1 files changed, 134 insertions, 53 deletions

diff --git a/drivers/gpu/drm/i915/gt/intel_workarounds.c b/drivers/gpu/drm/i915/gt/intel_workarounds.c
index 5726cd0a37e0..4a3bde7c9f21 100644
--- a/drivers/gpu/drm/i915/gt/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/gt/intel_workarounds.c
@@ -52,6 +52,37 @@
  * - Public functions to init or apply the given workaround type.
  */
 
+/*
+ * KBL revision ID ordering is bizarre; higher revision ID's map to lower
+ * steppings in some cases.  So rather than test against the revision ID
+ * directly, let's map that into our own range of increasing ID's that we
+ * can test against in a regular manner.
+ */
+
+const struct i915_rev_steppings kbl_revids[] = {
+	[0] = { .gt_stepping = KBL_REVID_A0, .disp_stepping = KBL_REVID_A0 },
+	[1] = { .gt_stepping = KBL_REVID_B0, .disp_stepping = KBL_REVID_B0 },
+	[2] = { .gt_stepping = KBL_REVID_C0, .disp_stepping = KBL_REVID_B0 },
+	[3] = { .gt_stepping = KBL_REVID_D0, .disp_stepping = KBL_REVID_B0 },
+	[4] = { .gt_stepping = KBL_REVID_F0, .disp_stepping = KBL_REVID_C0 },
+	[5] = { .gt_stepping = KBL_REVID_C0, .disp_stepping = KBL_REVID_B1 },
+	[6] = { .gt_stepping = KBL_REVID_D1, .disp_stepping = KBL_REVID_B1 },
+	[7] = { .gt_stepping = KBL_REVID_G0, .disp_stepping = KBL_REVID_C0 },
+};
+
+const struct i915_rev_steppings tgl_uy_revids[] = {
+	[0] = { .gt_stepping = TGL_REVID_A0, .disp_stepping = TGL_REVID_A0 },
+	[1] = { .gt_stepping = TGL_REVID_B0, .disp_stepping = TGL_REVID_C0 },
+	[2] = { .gt_stepping = TGL_REVID_B1, .disp_stepping = TGL_REVID_C0 },
+	[3] = { .gt_stepping = TGL_REVID_C0, .disp_stepping = TGL_REVID_D0 },
+};
+
+/* Same GT stepping between tgl_uy_revids and tgl_revids don't mean the same HW */
+const struct i915_rev_steppings tgl_revids[] = {
+	[0] = { .gt_stepping = TGL_REVID_A0, .disp_stepping = TGL_REVID_B0 },
+	[1] = { .gt_stepping = TGL_REVID_B0, .disp_stepping = TGL_REVID_D0 },
+};
+
 static void wa_init_start(struct i915_wa_list *wal, const char *name, const char *engine_name)
 {
 	wal->name = name;
@@ -100,8 +131,10 @@ static void _wa_add(struct i915_wa_list *wal, const struct i915_wa *wa)
 			return;
 		}
 
-		if (wal->list)
+		if (wal->list) {
 			memcpy(list, wal->list, sizeof(*wa) * wal->count);
+			kfree(wal->list);
+		}
 
 		wal->list = list;
 	}
@@ -470,7 +503,7 @@ static void kbl_ctx_workarounds_init(struct intel_engine_cs *engine,
 	gen9_ctx_workarounds_init(engine, wal);
 
 	/* WaToEnableHwFixForPushConstHWBug:kbl */
-	if (IS_KBL_REVID(i915, KBL_REVID_C0, REVID_FOREVER))
+	if (IS_KBL_GT_REVID(i915, KBL_REVID_C0, REVID_FOREVER))
 		WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2,
 				  GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION);
 
@@ -596,8 +629,8 @@ static void icl_ctx_workarounds_init(struct intel_engine_cs *engine,
 	wa_masked_en(wal, GEN9_ROW_CHICKEN4, GEN11_DIS_PICK_2ND_EU);
 }
 
-static void tgl_ctx_workarounds_init(struct intel_engine_cs *engine,
-				     struct i915_wa_list *wal)
+static void gen12_ctx_workarounds_init(struct intel_engine_cs *engine,
+				       struct i915_wa_list *wal)
 {
 	/*
 	 * Wa_1409142259:tgl
@@ -607,12 +640,28 @@ static void tgl_ctx_workarounds_init(struct intel_engine_cs *engine,
 	 * Wa_1409207793:tgl
 	 * Wa_1409178076:tgl
 	 * Wa_1408979724:tgl
+	 * Wa_14010443199:rkl
+	 * Wa_14010698770:rkl
 	 */
 	WA_SET_BIT_MASKED(GEN11_COMMON_SLICE_CHICKEN3,
 			  GEN12_DISABLE_CPS_AWARE_COLOR_PIPE);
 
+	/* WaDisableGPGPUMidThreadPreemption:gen12 */
+	WA_SET_FIELD_MASKED(GEN8_CS_CHICKEN1,
+			    GEN9_PREEMPT_GPGPU_LEVEL_MASK,
+			    GEN9_PREEMPT_GPGPU_THREAD_GROUP_LEVEL);
+}
+
+static void tgl_ctx_workarounds_init(struct intel_engine_cs *engine,
+				     struct i915_wa_list *wal)
+{
+	gen12_ctx_workarounds_init(engine, wal);
+
 	/*
-	 * Wa_1604555607:gen12 and Wa_1608008084:gen12
+	 * Wa_1604555607:tgl,rkl
+	 *
+	 * Note that the implementation of this workaround is further modified
+	 * according to the FF_MODE2 guidance given by Wa_1608008084:gen12.
 	 * FF_MODE2 register will return the wrong value when read. The default
 	 * value for this register is zero for all fields and there are no bit
 	 * masks. So instead of doing a RMW we should just write the GS Timer
@@ -623,11 +672,6 @@ static void tgl_ctx_workarounds_init(struct intel_engine_cs *engine,
 	       FF_MODE2_GS_TIMER_MASK | FF_MODE2_TDS_TIMER_MASK,
 	       FF_MODE2_GS_TIMER_224  | FF_MODE2_TDS_TIMER_128,
 	       0);
-
-	/* WaDisableGPGPUMidThreadPreemption:tgl */
-	WA_SET_FIELD_MASKED(GEN8_CS_CHICKEN1,
-			    GEN9_PREEMPT_GPGPU_LEVEL_MASK,
-			    GEN9_PREEMPT_GPGPU_THREAD_GROUP_LEVEL);
 }
 
 static void
@@ -642,8 +686,10 @@ __intel_engine_init_ctx_wa(struct intel_engine_cs *engine,
 
 	wa_init_start(wal, name, engine->name);
 
-	if (IS_GEN(i915, 12))
+	if (IS_ROCKETLAKE(i915) || IS_TIGERLAKE(i915))
 		tgl_ctx_workarounds_init(engine, wal);
+	else if (IS_GEN(i915, 12))
+		gen12_ctx_workarounds_init(engine, wal);
 	else if (IS_GEN(i915, 11))
 		icl_ctx_workarounds_init(engine, wal);
 	else if (IS_CANNONLAKE(i915))
@@ -995,7 +1041,7 @@ kbl_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal)
 	gen9_gt_workarounds_init(i915, wal);
 
 	/* WaDisableDynamicCreditSharing:kbl */
-	if (IS_KBL_REVID(i915, 0, KBL_REVID_B0))
+	if (IS_KBL_GT_REVID(i915, 0, KBL_REVID_B0))
 		wa_write_or(wal,
 			    GAMT_CHKN_BIT_REG,
 			    GAMT_CHKN_DISABLE_DYNAMIC_CREDIT_SHARING);
@@ -1176,18 +1222,25 @@ icl_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal)
 }
 
 static void
-tgl_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal)
+gen12_gt_workarounds_init(struct drm_i915_private *i915,
+			  struct i915_wa_list *wal)
 {
 	wa_init_mcr(i915, wal);
+}
+
+static void
+tgl_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal)
+{
+	gen12_gt_workarounds_init(i915, wal);
 
 	/* Wa_1409420604:tgl */
-	if (IS_TGL_REVID(i915, TGL_REVID_A0, TGL_REVID_A0))
+	if (IS_TGL_UY_GT_REVID(i915, TGL_REVID_A0, TGL_REVID_A0))
 		wa_write_or(wal,
 			    SUBSLICE_UNIT_LEVEL_CLKGATE2,
 			    CPSSUNIT_CLKGATE_DIS);
 
 	/* Wa_1607087056:tgl also know as BUG:1409180338 */
-	if (IS_TGL_REVID(i915, TGL_REVID_A0, TGL_REVID_A0))
+	if (IS_TGL_UY_GT_REVID(i915, TGL_REVID_A0, TGL_REVID_A0))
 		wa_write_or(wal,
 			    SLICE_UNIT_LEVEL_CLKGATE,
 			    L3_CLKGATE_DIS | L3_CR2X_CLKGATE_DIS);
@@ -1196,8 +1249,10 @@ tgl_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal)
 static void
 gt_init_workarounds(struct drm_i915_private *i915, struct i915_wa_list *wal)
 {
-	if (IS_GEN(i915, 12))
+	if (IS_TIGERLAKE(i915))
 		tgl_gt_workarounds_init(i915, wal);
+	else if (IS_GEN(i915, 12))
+		gen12_gt_workarounds_init(i915, wal);
 	else if (IS_GEN(i915, 11))
 		icl_gt_workarounds_init(i915, wal);
 	else if (IS_CANNONLAKE(i915))
@@ -1620,7 +1675,7 @@ rcs_engine_wa_init(struct intel_engine_cs *engine, struct i915_wa_list *wal)
 {
 	struct drm_i915_private *i915 = engine->i915;
 
-	if (IS_TGL_REVID(i915, TGL_REVID_A0, TGL_REVID_A0)) {
+	if (IS_TGL_UY_GT_REVID(i915, TGL_REVID_A0, TGL_REVID_A0)) {
 		/*
 		 * Wa_1607138336:tgl
 		 * Wa_1607063988:tgl
@@ -1630,18 +1685,6 @@ rcs_engine_wa_init(struct intel_engine_cs *engine, struct i915_wa_list *wal)
 			    GEN12_DISABLE_POSH_BUSY_FF_DOP_CG);
 
 		/*
-		 * Wa_1607030317:tgl
-		 * Wa_1607186500:tgl
-		 * Wa_1607297627:tgl there is 3 entries for this WA on BSpec, 2
-		 * of then says it is fixed on B0 the other one says it is
-		 * permanent
-		 */
-		wa_masked_en(wal,
-			     GEN6_RC_SLEEP_PSMI_CONTROL,
-			     GEN12_WAIT_FOR_EVENT_POWER_DOWN_DISABLE |
-			     GEN8_RC_SEMA_IDLE_MSG_DISABLE);
-
-		/*
 		 * Wa_1606679103:tgl
 		 * (see also Wa_1606682166:icl)
 		 */
@@ -1654,22 +1697,17 @@ rcs_engine_wa_init(struct intel_engine_cs *engine, struct i915_wa_list *wal)
 			    VSUNIT_CLKGATE_DIS_TGL);
 	}
 
-	if (IS_TIGERLAKE(i915)) {
-		/* Wa_1606931601:tgl */
+	if (IS_ROCKETLAKE(i915) || IS_TIGERLAKE(i915)) {
+		/* Wa_1606931601:tgl,rkl */
 		wa_masked_en(wal, GEN7_ROW_CHICKEN2, GEN12_DISABLE_EARLY_READ);
 
-		/* Wa_1409804808:tgl */
+		/* Wa_1409804808:tgl,rkl */
 		wa_masked_en(wal, GEN7_ROW_CHICKEN2,
 			     GEN12_PUSH_CONST_DEREF_HOLD_DIS);
 
-		/* Wa_1606700617:tgl */
-		wa_masked_en(wal,
-			     GEN9_CS_DEBUG_MODE1,
-			     FF_DOP_CLOCK_GATE_DISABLE);
-
 		/*
 		 * Wa_1409085225:tgl
-		 * Wa_14010229206:tgl
+		 * Wa_14010229206:tgl,rkl
 		 */
 		wa_masked_en(wal, GEN9_ROW_CHICKEN4, GEN12_DISABLE_TDL_PUSH);
 
@@ -1677,9 +1715,37 @@ rcs_engine_wa_init(struct intel_engine_cs *engine, struct i915_wa_list *wal)
 		 * Wa_1407928979:tgl A*
 		 * Wa_18011464164:tgl B0+
 		 * Wa_22010931296:tgl B0+
+		 * Wa_14010919138:rkl,tgl
 		 */
 		wa_write_or(wal, GEN7_FF_THREAD_MODE,
 			    GEN12_FF_TESSELATION_DOP_GATE_DISABLE);
+
+		/*
+		 * Wa_1607030317:tgl
+		 * Wa_1607186500:tgl
+		 * Wa_1607297627:tgl,rkl there are multiple entries for this
+		 * WA in the BSpec; some indicate this is an A0-only WA,
+		 * others indicate it applies to all steppings.
+		 */
+		wa_masked_en(wal,
+			     GEN6_RC_SLEEP_PSMI_CONTROL,
+			     GEN12_WAIT_FOR_EVENT_POWER_DOWN_DISABLE |
+			     GEN8_RC_SEMA_IDLE_MSG_DISABLE);
+
+		/*
+		 * Wa_1606700617:tgl
+		 * Wa_22010271021:tgl,rkl
+		 */
+		wa_masked_en(wal,
+			     GEN9_CS_DEBUG_MODE1,
+			     FF_DOP_CLOCK_GATE_DISABLE);
+	}
+
+	if (IS_GEN(i915, 12)) {
+		/* Wa_1406941453:gen12 */
+		wa_masked_en(wal,
+			     GEN10_SAMPLER_MODE,
+			     ENABLE_SMALLPL);
 	}
 
 	if (IS_GEN(i915, 11)) {
@@ -1898,7 +1964,7 @@ xcs_engine_wa_init(struct intel_engine_cs *engine, struct i915_wa_list *wal)
 	struct drm_i915_private *i915 = engine->i915;
 
 	/* WaKBLVECSSemaphoreWaitPoll:kbl */
-	if (IS_KBL_REVID(i915, KBL_REVID_A0, KBL_REVID_E0)) {
+	if (IS_KBL_GT_REVID(i915, KBL_REVID_A0, KBL_REVID_E0)) {
 		wa_write(wal,
 			 RING_SEMA_WAIT_POLL(engine->mmio_base),
 			 1);
@@ -2045,6 +2111,7 @@ static int engine_wa_list_verify(struct intel_context *ce,
 	const struct i915_wa *wa;
 	struct i915_request *rq;
 	struct i915_vma *vma;
+	struct i915_gem_ww_ctx ww;
 	unsigned int i;
 	u32 *results;
 	int err;
@@ -2057,29 +2124,34 @@ static int engine_wa_list_verify(struct intel_context *ce,
 		return PTR_ERR(vma);
 
 	intel_engine_pm_get(ce->engine);
-	rq = intel_context_create_request(ce);
-	intel_engine_pm_put(ce->engine);
+	i915_gem_ww_ctx_init(&ww, false);
+retry:
+	err = i915_gem_object_lock(vma->obj, &ww);
+	if (err == 0)
+		err = intel_context_pin_ww(ce, &ww);
+	if (err)
+		goto err_pm;
+
+	rq = i915_request_create(ce);
 	if (IS_ERR(rq)) {
 		err = PTR_ERR(rq);
-		goto err_vma;
+		goto err_unpin;
 	}
 
-	i915_vma_lock(vma);
 	err = i915_request_await_object(rq, vma->obj, true);
 	if (err == 0)
 		err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE);
-	i915_vma_unlock(vma);
-	if (err) {
-		i915_request_add(rq);
-		goto err_vma;
-	}
-
-	err = wa_list_srm(rq, wal, vma);
-	if (err)
-		goto err_vma;
+	if (err == 0)
+		err = wa_list_srm(rq, wal, vma);
 
 	i915_request_get(rq);
+	if (err)
+		i915_request_set_error_once(rq, err);
 	i915_request_add(rq);
+
+	if (err)
+		goto err_rq;
+
 	if (i915_request_wait(rq, 0, HZ / 5) < 0) {
 		err = -ETIME;
 		goto err_rq;
@@ -2104,7 +2176,16 @@ static int engine_wa_list_verify(struct intel_context *ce,
 
 err_rq:
 	i915_request_put(rq);
-err_vma:
+err_unpin:
+	intel_context_unpin(ce);
+err_pm:
+	if (err == -EDEADLK) {
+		err = i915_gem_ww_ctx_backoff(&ww);
+		if (!err)
+			goto retry;
+	}
+	i915_gem_ww_ctx_fini(&ww);
+	intel_engine_pm_put(ce->engine);
 	i915_vma_unpin(vma);
 	i915_vma_put(vma);
 	return err;