the support isn't actually there
@@ -1662,6 +2175,12 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)
 
 	setup_timer(&dev_priv->hangcheck_timer, i915_hangcheck_elapsed,
 		    (unsigned long) dev);
+
+	spin_lock(&mchdev_lock);
+	i915_mch_dev = dev_priv;
+	dev_priv->mchdev_lock = &mchdev_lock;
+	spin_unlock(&mchdev_lock);
+
 	return 0;
 
 out_workqueue_free:
@@ -1683,6 +2202,10 @@ int i915_driver_unload(struct drm_device *dev)
 
 	i915_destroy_error_state(dev);
 
+	spin_lock(&mchdev_lock);
+	i915_mch_dev = NULL;
+	spin_unlock(&mchdev_lock);
+
 	destroy_workqueue(dev_priv->wq);
 	del_timer_sync(&dev_priv->hangcheck_timer);
 
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 6bc0fc080f2b..91b3a1c20ef8 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -619,6 +619,18 @@ typedef struct drm_i915_private {
 	u8 cur_delay;
 	u8 min_delay;
 	u8 max_delay;
+	u8 fmax;
+	u8 fstart;
+
+ 	u64 last_count1;
+ 	unsigned long last_time1;
+ 	u64 last_count2;
+ 	struct timespec last_time2;
+ 	unsigned long gfx_power;
+ 	int c_m;
+ 	int r_t;
+ 	u8 corr;
+	spinlock_t *mchdev_lock;
 
 	enum no_fbc_reason no_fbc_reason;
 
@@ -802,6 +814,11 @@ extern int i915_emit_box(struct drm_device *dev,
 			 struct drm_clip_rect *boxes,
 			 int i, int DR1, int DR4);
 extern int i965_reset(struct drm_device *dev, u8 flags);
+extern unsigned long i915_chipset_val(struct drm_i915_private *dev_priv);
+extern unsigned long i915_mch_val(struct drm_i915_private *dev_priv);
+extern unsigned long i915_gfx_val(struct drm_i915_private *dev_priv);
+extern void i915_update_gfx_val(struct drm_i915_private *dev_priv);
+
 
 /* i915_irq.c */
 void i915_hangcheck_elapsed(unsigned long data);
@@ -1005,7 +1022,7 @@ extern void g4x_disable_fbc(struct drm_device *dev);
 extern void intel_disable_fbc(struct drm_device *dev);
 extern void intel_enable_fbc(struct drm_crtc *crtc, unsigned long interval);
 extern bool intel_fbc_enabled(struct drm_device *dev);
-
+extern bool ironlake_set_drps(struct drm_device *dev, u8 val);
 extern void intel_detect_pch (struct drm_device *dev);
 extern int intel_trans_dp_port_sel (struct drm_crtc *crtc);
 
@@ -1030,6 +1047,7 @@ extern int intel_trans_dp_port_sel (struct drm_crtc *crtc);
 #define I915_WRITE64(reg, val)	writeq(val, dev_priv->regs + (reg))
 #define I915_READ64(reg)	readq(dev_priv->regs + (reg))
 #define POSTING_READ(reg)	(void)I915_READ(reg)
+#define POSTING_READ16(reg)	(void)I915_READ16(reg)
 
 #define I915_VERBOSE 0
 
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index 0a3a5806a12e..b5dba4795f6f 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -278,10 +278,9 @@ static void i915_handle_rps_change(struct drm_device *dev)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	u32 busy_up, busy_down, max_avg, min_avg;
-	u16 rgvswctl;
 	u8 new_delay = dev_priv->cur_delay;
 
-	I915_WRITE(MEMINTRSTS, I915_READ(MEMINTRSTS) & ~MEMINT_EVAL_CHG);
+	I915_WRITE16(MEMINTRSTS, MEMINT_EVAL_CHG);
 	busy_up = I915_READ(RCPREVBSYTUPAVG);
 	busy_down = I915_READ(RCPREVBSYTDNAVG);
 	max_avg = I915_READ(RCBMAXAVG);
@@ -300,27 +299,8 @@ static void i915_handle_rps_change(struct drm_device *dev)
 			new_delay = dev_priv->min_delay;
 	}
 
-	DRM_DEBUG("rps change requested: %d -> %d\n",
-		  dev_priv->cur_delay, new_delay);
-
-	rgvswctl = I915_READ(MEMSWCTL);
-	if (rgvswctl & MEMCTL_CMD_STS) {
-		DRM_ERROR("gpu busy, RCS change rejected\n");
-		return; /* still busy with another command */
-	}
-
-	/* Program the new state */
-	rgvswctl = (MEMCTL_CMD_CHFREQ << MEMCTL_CMD_SHIFT) |
-		(new_delay << MEMCTL_FREQ_SHIFT) | MEMCTL_SFCAVM;
-	I915_WRITE(MEMSWCTL, rgvswctl);
-	POSTING_READ(MEMSWCTL);
-
-	rgvswctl |= MEMCTL_CMD_STS;
-	I915_WRITE(MEMSWCTL, rgvswctl);
-
-	dev_priv->cur_delay = new_delay;
-
-	DRM_DEBUG("rps changed\n");
+	if (ironlake_set_drps(dev, new_delay))
+		dev_priv->cur_delay = new_delay;
 
 	return;
 }
@@ -392,7 +372,7 @@ irqreturn_t ironlake_irq_handler(struct drm_device *dev)
 	}
 
 	if (de_iir & DE_PCU_EVENT) {
-		I915_WRITE(MEMINTRSTS, I915_READ(MEMINTRSTS));
+		I915_WRITE16(MEMINTRSTS, I915_READ(MEMINTRSTS));
 		i915_handle_rps_change(dev);
 	}
 
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index 7a726840efa7..df7224de0aed 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -838,6 +838,12 @@
 #define CLKCFG_MEM_800					(3 << 4)
 #define CLKCFG_MEM_MASK					(7 << 4)
 
+#define TR1			0x11006
+#define TSFS			0x11020
+#define   TSFS_SLOPE_MASK	0x0000ff00
+#define   TSFS_SLOPE_SHIFT	8
+#define   TSFS_INTR_MASK	0x000000ff
+
 #define CRSTANDVID		0x11100
 #define PXVFREQ_BASE		0x11110 /* P[0-15]VIDFREQ (0x1114c) (Ironlake) */
 #define   PXVFREQ_PX_MASK	0x7f000000
@@ -976,6 +982,41 @@
 #define   MEMSTAT_SRC_CTL_STDBY 3
 #define RCPREVBSYTUPAVG		0x113b8
 #define RCPREVBSYTDNAVG		0x113bc
+#define SDEW			0x1124c
+#define CSIEW0			0x11250
+#define CSIEW1			0x11254
+#define CSIEW2			0x11258
+#define PEW			0x1125c
+#define DEW			0x11270
+#define MCHAFE			0x112c0
+#define CSIEC			0x112e0
+#define DMIEC			0x112e4
+#define DDREC			0x112e8
+#define PEG0EC			0x112ec
+#define PEG1EC			0x112f0
+#define GFXEC			0x112f4
+#define RPPREVBSYTUPAVG		0x113b8
+#define RPPREVBSYTDNAVG		0x113bc
+#define ECR			0x11600
+#define   ECR_GPFE		(1<<31)
+#define   ECR_IMONE		(1<<30)
+#define   ECR_CAP_MASK		0x0000001f /* Event range, 0-31 */
+#define OGW0			0x11608
+#define OGW1			0x1160c
+#define EG0			0x11610
+#define EG1			0x11614
+#define EG2			0x11618
+#define EG3			0x1161c
+#define EG4			0x11620
+#define EG5			0x11624
+#define EG6			0x11628
+#define EG7			0x1162c
+#define PXW			0x11664
+#define PXWL			0x11680
+#define LCFUSE02		0x116c0
+#define   LCFUSE_HIV_MASK	0x000000ff
+#define CSIPLL0			0x12c10
+#define DDRMPLL1		0X12c20
 #define PEG_BAND_GAP_DATA	0x14d68
 
 /*
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 4c7c151114f7..e94d1db35364 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -4459,6 +4459,8 @@ static void intel_idle_update(struct work_struct *work)
 
 	mutex_lock(&dev->struct_mutex);
 
+	i915_update_gfx_val(dev_priv);
+
 	if (IS_I945G(dev) || IS_I945GM(dev)) {
 		DRM_DEBUG_DRIVER("enable memory self refresh on 945\n");
 		I915_WRITE(FW_BLC_SELF, FW_BLC_SELF_EN_MASK | FW_BLC_SELF_EN);
@@ -5045,10 +5047,32 @@ err_unref:
 	return NULL;
 }
 
+bool ironlake_set_drps(struct drm_device *dev, u8 val)
+{
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	u16 rgvswctl;
+
+	rgvswctl = I915_READ16(MEMSWCTL);
+	if (rgvswctl & MEMCTL_CMD_STS) {
+		DRM_DEBUG("gpu busy, RCS change rejected\n");
+		return false; /* still busy with another command */
+	}
+
+	rgvswctl = (MEMCTL_CMD_CHFREQ << MEMCTL_CMD_SHIFT) |
+		(val << MEMCTL_FREQ_SHIFT) | MEMCTL_SFCAVM;
+	I915_WRITE16(MEMSWCTL, rgvswctl);
+	POSTING_READ16(MEMSWCTL);
+
+	rgvswctl |= MEMCTL_CMD_STS;
+	I915_WRITE16(MEMSWCTL, rgvswctl);
+
+	return true;
+}
+
 void ironlake_enable_drps(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
-	u32 rgvmodectl = I915_READ(MEMMODECTL), rgvswctl;
+	u32 rgvmodectl = I915_READ(MEMMODECTL);
 	u8 fmax, fmin, fstart, vstart;
 	int i = 0;
 
@@ -5067,13 +5091,21 @@ void ironlake_enable_drps(struct drm_device *dev)
 	fmin = (rgvmodectl & MEMMODE_FMIN_MASK);
 	fstart = (rgvmodectl & MEMMODE_FSTART_MASK) >>
 		MEMMODE_FSTART_SHIFT;
+	fstart = fmax;
+
 	vstart = (I915_READ(PXVFREQ_BASE + (fstart * 4)) & PXVFREQ_PX_MASK) >>
 		PXVFREQ_PX_SHIFT;
 
-	dev_priv->max_delay = fstart; /* can't go to fmax w/o IPS */
+	dev_priv->fmax = fstart; /* IPS callback will increase this */
+	dev_priv->fstart = fstart;
+
+	dev_priv->max_delay = fmax;
 	dev_priv->min_delay = fmin;
 	dev_priv->cur_delay = fstart;
 
+	DRM_DEBUG_DRIVER("fmax: %d, fmin: %d, fstart: %d\n", fmax, fmin,
+			 fstart);
+
 	I915_WRITE(MEMINTREN, MEMINT_CX_SUPR_EN | MEMINT_EVAL_CHG_EN);
 
 	/*
@@ -5095,20 +5127,19 @@ void ironlake_enable_drps(struct drm_device *dev)
 	}
 	msleep(1);
 
-	rgvswctl = (MEMCTL_CMD_CHFREQ << MEMCTL_CMD_SHIFT) |
-		(fstart << MEMCTL_FREQ_SHIFT) | MEMCTL_SFCAVM;
-	I915_WRITE(MEMSWCTL, rgvswctl);
-	POSTING_READ(MEMSWCTL);
+	ironlake_set_drps(dev, fstart);
 
-	rgvswctl |= MEMCTL_CMD_STS;
-	I915_WRITE(MEMSWCTL, rgvswctl);
+	dev_priv->last_count1 = I915_READ(0x112e4) + I915_READ(0x112e8) +
+		I915_READ(0x112e0);
+	dev_priv->last_time1 = jiffies_to_msecs(jiffies);
+	dev_priv->last_count2 = I915_READ(0x112f4);
+	getrawmonotonic(&dev_priv->last_time2);
 }
 
 void ironlake_disable_drps(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
-	u32 rgvswctl;
-	u8 fstart;
+	u16 rgvswctl = I915_READ16(MEMSWCTL);
 
 	/* Ack interrupts, disable EFC interrupt */
 	I915_WRITE(MEMINTREN, I915_READ(MEMINTREN) & ~MEMINT_EVAL_CHG_EN);
@@ -5118,11 +5149,7 @@ void ironlake_disable_drps(struct drm_device *dev)
 	I915_WRITE(DEIMR, I915_READ(DEIMR) | DE_PCU_EVENT);
 
 	/* Go back to the starting frequency */
-	fstart = (I915_READ(MEMMODECTL) & MEMMODE_FSTART_MASK) >>
-		MEMMODE_FSTART_SHIFT;
-	rgvswctl = (MEMCTL_CMD_CHFREQ << MEMCTL_CMD_SHIFT) |
-		(fstart << MEMCTL_FREQ_SHIFT) | MEMCTL_SFCAVM;
-	I915_WRITE(MEMSWCTL, rgvswctl);
+	ironlake_set_drps(dev, dev_priv->fstart);
 	msleep(1);
 	rgvswctl |= MEMCTL_CMD_STS;
 	I915_WRITE(MEMSWCTL, rgvswctl);
@@ -5130,6 +5157,92 @@ void ironlake_disable_drps(struct drm_device *dev)
 
 }
 
+static unsigned long intel_pxfreq(u32 vidfreq)
+{
+	unsigned long freq;
+	int div = (vidfreq & 0x3f0000) >> 16;
+	int post = (vidfreq & 0x3000) >> 12;
+	int pre = (vidfreq & 0x7);
+
+	if (!pre)
+		return 0;
+
+	freq = ((div * 133333) / ((1<<post) * pre));
+
+	return freq;
+}
+
+void intel_init_emon(struct drm_device *dev)
+{
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	u32 lcfuse;
+	u8 pxw[16];
+	int i;
+
+	/* Disable to program */
+	I915_WRITE(ECR, 0);
+	POSTING_READ(ECR);
+
+	/* Program energy weights for various events */
+	I915_WRITE(SDEW, 0x15040d00);
+	I915_WRITE(CSIEW0, 0x007f0000);
+	I915_WRITE(CSIEW1, 0x1e220004);
+	I915_WRITE(CSIEW2, 0x04000004);
+
+	for (i = 0; i < 5; i++)
+		I915_WRITE(PEW + (i * 4), 0);
+	for (i = 0; i < 3; i++)
+		I915_WRITE(DEW + (i * 4), 0);
+
+	/* Program P-state weights to account for frequency power adjustment */
+	for (i = 0; i < 16; i++) {
+		u32 pxvidfreq = I915_READ(PXVFREQ_BASE + (i * 4));
+		unsigned long freq = intel_pxfreq(pxvidfreq);
+		unsigned long vid = (pxvidfreq & PXVFREQ_PX_MASK) >>
+			PXVFREQ_PX_SHIFT;
+		unsigned long val;
+
+		val = vid * vid;
+		val *= (freq / 1000);
+		val *= 255;
+		val /= (127*127*900);
+		if (val > 0xff)
+			DRM_ERROR("bad pxval: %ld\n", val);
+		pxw[i] = val;
+	}
+	/* Render standby states get 0 weight */
+	pxw[14] = 0;
+	pxw[15] = 0;
+
+	for (i = 0; i < 4; i++) {
+		u32 val = (pxw[i*4] << 24) | (pxw[(i*4)+1] << 16) |
+			(pxw[(i*4)+2] << 8) | (pxw[(i*4)+3]);
+		I915_WRITE(PXW + (i * 4), val);
+	}
+
+	/* Adjust magic regs to magic values (more experimental results) */
+	I915_WRITE(OGW0, 0);
+	I915_WRITE(OGW1, 0);
+	I915_WRITE(EG0, 0x00007f00);
+	I915_WRITE(EG1, 0x0000000e);
+	I915_WRITE(EG2, 0x000e0000);
+	I915_WRITE(EG3, 0x68000300);
+	I915_WRITE(EG4, 0x42000000);
+	I915_WRITE(EG5, 0x00140031);
+	I915_WRITE(EG6, 0);
+	I915_WRITE(EG7, 0);
+
+	for (i = 0; i < 8; i++)
+		I915_WRITE(PXWL + (i * 4), 0);
+
+	/* Enable PMON + select events */
+	I915_WRITE(ECR, 0x80000019);
+
+	lcfuse = I915_READ(LCFUSE02);
+
+	dev_priv->corr = (lcfuse & LCFUSE_HIV_MASK);
+}
+
 void intel_init_clock_gating(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
@@ -5376,8 +5489,10 @@ void intel_modeset_init(struct drm_device *dev)
 
 	intel_init_clock_gating(dev);
 
-	if (IS_IRONLAKE_M(dev))
+	if (IS_IRONLAKE_M(dev)) {
 		ironlake_enable_drps(dev);
+		intel_init_emon(dev);
+	}
 
 	INIT_WORK(&dev_priv->idle_work, intel_idle_update);
 	setup_timer(&dev_priv->idle_timer, intel_gpu_idle_timer,
-- 
cgit v1.2.3


From 778c35444f7bbb8f1816d40ada650e19c5da9c02 Mon Sep 17 00:00:00 2001
From: Daniel Vetter <daniel.vetter@ffwll.ch>
Date: Thu, 13 May 2010 11:49:44 +0200
Subject: drm/i915: combine all small integers into one single bitfield

This saves a whooping 7 dwords. Zero functional changes. Because
some of the refcounts are rather tightly calculated, I've put
BUG_ONs in the code to check for overflows.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_drv.h | 75 +++++++++++++++++++++++++++--------------
 drivers/gpu/drm/i915/i915_gem.c |  5 +++
 2 files changed, 54 insertions(+), 26 deletions(-)

(limited to 'drivers/gpu/drm/i915/i915_drv.h')

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 91b3a1c20ef8..cccf8019f65a 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -658,19 +658,64 @@ struct drm_i915_gem_object {
 	 * (has pending rendering), and is not set if it's on inactive (ready
 	 * to be unbound).
 	 */
-	int active;
+	unsigned int active : 1;
 
 	/**
 	 * This is set if the object has been written to since last bound
 	 * to the GTT
 	 */
-	int dirty;
+	unsigned int dirty : 1;
+
+	/**
+	 * Fence register bits (if any) for this object.  Will be set
+	 * as needed when mapped into the GTT.
+	 * Protected by dev->struct_mutex.
+	 *
+	 * Size: 4 bits for 16 fences + sign (for FENCE_REG_NONE)
+	 */
+	int fence_reg : 5;
+
+	/**
+	 * Used for checking the object doesn't appear more than once
+	 * in an execbuffer object list.
+	 */
+	unsigned int in_execbuffer : 1;
+
+	/**
+	 * Advice: are the backing pages purgeable?
+	 */
+	unsigned int madv : 2;
+
+	/**
+	 * Refcount for the pages array. With the current locking scheme, there
+	 * are at most two concurrent users: Binding a bo to the gtt and
+	 * pwrite/pread using physical addresses. So two bits for a maximum
+	 * of two users are enough.
+	 */
+	unsigned int pages_refcount : 2;
+#define DRM_I915_GEM_OBJECT_MAX_PAGES_REFCOUNT 0x3
+
+	/**
+	 * Current tiling mode for the object.
+	 */
+	unsigned int tiling_mode : 2;
+
+	/** How many users have pinned this object in GTT space. The following
+	 * users can each hold at most one reference: pwrite/pread, pin_ioctl
+	 * (via user_pin_count), execbuffer (objects are not allowed multiple
+	 * times for the same batchbuffer), and the framebuffer code. When
+	 * switching/pageflipping, the framebuffer code has at most two buffers
+	 * pinned per crtc.
+	 *
+	 * In the worst case this is 1 + 1 + 1 + 2*2 = 7. That would fit into 3
+	 * bits with absolutely no headroom. So use 4 bits. */
+	int pin_count : 4;
+#define DRM_I915_GEM_OBJECT_MAX_PIN_COUNT 0xf
 
 	/** AGP memory structure for our GTT binding. */
 	DRM_AGP_MEM *agp_mem;
 
 	struct page **pages;
-	int pages_refcount;
 
 	/**
 	 * Current offset of the object in GTT space.
@@ -687,21 +732,10 @@ struct drm_i915_gem_object {
 	 */
 	uint64_t mmap_offset;
 
-	/**
-	 * Fence register bits (if any) for this object.  Will be set
-	 * as needed when mapped into the GTT.
-	 * Protected by dev->struct_mutex.
-	 */
-	int fence_reg;
-
-	/** How many users have pinned this object in GTT space */
-	int pin_count;
-
 	/** Breadcrumb of last rendering to the buffer. */
 	uint32_t last_rendering_seqno;
 
-	/** Current tiling mode for the object. */
-	uint32_t tiling_mode;
+	/** Current tiling stride for the object, if it's tiled. */
 	uint32_t stride;
 
 	/** Record of address bit 17 of each page at last unbind. */
@@ -723,17 +757,6 @@ struct drm_i915_gem_object {
 	/** for phy allocated objects */
 	struct drm_i915_gem_phys_object *phys_obj;
 
-	/**
-	 * Used for checking the object doesn't appear more than once
-	 * in an execbuffer object list.
-	 */
-	int in_execbuffer;
-
-	/**
-	 * Advice: are the backing pages purgeable?
-	 */
-	int madv;
-
 	/**
 	 * Number of crtcs where this object is currently the fb, but
 	 * will be page flipped away on the next vblank.  When it
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index c51495f15718..bf703551343a 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -2247,6 +2247,9 @@ i915_gem_object_get_pages(struct drm_gem_object *obj,
 	struct inode *inode;
 	struct page *page;
 
+	BUG_ON(obj_priv->pages_refcount
+			== DRM_I915_GEM_OBJECT_MAX_PAGES_REFCOUNT);
+
 	if (obj_priv->pages_refcount++ != 0)
 		return 0;
 
@@ -4156,6 +4159,8 @@ i915_gem_object_pin(struct drm_gem_object *obj, uint32_t alignment)
 	struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
 	int ret;
 
+	BUG_ON(obj_priv->pin_count == DRM_I915_GEM_OBJECT_MAX_PIN_COUNT);
+
 	i915_verify_inactive(dev, __FILE__, __LINE__);
 	if (obj_priv->gtt_space == NULL) {
 		ret = i915_gem_object_bind_to_gtt(obj, alignment);
-- 
cgit v1.2.3


From 9553426372eef71c849499fb1d232f4b0577c0f9 Mon Sep 17 00:00:00 2001
From: Li Peng <peng.li@linux.intel.com>
Date: Tue, 18 May 2010 18:58:44 +0800
Subject: drm/i915: Add CxSR support on Pineview DDR3

Pineview with DDR3 memory has different latencies to enable CxSR.
This patch updates CxSR latency table to add Pineview DDR3 latency
configuration. It also adds one flag "is_ddr3" for checking DDR3
setting in MCHBAR.

Cc: Shaohua Li <shaohua.li@intel.com>
Cc: Zhao Yakui <yakui.zhao@intel.com>
Signed-off-by: Li Peng <peng.li@intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_dma.c      |  4 ++
 drivers/gpu/drm/i915/i915_drv.h      |  2 +-
 drivers/gpu/drm/i915/i915_reg.h      |  4 ++
 drivers/gpu/drm/i915/intel_display.c | 72 ++++++++++++++++++++++--------------
 4 files changed, 53 insertions(+), 29 deletions(-)

(limited to 'drivers/gpu/drm/i915/i915_drv.h')

diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 12e92f2cc3a7..a5f401664845 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -1491,6 +1491,10 @@ static void i915_pineview_get_mem_freq(struct drm_device *dev)
 		dev_priv->mem_freq = 800;
 		break;
 	}
+
+	/* detect pineview DDR3 setting */
+	tmp = I915_READ(CSHRDDR3CTL);
+	dev_priv->is_ddr3 = (tmp & CSHRDDR3CTL_DDR3) ? 1 : 0;
 }
 
 static void i915_ironlake_get_mem_freq(struct drm_device *dev)
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index cccf8019f65a..e6b4cab6565f 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -326,7 +326,7 @@ typedef struct drm_i915_private {
 	int fence_reg_start; /* 4 if userland hasn't ioctl'd us yet */
 	int num_fence_regs; /* 8 on pre-965, 16 otherwise */
 
-	unsigned int fsb_freq, mem_freq;
+	unsigned int fsb_freq, mem_freq, is_ddr3;
 
 	spinlock_t error_lock;
 	struct drm_i915_error_state *first_error;
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index 881fbe9a17f2..af7b10853e33 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -817,6 +817,10 @@
 #define DCC_CHANNEL_XOR_DISABLE				(1 << 10)
 #define DCC_CHANNEL_XOR_BIT_17				(1 << 9)
 
+/** Pineview MCH register contains DDR3 setting */
+#define CSHRDDR3CTL            0x101a8
+#define CSHRDDR3CTL_DDR3       (1 << 2)
+
 /** 965 MCH register controlling DRAM channel configuration */
 #define C0DRB3			0x10206
 #define C1DRB3			0x10606
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index f946332612ec..cfac4dd1d483 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -2640,6 +2640,7 @@ static unsigned long intel_calculate_wm(unsigned long clock_in_khz,
 
 struct cxsr_latency {
 	int is_desktop;
+	int is_ddr3;
 	unsigned long fsb_freq;
 	unsigned long mem_freq;
 	unsigned long display_sr;
@@ -2649,33 +2650,45 @@ struct cxsr_latency {
 };
 
 static struct cxsr_latency cxsr_latency_table[] = {
-	{1, 800, 400, 3382, 33382, 3983, 33983},    /* DDR2-400 SC */
-	{1, 800, 667, 3354, 33354, 3807, 33807},    /* DDR2-667 SC */
-	{1, 800, 800, 3347, 33347, 3763, 33763},    /* DDR2-800 SC */
-
-	{1, 667, 400, 3400, 33400, 4021, 34021},    /* DDR2-400 SC */
-	{1, 667, 667, 3372, 33372, 3845, 33845},    /* DDR2-667 SC */
-	{1, 667, 800, 3386, 33386, 3822, 33822},    /* DDR2-800 SC */
-
-	{1, 400, 400, 3472, 33472, 4173, 34173},    /* DDR2-400 SC */
-	{1, 400, 667, 3443, 33443, 3996, 33996},    /* DDR2-667 SC */
-	{1, 400, 800, 3430, 33430, 3946, 33946},    /* DDR2-800 SC */
-
-	{0, 800, 400, 3438, 33438, 4065, 34065},    /* DDR2-400 SC */
-	{0, 800, 667, 3410, 33410, 3889, 33889},    /* DDR2-667 SC */
-	{0, 800, 800, 3403, 33403, 3845, 33845},    /* DDR2-800 SC */
-
-	{0, 667, 400, 3456, 33456, 4103, 34106},    /* DDR2-400 SC */
-	{0, 667, 667, 3428, 33428, 3927, 33927},    /* DDR2-667 SC */
-	{0, 667, 800, 3443, 33443, 3905, 33905},    /* DDR2-800 SC */
-
-	{0, 400, 400, 3528, 33528, 4255, 34255},    /* DDR2-400 SC */
-	{0, 400, 667, 3500, 33500, 4079, 34079},    /* DDR2-667 SC */
-	{0, 400, 800, 3487, 33487, 4029, 34029},    /* DDR2-800 SC */
+	{1, 0, 800, 400, 3382, 33382, 3983, 33983},    /* DDR2-400 SC */
+	{1, 0, 800, 667, 3354, 33354, 3807, 33807},    /* DDR2-667 SC */
+	{1, 0, 800, 800, 3347, 33347, 3763, 33763},    /* DDR2-800 SC */
+	{1, 1, 800, 667, 6420, 36420, 6873, 36873},    /* DDR3-667 SC */
+	{1, 1, 800, 800, 5902, 35902, 6318, 36318},    /* DDR3-800 SC */
+
+	{1, 0, 667, 400, 3400, 33400, 4021, 34021},    /* DDR2-400 SC */
+	{1, 0, 667, 667, 3372, 33372, 3845, 33845},    /* DDR2-667 SC */
+	{1, 0, 667, 800, 3386, 33386, 3822, 33822},    /* DDR2-800 SC */
+	{1, 1, 667, 667, 6438, 36438, 6911, 36911},    /* DDR3-667 SC */
+	{1, 1, 667, 800, 5941, 35941, 6377, 36377},    /* DDR3-800 SC */
+
+	{1, 0, 400, 400, 3472, 33472, 4173, 34173},    /* DDR2-400 SC */
+	{1, 0, 400, 667, 3443, 33443, 3996, 33996},    /* DDR2-667 SC */
+	{1, 0, 400, 800, 3430, 33430, 3946, 33946},    /* DDR2-800 SC */
+	{1, 1, 400, 667, 6509, 36509, 7062, 37062},    /* DDR3-667 SC */
+	{1, 1, 400, 800, 5985, 35985, 6501, 36501},    /* DDR3-800 SC */
+
+	{0, 0, 800, 400, 3438, 33438, 4065, 34065},    /* DDR2-400 SC */
+	{0, 0, 800, 667, 3410, 33410, 3889, 33889},    /* DDR2-667 SC */
+	{0, 0, 800, 800, 3403, 33403, 3845, 33845},    /* DDR2-800 SC */
+	{0, 1, 800, 667, 6476, 36476, 6955, 36955},    /* DDR3-667 SC */
+	{0, 1, 800, 800, 5958, 35958, 6400, 36400},    /* DDR3-800 SC */
+
+	{0, 0, 667, 400, 3456, 33456, 4103, 34106},    /* DDR2-400 SC */
+	{0, 0, 667, 667, 3428, 33428, 3927, 33927},    /* DDR2-667 SC */
+	{0, 0, 667, 800, 3443, 33443, 3905, 33905},    /* DDR2-800 SC */
+	{0, 1, 667, 667, 6494, 36494, 6993, 36993},    /* DDR3-667 SC */
+	{0, 1, 667, 800, 5998, 35998, 6460, 36460},    /* DDR3-800 SC */
+
+	{0, 0, 400, 400, 3528, 33528, 4255, 34255},    /* DDR2-400 SC */
+	{0, 0, 400, 667, 3500, 33500, 4079, 34079},    /* DDR2-667 SC */
+	{0, 0, 400, 800, 3487, 33487, 4029, 34029},    /* DDR2-800 SC */
+	{0, 1, 400, 667, 6566, 36566, 7145, 37145},    /* DDR3-667 SC */
+	{0, 1, 400, 800, 6042, 36042, 6584, 36584},    /* DDR3-800 SC */
 };
 
-static struct cxsr_latency *intel_get_cxsr_latency(int is_desktop, int fsb,
-						   int mem)
+static struct cxsr_latency *intel_get_cxsr_latency(int is_desktop, int is_ddr3, 
+						   int fsb, int mem)
 {
 	int i;
 	struct cxsr_latency *latency;
@@ -2686,6 +2699,7 @@ static struct cxsr_latency *intel_get_cxsr_latency(int is_desktop, int fsb,
 	for (i = 0; i < ARRAY_SIZE(cxsr_latency_table); i++) {
 		latency = &cxsr_latency_table[i];
 		if (is_desktop == latency->is_desktop &&
+		    is_ddr3 == latency->is_ddr3 &&
 		    fsb == latency->fsb_freq && mem == latency->mem_freq)
 			return latency;
 	}
@@ -2800,8 +2814,8 @@ static void pineview_update_wm(struct drm_device *dev,  int planea_clock,
 	struct cxsr_latency *latency;
 	int sr_clock;
 
-	latency = intel_get_cxsr_latency(IS_PINEVIEW_G(dev), dev_priv->fsb_freq,
-					 dev_priv->mem_freq);
+	latency = intel_get_cxsr_latency(IS_PINEVIEW_G(dev), dev_priv->is_ddr3, 
+					 dev_priv->fsb_freq, dev_priv->mem_freq);
 	if (!latency) {
 		DRM_DEBUG_KMS("Unknown FSB/MEM found, disable CxSR\n");
 		pineview_disable_cxsr(dev);
@@ -5406,11 +5420,13 @@ static void intel_init_display(struct drm_device *dev)
 			dev_priv->display.update_wm = NULL;
 	} else if (IS_PINEVIEW(dev)) {
 		if (!intel_get_cxsr_latency(IS_PINEVIEW_G(dev),
+					    dev_priv->is_ddr3,
 					    dev_priv->fsb_freq,
 					    dev_priv->mem_freq)) {
 			DRM_INFO("failed to find known CxSR latency "
-				 "(found fsb freq %d, mem freq %d), "
+				 "(found ddr%s fsb freq %d, mem freq %d), "
 				 "disabling CxSR\n",
+				 (dev_priv->is_ddr3 == 1) ? "3": "2",
 				 dev_priv->fsb_freq, dev_priv->mem_freq);
 			/* Disable CxSR and never update its watermark again */
 			pineview_disable_cxsr(dev);
-- 
cgit v1.2.3


From e20f9c64c79e2282f9eb531509181965ec8f0a92 Mon Sep 17 00:00:00 2001
From: Eric Anholt <eric@anholt.net>
Date: Wed, 26 May 2010 14:51:06 -0700
Subject: drm/i915: Clean up leftover bits from hws move to ring structure.

Fixes /debug/dri/0/i915_gem_interrupt output for status page.

Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_debugfs.c | 6 +++---
 drivers/gpu/drm/i915/i915_dma.c     | 7 +++----
 drivers/gpu/drm/i915/i915_drv.h     | 3 ---
 3 files changed, 6 insertions(+), 10 deletions(-)

(limited to 'drivers/gpu/drm/i915/i915_drv.h')

diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index eb11e1e049cb..52510ad8b25d 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -144,7 +144,7 @@ static int i915_gem_seqno_info(struct seq_file *m, void *data)
 	struct drm_device *dev = node->minor->dev;
 	drm_i915_private_t *dev_priv = dev->dev_private;
 
-	if (dev_priv->hw_status_page != NULL) {
+	if (dev_priv->render_ring.status_page.page_addr != NULL) {
 		seq_printf(m, "Current sequence: %d\n",
 			   i915_get_gem_seqno(dev,  &dev_priv->render_ring));
 	} else {
@@ -196,7 +196,7 @@ static int i915_interrupt_info(struct seq_file *m, void *data)
 	}
 	seq_printf(m, "Interrupts received: %d\n",
 		   atomic_read(&dev_priv->irq_received));
-	if (dev_priv->hw_status_page != NULL) {
+	if (dev_priv->render_ring.status_page.page_addr != NULL) {
 		seq_printf(m, "Current sequence:    %d\n",
 			   i915_get_gem_seqno(dev,  &dev_priv->render_ring));
 	} else {
@@ -252,7 +252,7 @@ static int i915_hws_info(struct seq_file *m, void *data)
 	int i;
 	volatile u32 *hws;
 
-	hws = (volatile u32 *)dev_priv->hw_status_page;
+	hws = (volatile u32 *)dev_priv->render_ring.status_page.page_addr;
 	if (hws == NULL)
 		return 0;
 
diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index a5f401664845..6577798d3441 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -84,7 +84,6 @@ static void i915_free_hws(struct drm_device *dev)
 
 	if (dev_priv->render_ring.status_page.gfx_addr) {
 		dev_priv->render_ring.status_page.gfx_addr = 0;
-		dev_priv->status_gfx_addr = 0;
 		drm_core_ioremapfree(&dev_priv->hws_map, dev);
 	}
 
@@ -828,7 +827,7 @@ static int i915_set_status_page(struct drm_device *dev, void *data,
 	drm_core_ioremap_wc(&dev_priv->hws_map, dev);
 	if (dev_priv->hws_map.handle == NULL) {
 		i915_dma_cleanup(dev);
-		dev_priv->status_gfx_addr = 0;
+		ring->status_page.gfx_addr = 0;
 		DRM_ERROR("can not ioremap virtual address for"
 				" G33 hw status page\n");
 		return -ENOMEM;
@@ -838,9 +837,9 @@ static int i915_set_status_page(struct drm_device *dev, void *data,
 	I915_WRITE(HWS_PGA, ring->status_page.gfx_addr);
 
 	DRM_DEBUG_DRIVER("load hws HWS_PGA with gfx mem 0x%x\n",
-			dev_priv->status_gfx_addr);
+			 ring->status_page.gfx_addr);
 	DRM_DEBUG_DRIVER("load hws at %p\n",
-			dev_priv->hw_status_page);
+			 ring->status_page.page_addr);
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index e6b4cab6565f..9ed8ecd95801 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -238,14 +238,11 @@ typedef struct drm_i915_private {
 	struct intel_ring_buffer bsd_ring;
 
 	drm_dma_handle_t *status_page_dmah;
-	void *hw_status_page;
 	void *seqno_page;
 	dma_addr_t dma_status_page;
 	uint32_t counter;
-	unsigned int status_gfx_addr;
 	unsigned int seqno_gfx_addr;
 	drm_local_map_t hws_map;
-	struct drm_gem_object *hws_obj;
 	struct drm_gem_object *seqno_obj;
 	struct drm_gem_object *pwrctx;
 
-- 
cgit v1.2.3