From 73dec95e6ba37d8138bb111be5c9b8a1f3a622ae Mon Sep 17 00:00:00 2001
From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Date: Tue, 14 Feb 2017 11:32:42 +0000
Subject: [PATCH] drm/i915: Emit to ringbuffer directly

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
two and half kilobytes 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.

v2:
 * Change return from intel_ring_begin to error pointer by
   popular demand.
 * Move tail increment to intel_ring_advance to enable some
   error checking.

v3:
 * Move tail advance back into intel_ring_begin.
 * Rebase and tidy.

v4:
 * Complete rebase after a few months since v3.

v5:
 * Remove unecessary cast and fix !debug compile. (Chris Wilson)

v6:
 * Make intel_ring_offset take request as well.
 * Fix recording of request postfix plus a sprinkle of asserts.
   (Chris Wilson)

v7:
 * Use intel_ring_offset to get the postfix. (Chris Wilson)
 * Convert GVT code as well.

v8:
 * Rename *out++ to *cs++.

v9:
 * Fix GVT out to cs conversion in GVT.

v10:
 * Rebase for new intel_ring_begin in selftests.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Zhi Wang <zhi.a.wang@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Acked-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170214113242.29241-1-tvrtko.ursulin@linux.intel.com
---
 drivers/gpu/drm/i915/gvt/cmd_parser.c         |  36 +-
 drivers/gpu/drm/i915/i915_gem_context.c       |  76 +--
 drivers/gpu/drm/i915/i915_gem_execbuffer.c    |  40 +-
 drivers/gpu/drm/i915/i915_gem_gtt.c           |  69 +-
 drivers/gpu/drm/i915/i915_gem_request.c       |   8 +-
 drivers/gpu/drm/i915/intel_display.c          | 131 ++--
 drivers/gpu/drm/i915/intel_lrc.c              | 207 +++---
 drivers/gpu/drm/i915/intel_mocs.c             |  51 +-
 drivers/gpu/drm/i915/intel_overlay.c          |  79 +--
 drivers/gpu/drm/i915/intel_ringbuffer.c       | 592 ++++++++----------
 drivers/gpu/drm/i915/intel_ringbuffer.h       |  30 +-
 .../drm/i915/selftests/i915_gem_coherency.c   |  33 +-
 12 files changed, 625 insertions(+), 727 deletions(-)

diff --git a/drivers/gpu/drm/i915/gvt/cmd_parser.c b/drivers/gpu/drm/i915/gvt/cmd_parser.c
index 9a4b23c3ee97..c3d44c03157c 100644
--- a/drivers/gpu/drm/i915/gvt/cmd_parser.c
+++ b/drivers/gpu/drm/i915/gvt/cmd_parser.c
@@ -1513,7 +1513,7 @@ static int copy_gma_to_hva(struct intel_vgpu *vgpu, struct intel_vgpu_mm *mm,
 		len += copy_len;
 		gma += copy_len;
 	}
-	return 0;
+	return len;
 }
 
 
@@ -1630,7 +1630,7 @@ static int perform_bb_shadow(struct parser_exec_state *s)
 	ret = copy_gma_to_hva(s->vgpu, s->vgpu->gtt.ggtt_mm,
 			      gma, gma + bb_size,
 			      dst);
-	if (ret) {
+	if (ret < 0) {
 		gvt_err("fail to copy guest ring buffer\n");
 		goto unmap_src;
 	}
@@ -2594,11 +2594,8 @@ out:
 static int shadow_workload_ring_buffer(struct intel_vgpu_workload *workload)
 {
 	struct intel_vgpu *vgpu = workload->vgpu;
-	int ring_id = workload->ring_id;
-	struct i915_gem_context *shadow_ctx = vgpu->shadow_ctx;
-	struct intel_ring *ring = shadow_ctx->engine[ring_id].ring;
 	unsigned long gma_head, gma_tail, gma_top, guest_rb_size;
-	unsigned int copy_len = 0;
+	u32 *cs;
 	int ret;
 
 	guest_rb_size = _RING_CTL_BUF_SIZE(workload->rb_ctl);
@@ -2612,36 +2609,33 @@ static int shadow_workload_ring_buffer(struct intel_vgpu_workload *workload)
 	gma_top = workload->rb_start + guest_rb_size;
 
 	/* allocate shadow ring buffer */
-	ret = intel_ring_begin(workload->req, workload->rb_len / 4);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(workload->req, workload->rb_len / sizeof(u32));
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	/* get shadow ring buffer va */
-	workload->shadow_ring_buffer_va = ring->vaddr + ring->tail;
+	workload->shadow_ring_buffer_va = cs;
 
 	/* head > tail --> copy head <-> top */
 	if (gma_head > gma_tail) {
 		ret = copy_gma_to_hva(vgpu, vgpu->gtt.ggtt_mm,
-				gma_head, gma_top,
-				workload->shadow_ring_buffer_va);
-		if (ret) {
+				      gma_head, gma_top, cs);
+		if (ret < 0) {
 			gvt_err("fail to copy guest ring buffer\n");
 			return ret;
 		}
-		copy_len = gma_top - gma_head;
+		cs += ret / sizeof(u32);
 		gma_head = workload->rb_start;
 	}
 
 	/* copy head or start <-> tail */
-	ret = copy_gma_to_hva(vgpu, vgpu->gtt.ggtt_mm,
-			gma_head, gma_tail,
-			workload->shadow_ring_buffer_va + copy_len);
-	if (ret) {
+	ret = copy_gma_to_hva(vgpu, vgpu->gtt.ggtt_mm, gma_head, gma_tail, cs);
+	if (ret < 0) {
 		gvt_err("fail to copy guest ring buffer\n");
 		return ret;
 	}
-	ring->tail += workload->rb_len;
-	intel_ring_advance(ring);
+	cs += ret / sizeof(u32);
+	intel_ring_advance(workload->req, cs);
 	return 0;
 }
 
@@ -2695,7 +2689,7 @@ static int shadow_indirect_ctx(struct intel_shadow_wa_ctx *wa_ctx)
 				wa_ctx->workload->vgpu->gtt.ggtt_mm,
 				guest_gma, guest_gma + ctx_size,
 				map);
-	if (ret) {
+	if (ret < 0) {
 		gvt_err("fail to copy guest indirect ctx\n");
 		goto unmap_src;
 	}
diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c
index c73bf02870d5..9037356536dc 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.c
+++ b/drivers/gpu/drm/i915/i915_gem_context.c
@@ -596,10 +596,9 @@ 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;
 	enum intel_engine_id id;
-	u32 flags = hw_flags | MI_MM_SPACE_GTT;
+	u32 *cs, flags = hw_flags | MI_MM_SPACE_GTT;
 	const int num_rings =
 		/* Use an extended w/a on ivb+ if signalling from other rings */
 		i915.semaphores ?
@@ -629,99 +628,92 @@ 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);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, len);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	/* WaProgramMiArbOnOffAroundMiSetContext:ivb,vlv,hsw,bdw,chv */
 	if (INTEL_GEN(dev_priv) >= 7) {
-		intel_ring_emit(ring, MI_ARB_ON_OFF | MI_ARB_DISABLE);
+		*cs++ = 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));
+			*cs++ = MI_LOAD_REGISTER_IMM(num_rings);
 			for_each_engine(signaller, dev_priv, id) {
 				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));
+				*cs++ = i915_mmio_reg_offset(
+					   RING_PSMI_CTL(signaller->mmio_base));
+				*cs++ = _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);
+	*cs++ = MI_NOOP;
+	*cs++ = MI_SET_CONTEXT;
+	*cs++ = 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);
+	*cs++ = 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));
+			*cs++ = MI_LOAD_REGISTER_IMM(num_rings);
 			for_each_engine(signaller, dev_priv, id) {
 				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));
+				*cs++ = i915_mmio_reg_offset(last_reg);
+				*cs++ = _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);
+			*cs++ = MI_STORE_REGISTER_MEM | MI_SRM_LRM_GLOBAL_GTT;
+			*cs++ = i915_mmio_reg_offset(last_reg);
+			*cs++ = i915_ggtt_offset(engine->scratch);
+			*cs++ = MI_NOOP;
 		}
-		intel_ring_emit(ring, MI_ARB_ON_OFF | MI_ARB_ENABLE);
+		*cs++ = MI_ARB_ON_OFF | MI_ARB_ENABLE;
 	}
 
-	intel_ring_advance(ring);
+	intel_ring_advance(req, cs);
 
 	return ret;
 }
 
 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;
-	int i, ret;
+	u32 *cs, *remap_info = req->i915->l3_parity.remap_info[slice];
+	int i;
 
 	if (!remap_info)
 		return 0;
 
-	ret = intel_ring_begin(req, GEN7_L3LOG_SIZE/4 * 2 + 2);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, GEN7_L3LOG_SIZE/4 * 2 + 2);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	/*
 	 * Note: We do not worry about the concurrent register cacheline hang
 	 * 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));
+	*cs++ = 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]);
+		*cs++ = i915_mmio_reg_offset(GEN7_L3LOG(slice, i));
+		*cs++ = remap_info[i];
 	}
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*cs++ = MI_NOOP;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index 111b7f8c617a..da0846fe2ad6 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -1336,25 +1336,25 @@ 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;
-	int ret, i;
+	u32 *cs;
+	int i;
 
 	if (!IS_GEN7(req->i915) || req->engine->id != RCS) {
 		DRM_DEBUG("sol reset is gen7/rcs only\n");
 		return -EINVAL;
 	}
 
-	ret = intel_ring_begin(req, 4 * 3);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 4 * 3);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	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);
+		*cs++ = MI_LOAD_REGISTER_IMM(1);
+		*cs++ = i915_mmio_reg_offset(GEN7_SO_WRITE_OFFSET(i));
+		*cs++ = 0;
 	}
 
-	intel_ring_advance(ring);
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -1415,7 +1415,7 @@ execbuf_submit(struct i915_execbuffer_params *params,
 	struct drm_i915_private *dev_priv = params->request->i915;
 	u64 exec_start, exec_len;
 	int instp_mode;
-	u32 instp_mask;
+	u32 instp_mask, *cs;
 	int ret;
 
 	ret = i915_gem_execbuffer_move_to_gpu(params->request, vmas);
@@ -1461,17 +1461,15 @@ 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;
-
-		ret = intel_ring_begin(params->request, 4);
-		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);
+		cs = intel_ring_begin(params->request, 4);
+		if (IS_ERR(cs))
+			return PTR_ERR(cs);
+
+		*cs++ = MI_NOOP;
+		*cs++ = MI_LOAD_REGISTER_IMM(1);
+		*cs++ = i915_mmio_reg_offset(INSTPM);
+		*cs++ = instp_mask << 16 | instp_mode;
+		intel_ring_advance(params->request, cs);
 
 		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 eebbffdb9a0b..cb5ea5d38f11 100644
--- a/drivers/gpu/drm/i915/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.c
@@ -686,23 +686,22 @@ 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;
-	int ret;
+	u32 *cs;
 
 	BUG_ON(entry >= 4);
 
-	ret = intel_ring_begin(req, 6);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 6);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	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);
+	*cs++ = MI_LOAD_REGISTER_IMM(1);
+	*cs++ = i915_mmio_reg_offset(GEN8_RING_PDP_UDW(engine, entry));
+	*cs++ = upper_32_bits(addr);
+	*cs++ = MI_LOAD_REGISTER_IMM(1);
+	*cs++ = i915_mmio_reg_offset(GEN8_RING_PDP_LDW(engine, entry));
+	*cs++ = lower_32_bits(addr);
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -1730,8 +1729,8 @@ 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 *cs;
 	int ret;
 
 	/* NB: TLBs must be flushed and invalidated before a switch */
@@ -1739,17 +1738,17 @@ static int hsw_mm_switch(struct i915_hw_ppgtt *ppgtt,
 	if (ret)
 		return ret;
 
-	ret = intel_ring_begin(req, 6);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 6);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	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);
+	*cs++ = MI_LOAD_REGISTER_IMM(2);
+	*cs++ = i915_mmio_reg_offset(RING_PP_DIR_DCLV(engine));
+	*cs++ = PP_DIR_DCLV_2G;
+	*cs++ = i915_mmio_reg_offset(RING_PP_DIR_BASE(engine));
+	*cs++ = get_pd_offset(ppgtt);
+	*cs++ = MI_NOOP;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -1757,8 +1756,8 @@ 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 *cs;
 	int ret;
 
 	/* NB: TLBs must be flushed and invalidated before a switch */
@@ -1766,17 +1765,17 @@ static int gen7_mm_switch(struct i915_hw_ppgtt *ppgtt,
 	if (ret)
 		return ret;
 
-	ret = intel_ring_begin(req, 6);
-	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);
+	cs = intel_ring_begin(req, 6);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
+
+	*cs++ = MI_LOAD_REGISTER_IMM(2);
+	*cs++ = i915_mmio_reg_offset(RING_PP_DIR_DCLV(engine));
+	*cs++ = PP_DIR_DCLV_2G;
+	*cs++ = i915_mmio_reg_offset(RING_PP_DIR_BASE(engine));
+	*cs++ = get_pd_offset(ppgtt);
+	*cs++ = MI_NOOP;
+	intel_ring_advance(req, cs);
 
 	/* XXX: RCS is the only one to auto invalidate the TLBs? */
 	if (engine->id != RCS) {
diff --git a/drivers/gpu/drm/i915/i915_gem_request.c b/drivers/gpu/drm/i915/i915_gem_request.c
index 24571a5289a4..2f6cfa47dc61 100644
--- a/drivers/gpu/drm/i915/i915_gem_request.c
+++ b/drivers/gpu/drm/i915/i915_gem_request.c
@@ -824,6 +824,7 @@ void __i915_add_request(struct drm_i915_gem_request *request, bool flush_caches)
 	struct intel_ring *ring = request->ring;
 	struct intel_timeline *timeline = request->timeline;
 	struct drm_i915_gem_request *prev;
+	u32 *cs;
 	int err;
 
 	lockdep_assert_held(&request->i915->drm.struct_mutex);
@@ -862,10 +863,9 @@ void __i915_add_request(struct drm_i915_gem_request *request, bool flush_caches)
 	 * GPU processing the request, we never over-estimate the
 	 * position of the ring's HEAD.
 	 */
-	err = intel_ring_begin(request, engine->emit_breadcrumb_sz);
-	GEM_BUG_ON(err);
-	request->postfix = ring->tail;
-	ring->tail += engine->emit_breadcrumb_sz * sizeof(u32);
+	cs = intel_ring_begin(request, engine->emit_breadcrumb_sz);
+	GEM_BUG_ON(IS_ERR(cs));
+	request->postfix = intel_ring_offset(request, cs);
 
 	/* Seal the request and mark it as pending execution. Note that
 	 * we may inspect this state, without holding any locks, during
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 34c000859c0d..02e6bb2d614c 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -10158,14 +10158,12 @@ 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;
-	int ret;
+	u32 flip_mask, *cs;
 
-	ret = intel_ring_begin(req, 6);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 6);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	/* Can't queue multiple flips, so wait for the previous
 	 * one to finish before executing the next.
@@ -10174,13 +10172,12 @@ 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 */
+	*cs++ = MI_WAIT_FOR_EVENT | flip_mask;
+	*cs++ = MI_NOOP;
+	*cs++ = MI_DISPLAY_FLIP | MI_DISPLAY_FLIP_PLANE(intel_crtc->plane);
+	*cs++ = fb->pitches[0];
+	*cs++ = intel_crtc->flip_work->gtt_offset;
+	*cs++ = 0; /* aux display base address, unused */
 
 	return 0;
 }
@@ -10192,26 +10189,23 @@ 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;
-	int ret;
+	u32 flip_mask, *cs;
 
-	ret = intel_ring_begin(req, 6);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 6);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	if (intel_crtc->plane)
 		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);
+	*cs++ = MI_WAIT_FOR_EVENT | flip_mask;
+	*cs++ = MI_NOOP;
+	*cs++ = MI_DISPLAY_FLIP_I915 | MI_DISPLAY_FLIP_PLANE(intel_crtc->plane);
+	*cs++ = fb->pitches[0];
+	*cs++ = intel_crtc->flip_work->gtt_offset;
+	*cs++ = MI_NOOP;
 
 	return 0;
 }
@@ -10223,25 +10217,22 @@ 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);
-	uint32_t pf, pipesrc;
-	int ret;
+	u32 pf, pipesrc, *cs;
 
-	ret = intel_ring_begin(req, 4);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 4);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	/* i965+ uses the linear or tiled offsets from the
 	 * 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));
+	*cs++ = MI_DISPLAY_FLIP | MI_DISPLAY_FLIP_PLANE(intel_crtc->plane);
+	*cs++ = fb->pitches[0];
+	*cs++ = intel_crtc->flip_work->gtt_offset |
+		intel_fb_modifier_to_tiling(fb->modifier);
 
 	/* XXX Enabling the panel-fitter across page-flip is so far
 	 * untested on non-native modes, so ignore it for now.
@@ -10249,7 +10240,7 @@ 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);
+	*cs++ = pf | pipesrc;
 
 	return 0;
 }
@@ -10261,21 +10252,17 @@ 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);
-	uint32_t pf, pipesrc;
-	int ret;
+	u32 pf, pipesrc, *cs;
 
-	ret = intel_ring_begin(req, 4);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 4);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	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));
-	intel_ring_emit(ring, intel_crtc->flip_work->gtt_offset);
+	*cs++ = MI_DISPLAY_FLIP | MI_DISPLAY_FLIP_PLANE(intel_crtc->plane);
+	*cs++ = fb->pitches[0] | intel_fb_modifier_to_tiling(fb->modifier);
+	*cs++ = intel_crtc->flip_work->gtt_offset;
 
 	/* Contrary to the suggestions in the documentation,
 	 * "Enable Panel Fitter" does not seem to be required when page
@@ -10285,7 +10272,7 @@ 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);
+	*cs++ = pf | pipesrc;
 
 	return 0;
 }
@@ -10298,9 +10285,8 @@ static int intel_gen7_queue_flip(struct drm_device *dev,
 				 uint32_t flags)
 {
 	struct drm_i915_private *dev_priv = to_i915(dev);
-	struct intel_ring *ring = req->ring;
 	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
-	uint32_t plane_bit = 0;
+	u32 *cs, plane_bit = 0;
 	int len, ret;
 
 	switch (intel_crtc->plane) {
@@ -10344,9 +10330,9 @@ static int intel_gen7_queue_flip(struct drm_device *dev,
 	if (ret)
 		return ret;
 
-	ret = intel_ring_begin(req, len);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, len);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	/* Unmask the flip-done completion message. Note that the bspec says that
 	 * we should do this for both the BCS and RCS, and that we must not unmask
@@ -10358,31 +10344,28 @@ 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));
+		*cs++ = MI_LOAD_REGISTER_IMM(1);
+		*cs++ = i915_mmio_reg_offset(DERRMR);
+		*cs++ = ~(DERRMR_PIPEA_PRI_FLIP_DONE |
+			  DERRMR_PIPEB_PRI_FLIP_DONE |
+			  DERRMR_PIPEC_PRI_FLIP_DONE);
 		if (IS_GEN8(dev_priv))
-			intel_ring_emit(ring, MI_STORE_REGISTER_MEM_GEN8 |
-					      MI_SRM_LRM_GLOBAL_GTT);
+			*cs++ = 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);
+			*cs++ = MI_STORE_REGISTER_MEM | MI_SRM_LRM_GLOBAL_GTT;
+		*cs++ = i915_mmio_reg_offset(DERRMR);
+		*cs++ = i915_ggtt_offset(req->engine->scratch) + 256;
 		if (IS_GEN8(dev_priv)) {
-			intel_ring_emit(ring, 0);
-			intel_ring_emit(ring, MI_NOOP);
+			*cs++ = 0;
+			*cs++ = 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));
-	intel_ring_emit(ring, intel_crtc->flip_work->gtt_offset);
-	intel_ring_emit(ring, (MI_NOOP));
+	*cs++ = MI_DISPLAY_FLIP_I915 | plane_bit;
+	*cs++ = fb->pitches[0] | intel_fb_modifier_to_tiling(fb->modifier);
+	*cs++ = intel_crtc->flip_work->gtt_offset;
+	*cs++ = MI_NOOP;
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index af717e1356a9..f567d4b08863 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -834,6 +834,7 @@ static int execlists_request_alloc(struct drm_i915_gem_request *request)
 {
 	struct intel_engine_cs *engine = request->engine;
 	struct intel_context *ce = &request->ctx->engine[engine->id];
+	u32 *cs;
 	int ret;
 
 	GEM_BUG_ON(!ce->pin_count);
@@ -858,9 +859,11 @@ static int execlists_request_alloc(struct drm_i915_gem_request *request)
 			goto err;
 	}
 
-	ret = intel_ring_begin(request, 0);
-	if (ret)
+	cs = intel_ring_begin(request, 0);
+	if (IS_ERR(cs)) {
+		ret = PTR_ERR(cs);
 		goto err_unreserve;
+	}
 
 	if (!ce->initialised) {
 		ret = engine->init_context(request);
@@ -889,9 +892,9 @@ err:
 
 static int intel_logical_ring_workarounds_emit(struct drm_i915_gem_request *req)
 {
-	int ret, i;
-	struct intel_ring *ring = req->ring;
 	struct i915_workarounds *w = &req->i915->workarounds;
+	u32 *cs;
+	int ret, i;
 
 	if (w->count == 0)
 		return 0;
@@ -900,18 +903,18 @@ 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)
-		return ret;
+	cs = intel_ring_begin(req, w->count * 2 + 2);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(w->count));
+	*cs++ = 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);
+		*cs++ = i915_mmio_reg_offset(w->reg[i].addr);
+		*cs++ = w->reg[i].value;
 	}
-	intel_ring_emit(ring, MI_NOOP);
+	*cs++ = MI_NOOP;
 
-	intel_ring_advance(ring);
+	intel_ring_advance(req, cs);
 
 	ret = req->engine->emit_flush(req, EMIT_BARRIER);
 	if (ret)
@@ -1404,27 +1407,27 @@ static void reset_common_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 *cs;
+	int i;
 
-	ret = intel_ring_begin(req, num_lri_cmds * 2 + 2);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, num_lri_cmds * 2 + 2);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(num_lri_cmds));
+	*cs++ = 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));
+		*cs++ = i915_mmio_reg_offset(GEN8_RING_PDP_UDW(engine, i));
+		*cs++ = upper_32_bits(pd_daddr);
+		*cs++ = i915_mmio_reg_offset(GEN8_RING_PDP_LDW(engine, i));
+		*cs++ = lower_32_bits(pd_daddr);
 	}
 
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*cs++ = MI_NOOP;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -1433,8 +1436,8 @@ 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 *cs;
 	int ret;
 
 	/* Don't rely in hw updating PDPs, specially in lite-restore.
@@ -1455,19 +1458,17 @@ 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)
-		return ret;
+	cs = intel_ring_begin(req, 4);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	/* 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);
+	*cs++ = MI_BATCH_BUFFER_START_GEN8 | (ppgtt << 8) | (dispatch_flags &
+		I915_DISPATCH_RS ? MI_BATCH_RESOURCE_STREAMER : 0);
+	*cs++ = lower_32_bits(offset);
+	*cs++ = upper_32_bits(offset);
+	*cs++ = MI_NOOP;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -1488,13 +1489,11 @@ 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, *cs;
 
-	ret = intel_ring_begin(request, 4);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(request, 4);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	cmd = MI_FLUSH_DW + 1;
 
@@ -1511,13 +1510,11 @@ 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);
+	*cs++ = cmd;
+	*cs++ = I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT;
+	*cs++ = 0; /* upper addr */
+	*cs++ = 0; /* value */
+	intel_ring_advance(request, cs);
 
 	return 0;
 }
@@ -1525,13 +1522,11 @@ 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;
 	bool vf_flush_wa = false, dc_flush_wa = false;
-	u32 flags = 0;
-	int ret;
+	u32 *cs, flags = 0;
 	int len;
 
 	flags |= PIPE_CONTROL_CS_STALL;
@@ -1573,45 +1568,45 @@ 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)
-		return ret;
+	cs = intel_ring_begin(request, len);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	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);
+		*cs++ = GFX_OP_PIPE_CONTROL(6);
+		*cs++ = 0;
+		*cs++ = 0;
+		*cs++ = 0;
+		*cs++ = 0;
+		*cs++ = 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);
+		*cs++ = GFX_OP_PIPE_CONTROL(6);
+		*cs++ = PIPE_CONTROL_DC_FLUSH_ENABLE;
+		*cs++ = 0;
+		*cs++ = 0;
+		*cs++ = 0;
+		*cs++ = 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);
+	*cs++ = GFX_OP_PIPE_CONTROL(6);
+	*cs++ = flags;
+	*cs++ = scratch_addr;
+	*cs++ = 0;
+	*cs++ = 0;
+	*cs++ = 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);
+		*cs++ = GFX_OP_PIPE_CONTROL(6);
+		*cs++ = PIPE_CONTROL_CS_STALL;
+		*cs++ = 0;
+		*cs++ = 0;
+		*cs++ = 0;
+		*cs++ = 0;
 	}
 
-	intel_ring_advance(ring);
+	intel_ring_advance(request, cs);
 
 	return 0;
 }
@@ -1621,34 +1616,33 @@ static int gen8_emit_flush_render(struct drm_i915_gem_request *request,
  * used as a workaround for not being allowed to do lite
  * restore with HEAD==TAIL (WaIdleLiteRestore).
  */
-static void gen8_emit_wa_tail(struct drm_i915_gem_request *request, u32 *out)
+static void gen8_emit_wa_tail(struct drm_i915_gem_request *request, u32 *cs)
 {
-	*out++ = MI_NOOP;
-	*out++ = MI_NOOP;
-	request->wa_tail = intel_ring_offset(request->ring, out);
+	*cs++ = MI_NOOP;
+	*cs++ = MI_NOOP;
+	request->wa_tail = intel_ring_offset(request, cs);
 }
 
-static void gen8_emit_breadcrumb(struct drm_i915_gem_request *request,
-				 u32 *out)
+static void gen8_emit_breadcrumb(struct drm_i915_gem_request *request, u32 *cs)
 {
 	/* w/a: bit 5 needs to be zero for MI_FLUSH_DW address. */
 	BUILD_BUG_ON(I915_GEM_HWS_INDEX_ADDR & (1 << 5));
 
-	*out++ = (MI_FLUSH_DW + 1) | MI_FLUSH_DW_OP_STOREDW;
-	*out++ = intel_hws_seqno_address(request->engine) | MI_FLUSH_DW_USE_GTT;
-	*out++ = 0;
-	*out++ = request->global_seqno;
-	*out++ = MI_USER_INTERRUPT;
-	*out++ = MI_NOOP;
-	request->tail = intel_ring_offset(request->ring, out);
+	*cs++ = (MI_FLUSH_DW + 1) | MI_FLUSH_DW_OP_STOREDW;
+	*cs++ = intel_hws_seqno_address(request->engine) | MI_FLUSH_DW_USE_GTT;
+	*cs++ = 0;
+	*cs++ = request->global_seqno;
+	*cs++ = MI_USER_INTERRUPT;
+	*cs++ = MI_NOOP;
+	request->tail = intel_ring_offset(request, cs);
 
-	gen8_emit_wa_tail(request, out);
+	gen8_emit_wa_tail(request, cs);
 }
 
 static const int gen8_emit_breadcrumb_sz = 6 + WA_TAIL_DWORDS;
 
 static void gen8_emit_breadcrumb_render(struct drm_i915_gem_request *request,
-					u32 *out)
+					u32 *cs)
 {
 	/* We're using qword write, seqno should be aligned to 8 bytes. */
 	BUILD_BUG_ON(I915_GEM_HWS_INDEX & 1);
@@ -1657,20 +1651,19 @@ static void gen8_emit_breadcrumb_render(struct drm_i915_gem_request *request,
 	 * need a prior CS_STALL, which is emitted by the flush
 	 * following the batch.
 	 */
-	*out++ = GFX_OP_PIPE_CONTROL(6);
-	*out++ = (PIPE_CONTROL_GLOBAL_GTT_IVB |
-		  PIPE_CONTROL_CS_STALL |
-		  PIPE_CONTROL_QW_WRITE);
-	*out++ = intel_hws_seqno_address(request->engine);
-	*out++ = 0;
-	*out++ = request->global_seqno;
+	*cs++ = GFX_OP_PIPE_CONTROL(6);
+	*cs++ = PIPE_CONTROL_GLOBAL_GTT_IVB | PIPE_CONTROL_CS_STALL |
+		PIPE_CONTROL_QW_WRITE;
+	*cs++ = intel_hws_seqno_address(request->engine);
+	*cs++ = 0;
+	*cs++ = request->global_seqno;
 	/* We're thrashing one dword of HWS. */
-	*out++ = 0;
-	*out++ = MI_USER_INTERRUPT;
-	*out++ = MI_NOOP;
-	request->tail = intel_ring_offset(request->ring, out);
+	*cs++ = 0;
+	*cs++ = MI_USER_INTERRUPT;
+	*cs++ = MI_NOOP;
+	request->tail = intel_ring_offset(request, cs);
 
-	gen8_emit_wa_tail(request, out);
+	gen8_emit_wa_tail(request, cs);
 }
 
 static const int gen8_emit_breadcrumb_render_sz = 8 + WA_TAIL_DWORDS;
diff --git a/drivers/gpu/drm/i915/intel_mocs.c b/drivers/gpu/drm/i915/intel_mocs.c
index 773e36253e7c..92e461c68385 100644
--- a/drivers/gpu/drm/i915/intel_mocs.c
+++ b/drivers/gpu/drm/i915/intel_mocs.c
@@ -276,23 +276,22 @@ 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;
-	int ret;
+	u32 *cs;
 
 	if (WARN_ON(table->size > GEN9_NUM_MOCS_ENTRIES))
 		return -ENODEV;
 
-	ret = intel_ring_begin(req, 2 + 2 * GEN9_NUM_MOCS_ENTRIES);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 2 + 2 * GEN9_NUM_MOCS_ENTRIES);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(GEN9_NUM_MOCS_ENTRIES));
+	*cs++ = 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);
+		*cs++ = i915_mmio_reg_offset(mocs_register(engine, index));
+		*cs++ = table->table[index].control_value;
 	}
 
 	/*
@@ -304,12 +303,12 @@ 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);
+		*cs++ = i915_mmio_reg_offset(mocs_register(engine, index));
+		*cs++ = table->table[0].control_value;
 	}
 
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*cs++ = MI_NOOP;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -336,29 +335,27 @@ 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;
 	unsigned int i;
-	int ret;
+	u32 *cs;
 
 	if (WARN_ON(table->size > GEN9_NUM_MOCS_ENTRIES))
 		return -ENODEV;
 
-	ret = intel_ring_begin(req, 2 + GEN9_NUM_MOCS_ENTRIES);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 2 + GEN9_NUM_MOCS_ENTRIES);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	intel_ring_emit(ring,
-			MI_LOAD_REGISTER_IMM(GEN9_NUM_MOCS_ENTRIES / 2));
+	*cs++ = 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));
+		*cs++ = i915_mmio_reg_offset(GEN9_LNCFCMOCS(i));
+		*cs++ = 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));
+		*cs++ = i915_mmio_reg_offset(GEN9_LNCFCMOCS(i));
+		*cs++ = l3cc_combine(table, 2 * i, 0);
 		i++;
 	}
 
@@ -368,12 +365,12 @@ 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));
+		*cs++ = i915_mmio_reg_offset(GEN9_LNCFCMOCS(i));
+		*cs++ = l3cc_combine(table, 0, 0);
 	}
 
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*cs++ = MI_NOOP;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/i915/intel_overlay.c b/drivers/gpu/drm/i915/intel_overlay.c
index 0608fad7f593..5ef9f5bfb92c 100644
--- a/drivers/gpu/drm/i915/intel_overlay.c
+++ b/drivers/gpu/drm/i915/intel_overlay.c
@@ -267,8 +267,7 @@ 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;
-	int ret;
+	u32 *cs;
 
 	WARN_ON(overlay->active);
 	WARN_ON(IS_I830(dev_priv) && !(dev_priv->quirks & QUIRK_PIPEA_FORCE));
@@ -277,10 +276,10 @@ static int intel_overlay_on(struct intel_overlay *overlay)
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
-	ret = intel_ring_begin(req, 4);
-	if (ret) {
+	cs = intel_ring_begin(req, 4);
+	if (IS_ERR(cs)) {
 		i915_add_request_no_flush(req);
-		return ret;
+		return PTR_ERR(cs);
 	}
 
 	overlay->active = true;
@@ -288,12 +287,11 @@ static int intel_overlay_on(struct intel_overlay *overlay)
 	if (IS_I830(dev_priv))
 		i830_overlay_clock_gating(dev_priv, false);
 
-	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);
+	*cs++ = MI_OVERLAY_FLIP | MI_OVERLAY_ON;
+	*cs++ = overlay->flip_addr | OFC_UPDATE;
+	*cs++ = MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP;
+	*cs++ = MI_NOOP;
+	intel_ring_advance(req, cs);
 
 	return intel_overlay_do_wait_request(overlay, req, NULL);
 }
@@ -326,10 +324,8 @@ 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;
-	int ret;
+	u32 tmp, *cs;
 
 	WARN_ON(!overlay->active);
 
@@ -345,16 +341,15 @@ static int intel_overlay_continue(struct intel_overlay *overlay,
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
-	ret = intel_ring_begin(req, 2);
-	if (ret) {
+	cs = intel_ring_begin(req, 2);
+	if (IS_ERR(cs)) {
 		i915_add_request_no_flush(req);
-		return ret;
+		return PTR_ERR(cs);
 	}
 
-	ring = req->ring;
-	intel_ring_emit(ring, MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE);
-	intel_ring_emit(ring, flip_addr);
-	intel_ring_advance(ring);
+	*cs++ = MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE;
+	*cs++ = flip_addr;
+	intel_ring_advance(req, cs);
 
 	intel_overlay_flip_prepare(overlay, vma);
 
@@ -408,9 +403,7 @@ static void intel_overlay_off_tail(struct i915_gem_active *active,
 static int intel_overlay_off(struct intel_overlay *overlay)
 {
 	struct drm_i915_gem_request *req;
-	struct intel_ring *ring;
-	u32 flip_addr = overlay->flip_addr;
-	int ret;
+	u32 *cs, flip_addr = overlay->flip_addr;
 
 	WARN_ON(!overlay->active);
 
@@ -424,25 +417,23 @@ static int intel_overlay_off(struct intel_overlay *overlay)
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
-	ret = intel_ring_begin(req, 6);
-	if (ret) {
+	cs = intel_ring_begin(req, 6);
+	if (IS_ERR(cs)) {
 		i915_add_request_no_flush(req);
-		return ret;
+		return PTR_ERR(cs);
 	}
 
-	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);
+	*cs++ = MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE;
+	*cs++ = flip_addr;
+	*cs++ = MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP;
 
 	/* turn overlay off */
-	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);
+	*cs++ = MI_OVERLAY_FLIP | MI_OVERLAY_OFF;
+	*cs++ = flip_addr;
+	*cs++ = MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP;
 
-	intel_ring_advance(ring);
+	intel_ring_advance(req, cs);
 
 	intel_overlay_flip_prepare(overlay, NULL);
 
@@ -465,6 +456,7 @@ static int intel_overlay_recover_from_interrupt(struct intel_overlay *overlay)
 static int intel_overlay_release_old_vid(struct intel_overlay *overlay)
 {
 	struct drm_i915_private *dev_priv = overlay->i915;
+	u32 *cs;
 	int ret;
 
 	lockdep_assert_held(&dev_priv->drm.struct_mutex);
@@ -478,23 +470,20 @@ 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;
 
 		req = alloc_request(overlay);
 		if (IS_ERR(req))
 			return PTR_ERR(req);
 
-		ret = intel_ring_begin(req, 2);
-		if (ret) {
+		cs = intel_ring_begin(req, 2);
+		if (IS_ERR(cs)) {
 			i915_add_request_no_flush(req);
-			return ret;
+			return PTR_ERR(cs);
 		}
 
-		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);
+		*cs++ = MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP;
+		*cs++ = MI_NOOP;
+		intel_ring_advance(req, cs);
 
 		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 9fba5664376e..0b9030c36625 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -61,22 +61,20 @@ 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;
-	int ret;
+	u32 cmd, *cs;
 
 	cmd = MI_FLUSH;
 
 	if (mode & EMIT_INVALIDATE)
 		cmd |= MI_READ_FLUSH;
 
-	ret = intel_ring_begin(req, 2);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 2);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	intel_ring_emit(ring, cmd);
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*cs++ = cmd;
+	*cs++ = MI_NOOP;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -84,9 +82,7 @@ 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;
-	int ret;
+	u32 cmd, *cs;
 
 	/*
 	 * read/write caches:
@@ -123,13 +119,13 @@ gen4_render_ring_flush(struct drm_i915_gem_request *req, u32 mode)
 			cmd |= MI_INVALIDATE_ISP;
 	}
 
-	ret = intel_ring_begin(req, 2);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 2);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	intel_ring_emit(ring, cmd);
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*cs++ = cmd;
+	*cs++ = MI_NOOP;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -174,35 +170,33 @@ 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;
-	int ret;
-
-	ret = intel_ring_begin(req, 6);
-	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);
-
-	ret = intel_ring_begin(req, 6);
-	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);
+	u32 *cs;
+
+	cs = intel_ring_begin(req, 6);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
+
+	*cs++ = GFX_OP_PIPE_CONTROL(5);
+	*cs++ = PIPE_CONTROL_CS_STALL | PIPE_CONTROL_STALL_AT_SCOREBOARD;
+	*cs++ = scratch_addr | PIPE_CONTROL_GLOBAL_GTT;
+	*cs++ = 0; /* low dword */
+	*cs++ = 0; /* high dword */
+	*cs++ = MI_NOOP;
+	intel_ring_advance(req, cs);
+
+	cs = intel_ring_begin(req, 6);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
+
+	*cs++ = GFX_OP_PIPE_CONTROL(5);
+	*cs++ = PIPE_CONTROL_QW_WRITE;
+	*cs++ = scratch_addr | PIPE_CONTROL_GLOBAL_GTT;
+	*cs++ = 0;
+	*cs++ = 0;
+	*cs++ = MI_NOOP;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -210,10 +204,9 @@ 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 *cs, flags = 0;
 	int ret;
 
 	/* Force SNB workarounds for PIPE_CONTROL flushes */
@@ -247,15 +240,15 @@ 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);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 4);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	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);
+	*cs++ = GFX_OP_PIPE_CONTROL(4);
+	*cs++ = flags;
+	*cs++ = scratch_addr | PIPE_CONTROL_GLOBAL_GTT;
+	*cs++ = 0;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -263,20 +256,17 @@ 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;
-	int ret;
+	u32 *cs;
 
-	ret = intel_ring_begin(req, 4);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 4);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	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);
+	*cs++ = GFX_OP_PIPE_CONTROL(4);
+	*cs++ = PIPE_CONTROL_CS_STALL | PIPE_CONTROL_STALL_AT_SCOREBOARD;
+	*cs++ = 0;
+	*cs++ = 0;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -284,11 +274,9 @@ 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;
-	int ret;
+	u32 *cs, flags = 0;
 
 	/*
 	 * Ensure that any following seqno writes only happen when the render
@@ -332,15 +320,15 @@ 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);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 4);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	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);
+	*cs++ = GFX_OP_PIPE_CONTROL(4);
+	*cs++ = flags;
+	*cs++ = scratch_addr;
+	*cs++ = 0;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -349,20 +337,19 @@ static int
 gen8_emit_pipe_control(struct drm_i915_gem_request *req,
 		       u32 flags, u32 scratch_addr)
 {
-	struct intel_ring *ring = req->ring;
-	int ret;
+	u32 *cs;
 
-	ret = intel_ring_begin(req, 6);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 6);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	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);
+	*cs++ = GFX_OP_PIPE_CONTROL(6);
+	*cs++ = flags;
+	*cs++ = scratch_addr;
+	*cs++ = 0;
+	*cs++ = 0;
+	*cs++ = 0;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -659,8 +646,8 @@ static void reset_ring_common(struct intel_engine_cs *engine,
 
 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 *cs;
 	int ret, i;
 
 	if (w->count == 0)
@@ -670,18 +657,18 @@ 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));
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, (w->count * 2 + 2));
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(w->count));
+	*cs++ = 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);
+		*cs++ = i915_mmio_reg_offset(w->reg[i].addr);
+		*cs++ = w->reg[i].value;
 	}
-	intel_ring_emit(ring, MI_NOOP);
+	*cs++ = MI_NOOP;
 
-	intel_ring_advance(ring);
+	intel_ring_advance(req, cs);
 
 	ret = req->engine->emit_flush(req, EMIT_BARRIER);
 	if (ret)
@@ -1276,7 +1263,7 @@ static void render_ring_cleanup(struct intel_engine_cs *engine)
 	i915_vma_unpin_and_release(&dev_priv->semaphore);
 }
 
-static u32 *gen8_rcs_signal(struct drm_i915_gem_request *req, u32 *out)
+static u32 *gen8_rcs_signal(struct drm_i915_gem_request *req, u32 *cs)
 {
 	struct drm_i915_private *dev_priv = req->i915;
 	struct intel_engine_cs *waiter;
@@ -1287,23 +1274,22 @@ static u32 *gen8_rcs_signal(struct drm_i915_gem_request *req, u32 *out)
 		if (gtt_offset == MI_SEMAPHORE_SYNC_INVALID)
 			continue;
 
-		*out++ = GFX_OP_PIPE_CONTROL(6);
-		*out++ = (PIPE_CONTROL_GLOBAL_GTT_IVB |
-			  PIPE_CONTROL_QW_WRITE |
-			  PIPE_CONTROL_CS_STALL);
-		*out++ = lower_32_bits(gtt_offset);
-		*out++ = upper_32_bits(gtt_offset);
-		*out++ = req->global_seqno;
-		*out++ = 0;
-		*out++ = (MI_SEMAPHORE_SIGNAL |
-			  MI_SEMAPHORE_TARGET(waiter->hw_id));
-		*out++ = 0;
+		*cs++ = GFX_OP_PIPE_CONTROL(6);
+		*cs++ = PIPE_CONTROL_GLOBAL_GTT_IVB | PIPE_CONTROL_QW_WRITE |
+			PIPE_CONTROL_CS_STALL;
+		*cs++ = lower_32_bits(gtt_offset);
+		*cs++ = upper_32_bits(gtt_offset);
+		*cs++ = req->global_seqno;
+		*cs++ = 0;
+		*cs++ = MI_SEMAPHORE_SIGNAL |
+			MI_SEMAPHORE_TARGET(waiter->hw_id);
+		*cs++ = 0;
 	}
 
-	return out;
+	return cs;
 }
 
-static u32 *gen8_xcs_signal(struct drm_i915_gem_request *req, u32 *out)
+static u32 *gen8_xcs_signal(struct drm_i915_gem_request *req, u32 *cs)
 {
 	struct drm_i915_private *dev_priv = req->i915;
 	struct intel_engine_cs *waiter;
@@ -1314,19 +1300,19 @@ static u32 *gen8_xcs_signal(struct drm_i915_gem_request *req, u32 *out)
 		if (gtt_offset == MI_SEMAPHORE_SYNC_INVALID)
 			continue;
 
-		*out++ = (MI_FLUSH_DW + 1) | MI_FLUSH_DW_OP_STOREDW;
-		*out++ = lower_32_bits(gtt_offset) | MI_FLUSH_DW_USE_GTT;
-		*out++ = upper_32_bits(gtt_offset);
-		*out++ = req->global_seqno;
-		*out++ = (MI_SEMAPHORE_SIGNAL |
-			  MI_SEMAPHORE_TARGET(waiter->hw_id));
-		*out++ = 0;
+		*cs++ = (MI_FLUSH_DW + 1) | MI_FLUSH_DW_OP_STOREDW;
+		*cs++ = lower_32_bits(gtt_offset) | MI_FLUSH_DW_USE_GTT;
+		*cs++ = upper_32_bits(gtt_offset);
+		*cs++ = req->global_seqno;
+		*cs++ = MI_SEMAPHORE_SIGNAL |
+			MI_SEMAPHORE_TARGET(waiter->hw_id);
+		*cs++ = 0;
 	}
 
-	return out;
+	return cs;
 }
 
-static u32 *gen6_signal(struct drm_i915_gem_request *req, u32 *out)
+static u32 *gen6_signal(struct drm_i915_gem_request *req, u32 *cs)
 {
 	struct drm_i915_private *dev_priv = req->i915;
 	struct intel_engine_cs *engine;
@@ -1341,16 +1327,16 @@ static u32 *gen6_signal(struct drm_i915_gem_request *req, u32 *out)
 
 		mbox_reg = req->engine->semaphore.mbox.signal[engine->hw_id];
 		if (i915_mmio_reg_valid(mbox_reg)) {
-			*out++ = MI_LOAD_REGISTER_IMM(1);
-			*out++ = i915_mmio_reg_offset(mbox_reg);
-			*out++ = req->global_seqno;
+			*cs++ = MI_LOAD_REGISTER_IMM(1);
+			*cs++ = i915_mmio_reg_offset(mbox_reg);
+			*cs++ = req->global_seqno;
 			num_rings++;
 		}
 	}
 	if (num_rings & 1)
-		*out++ = MI_NOOP;
+		*cs++ = MI_NOOP;
 
-	return out;
+	return cs;
 }
 
 static void i9xx_submit_request(struct drm_i915_gem_request *request)
@@ -1362,15 +1348,14 @@ static void i9xx_submit_request(struct drm_i915_gem_request *request)
 	I915_WRITE_TAIL(request->engine, request->tail);
 }
 
-static void i9xx_emit_breadcrumb(struct drm_i915_gem_request *req,
-				 u32 *out)
+static void i9xx_emit_breadcrumb(struct drm_i915_gem_request *req, u32 *cs)
 {
-	*out++ = MI_STORE_DWORD_INDEX;
-	*out++ = I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT;
-	*out++ = req->global_seqno;
-	*out++ = MI_USER_INTERRUPT;
+	*cs++ = MI_STORE_DWORD_INDEX;
+	*cs++ = I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT;
+	*cs++ = req->global_seqno;
+	*cs++ = MI_USER_INTERRUPT;
 
-	req->tail = intel_ring_offset(req->ring, out);
+	req->tail = intel_ring_offset(req, cs);
 }
 
 static const int i9xx_emit_breadcrumb_sz = 4;
@@ -1383,34 +1368,32 @@ static const int i9xx_emit_breadcrumb_sz = 4;
  * Update the mailbox registers in the *other* rings with the current seqno.
  * This acts like a signal in the canonical semaphore.
  */
-static void gen6_sema_emit_breadcrumb(struct drm_i915_gem_request *req,
-				      u32 *out)
+static void gen6_sema_emit_breadcrumb(struct drm_i915_gem_request *req, u32 *cs)
 {
 	return i9xx_emit_breadcrumb(req,
-				    req->engine->semaphore.signal(req, out));
+				    req->engine->semaphore.signal(req, cs));
 }
 
 static void gen8_render_emit_breadcrumb(struct drm_i915_gem_request *req,
-					u32 *out)
+					u32 *cs)
 {
 	struct intel_engine_cs *engine = req->engine;
 
 	if (engine->semaphore.signal)
-		out = engine->semaphore.signal(req, out);
-
-	*out++ = GFX_OP_PIPE_CONTROL(6);
-	*out++ = (PIPE_CONTROL_GLOBAL_GTT_IVB |
-			       PIPE_CONTROL_CS_STALL |
-			       PIPE_CONTROL_QW_WRITE);
-	*out++ = intel_hws_seqno_address(engine);
-	*out++ = 0;
-	*out++ = req->global_seqno;
+		cs = engine->semaphore.signal(req, cs);
+
+	*cs++ = GFX_OP_PIPE_CONTROL(6);
+	*cs++ = PIPE_CONTROL_GLOBAL_GTT_IVB | PIPE_CONTROL_CS_STALL |
+		PIPE_CONTROL_QW_WRITE;
+	*cs++ = intel_hws_seqno_address(engine);
+	*cs++ = 0;
+	*cs++ = req->global_seqno;
 	/* We're thrashing one dword of HWS. */
-	*out++ = 0;
-	*out++ = MI_USER_INTERRUPT;
-	*out++ = MI_NOOP;
+	*cs++ = 0;
+	*cs++ = MI_USER_INTERRUPT;
+	*cs++ = MI_NOOP;
 
-	req->tail = intel_ring_offset(req->ring, out);
+	req->tail = intel_ring_offset(req, cs);
 }
 
 static const int gen8_render_emit_breadcrumb_sz = 8;
@@ -1427,24 +1410,21 @@ 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;
-	int ret;
+	u32 *cs;
 
-	ret = intel_ring_begin(req, 4);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 4);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	intel_ring_emit(ring,
-			MI_SEMAPHORE_WAIT |
-			MI_SEMAPHORE_GLOBAL_GTT |
-			MI_SEMAPHORE_SAD_GTE_SDD);
-	intel_ring_emit(ring, signal->global_seqno);
-	intel_ring_emit(ring, lower_32_bits(offset));
-	intel_ring_emit(ring, upper_32_bits(offset));
-	intel_ring_advance(ring);
+	*cs++ = MI_SEMAPHORE_WAIT | MI_SEMAPHORE_GLOBAL_GTT |
+		MI_SEMAPHORE_SAD_GTE_SDD;
+	*cs++ = signal->global_seqno;
+	*cs++ = lower_32_bits(offset);
+	*cs++ = upper_32_bits(offset);
+	intel_ring_advance(req, cs);
 
 	/* When the !RCS engines idle waiting upon a semaphore, they lose their
 	 * pagetables and we must reload them before executing the batch.
@@ -1461,28 +1441,27 @@ 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];
-	int ret;
+	u32 *cs;
 
 	WARN_ON(wait_mbox == MI_SEMAPHORE_SYNC_INVALID);
 
-	ret = intel_ring_begin(req, 4);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 4);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	intel_ring_emit(ring, dw1 | wait_mbox);
+	*cs++ = 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->global_seqno - 1);
-	intel_ring_emit(ring, 0);
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*cs++ = signal->global_seqno - 1;
+	*cs++ = 0;
+	*cs++ = MI_NOOP;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -1583,16 +1562,15 @@ 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;
-	int ret;
+	u32 *cs;
 
-	ret = intel_ring_begin(req, 2);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 2);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	intel_ring_emit(ring, MI_FLUSH);
-	intel_ring_emit(ring, MI_NOOP);
-	intel_ring_advance(ring);
+	*cs++ = MI_FLUSH;
+	*cs++ = MI_NOOP;
+	intel_ring_advance(req, cs);
 	return 0;
 }
 
@@ -1658,20 +1636,16 @@ i965_emit_bb_start(struct drm_i915_gem_request *req,
 		   u64 offset, u32 length,
 		   unsigned int dispatch_flags)
 {
-	struct intel_ring *ring = req->ring;
-	int ret;
+	u32 *cs;
 
-	ret = intel_ring_begin(req, 2);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 2);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	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);
+	*cs++ = MI_BATCH_BUFFER_START | MI_BATCH_GTT | (dispatch_flags &
+		I915_DISPATCH_SECURE ? 0 : MI_BATCH_NON_SECURE_I965);
+	*cs++ = offset;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -1685,59 +1659,56 @@ i830_emit_bb_start(struct drm_i915_gem_request *req,
 		   u64 offset, u32 len,
 		   unsigned int dispatch_flags)
 {
-	struct intel_ring *ring = req->ring;
-	u32 cs_offset = i915_ggtt_offset(req->engine->scratch);
-	int ret;
+	u32 *cs, cs_offset = i915_ggtt_offset(req->engine->scratch);
 
-	ret = intel_ring_begin(req, 6);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 6);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	/* 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);
-	intel_ring_advance(ring);
+	*cs++ = COLOR_BLT_CMD | BLT_WRITE_RGBA;
+	*cs++ = BLT_DEPTH_32 | BLT_ROP_COLOR_COPY | 4096;
+	*cs++ = I830_TLB_ENTRIES << 16 | 4; /* load each page */
+	*cs++ = cs_offset;
+	*cs++ = 0xdeadbeef;
+	*cs++ = MI_NOOP;
+	intel_ring_advance(req, cs);
 
 	if ((dispatch_flags & I915_DISPATCH_PINNED) == 0) {
 		if (len > I830_BATCH_LIMIT)
 			return -ENOSPC;
 
-		ret = intel_ring_begin(req, 6 + 2);
-		if (ret)
-			return ret;
+		cs = intel_ring_begin(req, 6 + 2);
+		if (IS_ERR(cs))
+			return PTR_ERR(cs);
 
 		/* Blit the batch (which has now all relocs applied) to the
 		 * 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);
-		intel_ring_advance(ring);
+		*cs++ = SRC_COPY_BLT_CMD | BLT_WRITE_RGBA;
+		*cs++ = BLT_DEPTH_32 | BLT_ROP_SRC_COPY | 4096;
+		*cs++ = DIV_ROUND_UP(len, 4096) << 16 | 4096;
+		*cs++ = cs_offset;
+		*cs++ = 4096;
+		*cs++ = offset;
+
+		*cs++ = MI_FLUSH;
+		*cs++ = MI_NOOP;
+		intel_ring_advance(req, cs);
 
 		/* ... and execute it. */
 		offset = cs_offset;
 	}
 
-	ret = intel_ring_begin(req, 2);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 2);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	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);
+	*cs++ = MI_BATCH_BUFFER_START | MI_BATCH_GTT;
+	*cs++ = offset | (dispatch_flags & I915_DISPATCH_SECURE ? 0 :
+		MI_BATCH_NON_SECURE);
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -1747,17 +1718,16 @@ i915_emit_bb_start(struct drm_i915_gem_request *req,
 		   u64 offset, u32 len,
 		   unsigned int dispatch_flags)
 {
-	struct intel_ring *ring = req->ring;
-	int ret;
+	u32 *cs;
 
-	ret = intel_ring_begin(req, 2);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 2);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	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);
+	*cs++ = MI_BATCH_BUFFER_START | MI_BATCH_GTT;
+	*cs++ = offset | (dispatch_flags & I915_DISPATCH_SECURE ? 0 :
+		MI_BATCH_NON_SECURE);
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -2165,7 +2135,7 @@ void intel_legacy_submission_resume(struct drm_i915_private *dev_priv)
 
 static int ring_request_alloc(struct drm_i915_gem_request *request)
 {
-	int ret;
+	u32 *cs;
 
 	GEM_BUG_ON(!request->ctx->engine[request->engine->id].pin_count);
 
@@ -2178,9 +2148,9 @@ static int ring_request_alloc(struct drm_i915_gem_request *request)
 	GEM_BUG_ON(!request->engine->buffer);
 	request->ring = request->engine->buffer;
 
-	ret = intel_ring_begin(request, 0);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(request, 0);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	request->reserved_space -= LEGACY_REQUEST_SIZE;
 	return 0;
@@ -2235,7 +2205,7 @@ 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)
+u32 *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;
@@ -2243,6 +2213,7 @@ int intel_ring_begin(struct drm_i915_gem_request *req, int num_dwords)
 	int bytes = num_dwords * sizeof(u32);
 	int total_bytes, wait_bytes;
 	bool need_wrap = false;
+	u32 *cs;
 
 	total_bytes = bytes + req->reserved_space;
 
@@ -2269,7 +2240,7 @@ int intel_ring_begin(struct drm_i915_gem_request *req, int num_dwords)
 	if (wait_bytes > ring->space) {
 		int ret = wait_for_space(req, wait_bytes);
 		if (unlikely(ret))
-			return ret;
+			return ERR_PTR(ret);
 	}
 
 	if (unlikely(need_wrap)) {
@@ -2282,32 +2253,34 @@ int intel_ring_begin(struct drm_i915_gem_request *req, int num_dwords)
 		ring->space -= remain_actual;
 	}
 
+	GEM_BUG_ON(ring->tail > ring->size - bytes);
+	cs = ring->vaddr + ring->tail;
+	ring->tail += bytes;
 	ring->space -= bytes;
 	GEM_BUG_ON(ring->space < 0);
-	GEM_DEBUG_EXEC(ring->advance = ring->tail + bytes);
-	return 0;
+
+	return cs;
 }
 
 /* Align the ring tail to a cacheline boundary */
 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);
-	int ret;
+		(req->ring->tail & (CACHELINE_BYTES - 1)) / sizeof(uint32_t);
+	u32 *cs;
 
 	if (num_dwords == 0)
 		return 0;
 
 	num_dwords = CACHELINE_BYTES / sizeof(uint32_t) - num_dwords;
-	ret = intel_ring_begin(req, num_dwords);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, num_dwords);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	while (num_dwords--)
-		intel_ring_emit(ring, MI_NOOP);
+		*cs++ = MI_NOOP;
 
-	intel_ring_advance(ring);
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -2351,13 +2324,11 @@ 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;
-	int ret;
+	u32 cmd, *cs;
 
-	ret = intel_ring_begin(req, 4);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 4);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	cmd = MI_FLUSH_DW;
 	if (INTEL_GEN(req->i915) >= 8)
@@ -2379,16 +2350,16 @@ 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);
+	*cs++ = cmd;
+	*cs++ = 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 */
+		*cs++ = 0; /* upper addr */
+		*cs++ = 0; /* value */
 	} else  {
-		intel_ring_emit(ring, 0);
-		intel_ring_emit(ring, MI_NOOP);
+		*cs++ = 0;
+		*cs++ = MI_NOOP;
 	}
-	intel_ring_advance(ring);
+	intel_ring_advance(req, cs);
 	return 0;
 }
 
@@ -2397,23 +2368,21 @@ 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);
-	int ret;
+	u32 *cs;
 
-	ret = intel_ring_begin(req, 4);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 4);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	/* 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);
+	*cs++ = MI_BATCH_BUFFER_START_GEN8 | (ppgtt << 8) | (dispatch_flags &
+		I915_DISPATCH_RS ? MI_BATCH_RESOURCE_STREAMER : 0);
+	*cs++ = lower_32_bits(offset);
+	*cs++ = upper_32_bits(offset);
+	*cs++ = MI_NOOP;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -2423,22 +2392,19 @@ hsw_emit_bb_start(struct drm_i915_gem_request *req,
 		  u64 offset, u32 len,
 		  unsigned int dispatch_flags)
 {
-	struct intel_ring *ring = req->ring;
-	int ret;
+	u32 *cs;
 
-	ret = intel_ring_begin(req, 2);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 2);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	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));
+	*cs++ = 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);
+	*cs++ = offset;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -2448,20 +2414,17 @@ gen6_emit_bb_start(struct drm_i915_gem_request *req,
 		   u64 offset, u32 len,
 		   unsigned int dispatch_flags)
 {
-	struct intel_ring *ring = req->ring;
-	int ret;
+	u32 *cs;
 
-	ret = intel_ring_begin(req, 2);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 2);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
-	intel_ring_emit(ring,
-			MI_BATCH_BUFFER_START |
-			(dispatch_flags & I915_DISPATCH_SECURE ?
-			 0 : MI_BATCH_NON_SECURE_I965));
+	*cs++ = 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);
+	*cs++ = offset;
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
@@ -2470,13 +2433,11 @@ 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;
-	int ret;
+	u32 cmd, *cs;
 
-	ret = intel_ring_begin(req, 4);
-	if (ret)
-		return ret;
+	cs = intel_ring_begin(req, 4);
+	if (IS_ERR(cs))
+		return PTR_ERR(cs);
 
 	cmd = MI_FLUSH_DW;
 	if (INTEL_GEN(req->i915) >= 8)
@@ -2497,17 +2458,16 @@ 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);
+	*cs++ = cmd;
+	*cs++ = 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 */
+		*cs++ = 0; /* upper addr */
+		*cs++ = 0; /* value */
 	} else  {
-		intel_ring_emit(ring, 0);
-		intel_ring_emit(ring, MI_NOOP);
+		*cs++ = 0;
+		*cs++ = MI_NOOP;
 	}
-	intel_ring_advance(ring);
+	intel_ring_advance(req, cs);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
index cc62e89010d3..4350713dbc58 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -145,7 +145,6 @@ struct intel_ring {
 
 	u32 head;
 	u32 tail;
-	GEM_DEBUG_DECL(u32 advance);
 
 	int space;
 	int size;
@@ -292,7 +291,7 @@ struct intel_engine_cs {
 #define I915_DISPATCH_PINNED BIT(1)
 #define I915_DISPATCH_RS     BIT(2)
 	void		(*emit_breadcrumb)(struct drm_i915_gem_request *req,
-					   u32 *out);
+					   u32 *cs);
 	int		emit_breadcrumb_sz;
 
 	/* Pass the request to the hardware queue (e.g. directly into
@@ -375,7 +374,7 @@ struct intel_engine_cs {
 		/* AKA wait() */
 		int	(*sync_to)(struct drm_i915_gem_request *req,
 				   struct drm_i915_gem_request *signal);
-		u32	*(*signal)(struct drm_i915_gem_request *req, u32 *out);
+		u32	*(*signal)(struct drm_i915_gem_request *req, u32 *cs);
 	} semaphore;
 
 	/* Execlists */
@@ -497,21 +496,12 @@ void intel_engine_cleanup(struct intel_engine_cs *engine);
 
 void intel_legacy_submission_resume(struct drm_i915_private *dev_priv);
 
-int __must_check intel_ring_begin(struct drm_i915_gem_request *req, int n);
 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)
-{
-	*(uint32_t *)(ring->vaddr + ring->tail) = data;
-	ring->tail += 4;
-}
+u32 __must_check *intel_ring_begin(struct drm_i915_gem_request *req, int n);
 
-static inline void intel_ring_emit_reg(struct intel_ring *ring, i915_reg_t reg)
-{
-	intel_ring_emit(ring, i915_mmio_reg_offset(reg));
-}
-
-static inline void intel_ring_advance(struct intel_ring *ring)
+static inline void
+intel_ring_advance(struct drm_i915_gem_request *req, u32 *cs)
 {
 	/* Dummy function.
 	 *
@@ -521,14 +511,16 @@ static inline void intel_ring_advance(struct intel_ring *ring)
 	 * reserved for the command packet (i.e. the value passed to
 	 * intel_ring_begin()).
 	 */
-	GEM_DEBUG_BUG_ON(ring->tail != ring->advance);
+	GEM_BUG_ON((req->ring->vaddr + req->ring->tail) != cs);
 }
 
-static inline u32 intel_ring_offset(struct intel_ring *ring, void *addr)
+static inline u32
+intel_ring_offset(struct drm_i915_gem_request *req, void *addr)
 {
 	/* Don't write ring->size (equivalent to 0) as that hangs some GPUs. */
-	u32 offset = addr - ring->vaddr;
-	return offset & (ring->size - 1);
+	u32 offset = addr - req->ring->vaddr;
+	GEM_BUG_ON(offset > req->ring->size);
+	return offset & (req->ring->size - 1);
 }
 
 int __intel_ring_space(int head, int tail, int size);
diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_coherency.c b/drivers/gpu/drm/i915/selftests/i915_gem_coherency.c
index 483cbe02d983..ad92c60290f1 100644
--- a/drivers/gpu/drm/i915/selftests/i915_gem_coherency.c
+++ b/drivers/gpu/drm/i915/selftests/i915_gem_coherency.c
@@ -186,6 +186,7 @@ static int gpu_set(struct drm_i915_gem_object *obj,
 	struct drm_i915_private *i915 = to_i915(obj->base.dev);
 	struct drm_i915_gem_request *rq;
 	struct i915_vma *vma;
+	u32 *cs;
 	int err;
 
 	err = i915_gem_object_set_to_gtt_domain(obj, true);
@@ -202,30 +203,30 @@ static int gpu_set(struct drm_i915_gem_object *obj,
 		return PTR_ERR(rq);
 	}
 
-	err = intel_ring_begin(rq, 4);
-	if (err) {
+	cs = intel_ring_begin(rq, 4);
+	if (IS_ERR(cs)) {
 		__i915_add_request(rq, false);
 		i915_vma_unpin(vma);
-		return err;
+		return PTR_ERR(cs);
 	}
 
 	if (INTEL_GEN(i915) >= 8) {
-		intel_ring_emit(rq->ring, MI_STORE_DWORD_IMM_GEN4 | 1 << 22);
-		intel_ring_emit(rq->ring, lower_32_bits(i915_ggtt_offset(vma) + offset));
-		intel_ring_emit(rq->ring, upper_32_bits(i915_ggtt_offset(vma) + offset));
-		intel_ring_emit(rq->ring, v);
+		*cs++ = MI_STORE_DWORD_IMM_GEN4 | 1 << 22;
+		*cs++ = lower_32_bits(i915_ggtt_offset(vma) + offset);
+		*cs++ = upper_32_bits(i915_ggtt_offset(vma) + offset);
+		*cs++ = v;
 	} else if (INTEL_GEN(i915) >= 4) {
-		intel_ring_emit(rq->ring, MI_STORE_DWORD_IMM_GEN4 | 1 << 22);
-		intel_ring_emit(rq->ring, 0);
-		intel_ring_emit(rq->ring, i915_ggtt_offset(vma) + offset);
-		intel_ring_emit(rq->ring, v);
+		*cs++ = MI_STORE_DWORD_IMM_GEN4 | 1 << 22;
+		*cs++ = 0;
+		*cs++ = i915_ggtt_offset(vma) + offset;
+		*cs++ = v;
 	} else {
-		intel_ring_emit(rq->ring, MI_STORE_DWORD_IMM | 1 << 22);
-		intel_ring_emit(rq->ring, i915_ggtt_offset(vma) + offset);
-		intel_ring_emit(rq->ring, v);
-		intel_ring_emit(rq->ring, MI_NOOP);
+		*cs++ = MI_STORE_DWORD_IMM | 1 << 22;
+		*cs++ = i915_ggtt_offset(vma) + offset;
+		*cs++ = v;
+		*cs++ = MI_NOOP;
 	}
-	intel_ring_advance(rq->ring);
+	intel_ring_advance(rq, cs);
 
 	i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE);
 	i915_vma_unpin(vma);
-- 
2.20.1