[RFC] drm/i915: Emit to ringbuffer directly

* [RFC] drm/i915: Emit to ringbuffer directly
@ 2016-09-08 15:12 Tvrtko Ursulin
  2016-09-08 15:54 ` ✗ Fi.CI.BAT: failure for " Patchwork
                   ` (3 more replies)
  0 siblings, 4 replies; 15+ messages in thread
From: Tvrtko Ursulin @ 2016-09-08 15:12 UTC (permalink / raw)
  To: Intel-gfx

From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>

This removes the usage of intel_ring_emit in favour of
directly writing to the ring buffer.

intel_ring_emit was preventing the compiler for optimising
fetch and increment of the current ring buffer pointer and
therefore generating very verbose code for every write.

It had no useful purpose since all ringbuffer operations
are started and ended with intel_ring_begin and
intel_ring_advance respectively, with no bail out in the
middle possible, so it is fine to increment the tail in
intel_ring_begin and let the code manage the pointer
itself.

Useless instruction removal amounts to approximately
2384 bytes of saved text on my build.

Not sure if this has any measurable performance
implications but executing a ton of useless instructions
on fast paths cannot be good.

Patch is not fully polished, but it compiles and runs
on Gen9 at least.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
---
 drivers/gpu/drm/i915/i915_gem_context.c    |  62 ++--
 drivers/gpu/drm/i915/i915_gem_execbuffer.c |  27 +-
 drivers/gpu/drm/i915/i915_gem_gtt.c        |  57 ++--
 drivers/gpu/drm/i915/intel_display.c       | 113 ++++---
 drivers/gpu/drm/i915/intel_lrc.c           | 223 +++++++-------
 drivers/gpu/drm/i915/intel_mocs.c          |  43 +--
 drivers/gpu/drm/i915/intel_overlay.c       |  69 ++---
 drivers/gpu/drm/i915/intel_ringbuffer.c    | 480 +++++++++++++++--------------
 drivers/gpu/drm/i915/intel_ringbuffer.h    |  19 +-
 9 files changed, 555 insertions(+), 538 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c
index 35950ee46a1d..c9b61953f23b 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.c
+++ b/drivers/gpu/drm/i915/i915_gem_context.c
@@ -577,7 +577,6 @@ static inline int
 mi_set_context(struct drm_i915_gem_request *req, u32 hw_flags)
 {
 	struct drm_i915_private *dev_priv = req->i915;
-	struct intel_ring *ring = req->ring;
 	struct intel_engine_cs *engine = req->engine;
 	u32 flags = hw_flags | MI_MM_SPACE_GTT;
 	const int num_rings =
@@ -585,6 +584,7 @@ mi_set_context(struct drm_i915_gem_request *req, u32 hw_flags)
 		i915.semaphores ?
 		INTEL_INFO(dev_priv)->num_rings - 1 :
 		0;
+	u32 *rbuf;
 	int len, ret;
 
 	/* w/a: If Flush TLB Invalidation Mode is enabled, driver must do a TLB
@@ -609,70 +609,61 @@ mi_set_context(struct drm_i915_gem_request *req, u32 hw_flags)
 	if (INTEL_GEN(dev_priv) >= 7)
 		len += 2 + (num_rings ? 4*num_rings + 6 : 0);
 
-	ret = intel_ring_begin(req, len);
+	ret = intel_ring_begin(req, len, &rbuf);
 	if (ret)
 		return ret;
 
 	/* WaProgramMiArbOnOffAroundMiSetContext:ivb,vlv,hsw,bdw,chv */
 	if (INTEL_GEN(dev_priv) >= 7) {
-		intel_ring_emit(ring, MI_ARB_ON_OFF | MI_ARB_DISABLE);
+		*rbuf++ = MI_ARB_ON_OFF | MI_ARB_DISABLE;
 		if (num_rings) {
 			struct intel_engine_cs *signaller;
 
-			intel_ring_emit(ring,
-					MI_LOAD_REGISTER_IMM(num_rings));
+			*rbuf++ = MI_LOAD_REGISTER_IMM(num_rings);
 			for_each_engine(signaller, dev_priv) {
 				if (signaller == engine)
 					continue;
 
-				intel_ring_emit_reg(ring,
-						    RING_PSMI_CTL(signaller->mmio_base));
-				intel_ring_emit(ring,
-						_MASKED_BIT_ENABLE(GEN6_PSMI_SLEEP_MSG_DISABLE));
+				*rbuf++ = RING_PSMI_CTL(signaller->mmio_base).reg;
+				*rbuf++ = _MASKED_BIT_ENABLE(GEN6_PSMI_SLEEP_MSG_DISABLE);
 			}
 		}
 	}
 
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_emit(ring, MI_SET_CONTEXT);
-	intel_ring_emit(ring,
-			i915_ggtt_offset(req->ctx->engine[RCS].state) | flags);
+	*rbuf++ = MI_NOOP;
+	*rbuf++ = MI_SET_CONTEXT;
+	*rbuf++ = i915_ggtt_offset(req->ctx->engine[RCS].state) | flags;
 	/*
 	 * w/a: MI_SET_CONTEXT must always be followed by MI_NOOP
 	 * WaMiSetContext_Hang:snb,ivb,vlv
 	 */
-	intel_ring_emit(ring, MI_NOOP);
+	*rbuf++ = MI_NOOP;
 
 	if (INTEL_GEN(dev_priv) >= 7) {
 		if (num_rings) {
 			struct intel_engine_cs *signaller;
 			i915_reg_t last_reg = {}; /* keep gcc quiet */
 
-			intel_ring_emit(ring,
-					MI_LOAD_REGISTER_IMM(num_rings));
+			*rbuf++ = MI_LOAD_REGISTER_IMM(num_rings);
 			for_each_engine(signaller, dev_priv) {
 				if (signaller == engine)
 					continue;
 
 				last_reg = RING_PSMI_CTL(signaller->mmio_base);
-				intel_ring_emit_reg(ring, last_reg);
-				intel_ring_emit(ring,
-						_MASKED_BIT_DISABLE(GEN6_PSMI_SLEEP_MSG_DISABLE));
+				*rbuf++ = last_reg.reg;
+				*rbuf++ = _MASKED_BIT_DISABLE(GEN6_PSMI_SLEEP_MSG_DISABLE);
 			}
 
 			/* Insert a delay before the next switch! */
-			intel_ring_emit(ring,
-					MI_STORE_REGISTER_MEM |
-					MI_SRM_LRM_GLOBAL_GTT);
-			intel_ring_emit_reg(ring, last_reg);
-			intel_ring_emit(ring,
-					i915_ggtt_offset(engine->scratch));
-			intel_ring_emit(ring, MI_NOOP);
+			*rbuf++ = MI_STORE_REGISTER_MEM | MI_SRM_LRM_GLOBAL_GTT;
+			*rbuf++ = last_reg.reg;
+			*rbuf++ = i915_ggtt_offset(engine->scratch);
+			*rbuf++ = MI_NOOP;
 		}
-		intel_ring_emit(ring, MI_ARB_ON_OFF | MI_ARB_ENABLE);
+		*rbuf++ = MI_ARB_ON_OFF | MI_ARB_ENABLE;
 	}
 
-	intel_ring_advance(ring);
+	intel_ring_advance(req->ring);
 
 	return ret;
 }
@@ -680,13 +671,13 @@ mi_set_context(struct drm_i915_gem_request *req, u32 hw_flags)
 static int remap_l3(struct drm_i915_gem_request *req, int slice)
 {
 	u32 *remap_info = req->i915->l3_parity.remap_info[slice];
-	struct intel_ring *ring = req->ring;
+	u32 *rbuf;
 	int i, ret;
 
 	if (!remap_info)
 		return 0;
 
-	ret = intel_ring_begin(req, GEN7_L3LOG_SIZE/4 * 2 + 2);
+	ret = intel_ring_begin(req, GEN7_L3LOG_SIZE/4 * 2 + 2, &rbuf);
 	if (ret)
 		return ret;
 
@@ -695,13 +686,14 @@ static int remap_l3(struct drm_i915_gem_request *req, int slice)
 	 * here because no other code should access these registers other than
 	 * at initialization time.
 	 */
-	intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(GEN7_L3LOG_SIZE/4));
+	*rbuf++ = MI_LOAD_REGISTER_IMM(GEN7_L3LOG_SIZE/4);
 	for (i = 0; i < GEN7_L3LOG_SIZE/4; i++) {
-		intel_ring_emit_reg(ring, GEN7_L3LOG(slice, i));
-		intel_ring_emit(ring, remap_info[i]);
+		*rbuf++ = GEN7_L3LOG(slice, i).reg;
+		*rbuf++ = remap_info[i];
 	}
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*rbuf++ = MI_NOOP;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index 9432d4ce9ffb..d9a226ea768c 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -1358,7 +1358,7 @@ i915_gem_execbuffer_move_to_active(struct list_head *vmas,
 static int
 i915_reset_gen7_sol_offsets(struct drm_i915_gem_request *req)
 {
-	struct intel_ring *ring = req->ring;
+	u32 *rbuf;
 	int ret, i;
 
 	if (!IS_GEN7(req->i915) || req->engine->id != RCS) {
@@ -1366,17 +1366,17 @@ i915_reset_gen7_sol_offsets(struct drm_i915_gem_request *req)
 		return -EINVAL;
 	}
 
-	ret = intel_ring_begin(req, 4 * 3);
+	ret = intel_ring_begin(req, 4 * 3, &rbuf);
 	if (ret)
 		return ret;
 
 	for (i = 0; i < 4; i++) {
-		intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(1));
-		intel_ring_emit_reg(ring, GEN7_SO_WRITE_OFFSET(i));
-		intel_ring_emit(ring, 0);
+		*rbuf++ = MI_LOAD_REGISTER_IMM(1);
+		*rbuf++ = GEN7_SO_WRITE_OFFSET(i).reg;
+		*rbuf++ = 0;
 	}
 
-	intel_ring_advance(ring);
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -1483,17 +1483,18 @@ execbuf_submit(struct i915_execbuffer_params *params,
 
 	if (params->engine->id == RCS &&
 	    instp_mode != dev_priv->relative_constants_mode) {
-		struct intel_ring *ring = params->request->ring;
+		u32 *rbuf;
 
-		ret = intel_ring_begin(params->request, 4);
+		ret = intel_ring_begin(params->request, 4, &rbuf);
 		if (ret)
 			return ret;
 
-		intel_ring_emit(ring, MI_NOOP);
-		intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(1));
-		intel_ring_emit_reg(ring, INSTPM);
-		intel_ring_emit(ring, instp_mask << 16 | instp_mode);
-		intel_ring_advance(ring);
+		*rbuf++ = MI_NOOP;
+		*rbuf++ = MI_LOAD_REGISTER_IMM(1);
+		*rbuf++ = INSTPM.reg;
+		*rbuf++ = instp_mask << 16 | instp_mode;
+
+		intel_ring_advance(params->request->ring);
 
 		dev_priv->relative_constants_mode = instp_mode;
 	}
diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c
index e16c38086abe..d5b5cea08263 100644
--- a/drivers/gpu/drm/i915/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.c
@@ -663,23 +663,24 @@ static int gen8_write_pdp(struct drm_i915_gem_request *req,
 			  unsigned entry,
 			  dma_addr_t addr)
 {
-	struct intel_ring *ring = req->ring;
 	struct intel_engine_cs *engine = req->engine;
+	u32 *rbuf;
 	int ret;
 
 	BUG_ON(entry >= 4);
 
-	ret = intel_ring_begin(req, 6);
+	ret = intel_ring_begin(req, 6, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(1));
-	intel_ring_emit_reg(ring, GEN8_RING_PDP_UDW(engine, entry));
-	intel_ring_emit(ring, upper_32_bits(addr));
-	intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(1));
-	intel_ring_emit_reg(ring, GEN8_RING_PDP_LDW(engine, entry));
-	intel_ring_emit(ring, lower_32_bits(addr));
-	intel_ring_advance(ring);
+	*rbuf++ = MI_LOAD_REGISTER_IMM(1);
+	*rbuf++ = GEN8_RING_PDP_UDW(engine, entry).reg;
+	*rbuf++ = upper_32_bits(addr);
+	*rbuf++ = MI_LOAD_REGISTER_IMM(1);
+	*rbuf++ = GEN8_RING_PDP_LDW(engine, entry).reg;
+	*rbuf++ = lower_32_bits(addr);
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -1655,8 +1656,8 @@ static uint32_t get_pd_offset(struct i915_hw_ppgtt *ppgtt)
 static int hsw_mm_switch(struct i915_hw_ppgtt *ppgtt,
 			 struct drm_i915_gem_request *req)
 {
-	struct intel_ring *ring = req->ring;
 	struct intel_engine_cs *engine = req->engine;
+	u32 *rbuf;
 	int ret;
 
 	/* NB: TLBs must be flushed and invalidated before a switch */
@@ -1664,17 +1665,18 @@ static int hsw_mm_switch(struct i915_hw_ppgtt *ppgtt,
 	if (ret)
 		return ret;
 
-	ret = intel_ring_begin(req, 6);
+	ret = intel_ring_begin(req, 6, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(2));
-	intel_ring_emit_reg(ring, RING_PP_DIR_DCLV(engine));
-	intel_ring_emit(ring, PP_DIR_DCLV_2G);
-	intel_ring_emit_reg(ring, RING_PP_DIR_BASE(engine));
-	intel_ring_emit(ring, get_pd_offset(ppgtt));
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*rbuf++ = MI_LOAD_REGISTER_IMM(2);
+	*rbuf++ = RING_PP_DIR_DCLV(engine).reg;
+	*rbuf++ = PP_DIR_DCLV_2G;
+	*rbuf++ = RING_PP_DIR_BASE(engine).reg;
+	*rbuf++ = get_pd_offset(ppgtt);
+	*rbuf++ = MI_NOOP;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -1682,8 +1684,8 @@ static int hsw_mm_switch(struct i915_hw_ppgtt *ppgtt,
 static int gen7_mm_switch(struct i915_hw_ppgtt *ppgtt,
 			  struct drm_i915_gem_request *req)
 {
-	struct intel_ring *ring = req->ring;
 	struct intel_engine_cs *engine = req->engine;
+	u32 *rbuf;
 	int ret;
 
 	/* NB: TLBs must be flushed and invalidated before a switch */
@@ -1691,17 +1693,18 @@ static int gen7_mm_switch(struct i915_hw_ppgtt *ppgtt,
 	if (ret)
 		return ret;
 
-	ret = intel_ring_begin(req, 6);
+	ret = intel_ring_begin(req, 6, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(2));
-	intel_ring_emit_reg(ring, RING_PP_DIR_DCLV(engine));
-	intel_ring_emit(ring, PP_DIR_DCLV_2G);
-	intel_ring_emit_reg(ring, RING_PP_DIR_BASE(engine));
-	intel_ring_emit(ring, get_pd_offset(ppgtt));
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*rbuf++ = MI_LOAD_REGISTER_IMM(2);
+	*rbuf++ = RING_PP_DIR_DCLV(engine).reg;
+	*rbuf++ = PP_DIR_DCLV_2G;
+	*rbuf++ = RING_PP_DIR_BASE(engine).reg;
+	*rbuf++ = get_pd_offset(ppgtt);
+	*rbuf++ = MI_NOOP;
+
+	intel_ring_advance(req->ring);
 
 	/* XXX: RCS is the only one to auto invalidate the TLBs? */
 	if (engine->id != RCS) {
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 0c65212781e4..110252644287 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -11679,12 +11679,11 @@ static int intel_gen2_queue_flip(struct drm_device *dev,
 				 struct drm_i915_gem_request *req,
 				 uint32_t flags)
 {
-	struct intel_ring *ring = req->ring;
 	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
-	u32 flip_mask;
+	u32 flip_mask, *rbuf;
 	int ret;
 
-	ret = intel_ring_begin(req, 6);
+	ret = intel_ring_begin(req, 6, &rbuf);
 	if (ret)
 		return ret;
 
@@ -11695,13 +11694,14 @@ static int intel_gen2_queue_flip(struct drm_device *dev,
 		flip_mask = MI_WAIT_FOR_PLANE_B_FLIP;
 	else
 		flip_mask = MI_WAIT_FOR_PLANE_A_FLIP;
-	intel_ring_emit(ring, MI_WAIT_FOR_EVENT | flip_mask);
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_emit(ring, MI_DISPLAY_FLIP |
-			MI_DISPLAY_FLIP_PLANE(intel_crtc->plane));
-	intel_ring_emit(ring, fb->pitches[0]);
-	intel_ring_emit(ring, intel_crtc->flip_work->gtt_offset);
-	intel_ring_emit(ring, 0); /* aux display base address, unused */
+
+	*rbuf++ = MI_WAIT_FOR_EVENT | flip_mask;
+	*rbuf++ = MI_NOOP;
+	*rbuf++ = MI_DISPLAY_FLIP |
+		  MI_DISPLAY_FLIP_PLANE(intel_crtc->plane);
+	*rbuf++ = fb->pitches[0];
+	*rbuf++ = intel_crtc->flip_work->gtt_offset;
+	*rbuf++ = 0; /* aux display base address, unused */
 
 	return 0;
 }
@@ -11713,12 +11713,11 @@ static int intel_gen3_queue_flip(struct drm_device *dev,
 				 struct drm_i915_gem_request *req,
 				 uint32_t flags)
 {
-	struct intel_ring *ring = req->ring;
 	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
-	u32 flip_mask;
+	u32 flip_mask, *rbuf;
 	int ret;
 
-	ret = intel_ring_begin(req, 6);
+	ret = intel_ring_begin(req, 6, &rbuf);
 	if (ret)
 		return ret;
 
@@ -11726,13 +11725,14 @@ static int intel_gen3_queue_flip(struct drm_device *dev,
 		flip_mask = MI_WAIT_FOR_PLANE_B_FLIP;
 	else
 		flip_mask = MI_WAIT_FOR_PLANE_A_FLIP;
-	intel_ring_emit(ring, MI_WAIT_FOR_EVENT | flip_mask);
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_emit(ring, MI_DISPLAY_FLIP_I915 |
-			MI_DISPLAY_FLIP_PLANE(intel_crtc->plane));
-	intel_ring_emit(ring, fb->pitches[0]);
-	intel_ring_emit(ring, intel_crtc->flip_work->gtt_offset);
-	intel_ring_emit(ring, MI_NOOP);
+
+	*rbuf++ = MI_WAIT_FOR_EVENT | flip_mask;
+	*rbuf++ = MI_NOOP;
+	*rbuf++ = MI_DISPLAY_FLIP_I915 |
+		  MI_DISPLAY_FLIP_PLANE(intel_crtc->plane);
+	*rbuf++ = fb->pitches[0];
+	*rbuf++ = intel_crtc->flip_work->gtt_offset;
+	*rbuf++ = MI_NOOP;
 
 	return 0;
 }
@@ -11744,13 +11744,13 @@ static int intel_gen4_queue_flip(struct drm_device *dev,
 				 struct drm_i915_gem_request *req,
 				 uint32_t flags)
 {
-	struct intel_ring *ring = req->ring;
 	struct drm_i915_private *dev_priv = to_i915(dev);
 	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
+	u32 *rbuf;
 	uint32_t pf, pipesrc;
 	int ret;
 
-	ret = intel_ring_begin(req, 4);
+	ret = intel_ring_begin(req, 4, &rbuf);
 	if (ret)
 		return ret;
 
@@ -11758,11 +11758,11 @@ static int intel_gen4_queue_flip(struct drm_device *dev,
 	 * Display Registers (which do not change across a page-flip)
 	 * so we need only reprogram the base address.
 	 */
-	intel_ring_emit(ring, MI_DISPLAY_FLIP |
-			MI_DISPLAY_FLIP_PLANE(intel_crtc->plane));
-	intel_ring_emit(ring, fb->pitches[0]);
-	intel_ring_emit(ring, intel_crtc->flip_work->gtt_offset |
-			intel_fb_modifier_to_tiling(fb->modifier[0]));
+	*rbuf++ = MI_DISPLAY_FLIP |
+		  MI_DISPLAY_FLIP_PLANE(intel_crtc->plane);
+	*rbuf++ = fb->pitches[0];
+	*rbuf++ = intel_crtc->flip_work->gtt_offset |
+		  intel_fb_modifier_to_tiling(fb->modifier[0]);
 
 	/* XXX Enabling the panel-fitter across page-flip is so far
 	 * untested on non-native modes, so ignore it for now.
@@ -11770,7 +11770,7 @@ static int intel_gen4_queue_flip(struct drm_device *dev,
 	 */
 	pf = 0;
 	pipesrc = I915_READ(PIPESRC(intel_crtc->pipe)) & 0x0fff0fff;
-	intel_ring_emit(ring, pf | pipesrc);
+	*rbuf++ = pf | pipesrc;
 
 	return 0;
 }
@@ -11782,21 +11782,21 @@ static int intel_gen6_queue_flip(struct drm_device *dev,
 				 struct drm_i915_gem_request *req,
 				 uint32_t flags)
 {
-	struct intel_ring *ring = req->ring;
 	struct drm_i915_private *dev_priv = to_i915(dev);
 	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
+	u32 *rbuf;
 	uint32_t pf, pipesrc;
 	int ret;
 
-	ret = intel_ring_begin(req, 4);
+	ret = intel_ring_begin(req, 4, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, MI_DISPLAY_FLIP |
-			MI_DISPLAY_FLIP_PLANE(intel_crtc->plane));
-	intel_ring_emit(ring, fb->pitches[0] |
-			intel_fb_modifier_to_tiling(fb->modifier[0]));
-	intel_ring_emit(ring, intel_crtc->flip_work->gtt_offset);
+	*rbuf++ = MI_DISPLAY_FLIP |
+		  MI_DISPLAY_FLIP_PLANE(intel_crtc->plane);
+	*rbuf++ = fb->pitches[0] |
+		  intel_fb_modifier_to_tiling(fb->modifier[0]);
+	*rbuf++ = intel_crtc->flip_work->gtt_offset;
 
 	/* Contrary to the suggestions in the documentation,
 	 * "Enable Panel Fitter" does not seem to be required when page
@@ -11806,7 +11806,7 @@ static int intel_gen6_queue_flip(struct drm_device *dev,
 	 */
 	pf = 0;
 	pipesrc = I915_READ(PIPESRC(intel_crtc->pipe)) & 0x0fff0fff;
-	intel_ring_emit(ring, pf | pipesrc);
+	*rbuf++ = pf | pipesrc;
 
 	return 0;
 }
@@ -11818,8 +11818,8 @@ static int intel_gen7_queue_flip(struct drm_device *dev,
 				 struct drm_i915_gem_request *req,
 				 uint32_t flags)
 {
-	struct intel_ring *ring = req->ring;
 	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
+	u32 *rbuf;
 	uint32_t plane_bit = 0;
 	int len, ret;
 
@@ -11864,7 +11864,7 @@ static int intel_gen7_queue_flip(struct drm_device *dev,
 	if (ret)
 		return ret;
 
-	ret = intel_ring_begin(req, len);
+	ret = intel_ring_begin(req, len, &rbuf);
 	if (ret)
 		return ret;
 
@@ -11878,31 +11878,30 @@ static int intel_gen7_queue_flip(struct drm_device *dev,
 	 * to zero does lead to lockups within MI_DISPLAY_FLIP.
 	 */
 	if (req->engine->id == RCS) {
-		intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(1));
-		intel_ring_emit_reg(ring, DERRMR);
-		intel_ring_emit(ring, ~(DERRMR_PIPEA_PRI_FLIP_DONE |
-					  DERRMR_PIPEB_PRI_FLIP_DONE |
-					  DERRMR_PIPEC_PRI_FLIP_DONE));
+		*rbuf++ = MI_LOAD_REGISTER_IMM(1);
+		*rbuf++ = DERRMR.reg;
+		*rbuf++ = ~(DERRMR_PIPEA_PRI_FLIP_DONE |
+			   DERRMR_PIPEB_PRI_FLIP_DONE |
+			   DERRMR_PIPEC_PRI_FLIP_DONE);
 		if (IS_GEN8(dev))
-			intel_ring_emit(ring, MI_STORE_REGISTER_MEM_GEN8 |
-					      MI_SRM_LRM_GLOBAL_GTT);
+			*rbuf++ = MI_STORE_REGISTER_MEM_GEN8 |
+				  MI_SRM_LRM_GLOBAL_GTT;
 		else
-			intel_ring_emit(ring, MI_STORE_REGISTER_MEM |
-					      MI_SRM_LRM_GLOBAL_GTT);
-		intel_ring_emit_reg(ring, DERRMR);
-		intel_ring_emit(ring,
-				i915_ggtt_offset(req->engine->scratch) + 256);
+			*rbuf++ = MI_STORE_REGISTER_MEM |
+				  MI_SRM_LRM_GLOBAL_GTT;
+		*rbuf++ = DERRMR.reg;
+		*rbuf++ = i915_ggtt_offset(req->engine->scratch) + 256;
 		if (IS_GEN8(dev)) {
-			intel_ring_emit(ring, 0);
-			intel_ring_emit(ring, MI_NOOP);
+			*rbuf++ = 0;
+			*rbuf++ = MI_NOOP;
 		}
 	}
 
-	intel_ring_emit(ring, MI_DISPLAY_FLIP_I915 | plane_bit);
-	intel_ring_emit(ring, fb->pitches[0] |
-			intel_fb_modifier_to_tiling(fb->modifier[0]));
-	intel_ring_emit(ring, intel_crtc->flip_work->gtt_offset);
-	intel_ring_emit(ring, (MI_NOOP));
+	*rbuf++ = MI_DISPLAY_FLIP_I915 | plane_bit;
+	*rbuf++ = fb->pitches[0] |
+		  intel_fb_modifier_to_tiling(fb->modifier[0]);
+	*rbuf++ = intel_crtc->flip_work->gtt_offset;
+	*rbuf++ = MI_NOOP;
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index 92bfe47ad33c..e8846523be4b 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -677,7 +677,7 @@ int intel_logical_ring_alloc_request_extras(struct drm_i915_gem_request *request
 	if (ret)
 		return ret;
 
-	ret = intel_ring_begin(request, 0);
+	ret = __intel_ring_begin(request, 0);
 	if (ret)
 		goto err_unpin;
 
@@ -705,6 +705,13 @@ err_unpin:
 }
 
 /*
+ * Reserve space for 2 NOOPs at the end of each request to be
+ * used as a workaround for not being allowed to do lite
+ * restore with HEAD==TAIL (WaIdleLiteRestore).
+ */
+#define WA_TAIL_DWORDS 2
+
+/*
  * intel_logical_ring_advance() - advance the tail and prepare for submission
  * @request: Request to advance the logical ringbuffer of.
  *
@@ -714,13 +721,14 @@ err_unpin:
  * point, the tail *inside* the context is updated and the ELSP written to.
  */
 static int
-intel_logical_ring_advance(struct drm_i915_gem_request *request)
+intel_logical_ring_advance(struct drm_i915_gem_request *request, u32 *rbuf)
 {
 	struct intel_ring *ring = request->ring;
 	struct intel_engine_cs *engine = request->engine;
 
 	intel_ring_advance(ring);
-	request->tail = ring->tail;
+
+	request->tail = ring->tail - WA_TAIL_DWORDS * sizeof(u32);
 
 	/*
 	 * Here we add two extra NOOPs as padding to avoid
@@ -728,8 +736,9 @@ intel_logical_ring_advance(struct drm_i915_gem_request *request)
 	 *
 	 * Caller must reserve WA_TAIL_DWORDS for us!
 	 */
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_emit(ring, MI_NOOP);
+	*rbuf++ = MI_NOOP;
+	*rbuf++ = MI_NOOP;
+
 	intel_ring_advance(ring);
 
 	/* We keep the previous context alive until we retire the following
@@ -837,7 +846,7 @@ void intel_lr_context_unpin(struct i915_gem_context *ctx,
 static int intel_logical_ring_workarounds_emit(struct drm_i915_gem_request *req)
 {
 	int ret, i;
-	struct intel_ring *ring = req->ring;
+	u32 *rbuf;
 	struct i915_workarounds *w = &req->i915->workarounds;
 
 	if (w->count == 0)
@@ -847,18 +856,18 @@ static int intel_logical_ring_workarounds_emit(struct drm_i915_gem_request *req)
 	if (ret)
 		return ret;
 
-	ret = intel_ring_begin(req, w->count * 2 + 2);
-	if (ret)
+	ret = intel_ring_begin(req, w->count * 2 + 2, &rbuf);
+	if (unlikely(ret))
 		return ret;
 
-	intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(w->count));
+	*rbuf++ = MI_LOAD_REGISTER_IMM(w->count);
 	for (i = 0; i < w->count; i++) {
-		intel_ring_emit_reg(ring, w->reg[i].addr);
-		intel_ring_emit(ring, w->reg[i].value);
+		*rbuf++ = w->reg[i].addr.reg;
+		*rbuf++ = w->reg[i].value;
 	}
-	intel_ring_emit(ring, MI_NOOP);
+	*rbuf++ = MI_NOOP;
 
-	intel_ring_advance(ring);
+	intel_ring_advance(req->ring);
 
 	ret = req->engine->emit_flush(req, EMIT_BARRIER);
 	if (ret)
@@ -1360,27 +1369,27 @@ static int gen9_init_render_ring(struct intel_engine_cs *engine)
 static int intel_logical_ring_emit_pdps(struct drm_i915_gem_request *req)
 {
 	struct i915_hw_ppgtt *ppgtt = req->ctx->ppgtt;
-	struct intel_ring *ring = req->ring;
 	struct intel_engine_cs *engine = req->engine;
 	const int num_lri_cmds = GEN8_LEGACY_PDPES * 2;
-	int i, ret;
+	u32 *rbuf;
+	int ret, i;
 
-	ret = intel_ring_begin(req, num_lri_cmds * 2 + 2);
-	if (ret)
+	ret = intel_ring_begin(req, num_lri_cmds * 2 + 2, &rbuf);
+	if (unlikely(ret))
 		return ret;
 
-	intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(num_lri_cmds));
+	*rbuf++ = MI_LOAD_REGISTER_IMM(num_lri_cmds);
 	for (i = GEN8_LEGACY_PDPES - 1; i >= 0; i--) {
 		const dma_addr_t pd_daddr = i915_page_dir_dma_addr(ppgtt, i);
 
-		intel_ring_emit_reg(ring, GEN8_RING_PDP_UDW(engine, i));
-		intel_ring_emit(ring, upper_32_bits(pd_daddr));
-		intel_ring_emit_reg(ring, GEN8_RING_PDP_LDW(engine, i));
-		intel_ring_emit(ring, lower_32_bits(pd_daddr));
+		*rbuf++ = GEN8_RING_PDP_UDW(engine, i).reg;
+		*rbuf++ = upper_32_bits(pd_daddr);
+		*rbuf++ = GEN8_RING_PDP_LDW(engine, i).reg;
+		*rbuf++ = lower_32_bits(pd_daddr);
 	}
+	*rbuf++ = MI_NOOP;
 
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -1389,8 +1398,8 @@ static int gen8_emit_bb_start(struct drm_i915_gem_request *req,
 			      u64 offset, u32 len,
 			      unsigned int dispatch_flags)
 {
-	struct intel_ring *ring = req->ring;
 	bool ppgtt = !(dispatch_flags & I915_DISPATCH_SECURE);
+	u32 *rbuf;
 	int ret;
 
 	/* Don't rely in hw updating PDPs, specially in lite-restore.
@@ -1411,19 +1420,20 @@ static int gen8_emit_bb_start(struct drm_i915_gem_request *req,
 		req->ctx->ppgtt->pd_dirty_rings &= ~intel_engine_flag(req->engine);
 	}
 
-	ret = intel_ring_begin(req, 4);
-	if (ret)
+	ret = intel_ring_begin(req, 4, &rbuf);
+	if (unlikely(ret))
 		return ret;
 
 	/* FIXME(BDW): Address space and security selectors. */
-	intel_ring_emit(ring, MI_BATCH_BUFFER_START_GEN8 |
-			(ppgtt<<8) |
-			(dispatch_flags & I915_DISPATCH_RS ?
-			 MI_BATCH_RESOURCE_STREAMER : 0));
-	intel_ring_emit(ring, lower_32_bits(offset));
-	intel_ring_emit(ring, upper_32_bits(offset));
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*rbuf++ = MI_BATCH_BUFFER_START_GEN8 |
+		  (ppgtt << 8) |
+		  (dispatch_flags & I915_DISPATCH_RS ?
+		  MI_BATCH_RESOURCE_STREAMER : 0);
+	*rbuf++ = lower_32_bits(offset);
+	*rbuf++ = upper_32_bits(offset);
+	*rbuf++ = MI_NOOP;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -1444,12 +1454,11 @@ static void gen8_logical_ring_disable_irq(struct intel_engine_cs *engine)
 
 static int gen8_emit_flush(struct drm_i915_gem_request *request, u32 mode)
 {
-	struct intel_ring *ring = request->ring;
-	u32 cmd;
 	int ret;
+	u32 cmd, *rbuf;
 
-	ret = intel_ring_begin(request, 4);
-	if (ret)
+	ret = intel_ring_begin(request, 4, &rbuf);
+	if (unlikely(ret))
 		return ret;
 
 	cmd = MI_FLUSH_DW + 1;
@@ -1467,13 +1476,12 @@ static int gen8_emit_flush(struct drm_i915_gem_request *request, u32 mode)
 			cmd |= MI_INVALIDATE_BSD;
 	}
 
-	intel_ring_emit(ring, cmd);
-	intel_ring_emit(ring,
-			I915_GEM_HWS_SCRATCH_ADDR |
-			MI_FLUSH_DW_USE_GTT);
-	intel_ring_emit(ring, 0); /* upper addr */
-	intel_ring_emit(ring, 0); /* value */
-	intel_ring_advance(ring);
+	*rbuf++ = cmd;
+	*rbuf++ = I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT;
+	*rbuf++ = 0; /* upper addr */
+	*rbuf++ = 0; /* value */
+
+	intel_ring_advance(request->ring);
 
 	return 0;
 }
@@ -1481,14 +1489,11 @@ static int gen8_emit_flush(struct drm_i915_gem_request *request, u32 mode)
 static int gen8_emit_flush_render(struct drm_i915_gem_request *request,
 				  u32 mode)
 {
-	struct intel_ring *ring = request->ring;
-	struct intel_engine_cs *engine = request->engine;
 	u32 scratch_addr =
-		i915_ggtt_offset(engine->scratch) + 2 * CACHELINE_BYTES;
+	       i915_ggtt_offset(request->engine->scratch) + 2 * CACHELINE_BYTES;
 	bool vf_flush_wa = false, dc_flush_wa = false;
-	u32 flags = 0;
-	int ret;
-	int len;
+	u32 *rbuf, flags = 0;
+	int ret, len;
 
 	flags |= PIPE_CONTROL_CS_STALL;
 
@@ -1529,45 +1534,45 @@ static int gen8_emit_flush_render(struct drm_i915_gem_request *request,
 	if (dc_flush_wa)
 		len += 12;
 
-	ret = intel_ring_begin(request, len);
-	if (ret)
+	ret = intel_ring_begin(request, len, &rbuf);
+	if (unlikely(ret))
 		return ret;
 
 	if (vf_flush_wa) {
-		intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(6));
-		intel_ring_emit(ring, 0);
-		intel_ring_emit(ring, 0);
-		intel_ring_emit(ring, 0);
-		intel_ring_emit(ring, 0);
-		intel_ring_emit(ring, 0);
+		*rbuf++ = GFX_OP_PIPE_CONTROL(6);
+		*rbuf++ = 0;
+		*rbuf++ = 0;
+		*rbuf++ = 0;
+		*rbuf++ = 0;
+		*rbuf++ = 0;
 	}
 
 	if (dc_flush_wa) {
-		intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(6));
-		intel_ring_emit(ring, PIPE_CONTROL_DC_FLUSH_ENABLE);
-		intel_ring_emit(ring, 0);
-		intel_ring_emit(ring, 0);
-		intel_ring_emit(ring, 0);
-		intel_ring_emit(ring, 0);
+		*rbuf++ = GFX_OP_PIPE_CONTROL(6);
+		*rbuf++ = PIPE_CONTROL_DC_FLUSH_ENABLE;
+		*rbuf++ = 0;
+		*rbuf++ = 0;
+		*rbuf++ = 0;
+		*rbuf++ = 0;
 	}
 
-	intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(6));
-	intel_ring_emit(ring, flags);
-	intel_ring_emit(ring, scratch_addr);
-	intel_ring_emit(ring, 0);
-	intel_ring_emit(ring, 0);
-	intel_ring_emit(ring, 0);
+	*rbuf++ = GFX_OP_PIPE_CONTROL(6);
+	*rbuf++ = flags;
+	*rbuf++ = scratch_addr;
+	*rbuf++ = 0;
+	*rbuf++ = 0;
+	*rbuf++ = 0;
 
 	if (dc_flush_wa) {
-		intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(6));
-		intel_ring_emit(ring, PIPE_CONTROL_CS_STALL);
-		intel_ring_emit(ring, 0);
-		intel_ring_emit(ring, 0);
-		intel_ring_emit(ring, 0);
-		intel_ring_emit(ring, 0);
+		*rbuf++ = GFX_OP_PIPE_CONTROL(6);
+		*rbuf++ = PIPE_CONTROL_CS_STALL;
+		*rbuf++ = 0;
+		*rbuf++ = 0;
+		*rbuf++ = 0;
+		*rbuf++ = 0;
 	}
 
-	intel_ring_advance(ring);
+	intel_ring_advance(request->ring);
 
 	return 0;
 }
@@ -1587,43 +1592,36 @@ static void bxt_a_seqno_barrier(struct intel_engine_cs *engine)
 	intel_flush_status_page(engine, I915_GEM_HWS_INDEX);
 }
 
-/*
- * Reserve space for 2 NOOPs at the end of each request to be
- * used as a workaround for not being allowed to do lite
- * restore with HEAD==TAIL (WaIdleLiteRestore).
- */
-#define WA_TAIL_DWORDS 2
-
 static int gen8_emit_request(struct drm_i915_gem_request *request)
 {
-	struct intel_ring *ring = request->ring;
 	int ret;
+	u32 *rbuf;
 
-	ret = intel_ring_begin(request, 6 + WA_TAIL_DWORDS);
-	if (ret)
+	ret = intel_ring_begin(request, 6 + WA_TAIL_DWORDS, &rbuf);
+	if (unlikely(ret))
 		return ret;
 
 	/* w/a: bit 5 needs to be zero for MI_FLUSH_DW address. */
 	BUILD_BUG_ON(I915_GEM_HWS_INDEX_ADDR & (1 << 5));
 
-	intel_ring_emit(ring, (MI_FLUSH_DW + 1) | MI_FLUSH_DW_OP_STOREDW);
-	intel_ring_emit(ring,
-			intel_hws_seqno_address(request->engine) |
-			MI_FLUSH_DW_USE_GTT);
-	intel_ring_emit(ring, 0);
-	intel_ring_emit(ring, request->fence.seqno);
-	intel_ring_emit(ring, MI_USER_INTERRUPT);
-	intel_ring_emit(ring, MI_NOOP);
-	return intel_logical_ring_advance(request);
+	*rbuf++ = (MI_FLUSH_DW + 1) | MI_FLUSH_DW_OP_STOREDW;
+	*rbuf++ =  intel_hws_seqno_address(request->engine) |
+		   MI_FLUSH_DW_USE_GTT;
+	*rbuf++ = 0;
+	*rbuf++ = request->fence.seqno;
+	*rbuf++ = MI_USER_INTERRUPT;
+	*rbuf++ = MI_NOOP;
+
+	return intel_logical_ring_advance(request, rbuf);
 }
 
 static int gen8_emit_request_render(struct drm_i915_gem_request *request)
 {
-	struct intel_ring *ring = request->ring;
 	int ret;
+	u32 *rbuf;
 
-	ret = intel_ring_begin(request, 8 + WA_TAIL_DWORDS);
-	if (ret)
+	ret = intel_ring_begin(request, 8 + WA_TAIL_DWORDS, &rbuf);
+	if (unlikely(ret))
 		return ret;
 
 	/* We're using qword write, seqno should be aligned to 8 bytes. */
@@ -1633,19 +1631,18 @@ static int gen8_emit_request_render(struct drm_i915_gem_request *request)
 	 * need a prior CS_STALL, which is emitted by the flush
 	 * following the batch.
 	 */
-	intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(6));
-	intel_ring_emit(ring,
-			(PIPE_CONTROL_GLOBAL_GTT_IVB |
-			 PIPE_CONTROL_CS_STALL |
-			 PIPE_CONTROL_QW_WRITE));
-	intel_ring_emit(ring, intel_hws_seqno_address(request->engine));
-	intel_ring_emit(ring, 0);
-	intel_ring_emit(ring, i915_gem_request_get_seqno(request));
+	*rbuf++ = GFX_OP_PIPE_CONTROL(6);
+	*rbuf++ = PIPE_CONTROL_GLOBAL_GTT_IVB | PIPE_CONTROL_CS_STALL |
+		  PIPE_CONTROL_QW_WRITE;
+	*rbuf++ = intel_hws_seqno_address(request->engine);
+	*rbuf++ = 0;
+	*rbuf++ = i915_gem_request_get_seqno(request);
 	/* We're thrashing one dword of HWS. */
-	intel_ring_emit(ring, 0);
-	intel_ring_emit(ring, MI_USER_INTERRUPT);
-	intel_ring_emit(ring, MI_NOOP);
-	return intel_logical_ring_advance(request);
+	*rbuf++ = 0;
+	*rbuf++ = MI_USER_INTERRUPT;
+	*rbuf++ = MI_NOOP;
+
+	return intel_logical_ring_advance(request, rbuf);
 }
 
 static int gen8_init_rcs_context(struct drm_i915_gem_request *req)
diff --git a/drivers/gpu/drm/i915/intel_mocs.c b/drivers/gpu/drm/i915/intel_mocs.c
index 80bb9247ce66..7080d3225f6b 100644
--- a/drivers/gpu/drm/i915/intel_mocs.c
+++ b/drivers/gpu/drm/i915/intel_mocs.c
@@ -276,23 +276,23 @@ int intel_mocs_init_engine(struct intel_engine_cs *engine)
 static int emit_mocs_control_table(struct drm_i915_gem_request *req,
 				   const struct drm_i915_mocs_table *table)
 {
-	struct intel_ring *ring = req->ring;
 	enum intel_engine_id engine = req->engine->id;
 	unsigned int index;
+	u32 *rbuf;
 	int ret;
 
 	if (WARN_ON(table->size > GEN9_NUM_MOCS_ENTRIES))
 		return -ENODEV;
 
-	ret = intel_ring_begin(req, 2 + 2 * GEN9_NUM_MOCS_ENTRIES);
+	ret = intel_ring_begin(req, 2 + 2 * GEN9_NUM_MOCS_ENTRIES, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(GEN9_NUM_MOCS_ENTRIES));
+	*rbuf++ = MI_LOAD_REGISTER_IMM(GEN9_NUM_MOCS_ENTRIES);
 
 	for (index = 0; index < table->size; index++) {
-		intel_ring_emit_reg(ring, mocs_register(engine, index));
-		intel_ring_emit(ring, table->table[index].control_value);
+		*rbuf++ = mocs_register(engine, index).reg;
+		*rbuf++ = table->table[index].control_value;
 	}
 
 	/*
@@ -304,12 +304,13 @@ static int emit_mocs_control_table(struct drm_i915_gem_request *req,
 	 * that value to all the used entries.
 	 */
 	for (; index < GEN9_NUM_MOCS_ENTRIES; index++) {
-		intel_ring_emit_reg(ring, mocs_register(engine, index));
-		intel_ring_emit(ring, table->table[0].control_value);
+		*rbuf++ = mocs_register(engine, index).reg;
+		*rbuf++ = table->table[0].control_value;
 	}
 
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*rbuf++ = MI_NOOP;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -336,29 +337,28 @@ static inline u32 l3cc_combine(const struct drm_i915_mocs_table *table,
 static int emit_mocs_l3cc_table(struct drm_i915_gem_request *req,
 				const struct drm_i915_mocs_table *table)
 {
-	struct intel_ring *ring = req->ring;
+	u32 *rbuf;
 	unsigned int i;
 	int ret;
 
 	if (WARN_ON(table->size > GEN9_NUM_MOCS_ENTRIES))
 		return -ENODEV;
 
-	ret = intel_ring_begin(req, 2 + GEN9_NUM_MOCS_ENTRIES);
+	ret = intel_ring_begin(req, 2 + GEN9_NUM_MOCS_ENTRIES, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring,
-			MI_LOAD_REGISTER_IMM(GEN9_NUM_MOCS_ENTRIES / 2));
+	*rbuf++ = MI_LOAD_REGISTER_IMM(GEN9_NUM_MOCS_ENTRIES / 2);
 
 	for (i = 0; i < table->size/2; i++) {
-		intel_ring_emit_reg(ring, GEN9_LNCFCMOCS(i));
-		intel_ring_emit(ring, l3cc_combine(table, 2*i, 2*i+1));
+		*rbuf++ = GEN9_LNCFCMOCS(i).reg;
+		*rbuf++ = l3cc_combine(table, 2*i, 2*i+1);
 	}
 
 	if (table->size & 0x01) {
 		/* Odd table size - 1 left over */
-		intel_ring_emit_reg(ring, GEN9_LNCFCMOCS(i));
-		intel_ring_emit(ring, l3cc_combine(table, 2*i, 0));
+		*rbuf++ = GEN9_LNCFCMOCS(i).reg;
+		*rbuf++ = l3cc_combine(table, 2*i, 0);
 		i++;
 	}
 
@@ -368,12 +368,13 @@ static int emit_mocs_l3cc_table(struct drm_i915_gem_request *req,
 	 * they are reserved by the hardware.
 	 */
 	for (; i < GEN9_NUM_MOCS_ENTRIES / 2; i++) {
-		intel_ring_emit_reg(ring, GEN9_LNCFCMOCS(i));
-		intel_ring_emit(ring, l3cc_combine(table, 0, 0));
+		*rbuf++ = GEN9_LNCFCMOCS(i).reg;
+		*rbuf++ = l3cc_combine(table, 0, 0);
 	}
 
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*rbuf++ = MI_NOOP;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/i915/intel_overlay.c b/drivers/gpu/drm/i915/intel_overlay.c
index a24bc8c7889f..cf49e5855df7 100644
--- a/drivers/gpu/drm/i915/intel_overlay.c
+++ b/drivers/gpu/drm/i915/intel_overlay.c
@@ -243,7 +243,7 @@ static int intel_overlay_on(struct intel_overlay *overlay)
 {
 	struct drm_i915_private *dev_priv = overlay->i915;
 	struct drm_i915_gem_request *req;
-	struct intel_ring *ring;
+	u32 *rbuf;
 	int ret;
 
 	WARN_ON(overlay->active);
@@ -253,7 +253,7 @@ static int intel_overlay_on(struct intel_overlay *overlay)
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
-	ret = intel_ring_begin(req, 4);
+	ret = intel_ring_begin(req, 4, &rbuf);
 	if (ret) {
 		i915_add_request_no_flush(req);
 		return ret;
@@ -261,12 +261,12 @@ static int intel_overlay_on(struct intel_overlay *overlay)
 
 	overlay->active = true;
 
-	ring = req->ring;
-	intel_ring_emit(ring, MI_OVERLAY_FLIP | MI_OVERLAY_ON);
-	intel_ring_emit(ring, overlay->flip_addr | OFC_UPDATE);
-	intel_ring_emit(ring, MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP);
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*rbuf++ = MI_OVERLAY_FLIP | MI_OVERLAY_ON;
+	*rbuf++ = overlay->flip_addr | OFC_UPDATE;
+	*rbuf++ = MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP;
+	*rbuf++ = MI_NOOP;
+
+	intel_ring_advance(req->ring);
 
 	return intel_overlay_do_wait_request(overlay, req, NULL);
 }
@@ -277,9 +277,8 @@ static int intel_overlay_continue(struct intel_overlay *overlay,
 {
 	struct drm_i915_private *dev_priv = overlay->i915;
 	struct drm_i915_gem_request *req;
-	struct intel_ring *ring;
 	u32 flip_addr = overlay->flip_addr;
-	u32 tmp;
+	u32 *rbuf, tmp;
 	int ret;
 
 	WARN_ON(!overlay->active);
@@ -296,16 +295,16 @@ static int intel_overlay_continue(struct intel_overlay *overlay,
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
-	ret = intel_ring_begin(req, 2);
+	ret = intel_ring_begin(req, 2, &rbuf);
 	if (ret) {
 		i915_add_request_no_flush(req);
 		return ret;
 	}
 
-	ring = req->ring;
-	intel_ring_emit(ring, MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE);
-	intel_ring_emit(ring, flip_addr);
-	intel_ring_advance(ring);
+	*rbuf++ = MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE;
+	*rbuf++ = flip_addr;
+
+	intel_ring_advance(req->ring);
 
 	intel_overlay_submit_request(overlay, req, NULL);
 
@@ -355,8 +354,8 @@ static int intel_overlay_off(struct intel_overlay *overlay)
 {
 	struct drm_i915_private *dev_priv = overlay->i915;
 	struct drm_i915_gem_request *req;
-	struct intel_ring *ring;
 	u32 flip_addr = overlay->flip_addr;
+	u32 *rbuf;
 	int ret;
 
 	WARN_ON(!overlay->active);
@@ -371,31 +370,30 @@ static int intel_overlay_off(struct intel_overlay *overlay)
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
-	ret = intel_ring_begin(req, 6);
+	ret = intel_ring_begin(req, 6, &rbuf);
 	if (ret) {
 		i915_add_request_no_flush(req);
 		return ret;
 	}
 
-	ring = req->ring;
 	/* wait for overlay to go idle */
-	intel_ring_emit(ring, MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE);
-	intel_ring_emit(ring, flip_addr);
-	intel_ring_emit(ring, MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP);
+	*rbuf++ = MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE;
+	*rbuf++ = flip_addr;
+	*rbuf++ = MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP;
 	/* turn overlay off */
 	if (IS_I830(dev_priv)) {
 		/* Workaround: Don't disable the overlay fully, since otherwise
 		 * it dies on the next OVERLAY_ON cmd. */
-		intel_ring_emit(ring, MI_NOOP);
-		intel_ring_emit(ring, MI_NOOP);
-		intel_ring_emit(ring, MI_NOOP);
+		*rbuf++ = MI_NOOP;
+		*rbuf++ = MI_NOOP;
+		*rbuf++ = MI_NOOP;
 	} else {
-		intel_ring_emit(ring, MI_OVERLAY_FLIP | MI_OVERLAY_OFF);
-		intel_ring_emit(ring, flip_addr);
-		intel_ring_emit(ring,
-				MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP);
+		*rbuf++ = MI_OVERLAY_FLIP | MI_OVERLAY_OFF;
+		*rbuf++ = flip_addr;
+		*rbuf++ = MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP;
 	}
-	intel_ring_advance(ring);
+
+	intel_ring_advance(req->ring);
 
 	return intel_overlay_do_wait_request(overlay, req,
 					     intel_overlay_off_tail);
@@ -429,23 +427,22 @@ static int intel_overlay_release_old_vid(struct intel_overlay *overlay)
 	if (I915_READ(ISR) & I915_OVERLAY_PLANE_FLIP_PENDING_INTERRUPT) {
 		/* synchronous slowpath */
 		struct drm_i915_gem_request *req;
-		struct intel_ring *ring;
+		u32 *rbuf;
 
 		req = alloc_request(overlay);
 		if (IS_ERR(req))
 			return PTR_ERR(req);
 
-		ret = intel_ring_begin(req, 2);
+		ret = intel_ring_begin(req, 2, &rbuf);
 		if (ret) {
 			i915_add_request_no_flush(req);
 			return ret;
 		}
 
-		ring = req->ring;
-		intel_ring_emit(ring,
-				MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP);
-		intel_ring_emit(ring, MI_NOOP);
-		intel_ring_advance(ring);
+		*rbuf++ = MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP;
+		*rbuf++ = MI_NOOP;
+
+		intel_ring_advance(req->ring);
 
 		ret = intel_overlay_do_wait_request(overlay, req,
 						    intel_overlay_release_old_vid_tail);
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index fd8fcc6ec970..d7892538d130 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -61,8 +61,7 @@ void intel_ring_update_space(struct intel_ring *ring)
 static int
 gen2_render_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 {
-	struct intel_ring *ring = req->ring;
-	u32 cmd;
+	u32 cmd, *rbuf;
 	int ret;
 
 	cmd = MI_FLUSH;
@@ -70,13 +69,14 @@ gen2_render_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 	if (mode & EMIT_INVALIDATE)
 		cmd |= MI_READ_FLUSH;
 
-	ret = intel_ring_begin(req, 2);
+	ret = intel_ring_begin(req, 2, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, cmd);
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*rbuf++ = cmd;
+	*rbuf++ = MI_NOOP;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -84,8 +84,7 @@ gen2_render_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 static int
 gen4_render_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 {
-	struct intel_ring *ring = req->ring;
-	u32 cmd;
+	u32 cmd, *rbuf;
 	int ret;
 
 	/*
@@ -123,13 +122,14 @@ gen4_render_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 			cmd |= MI_INVALIDATE_ISP;
 	}
 
-	ret = intel_ring_begin(req, 2);
+	ret = intel_ring_begin(req, 2, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, cmd);
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*rbuf++ = cmd;
+	*rbuf++ = MI_NOOP;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -174,35 +174,37 @@ gen4_render_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 static int
 intel_emit_post_sync_nonzero_flush(struct drm_i915_gem_request *req)
 {
-	struct intel_ring *ring = req->ring;
 	u32 scratch_addr =
 		i915_ggtt_offset(req->engine->scratch) + 2 * CACHELINE_BYTES;
+	u32 *rbuf;
 	int ret;
 
-	ret = intel_ring_begin(req, 6);
+	ret = intel_ring_begin(req, 6, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(5));
-	intel_ring_emit(ring, PIPE_CONTROL_CS_STALL |
-			PIPE_CONTROL_STALL_AT_SCOREBOARD);
-	intel_ring_emit(ring, scratch_addr | PIPE_CONTROL_GLOBAL_GTT);
-	intel_ring_emit(ring, 0); /* low dword */
-	intel_ring_emit(ring, 0); /* high dword */
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*rbuf++ = GFX_OP_PIPE_CONTROL(5);
+	*rbuf++ = PIPE_CONTROL_CS_STALL |
+		  PIPE_CONTROL_STALL_AT_SCOREBOARD;
+	*rbuf++ = scratch_addr | PIPE_CONTROL_GLOBAL_GTT;
+	*rbuf++ = 0; /* low dword */
+	*rbuf++ = 0; /* high dword */
+	*rbuf++ = MI_NOOP;
+
+	intel_ring_advance(req->ring);
 
-	ret = intel_ring_begin(req, 6);
+	ret = intel_ring_begin(req, 6, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(5));
-	intel_ring_emit(ring, PIPE_CONTROL_QW_WRITE);
-	intel_ring_emit(ring, scratch_addr | PIPE_CONTROL_GLOBAL_GTT);
-	intel_ring_emit(ring, 0);
-	intel_ring_emit(ring, 0);
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*rbuf++ = GFX_OP_PIPE_CONTROL(5);
+	*rbuf++ = PIPE_CONTROL_QW_WRITE;
+	*rbuf++ = scratch_addr | PIPE_CONTROL_GLOBAL_GTT;
+	*rbuf++ = 0;
+	*rbuf++ = 0;
+	*rbuf++ = MI_NOOP;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -210,10 +212,9 @@ intel_emit_post_sync_nonzero_flush(struct drm_i915_gem_request *req)
 static int
 gen6_render_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 {
-	struct intel_ring *ring = req->ring;
 	u32 scratch_addr =
 		i915_ggtt_offset(req->engine->scratch) + 2 * CACHELINE_BYTES;
-	u32 flags = 0;
+	u32 *rbuf, flags = 0;
 	int ret;
 
 	/* Force SNB workarounds for PIPE_CONTROL flushes */
@@ -247,15 +248,16 @@ gen6_render_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 		flags |= PIPE_CONTROL_QW_WRITE | PIPE_CONTROL_CS_STALL;
 	}
 
-	ret = intel_ring_begin(req, 4);
+	ret = intel_ring_begin(req, 4, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(4));
-	intel_ring_emit(ring, flags);
-	intel_ring_emit(ring, scratch_addr | PIPE_CONTROL_GLOBAL_GTT);
-	intel_ring_emit(ring, 0);
-	intel_ring_advance(ring);
+	*rbuf++ = GFX_OP_PIPE_CONTROL(4);
+	*rbuf++ = flags;
+	*rbuf++ = scratch_addr | PIPE_CONTROL_GLOBAL_GTT;
+	*rbuf++ = 0;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -263,20 +265,20 @@ gen6_render_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 static int
 gen7_render_ring_cs_stall_wa(struct drm_i915_gem_request *req)
 {
-	struct intel_ring *ring = req->ring;
+	u32 *rbuf;
 	int ret;
 
-	ret = intel_ring_begin(req, 4);
+	ret = intel_ring_begin(req, 4, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(4));
-	intel_ring_emit(ring,
-			PIPE_CONTROL_CS_STALL |
-			PIPE_CONTROL_STALL_AT_SCOREBOARD);
-	intel_ring_emit(ring, 0);
-	intel_ring_emit(ring, 0);
-	intel_ring_advance(ring);
+	*rbuf++ = GFX_OP_PIPE_CONTROL(4);
+	*rbuf++ = PIPE_CONTROL_CS_STALL |
+		  PIPE_CONTROL_STALL_AT_SCOREBOARD;
+	*rbuf++ = 0;
+	*rbuf++ = 0;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -284,10 +286,9 @@ gen7_render_ring_cs_stall_wa(struct drm_i915_gem_request *req)
 static int
 gen7_render_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 {
-	struct intel_ring *ring = req->ring;
 	u32 scratch_addr =
 		i915_ggtt_offset(req->engine->scratch) + 2 * CACHELINE_BYTES;
-	u32 flags = 0;
+	u32 *rbuf, flags = 0;
 	int ret;
 
 	/*
@@ -332,15 +333,16 @@ gen7_render_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 		gen7_render_ring_cs_stall_wa(req);
 	}
 
-	ret = intel_ring_begin(req, 4);
+	ret = intel_ring_begin(req, 4, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(4));
-	intel_ring_emit(ring, flags);
-	intel_ring_emit(ring, scratch_addr);
-	intel_ring_emit(ring, 0);
-	intel_ring_advance(ring);
+	*rbuf++ = GFX_OP_PIPE_CONTROL(4);
+	*rbuf++ = flags;
+	*rbuf++ = scratch_addr;
+	*rbuf++ = 0;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -349,20 +351,21 @@ static int
 gen8_emit_pipe_control(struct drm_i915_gem_request *req,
 		       u32 flags, u32 scratch_addr)
 {
-	struct intel_ring *ring = req->ring;
+	u32 *rbuf;
 	int ret;
 
-	ret = intel_ring_begin(req, 6);
+	ret = intel_ring_begin(req, 6, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(6));
-	intel_ring_emit(ring, flags);
-	intel_ring_emit(ring, scratch_addr);
-	intel_ring_emit(ring, 0);
-	intel_ring_emit(ring, 0);
-	intel_ring_emit(ring, 0);
-	intel_ring_advance(ring);
+	*rbuf++ = GFX_OP_PIPE_CONTROL(6);
+	*rbuf++ = flags;
+	*rbuf++ = scratch_addr;
+	*rbuf++ = 0;
+	*rbuf++ = 0;
+	*rbuf++ = 0;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -615,8 +618,8 @@ out:
 
 static int intel_ring_workarounds_emit(struct drm_i915_gem_request *req)
 {
-	struct intel_ring *ring = req->ring;
 	struct i915_workarounds *w = &req->i915->workarounds;
+	u32 *rbuf;
 	int ret, i;
 
 	if (w->count == 0)
@@ -626,18 +629,18 @@ static int intel_ring_workarounds_emit(struct drm_i915_gem_request *req)
 	if (ret)
 		return ret;
 
-	ret = intel_ring_begin(req, (w->count * 2 + 2));
+	ret = intel_ring_begin(req, w->count * 2 + 2, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(w->count));
+	*rbuf++ = MI_LOAD_REGISTER_IMM(w->count);
 	for (i = 0; i < w->count; i++) {
-		intel_ring_emit_reg(ring, w->reg[i].addr);
-		intel_ring_emit(ring, w->reg[i].value);
+		*rbuf++ = w->reg[i].addr.reg;
+		*rbuf++ = w->reg[i].value;
 	}
-	intel_ring_emit(ring, MI_NOOP);
+	*rbuf++ = MI_NOOP;
 
-	intel_ring_advance(ring);
+	intel_ring_advance(req->ring);
 
 	ret = req->engine->emit_flush(req, EMIT_BARRIER);
 	if (ret)
@@ -1263,14 +1266,14 @@ static void render_ring_cleanup(struct intel_engine_cs *engine)
 
 static int gen8_rcs_signal(struct drm_i915_gem_request *req)
 {
-	struct intel_ring *ring = req->ring;
 	struct drm_i915_private *dev_priv = req->i915;
 	struct intel_engine_cs *waiter;
 	enum intel_engine_id id;
+	u32 *rbuf;
 	int ret, num_rings;
 
 	num_rings = INTEL_INFO(dev_priv)->num_rings;
-	ret = intel_ring_begin(req, (num_rings-1) * 8);
+	ret = intel_ring_begin(req, (num_rings-1) * 8, &rbuf);
 	if (ret)
 		return ret;
 
@@ -1279,35 +1282,33 @@ static int gen8_rcs_signal(struct drm_i915_gem_request *req)
 		if (gtt_offset == MI_SEMAPHORE_SYNC_INVALID)
 			continue;
 
-		intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(6));
-		intel_ring_emit(ring,
-				PIPE_CONTROL_GLOBAL_GTT_IVB |
-				PIPE_CONTROL_QW_WRITE |
-				PIPE_CONTROL_CS_STALL);
-		intel_ring_emit(ring, lower_32_bits(gtt_offset));
-		intel_ring_emit(ring, upper_32_bits(gtt_offset));
-		intel_ring_emit(ring, req->fence.seqno);
-		intel_ring_emit(ring, 0);
-		intel_ring_emit(ring,
-				MI_SEMAPHORE_SIGNAL |
-				MI_SEMAPHORE_TARGET(waiter->hw_id));
-		intel_ring_emit(ring, 0);
+		*rbuf++ = GFX_OP_PIPE_CONTROL(6);
+		*rbuf++ = PIPE_CONTROL_GLOBAL_GTT_IVB |
+			  PIPE_CONTROL_QW_WRITE |
+			  PIPE_CONTROL_CS_STALL;
+		*rbuf++ = lower_32_bits(gtt_offset);
+		*rbuf++ = upper_32_bits(gtt_offset);
+		*rbuf++ = req->fence.seqno;
+		*rbuf++ = 0;
+		*rbuf++ = MI_SEMAPHORE_SIGNAL |
+			  MI_SEMAPHORE_TARGET(waiter->hw_id);
+		*rbuf++ = 0;
 	}
-	intel_ring_advance(ring);
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
 
 static int gen8_xcs_signal(struct drm_i915_gem_request *req)
 {
-	struct intel_ring *ring = req->ring;
 	struct drm_i915_private *dev_priv = req->i915;
 	struct intel_engine_cs *waiter;
 	enum intel_engine_id id;
+	u32 *rbuf;
 	int ret, num_rings;
 
 	num_rings = INTEL_INFO(dev_priv)->num_rings;
-	ret = intel_ring_begin(req, (num_rings-1) * 6);
+	ret = intel_ring_begin(req, (num_rings-1) * 6, &rbuf);
 	if (ret)
 		return ret;
 
@@ -1316,32 +1317,28 @@ static int gen8_xcs_signal(struct drm_i915_gem_request *req)
 		if (gtt_offset == MI_SEMAPHORE_SYNC_INVALID)
 			continue;
 
-		intel_ring_emit(ring,
-				(MI_FLUSH_DW + 1) | MI_FLUSH_DW_OP_STOREDW);
-		intel_ring_emit(ring,
-				lower_32_bits(gtt_offset) |
-				MI_FLUSH_DW_USE_GTT);
-		intel_ring_emit(ring, upper_32_bits(gtt_offset));
-		intel_ring_emit(ring, req->fence.seqno);
-		intel_ring_emit(ring,
-				MI_SEMAPHORE_SIGNAL |
-				MI_SEMAPHORE_TARGET(waiter->hw_id));
-		intel_ring_emit(ring, 0);
+		*rbuf++ = (MI_FLUSH_DW + 1) | MI_FLUSH_DW_OP_STOREDW;
+		*rbuf++ = lower_32_bits(gtt_offset) | MI_FLUSH_DW_USE_GTT;
+		*rbuf++ = upper_32_bits(gtt_offset);
+		*rbuf++ = req->fence.seqno;
+		*rbuf++ = MI_SEMAPHORE_SIGNAL |
+			  MI_SEMAPHORE_TARGET(waiter->hw_id);
+		*rbuf++ = 0;
 	}
-	intel_ring_advance(ring);
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
 
 static int gen6_signal(struct drm_i915_gem_request *req)
 {
-	struct intel_ring *ring = req->ring;
 	struct drm_i915_private *dev_priv = req->i915;
 	struct intel_engine_cs *engine;
+	u32 *rbuf;
 	int ret, num_rings;
 
 	num_rings = INTEL_INFO(dev_priv)->num_rings;
-	ret = intel_ring_begin(req, round_up((num_rings-1) * 3, 2));
+	ret = intel_ring_begin(req, round_up((num_rings-1) * 3, 2), &rbuf);
 	if (ret)
 		return ret;
 
@@ -1353,16 +1350,17 @@ static int gen6_signal(struct drm_i915_gem_request *req)
 
 		mbox_reg = req->engine->semaphore.mbox.signal[engine->hw_id];
 		if (i915_mmio_reg_valid(mbox_reg)) {
-			intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(1));
-			intel_ring_emit_reg(ring, mbox_reg);
-			intel_ring_emit(ring, req->fence.seqno);
+			*rbuf++ = MI_LOAD_REGISTER_IMM(1);
+			*rbuf++ = mbox_reg.reg;
+			*rbuf++ = req->fence.seqno;
 		}
 	}
 
 	/* If num_dwords was rounded, make sure the tail pointer is correct */
 	if (num_rings % 2 == 0)
-		intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+		*rbuf++ = MI_NOOP;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -1378,16 +1376,18 @@ static void i9xx_submit_request(struct drm_i915_gem_request *request)
 static int i9xx_emit_request(struct drm_i915_gem_request *req)
 {
 	struct intel_ring *ring = req->ring;
+	u32 *rbuf;
 	int ret;
 
-	ret = intel_ring_begin(req, 4);
+	ret = intel_ring_begin(req, 4, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, MI_STORE_DWORD_INDEX);
-	intel_ring_emit(ring, I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT);
-	intel_ring_emit(ring, req->fence.seqno);
-	intel_ring_emit(ring, MI_USER_INTERRUPT);
+	*rbuf++ = MI_STORE_DWORD_INDEX;
+	*rbuf++ = I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT;
+	*rbuf++ = req->fence.seqno;
+	*rbuf++ = MI_USER_INTERRUPT;
+
 	intel_ring_advance(ring);
 
 	req->tail = ring->tail;
@@ -1418,6 +1418,7 @@ static int gen8_render_emit_request(struct drm_i915_gem_request *req)
 {
 	struct intel_engine_cs *engine = req->engine;
 	struct intel_ring *ring = req->ring;
+	u32 *rbuf;
 	int ret;
 
 	if (engine->semaphore.signal) {
@@ -1426,21 +1427,22 @@ static int gen8_render_emit_request(struct drm_i915_gem_request *req)
 			return ret;
 	}
 
-	ret = intel_ring_begin(req, 8);
+	ret = intel_ring_begin(req, 8, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(6));
-	intel_ring_emit(ring, (PIPE_CONTROL_GLOBAL_GTT_IVB |
-			       PIPE_CONTROL_CS_STALL |
-			       PIPE_CONTROL_QW_WRITE));
-	intel_ring_emit(ring, intel_hws_seqno_address(engine));
-	intel_ring_emit(ring, 0);
-	intel_ring_emit(ring, i915_gem_request_get_seqno(req));
+	*rbuf++ = GFX_OP_PIPE_CONTROL(6);
+	*rbuf++ = (PIPE_CONTROL_GLOBAL_GTT_IVB |
+		  PIPE_CONTROL_CS_STALL |
+		  PIPE_CONTROL_QW_WRITE);
+	*rbuf++ = intel_hws_seqno_address(engine);
+	*rbuf++ = 0;
+	*rbuf++ = i915_gem_request_get_seqno(req);
 	/* We're thrashing one dword of HWS. */
-	intel_ring_emit(ring, 0);
-	intel_ring_emit(ring, MI_USER_INTERRUPT);
-	intel_ring_emit(ring, MI_NOOP);
+	*rbuf++ = 0;
+	*rbuf++ = MI_USER_INTERRUPT;
+	*rbuf++ = MI_NOOP;
+
 	intel_ring_advance(ring);
 
 	req->tail = ring->tail;
@@ -1460,24 +1462,24 @@ static int
 gen8_ring_sync_to(struct drm_i915_gem_request *req,
 		  struct drm_i915_gem_request *signal)
 {
-	struct intel_ring *ring = req->ring;
 	struct drm_i915_private *dev_priv = req->i915;
 	u64 offset = GEN8_WAIT_OFFSET(req->engine, signal->engine->id);
 	struct i915_hw_ppgtt *ppgtt;
+	u32 *rbuf;
 	int ret;
 
-	ret = intel_ring_begin(req, 4);
+	ret = intel_ring_begin(req, 4, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring,
-			MI_SEMAPHORE_WAIT |
-			MI_SEMAPHORE_GLOBAL_GTT |
-			MI_SEMAPHORE_SAD_GTE_SDD);
-	intel_ring_emit(ring, signal->fence.seqno);
-	intel_ring_emit(ring, lower_32_bits(offset));
-	intel_ring_emit(ring, upper_32_bits(offset));
-	intel_ring_advance(ring);
+	*rbuf++ = MI_SEMAPHORE_WAIT |
+		  MI_SEMAPHORE_GLOBAL_GTT |
+		  MI_SEMAPHORE_SAD_GTE_SDD;
+	*rbuf++ = signal->fence.seqno;
+	*rbuf++ = lower_32_bits(offset);
+	*rbuf++ = upper_32_bits(offset);
+
+	intel_ring_advance(req->ring);
 
 	/* When the !RCS engines idle waiting upon a semaphore, they lose their
 	 * pagetables and we must reload them before executing the batch.
@@ -1494,28 +1496,29 @@ static int
 gen6_ring_sync_to(struct drm_i915_gem_request *req,
 		  struct drm_i915_gem_request *signal)
 {
-	struct intel_ring *ring = req->ring;
 	u32 dw1 = MI_SEMAPHORE_MBOX |
 		  MI_SEMAPHORE_COMPARE |
 		  MI_SEMAPHORE_REGISTER;
 	u32 wait_mbox = signal->engine->semaphore.mbox.wait[req->engine->hw_id];
+	u32 *rbuf;
 	int ret;
 
 	WARN_ON(wait_mbox == MI_SEMAPHORE_SYNC_INVALID);
 
-	ret = intel_ring_begin(req, 4);
+	ret = intel_ring_begin(req, 4, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, dw1 | wait_mbox);
+	*rbuf++ = dw1 | wait_mbox;
 	/* Throughout all of the GEM code, seqno passed implies our current
 	 * seqno is >= the last seqno executed. However for hardware the
 	 * comparison is strictly greater than.
 	 */
-	intel_ring_emit(ring, signal->fence.seqno - 1);
-	intel_ring_emit(ring, 0);
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*rbuf++ = signal->fence.seqno - 1;
+	*rbuf++ = 0;
+	*rbuf++ = MI_NOOP;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -1616,16 +1619,17 @@ i8xx_irq_disable(struct intel_engine_cs *engine)
 static int
 bsd_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 {
-	struct intel_ring *ring = req->ring;
+	u32 *rbuf;
 	int ret;
 
-	ret = intel_ring_begin(req, 2);
+	ret = intel_ring_begin(req, 2, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, MI_FLUSH);
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*rbuf++ = MI_FLUSH;
+	*rbuf++ = MI_NOOP;
+
+	intel_ring_advance(req->ring);
 	return 0;
 }
 
@@ -1691,20 +1695,20 @@ i965_emit_bb_start(struct drm_i915_gem_request *req,
 		   u64 offset, u32 length,
 		   unsigned int dispatch_flags)
 {
-	struct intel_ring *ring = req->ring;
+	u32 *rbuf;
 	int ret;
 
-	ret = intel_ring_begin(req, 2);
+	ret = intel_ring_begin(req, 2, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring,
-			MI_BATCH_BUFFER_START |
-			MI_BATCH_GTT |
-			(dispatch_flags & I915_DISPATCH_SECURE ?
-			 0 : MI_BATCH_NON_SECURE_I965));
-	intel_ring_emit(ring, offset);
-	intel_ring_advance(ring);
+	*rbuf++ = MI_BATCH_BUFFER_START |
+		  MI_BATCH_GTT |
+		  (dispatch_flags & I915_DISPATCH_SECURE ?
+		  0 : MI_BATCH_NON_SECURE_I965);
+	*rbuf++ = offset;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -1720,26 +1724,28 @@ i830_emit_bb_start(struct drm_i915_gem_request *req,
 {
 	struct intel_ring *ring = req->ring;
 	u32 cs_offset = i915_ggtt_offset(req->engine->scratch);
+	u32 *rbuf;
 	int ret;
 
-	ret = intel_ring_begin(req, 6);
+	ret = intel_ring_begin(req, 6, &rbuf);
 	if (ret)
 		return ret;
 
 	/* Evict the invalid PTE TLBs */
-	intel_ring_emit(ring, COLOR_BLT_CMD | BLT_WRITE_RGBA);
-	intel_ring_emit(ring, BLT_DEPTH_32 | BLT_ROP_COLOR_COPY | 4096);
-	intel_ring_emit(ring, I830_TLB_ENTRIES << 16 | 4); /* load each page */
-	intel_ring_emit(ring, cs_offset);
-	intel_ring_emit(ring, 0xdeadbeef);
-	intel_ring_emit(ring, MI_NOOP);
+	*rbuf++ = COLOR_BLT_CMD | BLT_WRITE_RGBA;
+	*rbuf++ = BLT_DEPTH_32 | BLT_ROP_COLOR_COPY | 4096;
+	*rbuf++ = I830_TLB_ENTRIES << 16 | 4; /* load each page */
+	*rbuf++ = cs_offset;
+	*rbuf++ = 0xdeadbeef;
+	*rbuf++ = MI_NOOP;
+
 	intel_ring_advance(ring);
 
 	if ((dispatch_flags & I915_DISPATCH_PINNED) == 0) {
 		if (len > I830_BATCH_LIMIT)
 			return -ENOSPC;
 
-		ret = intel_ring_begin(req, 6 + 2);
+		ret = intel_ring_begin(req, 6 + 2, &rbuf);
 		if (ret)
 			return ret;
 
@@ -1747,29 +1753,30 @@ i830_emit_bb_start(struct drm_i915_gem_request *req,
 		 * stable batch scratch bo area (so that the CS never
 		 * stumbles over its tlb invalidation bug) ...
 		 */
-		intel_ring_emit(ring, SRC_COPY_BLT_CMD | BLT_WRITE_RGBA);
-		intel_ring_emit(ring,
-				BLT_DEPTH_32 | BLT_ROP_SRC_COPY | 4096);
-		intel_ring_emit(ring, DIV_ROUND_UP(len, 4096) << 16 | 4096);
-		intel_ring_emit(ring, cs_offset);
-		intel_ring_emit(ring, 4096);
-		intel_ring_emit(ring, offset);
-
-		intel_ring_emit(ring, MI_FLUSH);
-		intel_ring_emit(ring, MI_NOOP);
+		*rbuf++ = SRC_COPY_BLT_CMD | BLT_WRITE_RGBA;
+		*rbuf++ = BLT_DEPTH_32 | BLT_ROP_SRC_COPY | 4096;
+		*rbuf++ = DIV_ROUND_UP(len, 4096) << 16 | 4096;
+		*rbuf++ = cs_offset;
+		*rbuf++ = 4096;
+		*rbuf++ = offset;
+
+		*rbuf++ = MI_FLUSH;
+		*rbuf++ = MI_NOOP;
+
 		intel_ring_advance(ring);
 
 		/* ... and execute it. */
 		offset = cs_offset;
 	}
 
-	ret = intel_ring_begin(req, 2);
+	ret = intel_ring_begin(req, 2, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, MI_BATCH_BUFFER_START | MI_BATCH_GTT);
-	intel_ring_emit(ring, offset | (dispatch_flags & I915_DISPATCH_SECURE ?
-					0 : MI_BATCH_NON_SECURE));
+	*rbuf++ = MI_BATCH_BUFFER_START | MI_BATCH_GTT;
+	*rbuf++ = offset | (dispatch_flags & I915_DISPATCH_SECURE ?
+		  0 : MI_BATCH_NON_SECURE);
+
 	intel_ring_advance(ring);
 
 	return 0;
@@ -1780,17 +1787,18 @@ i915_emit_bb_start(struct drm_i915_gem_request *req,
 		   u64 offset, u32 len,
 		   unsigned int dispatch_flags)
 {
-	struct intel_ring *ring = req->ring;
+	u32 *rbuf;
 	int ret;
 
-	ret = intel_ring_begin(req, 2);
+	ret = intel_ring_begin(req, 2, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring, MI_BATCH_BUFFER_START | MI_BATCH_GTT);
-	intel_ring_emit(ring, offset | (dispatch_flags & I915_DISPATCH_SECURE ?
-					0 : MI_BATCH_NON_SECURE));
-	intel_ring_advance(ring);
+	*rbuf++ = MI_BATCH_BUFFER_START | MI_BATCH_GTT;
+	*rbuf++ = offset | (dispatch_flags & I915_DISPATCH_SECURE ?
+		  0 : MI_BATCH_NON_SECURE);
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -2181,7 +2189,7 @@ int intel_ring_alloc_request_extras(struct drm_i915_gem_request *request)
 
 	request->ring = request->engine->buffer;
 
-	ret = intel_ring_begin(request, 0);
+	ret = __intel_ring_begin(request, 0);
 	if (ret)
 		return ret;
 
@@ -2237,7 +2245,7 @@ static int wait_for_space(struct drm_i915_gem_request *req, int bytes)
 	return 0;
 }
 
-int intel_ring_begin(struct drm_i915_gem_request *req, int num_dwords)
+int __intel_ring_begin(struct drm_i915_gem_request *req, int num_dwords)
 {
 	struct intel_ring *ring = req->ring;
 	int remain_actual = ring->size - ring->tail;
@@ -2295,18 +2303,19 @@ int intel_ring_cacheline_align(struct drm_i915_gem_request *req)
 	struct intel_ring *ring = req->ring;
 	int num_dwords =
 		(ring->tail & (CACHELINE_BYTES - 1)) / sizeof(uint32_t);
+	u32 *rbuf;
 	int ret;
 
 	if (num_dwords == 0)
 		return 0;
 
 	num_dwords = CACHELINE_BYTES / sizeof(uint32_t) - num_dwords;
-	ret = intel_ring_begin(req, num_dwords);
+	ret = intel_ring_begin(req, num_dwords, &rbuf);
 	if (ret)
 		return ret;
 
 	while (num_dwords--)
-		intel_ring_emit(ring, MI_NOOP);
+		*rbuf++ = MI_NOOP;
 
 	intel_ring_advance(ring);
 
@@ -2352,11 +2361,10 @@ static void gen6_bsd_submit_request(struct drm_i915_gem_request *request)
 
 static int gen6_bsd_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 {
-	struct intel_ring *ring = req->ring;
-	uint32_t cmd;
+	u32 cmd, *rbuf;
 	int ret;
 
-	ret = intel_ring_begin(req, 4);
+	ret = intel_ring_begin(req, 4, &rbuf);
 	if (ret)
 		return ret;
 
@@ -2380,16 +2388,18 @@ static int gen6_bsd_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 	if (mode & EMIT_INVALIDATE)
 		cmd |= MI_INVALIDATE_TLB | MI_INVALIDATE_BSD;
 
-	intel_ring_emit(ring, cmd);
-	intel_ring_emit(ring, I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT);
+	*rbuf++ = cmd;
+	*rbuf++ = I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT;
 	if (INTEL_GEN(req->i915) >= 8) {
-		intel_ring_emit(ring, 0); /* upper addr */
-		intel_ring_emit(ring, 0); /* value */
+		*rbuf++ = 0; /* upper addr */
+		*rbuf++ = 0; /* value */
 	} else  {
-		intel_ring_emit(ring, 0);
-		intel_ring_emit(ring, MI_NOOP);
+		*rbuf++ = 0;
+		*rbuf++ = MI_NOOP;
 	}
-	intel_ring_advance(ring);
+
+	intel_ring_advance(req->ring);
+
 	return 0;
 }
 
@@ -2398,23 +2408,24 @@ gen8_emit_bb_start(struct drm_i915_gem_request *req,
 		   u64 offset, u32 len,
 		   unsigned int dispatch_flags)
 {
-	struct intel_ring *ring = req->ring;
 	bool ppgtt = USES_PPGTT(req->i915) &&
 			!(dispatch_flags & I915_DISPATCH_SECURE);
+	u32 *rbuf;
 	int ret;
 
-	ret = intel_ring_begin(req, 4);
+	ret = intel_ring_begin(req, 4, &rbuf);
 	if (ret)
 		return ret;
 
 	/* FIXME(BDW): Address space and security selectors. */
-	intel_ring_emit(ring, MI_BATCH_BUFFER_START_GEN8 | (ppgtt<<8) |
-			(dispatch_flags & I915_DISPATCH_RS ?
-			 MI_BATCH_RESOURCE_STREAMER : 0));
-	intel_ring_emit(ring, lower_32_bits(offset));
-	intel_ring_emit(ring, upper_32_bits(offset));
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*rbuf++ = MI_BATCH_BUFFER_START_GEN8 | (ppgtt<<8) |
+		  (dispatch_flags & I915_DISPATCH_RS ?
+		  MI_BATCH_RESOURCE_STREAMER : 0);
+	*rbuf++ = lower_32_bits(offset);
+	*rbuf++ = upper_32_bits(offset);
+	*rbuf++ = MI_NOOP;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -2424,22 +2435,22 @@ hsw_emit_bb_start(struct drm_i915_gem_request *req,
 		  u64 offset, u32 len,
 		  unsigned int dispatch_flags)
 {
-	struct intel_ring *ring = req->ring;
+	u32 *rbuf;
 	int ret;
 
-	ret = intel_ring_begin(req, 2);
+	ret = intel_ring_begin(req, 2, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring,
-			MI_BATCH_BUFFER_START |
-			(dispatch_flags & I915_DISPATCH_SECURE ?
-			 0 : MI_BATCH_PPGTT_HSW | MI_BATCH_NON_SECURE_HSW) |
-			(dispatch_flags & I915_DISPATCH_RS ?
-			 MI_BATCH_RESOURCE_STREAMER : 0));
+	*rbuf++ = MI_BATCH_BUFFER_START |
+		  (dispatch_flags & I915_DISPATCH_SECURE ?
+		  0 : MI_BATCH_PPGTT_HSW | MI_BATCH_NON_SECURE_HSW) |
+		  (dispatch_flags & I915_DISPATCH_RS ?
+		  MI_BATCH_RESOURCE_STREAMER : 0);
 	/* bit0-7 is the length on GEN6+ */
-	intel_ring_emit(ring, offset);
-	intel_ring_advance(ring);
+	*rbuf++ = offset;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -2449,20 +2460,20 @@ gen6_emit_bb_start(struct drm_i915_gem_request *req,
 		   u64 offset, u32 len,
 		   unsigned int dispatch_flags)
 {
-	struct intel_ring *ring = req->ring;
+	u32 *rbuf;
 	int ret;
 
-	ret = intel_ring_begin(req, 2);
+	ret = intel_ring_begin(req, 2, &rbuf);
 	if (ret)
 		return ret;
 
-	intel_ring_emit(ring,
-			MI_BATCH_BUFFER_START |
-			(dispatch_flags & I915_DISPATCH_SECURE ?
-			 0 : MI_BATCH_NON_SECURE_I965));
+	*rbuf++ = MI_BATCH_BUFFER_START |
+		  (dispatch_flags & I915_DISPATCH_SECURE ?
+		  0 : MI_BATCH_NON_SECURE_I965);
 	/* bit0-7 is the length on GEN6+ */
-	intel_ring_emit(ring, offset);
-	intel_ring_advance(ring);
+	*rbuf++ = offset;
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
@@ -2471,11 +2482,10 @@ gen6_emit_bb_start(struct drm_i915_gem_request *req,
 
 static int gen6_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 {
-	struct intel_ring *ring = req->ring;
-	uint32_t cmd;
+	u32 cmd, *rbuf;
 	int ret;
 
-	ret = intel_ring_begin(req, 4);
+	ret = intel_ring_begin(req, 4, &rbuf);
 	if (ret)
 		return ret;
 
@@ -2498,17 +2508,17 @@ static int gen6_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 	 */
 	if (mode & EMIT_INVALIDATE)
 		cmd |= MI_INVALIDATE_TLB;
-	intel_ring_emit(ring, cmd);
-	intel_ring_emit(ring,
-			I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT);
+	*rbuf++ = cmd;
+	*rbuf++ = I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT;
 	if (INTEL_GEN(req->i915) >= 8) {
-		intel_ring_emit(ring, 0); /* upper addr */
-		intel_ring_emit(ring, 0); /* value */
+		*rbuf++ = 0; /* upper addr */
+		*rbuf++ = 0; /* value */
 	} else  {
-		intel_ring_emit(ring, 0);
-		intel_ring_emit(ring, MI_NOOP);
+		*rbuf++ = 0;
+		*rbuf++ = MI_NOOP;
 	}
-	intel_ring_advance(ring);
+
+	intel_ring_advance(req->ring);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
index 84aea549de5d..1182116f20a0 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -443,7 +443,24 @@ void intel_engine_cleanup(struct intel_engine_cs *engine);
 
 int intel_ring_alloc_request_extras(struct drm_i915_gem_request *request);
 
-int __must_check intel_ring_begin(struct drm_i915_gem_request *req, int n);
+int __must_check __intel_ring_begin(struct drm_i915_gem_request *req, int n);
+
+static inline int __must_check
+intel_ring_begin(struct drm_i915_gem_request *req, int n, u32 **rbuf)
+{
+	struct intel_ring *ring = req->ring;
+	int ret;
+
+	ret = __intel_ring_begin(req, n);
+	if (unlikely(ret))
+		return ret;
+
+	*rbuf = (u32 *)(ring->vaddr + ring->tail);
+	ring->tail += n * sizeof(u32);
+
+	return 0;
+}
+
 int __must_check intel_ring_cacheline_align(struct drm_i915_gem_request *req);
 
 static inline void intel_ring_emit(struct intel_ring *ring, u32 data)
-- 
1.9.1

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

^ permalink raw reply related	[flat|nested] 15+ messages in thread