drm/i915/tgl: Extend MI_SEMAPHORE_WAIT

On Tigerlake, MI_SEMAPHORE_WAIT grew an extra dword, so be sure to
update the length field and emit that extra parameter and any padding
noop as required.

v2: Define the token shift while we are adding the updated MI_SEMAPHORE_WAIT
v3: Use int instead of bool in the addition so that readers are not left
wondering about the intricacies of the C spec. Now they just have to
worry what the integer value of a boolean operation is...
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Michal Winiarski <michal.winiarski@intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190917123055.28965-1-chris@chris-wilson.co.uk

drivers/gpu/drm/i915/gt/intel_gpu_commands.h

@@ -112,6 +112,7 @@
 #define MI_SEMAPHORE_SIGNAL	MI_INSTR(0x1b, 0) /* GEN8+ */
 #define   MI_SEMAPHORE_TARGET(engine)	((engine)<<15)
 #define MI_SEMAPHORE_WAIT	MI_INSTR(0x1c, 2) /* GEN8+ */
+#define MI_SEMAPHORE_WAIT_TOKEN	MI_INSTR(0x1c, 3) /* GEN12+ */
 #define   MI_SEMAPHORE_POLL		(1 << 15)
 #define   MI_SEMAPHORE_SAD_GT_SDD	(0 << 12)
 #define   MI_SEMAPHORE_SAD_GTE_SDD	(1 << 12)
@@ -119,6 +120,8 @@
 #define   MI_SEMAPHORE_SAD_LTE_SDD	(3 << 12)
 #define   MI_SEMAPHORE_SAD_EQ_SDD	(4 << 12)
 #define   MI_SEMAPHORE_SAD_NEQ_SDD	(5 << 12)
+#define   MI_SEMAPHORE_TOKEN_MASK	REG_GENMASK(9, 5)
+#define   MI_SEMAPHORE_TOKEN_SHIFT	5
 #define MI_STORE_DWORD_IMM	MI_INSTR(0x20, 1)
 #define MI_STORE_DWORD_IMM_GEN4	MI_INSTR(0x20, 2)
 #define   MI_MEM_VIRTUAL	(1 << 22) /* 945,g33,965 */

drivers/gpu/drm/i915/gt/intel_lrc.c

@@ -2879,6 +2879,22 @@ static u32 *gen8_emit_fini_breadcrumb_rcs(struct i915_request *request, u32 *cs)
 	return gen8_emit_fini_breadcrumb_footer(request, cs);
 }
 
+static u32 *
+gen11_emit_fini_breadcrumb_rcs(struct i915_request *request, u32 *cs)
+{
+	cs = gen8_emit_ggtt_write_rcs(cs,
+				      request->fence.seqno,
+				      request->timeline->hwsp_offset,
+				      PIPE_CONTROL_CS_STALL |
+				      PIPE_CONTROL_TILE_CACHE_FLUSH |
+				      PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH |
+				      PIPE_CONTROL_DEPTH_CACHE_FLUSH |
+				      PIPE_CONTROL_DC_FLUSH_ENABLE |
+				      PIPE_CONTROL_FLUSH_ENABLE);
+
+	return gen8_emit_fini_breadcrumb_footer(request, cs);
+}
+
 /*
  * Note that the CS instruction pre-parser will not stall on the breadcrumb
  * flush and will continue pre-fetching the instructions after it before the
@@ -2897,8 +2913,49 @@ static u32 *gen8_emit_fini_breadcrumb_rcs(struct i915_request *request, u32 *cs)
  * All the above applies only to the instructions themselves. Non-inline data
  * used by the instructions is not pre-fetched.
  */
-static u32 *gen11_emit_fini_breadcrumb_rcs(struct i915_request *request,
-					   u32 *cs)
+
+static u32 *gen12_emit_preempt_busywait(struct i915_request *request, u32 *cs)
+{
+	*cs++ = MI_SEMAPHORE_WAIT_TOKEN |
+		MI_SEMAPHORE_GLOBAL_GTT |
+		MI_SEMAPHORE_POLL |
+		MI_SEMAPHORE_SAD_EQ_SDD;
+	*cs++ = 0;
+	*cs++ = intel_hws_preempt_address(request->engine);
+	*cs++ = 0;
+	*cs++ = 0;
+	*cs++ = MI_NOOP;
+
+	return cs;
+}
+
+static __always_inline u32*
+gen12_emit_fini_breadcrumb_footer(struct i915_request *request, u32 *cs)
+{
+	*cs++ = MI_USER_INTERRUPT;
+
+	*cs++ = MI_ARB_ON_OFF | MI_ARB_ENABLE;
+	if (intel_engine_has_semaphores(request->engine))
+		cs = gen12_emit_preempt_busywait(request, cs);
+
+	request->tail = intel_ring_offset(request, cs);
+	assert_ring_tail_valid(request->ring, request->tail);
+
+	return gen8_emit_wa_tail(request, cs);
+}
+
+static u32 *gen12_emit_fini_breadcrumb(struct i915_request *request, u32 *cs)
+{
+	cs = gen8_emit_ggtt_write(cs,
+				  request->fence.seqno,
+				  request->timeline->hwsp_offset,
+				  0);
+
+	return gen12_emit_fini_breadcrumb_footer(request, cs);
+}
+
+static u32 *
+gen12_emit_fini_breadcrumb_rcs(struct i915_request *request, u32 *cs)
 {
 	cs = gen8_emit_ggtt_write_rcs(cs,
 				      request->fence.seqno,
@@ -2910,7 +2967,7 @@ static u32 *gen11_emit_fini_breadcrumb_rcs(struct i915_request *request,
 				      PIPE_CONTROL_DC_FLUSH_ENABLE |
 				      PIPE_CONTROL_FLUSH_ENABLE);
 
-	return gen8_emit_fini_breadcrumb_footer(request, cs);
+	return gen12_emit_fini_breadcrumb_footer(request, cs);
 }
 
 static void execlists_park(struct intel_engine_cs *engine)
@@ -2939,9 +2996,6 @@ void intel_execlists_set_default_submission(struct intel_engine_cs *engine)
 			engine->flags |= I915_ENGINE_HAS_PREEMPTION;
 	}
 
-	if (INTEL_GEN(engine->i915) >= 12) /* XXX disabled for debugging */
-		engine->flags &= ~I915_ENGINE_HAS_SEMAPHORES;
-
 	if (engine->class != COPY_ENGINE_CLASS && INTEL_GEN(engine->i915) >= 12)
 		engine->flags |= I915_ENGINE_HAS_RELATIVE_MMIO;
 }
@@ -2971,6 +3025,8 @@ logical_ring_default_vfuncs(struct intel_engine_cs *engine)
 	engine->emit_flush = gen8_emit_flush;
 	engine->emit_init_breadcrumb = gen8_emit_init_breadcrumb;
 	engine->emit_fini_breadcrumb = gen8_emit_fini_breadcrumb;
+	if (INTEL_GEN(engine->i915) >= 12)
+		engine->emit_fini_breadcrumb = gen12_emit_fini_breadcrumb;
 
 	engine->set_default_submission = intel_execlists_set_default_submission;
 
@@ -3016,6 +3072,9 @@ static void rcs_submission_override(struct intel_engine_cs *engine)
 {
 	switch (INTEL_GEN(engine->i915)) {
 	case 12:
+		engine->emit_flush = gen11_emit_flush_render;
+		engine->emit_fini_breadcrumb = gen12_emit_fini_breadcrumb_rcs;
+		break;
 	case 11:
 		engine->emit_flush = gen11_emit_flush_render;
 		engine->emit_fini_breadcrumb = gen11_emit_fini_breadcrumb_rcs;

drivers/gpu/drm/i915/i915_pci.c

@@ -797,7 +797,6 @@ static const struct intel_device_info intel_tigerlake_12_info = {
 	.display.has_modular_fia = 1,
 	.engine_mask =
 		BIT(RCS0) | BIT(BCS0) | BIT(VECS0) | BIT(VCS0) | BIT(VCS2),
-	.has_logical_ring_preemption = false, /* XXX disabled for debugging */
 	.engine_mask = BIT(RCS0), /* XXX reduced for debugging */
 };
 

drivers/gpu/drm/i915/i915_request.c

@@ -783,7 +783,9 @@ emit_semaphore_wait(struct i915_request *to,
 		    struct i915_request *from,
 		    gfp_t gfp)
 {
+	const int has_token = INTEL_GEN(to->i915) >= 12;
 	u32 hwsp_offset;
+	int len;
 	u32 *cs;
 	int err;
 
@@ -810,7 +812,11 @@ emit_semaphore_wait(struct i915_request *to,
 	if (err)
 		return err;
 
-	cs = intel_ring_begin(to, 4);
+	len = 4;
+	if (has_token)
+		len += 2;
+
+	cs = intel_ring_begin(to, len);
 	if (IS_ERR(cs))
 		return PTR_ERR(cs);
 
@@ -822,13 +828,18 @@ emit_semaphore_wait(struct i915_request *to,
 	 * (post-wrap) values than they were expecting (and so wait
 	 * forever).
 	 */
-	*cs++ = MI_SEMAPHORE_WAIT |
+	*cs++ = (MI_SEMAPHORE_WAIT |
 		 MI_SEMAPHORE_GLOBAL_GTT |
 		 MI_SEMAPHORE_POLL |
-		MI_SEMAPHORE_SAD_GTE_SDD;
+		 MI_SEMAPHORE_SAD_GTE_SDD) +
+		has_token;
 	*cs++ = from->fence.seqno;
 	*cs++ = hwsp_offset;
 	*cs++ = 0;
+	if (has_token) {
+		*cs++ = 0;
+		*cs++ = MI_NOOP;
+	}
 
 	intel_ring_advance(to, cs);
 	to->sched.semaphores |= from->engine->mask;