Skip to contentSkip to conversation

drm/i915/perf: Allow non-privileged access when OA buffer is not sampled

PATCH Superseded Umesh Nerlige Ramappa 3 participants 7 messages Nov 14, 2019 – Nov 18, 2019

Subsystems (2)

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/2 drm/i915/perf: Allow non-privileged access when OA buffer is not sampled

SAMPLE_OA_REPORT enables sampling of OA reports from the OA buffer. Since reports from OA buffer had system wide visibility, collecting samples from the OA buffer was a privileged operation on previous platforms. Prior to TGL, it was also necessary to sample the OA buffer to normalize reports from MI REPORT PERF COUNT.

TGL has a dedicated OAR unit to sample perf reports for a specific render context. This removes the necessity to sample OA buffer.

- If not sampling the OA buffer, allow non-privileged access. An earlier patch allows the non-privilege access:
https://patchwork.freedesktop.org/patch/337716/?series=68582&rev=1
- Clear up the path for non-privileged access in this patch

---
 drivers/gpu/drm/i915/i915_perf.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_perf.c +3 -2
index 31e47ee23357..221c1090ae93 100644
@@ -2718,7 +2718,8 @@ static int i915_oa_stream_init(struct i915_perf_stream *stream,
 		return -EINVAL;
 	}
 
-	if (!(props->sample_flags & SAMPLE_OA_REPORT)) {
+	if (!(props->sample_flags & SAMPLE_OA_REPORT) &&
+	    (INTEL_GEN(perf->i915) < 12 || !stream->ctx)) {
 		DRM_DEBUG("Only OA report sampling supported\n");
 		return -EINVAL;
 	}
@@ -2750,7 +2751,7 @@ static int i915_oa_stream_init(struct i915_perf_stream *stream,
 
 	format_size = perf->oa_formats[props->oa_format].size;
 
-	stream->sample_flags |= SAMPLE_OA_REPORT;
+	stream->sample_flags = props->sample_flags;
 	stream->sample_size += format_size;
 
 	stream->oa_buffer.format_size = format_size;
Signature · 6 lines
2.20.1

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#1 lore raw

PATCH 2/2 drm/i915/perf: Configure OAR for specific context

Gen12 supports saving/restoring render counters per context. Apply OAR configuration only for the context that is passed in to perf.

v2:
- Fix OACTXCONTROL value to only stop/resume counters.
- Remove gen12_update_reg_state_unlocked as power state is already applied by the caller.

---
 drivers/gpu/drm/i915/i915_perf.c | 193 +++++++++++++++++--------------
 1 file changed, 108 insertions(+), 85 deletions(-)
drivers/gpu/drm/i915/i915_perf.c +108 -85
index 221c1090ae93..2f0be5fbef4b 100644
@@ -2078,20 +2078,12 @@ gen8_update_reg_state_unlocked(const struct intel_context *ce,
 	u32 *reg_state = ce->lrc_reg_state;
 	int i;
 
-	if (IS_GEN(stream->perf->i915, 12)) {
-		u32 format = stream->oa_buffer.format;
+	reg_state[ctx_oactxctrl + 1] =
+		(stream->period_exponent << GEN8_OA_TIMER_PERIOD_SHIFT) |
+		(stream->periodic ? GEN8_OA_TIMER_ENABLE : 0) |
+		GEN8_OA_COUNTER_RESUME;
 
-		reg_state[ctx_oactxctrl + 1] =
-			(format << GEN12_OAR_OACONTROL_COUNTER_FORMAT_SHIFT) |
-			(stream->oa_config ? GEN12_OAR_OACONTROL_COUNTER_ENABLE : 0);
-	} else {
-		reg_state[ctx_oactxctrl + 1] =
-			(stream->period_exponent << GEN8_OA_TIMER_PERIOD_SHIFT) |
-			(stream->periodic ? GEN8_OA_TIMER_ENABLE : 0) |
-			GEN8_OA_COUNTER_RESUME;
-	}
-
-	for (i = 0; !!ctx_flexeu0 && i < ARRAY_SIZE(flex_regs); i++)
+	for (i = 0; i < ARRAY_SIZE(flex_regs); i++)
 		reg_state[ctx_flexeu0 + i * 2 + 1] =
 			oa_config_flex_reg(stream->oa_config, flex_regs[i]);
 
@@ -2224,34 +2216,49 @@ static int gen8_configure_context(struct i915_gem_context *ctx,
 	return err;
 }
 
-static int gen12_emit_oar_config(struct intel_context *ce, bool enable)
+static int gen12_configure_oar_context(struct i915_perf_stream *stream, bool enable)
 {
-	struct i915_request *rq;
-	u32 *cs;
-	int err = 0;
-
-	rq = i915_request_create(ce);
-	if (IS_ERR(rq))
-		return PTR_ERR(rq);
+	int err;
+	struct intel_context *ce = stream->pinned_ctx;
+	struct flex regs_context[] = {
+		{
+			GEN8_OACTXCONTROL,
+			stream->perf->ctx_oactxctrl_offset + 1,
+			enable ? GEN8_OA_COUNTER_RESUME : 0,
+		},
+	};
+	struct flex regs_lri[] = {
+		{
+			GEN12_OAR_OACONTROL,
+		},
+		{
+			RING_CONTEXT_CONTROL(ce->engine->mmio_base),
+		},
+	};
+	u32 format = stream->oa_buffer.format;
 
-	cs = intel_ring_begin(rq, 4);
-	if (IS_ERR(cs)) {
-		err = PTR_ERR(cs);
-		goto out;
-	}
+	regs_lri[0].value =
+		(format << GEN12_OAR_OACONTROL_COUNTER_FORMAT_SHIFT) |
+		(enable ? GEN12_OAR_OACONTROL_COUNTER_ENABLE : 0);
 
-	*cs++ = MI_LOAD_REGISTER_IMM(1);
-	*cs++ = i915_mmio_reg_offset(RING_CONTEXT_CONTROL(ce->engine->mmio_base));
-	*cs++ = _MASKED_FIELD(GEN12_CTX_CTRL_OAR_CONTEXT_ENABLE,
-			      enable ? GEN12_CTX_CTRL_OAR_CONTEXT_ENABLE : 0);
-	*cs++ = MI_NOOP;
+	regs_lri[1].value =
+		_MASKED_FIELD(GEN12_CTX_CTRL_OAR_CONTEXT_ENABLE,
+			      enable ?
+			      GEN12_CTX_CTRL_OAR_CONTEXT_ENABLE :
+			      0);
 
-	intel_ring_advance(rq, cs);
+	/* Modify the context image of pinned context with regs_context*/
+	err = intel_context_lock_pinned(ce);
+	if (err)
+		return err;
 
-out:
-	i915_request_add(rq);
+	err = gen8_modify_context(ce, regs_context, ARRAY_SIZE(regs_context));
+	intel_context_unlock_pinned(ce);
+	if (err)
+		return err;
 
-	return err;
+	/* Apply regs_lri using LRI with pinned context */
+	return gen8_modify_self(ce, regs_lri, ARRAY_SIZE(regs_lri));
 }
 
 /*
@@ -2277,53 +2284,16 @@ static int gen12_emit_oar_config(struct intel_context *ce, bool enable)
  *   per-context OA state.
  *
  * Note: it's only the RCS/Render context that has any OA state.
+ * Note: the first flex register passed must always be R_PWR_CLK_STATE
  */
-static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
-				      const struct i915_oa_config *oa_config)
+static int oa_configure_all_contexts(struct i915_perf_stream *stream,
+				     struct flex *regs,
+				     size_t num_regs)
 {
 	struct drm_i915_private *i915 = stream->perf->i915;
-	/* The MMIO offsets for Flex EU registers aren't contiguous */
-	const u32 ctx_flexeu0 = stream->perf->ctx_flexeu0_offset;
-#define ctx_flexeuN(N) (ctx_flexeu0 + 2 * (N) + 1)
-	struct flex regs[] = {
-		{
-			GEN8_R_PWR_CLK_STATE,
-			CTX_R_PWR_CLK_STATE,
-		},
-		{
-			IS_GEN(i915, 12) ?
-			GEN12_OAR_OACONTROL : GEN8_OACTXCONTROL,
-			stream->perf->ctx_oactxctrl_offset + 1,
-		},
-		{ EU_PERF_CNTL0, ctx_flexeuN(0) },
-		{ EU_PERF_CNTL1, ctx_flexeuN(1) },
-		{ EU_PERF_CNTL2, ctx_flexeuN(2) },
-		{ EU_PERF_CNTL3, ctx_flexeuN(3) },
-		{ EU_PERF_CNTL4, ctx_flexeuN(4) },
-		{ EU_PERF_CNTL5, ctx_flexeuN(5) },
-		{ EU_PERF_CNTL6, ctx_flexeuN(6) },
-	};
-#undef ctx_flexeuN
 	struct intel_engine_cs *engine;
 	struct i915_gem_context *ctx, *cn;
-	size_t array_size = IS_GEN(i915, 12) ? 2 : ARRAY_SIZE(regs);
-	int i, err;
-
-	if (IS_GEN(i915, 12)) {
-		u32 format = stream->oa_buffer.format;
-
-		regs[1].value =
-			(format << GEN12_OAR_OACONTROL_COUNTER_FORMAT_SHIFT) |
-			(oa_config ? GEN12_OAR_OACONTROL_COUNTER_ENABLE : 0);
-	} else {
-		regs[1].value =
-			(stream->period_exponent << GEN8_OA_TIMER_PERIOD_SHIFT) |
-			(stream->periodic ? GEN8_OA_TIMER_ENABLE : 0) |
-			GEN8_OA_COUNTER_RESUME;
-	}
-
-	for (i = 2; !!ctx_flexeu0 && i < array_size; i++)
-		regs[i].value = oa_config_flex_reg(oa_config, regs[i].reg);
+	int err;
 
 	lockdep_assert_held(&stream->perf->lock);
 
@@ -2353,7 +2323,7 @@ static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
 
 		spin_unlock(&i915->gem.contexts.lock);
 
-		err = gen8_configure_context(ctx, regs, array_size);
+		err = gen8_configure_context(ctx, regs, num_regs);
 		if (err) {
 			i915_gem_context_put(ctx);
 			return err;
@@ -2378,7 +2348,7 @@ static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
 
 		regs[0].value = intel_sseu_make_rpcs(i915, &ce->sseu);
 
-		err = gen8_modify_self(ce, regs, array_size);
+		err = gen8_modify_self(ce, regs, num_regs);
 		if (err)
 			return err;
 	}
@@ -2386,6 +2356,56 @@ static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
 	return 0;
 }
 
+static int gen12_configure_all_contexts(struct i915_perf_stream *stream,
+					const struct i915_oa_config *oa_config)
+{
+	struct flex regs[] = {
+		{
+			GEN8_R_PWR_CLK_STATE,
+			CTX_R_PWR_CLK_STATE,
+		},
+	};
+
+	return oa_configure_all_contexts(stream, regs, ARRAY_SIZE(regs));
+}
+
+static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
+				      const struct i915_oa_config *oa_config)
+{
+	/* The MMIO offsets for Flex EU registers aren't contiguous */
+	const u32 ctx_flexeu0 = stream->perf->ctx_flexeu0_offset;
+#define ctx_flexeuN(N) (ctx_flexeu0 + 2 * (N) + 1)
+	struct flex regs[] = {
+		{
+			GEN8_R_PWR_CLK_STATE,
+			CTX_R_PWR_CLK_STATE,
+		},
+		{
+			GEN8_OACTXCONTROL,
+			stream->perf->ctx_oactxctrl_offset + 1,
+		},
+		{ EU_PERF_CNTL0, ctx_flexeuN(0) },
+		{ EU_PERF_CNTL1, ctx_flexeuN(1) },
+		{ EU_PERF_CNTL2, ctx_flexeuN(2) },
+		{ EU_PERF_CNTL3, ctx_flexeuN(3) },
+		{ EU_PERF_CNTL4, ctx_flexeuN(4) },
+		{ EU_PERF_CNTL5, ctx_flexeuN(5) },
+		{ EU_PERF_CNTL6, ctx_flexeuN(6) },
+	};
+#undef ctx_flexeuN
+	int i;
+
+	regs[1].value =
+		(stream->period_exponent << GEN8_OA_TIMER_PERIOD_SHIFT) |
+		(stream->periodic ? GEN8_OA_TIMER_ENABLE : 0) |
+		GEN8_OA_COUNTER_RESUME;
+
+	for (i = 2; i < ARRAY_SIZE(regs); i++)
+		regs[i].value = oa_config_flex_reg(oa_config, regs[i].reg);
+
+	return oa_configure_all_contexts(stream, regs, ARRAY_SIZE(regs));
+}
+
 static int gen8_enable_metric_set(struct i915_perf_stream *stream)
 {
 	struct intel_uncore *uncore = stream->uncore;
@@ -2469,7 +2489,7 @@ static int gen12_enable_metric_set(struct i915_perf_stream *stream)
 	 * to make sure all slices/subslices are ON before writing to NOA
 	 * registers.
 	 */
-	ret = lrc_configure_all_contexts(stream, oa_config);
+	ret = gen12_configure_all_contexts(stream, oa_config);
 	if (ret)
 		return ret;
 
@@ -2479,8 +2499,7 @@ static int gen12_enable_metric_set(struct i915_perf_stream *stream)
 	 * requested this.
 	 */
 	if (stream->ctx) {
-		ret = gen12_emit_oar_config(stream->pinned_ctx,
-					    oa_config != NULL);
+		ret = gen12_configure_oar_context(stream, oa_config != NULL);
 		if (ret)
 			return ret;
 	}
@@ -2514,11 +2533,11 @@ static void gen12_disable_metric_set(struct i915_perf_stream *stream)
 	struct intel_uncore *uncore = stream->uncore;
 
 	/* Reset all contexts' slices/subslices configurations. */
-	lrc_configure_all_contexts(stream, NULL);
+	gen12_configure_all_contexts(stream, NULL);
 
 	/* disable the context save/restore or OAR counters */
 	if (stream->ctx)
-		gen12_emit_oar_config(stream->pinned_ctx, false);
+		gen12_configure_oar_context(stream, false);
 
 	/* Make sure we disable noa to save power. */
 	intel_uncore_rmw(uncore, RPM_CONFIG1, GEN10_GT_NOA_ENABLE, 0);
@@ -2860,7 +2879,11 @@ void i915_oa_init_reg_state(const struct intel_context *ce,
 		return;
 
 	stream = engine->i915->perf.exclusive_stream;
-	if (stream)
+	/*
+	 * For gen12, only CTX_R_PWR_CLK_STATE needs update, but the caller
+	 * is already doing that, so nothing to be done for gen12 here.
+	 */
+	if (stream && INTEL_GEN(stream->perf->i915) < 12)
 		gen8_update_reg_state_unlocked(ce, stream);
 }
 
Signature · 6 lines
2.20.1

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#2 lore raw

Re: [PATCH 2/2] drm/i915/perf: Configure OAR for specific context

Quoted 9 lines · Umesh Nerlige Ramappa

Gen12 supports saving/restoring render counters per context. Apply OAR configuration only for the context that is passed in to perf.

v2:
- Fix OACTXCONTROL value to only stop/resume counters.
- Remove gen12_update_reg_state_unlocked as power state is already applied by the caller.

IGT Tests in this series:
https://patchwork.freedesktop.org/series/69322/

gen12-mi-rpc
gen12-unprivileged-single-ctx-counters

Thanks,
Umesh

Quoted 293 lines

---
drivers/gpu/drm/i915/i915_perf.c | 193 +++++++++++++++++--------------

1 file changed, 108 insertions(+), 85 deletions(-)
drivers/gpu/drm/i915/i915_perf.c +108 -85
index 221c1090ae93..2f0be5fbef4b 100644
@@ -2078,20 +2078,12 @@ gen8_update_reg_state_unlocked(const struct intel_context *ce,
 	u32 *reg_state = ce->lrc_reg_state;
 	int i;
 
-	if (IS_GEN(stream->perf->i915, 12)) {
-		u32 format = stream->oa_buffer.format;
+	reg_state[ctx_oactxctrl + 1] =
+		(stream->period_exponent << GEN8_OA_TIMER_PERIOD_SHIFT) |
+		(stream->periodic ? GEN8_OA_TIMER_ENABLE : 0) |
+		GEN8_OA_COUNTER_RESUME;
 
-		reg_state[ctx_oactxctrl + 1] =
-			(format << GEN12_OAR_OACONTROL_COUNTER_FORMAT_SHIFT) |
-			(stream->oa_config ? GEN12_OAR_OACONTROL_COUNTER_ENABLE : 0);
-	} else {
-		reg_state[ctx_oactxctrl + 1] =
-			(stream->period_exponent << GEN8_OA_TIMER_PERIOD_SHIFT) |
-			(stream->periodic ? GEN8_OA_TIMER_ENABLE : 0) |
-			GEN8_OA_COUNTER_RESUME;
-	}
-
-	for (i = 0; !!ctx_flexeu0 && i < ARRAY_SIZE(flex_regs); i++)
+	for (i = 0; i < ARRAY_SIZE(flex_regs); i++)
 		reg_state[ctx_flexeu0 + i * 2 + 1] =
 			oa_config_flex_reg(stream->oa_config, flex_regs[i]);
 
@@ -2224,34 +2216,49 @@ static int gen8_configure_context(struct i915_gem_context *ctx,
 	return err;
 }
 
-static int gen12_emit_oar_config(struct intel_context *ce, bool enable)
+static int gen12_configure_oar_context(struct i915_perf_stream *stream, bool enable)
 {
-	struct i915_request *rq;
-	u32 *cs;
-	int err = 0;
-
-	rq = i915_request_create(ce);
-	if (IS_ERR(rq))
-		return PTR_ERR(rq);
+	int err;
+	struct intel_context *ce = stream->pinned_ctx;
+	struct flex regs_context[] = {
+		{
+			GEN8_OACTXCONTROL,
+			stream->perf->ctx_oactxctrl_offset + 1,
+			enable ? GEN8_OA_COUNTER_RESUME : 0,
+		},
+	};
+	struct flex regs_lri[] = {
+		{
+			GEN12_OAR_OACONTROL,
+		},
+		{
+			RING_CONTEXT_CONTROL(ce->engine->mmio_base),
+		},
+	};
+	u32 format = stream->oa_buffer.format;
 
-	cs = intel_ring_begin(rq, 4);
-	if (IS_ERR(cs)) {
-		err = PTR_ERR(cs);
-		goto out;
-	}
+	regs_lri[0].value =
+		(format << GEN12_OAR_OACONTROL_COUNTER_FORMAT_SHIFT) |
+		(enable ? GEN12_OAR_OACONTROL_COUNTER_ENABLE : 0);
 
-	*cs++ = MI_LOAD_REGISTER_IMM(1);
-	*cs++ = i915_mmio_reg_offset(RING_CONTEXT_CONTROL(ce->engine->mmio_base));
-	*cs++ = _MASKED_FIELD(GEN12_CTX_CTRL_OAR_CONTEXT_ENABLE,
-			      enable ? GEN12_CTX_CTRL_OAR_CONTEXT_ENABLE : 0);
-	*cs++ = MI_NOOP;
+	regs_lri[1].value =
+		_MASKED_FIELD(GEN12_CTX_CTRL_OAR_CONTEXT_ENABLE,
+			      enable ?
+			      GEN12_CTX_CTRL_OAR_CONTEXT_ENABLE :
+			      0);
 
-	intel_ring_advance(rq, cs);
+	/* Modify the context image of pinned context with regs_context*/
+	err = intel_context_lock_pinned(ce);
+	if (err)
+		return err;
 
-out:
-	i915_request_add(rq);
+	err = gen8_modify_context(ce, regs_context, ARRAY_SIZE(regs_context));
+	intel_context_unlock_pinned(ce);
+	if (err)
+		return err;
 
-	return err;
+	/* Apply regs_lri using LRI with pinned context */
+	return gen8_modify_self(ce, regs_lri, ARRAY_SIZE(regs_lri));
 }
 
 /*
@@ -2277,53 +2284,16 @@ static int gen12_emit_oar_config(struct intel_context *ce, bool enable)
 *   per-context OA state.
 *
 * Note: it's only the RCS/Render context that has any OA state.
+ * Note: the first flex register passed must always be R_PWR_CLK_STATE
 */
-static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
-				      const struct i915_oa_config *oa_config)
+static int oa_configure_all_contexts(struct i915_perf_stream *stream,
+				     struct flex *regs,
+				     size_t num_regs)
 {
 	struct drm_i915_private *i915 = stream->perf->i915;
-	/* The MMIO offsets for Flex EU registers aren't contiguous */
-	const u32 ctx_flexeu0 = stream->perf->ctx_flexeu0_offset;
-#define ctx_flexeuN(N) (ctx_flexeu0 + 2 * (N) + 1)
-	struct flex regs[] = {
-		{
-			GEN8_R_PWR_CLK_STATE,
-			CTX_R_PWR_CLK_STATE,
-		},
-		{
-			IS_GEN(i915, 12) ?
-			GEN12_OAR_OACONTROL : GEN8_OACTXCONTROL,
-			stream->perf->ctx_oactxctrl_offset + 1,
-		},
-		{ EU_PERF_CNTL0, ctx_flexeuN(0) },
-		{ EU_PERF_CNTL1, ctx_flexeuN(1) },
-		{ EU_PERF_CNTL2, ctx_flexeuN(2) },
-		{ EU_PERF_CNTL3, ctx_flexeuN(3) },
-		{ EU_PERF_CNTL4, ctx_flexeuN(4) },
-		{ EU_PERF_CNTL5, ctx_flexeuN(5) },
-		{ EU_PERF_CNTL6, ctx_flexeuN(6) },
-	};
-#undef ctx_flexeuN
 	struct intel_engine_cs *engine;
 	struct i915_gem_context *ctx, *cn;
-	size_t array_size = IS_GEN(i915, 12) ? 2 : ARRAY_SIZE(regs);
-	int i, err;
-
-	if (IS_GEN(i915, 12)) {
-		u32 format = stream->oa_buffer.format;
-
-		regs[1].value =
-			(format << GEN12_OAR_OACONTROL_COUNTER_FORMAT_SHIFT) |
-			(oa_config ? GEN12_OAR_OACONTROL_COUNTER_ENABLE : 0);
-	} else {
-		regs[1].value =
-			(stream->period_exponent << GEN8_OA_TIMER_PERIOD_SHIFT) |
-			(stream->periodic ? GEN8_OA_TIMER_ENABLE : 0) |
-			GEN8_OA_COUNTER_RESUME;
-	}
-
-	for (i = 2; !!ctx_flexeu0 && i < array_size; i++)
-		regs[i].value = oa_config_flex_reg(oa_config, regs[i].reg);
+	int err;
 
 	lockdep_assert_held(&stream->perf->lock);
 
@@ -2353,7 +2323,7 @@ static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
 
 		spin_unlock(&i915->gem.contexts.lock);
 
-		err = gen8_configure_context(ctx, regs, array_size);
+		err = gen8_configure_context(ctx, regs, num_regs);
 		if (err) {
 			i915_gem_context_put(ctx);
 			return err;
@@ -2378,7 +2348,7 @@ static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
 
 		regs[0].value = intel_sseu_make_rpcs(i915, &ce->sseu);
 
-		err = gen8_modify_self(ce, regs, array_size);
+		err = gen8_modify_self(ce, regs, num_regs);
 		if (err)
 			return err;
 	}
@@ -2386,6 +2356,56 @@ static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
 	return 0;
 }
 
+static int gen12_configure_all_contexts(struct i915_perf_stream *stream,
+					const struct i915_oa_config *oa_config)
+{
+	struct flex regs[] = {
+		{
+			GEN8_R_PWR_CLK_STATE,
+			CTX_R_PWR_CLK_STATE,
+		},
+	};
+
+	return oa_configure_all_contexts(stream, regs, ARRAY_SIZE(regs));
+}
+
+static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
+				      const struct i915_oa_config *oa_config)
+{
+	/* The MMIO offsets for Flex EU registers aren't contiguous */
+	const u32 ctx_flexeu0 = stream->perf->ctx_flexeu0_offset;
+#define ctx_flexeuN(N) (ctx_flexeu0 + 2 * (N) + 1)
+	struct flex regs[] = {
+		{
+			GEN8_R_PWR_CLK_STATE,
+			CTX_R_PWR_CLK_STATE,
+		},
+		{
+			GEN8_OACTXCONTROL,
+			stream->perf->ctx_oactxctrl_offset + 1,
+		},
+		{ EU_PERF_CNTL0, ctx_flexeuN(0) },
+		{ EU_PERF_CNTL1, ctx_flexeuN(1) },
+		{ EU_PERF_CNTL2, ctx_flexeuN(2) },
+		{ EU_PERF_CNTL3, ctx_flexeuN(3) },
+		{ EU_PERF_CNTL4, ctx_flexeuN(4) },
+		{ EU_PERF_CNTL5, ctx_flexeuN(5) },
+		{ EU_PERF_CNTL6, ctx_flexeuN(6) },
+	};
+#undef ctx_flexeuN
+	int i;
+
+	regs[1].value =
+		(stream->period_exponent << GEN8_OA_TIMER_PERIOD_SHIFT) |
+		(stream->periodic ? GEN8_OA_TIMER_ENABLE : 0) |
+		GEN8_OA_COUNTER_RESUME;
+
+	for (i = 2; i < ARRAY_SIZE(regs); i++)
+		regs[i].value = oa_config_flex_reg(oa_config, regs[i].reg);
+
+	return oa_configure_all_contexts(stream, regs, ARRAY_SIZE(regs));
+}
+
 static int gen8_enable_metric_set(struct i915_perf_stream *stream)
 {
 	struct intel_uncore *uncore = stream->uncore;
@@ -2469,7 +2489,7 @@ static int gen12_enable_metric_set(struct i915_perf_stream *stream)
 	 * to make sure all slices/subslices are ON before writing to NOA
 	 * registers.
 	 */
-	ret = lrc_configure_all_contexts(stream, oa_config);
+	ret = gen12_configure_all_contexts(stream, oa_config);
 	if (ret)
 		return ret;
 
@@ -2479,8 +2499,7 @@ static int gen12_enable_metric_set(struct i915_perf_stream *stream)
 	 * requested this.
 	 */
 	if (stream->ctx) {
-		ret = gen12_emit_oar_config(stream->pinned_ctx,
-					    oa_config != NULL);
+		ret = gen12_configure_oar_context(stream, oa_config != NULL);
 		if (ret)
 			return ret;
 	}
@@ -2514,11 +2533,11 @@ static void gen12_disable_metric_set(struct i915_perf_stream *stream)
 	struct intel_uncore *uncore = stream->uncore;
 
 	/* Reset all contexts' slices/subslices configurations. */
-	lrc_configure_all_contexts(stream, NULL);
+	gen12_configure_all_contexts(stream, NULL);
 
 	/* disable the context save/restore or OAR counters */
 	if (stream->ctx)
-		gen12_emit_oar_config(stream->pinned_ctx, false);
+		gen12_configure_oar_context(stream, false);
 
 	/* Make sure we disable noa to save power. */
 	intel_uncore_rmw(uncore, RPM_CONFIG1, GEN10_GT_NOA_ENABLE, 0);
@@ -2860,7 +2879,11 @@ void i915_oa_init_reg_state(const struct intel_context *ce,
 		return;
 
 	stream = engine->i915->perf.exclusive_stream;
-	if (stream)
+	/*
+	 * For gen12, only CTX_R_PWR_CLK_STATE needs update, but the caller
+	 * is already doing that, so nothing to be done for gen12 here.
+	 */
+	if (stream && INTEL_GEN(stream->perf->i915) < 12)
 		gen8_update_reg_state_unlocked(ce, stream);
 }
 
Signature · 6 lines
2.20.1

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#3 lore raw

Re: [PATCH 2/2] drm/i915/perf: Configure OAR for specific context

In reply to Umesh Nerlige Ramappa
Quoted 56 lines · Umesh Nerlige Ramappa

Gen12 supports saving/restoring render counters per context. Apply OAR configuration only for the context that is passed in to perf.

v2:
- Fix OACTXCONTROL value to only stop/resume counters.
- Remove gen12_update_reg_state_unlocked as power state is already applied by the caller.

---
  drivers/gpu/drm/i915/i915_perf.c | 193 +++++++++++++++++--------------
  1 file changed, 108 insertions(+), 85 deletions(-)
drivers/gpu/drm/i915/i915_perf.c +6 -20
index 221c1090ae93..2f0be5fbef4b 100644
@@ -2078,20 +2078,12 @@ gen8_update_reg_state_unlocked(const struct intel_context *ce,
  	u32 *reg_state = ce->lrc_reg_state;
  	int i;
 
-	if (IS_GEN(stream->perf->i915, 12)) {
-		u32 format = stream->oa_buffer.format;
+	reg_state[ctx_oactxctrl + 1] =
+		(stream->period_exponent << GEN8_OA_TIMER_PERIOD_SHIFT) |
+		(stream->periodic ? GEN8_OA_TIMER_ENABLE : 0) |
+		GEN8_OA_COUNTER_RESUME;
 
-		reg_state[ctx_oactxctrl + 1] =
-			(format << GEN12_OAR_OACONTROL_COUNTER_FORMAT_SHIFT) |
-			(stream->oa_config ? GEN12_OAR_OACONTROL_COUNTER_ENABLE : 0);
-	} else {
-		reg_state[ctx_oactxctrl + 1] =
-			(stream->period_exponent << GEN8_OA_TIMER_PERIOD_SHIFT) |
-			(stream->periodic ? GEN8_OA_TIMER_ENABLE : 0) |
-			GEN8_OA_COUNTER_RESUME;
-	}
-
-	for (i = 0; !!ctx_flexeu0 && i < ARRAY_SIZE(flex_regs); i++)
+	for (i = 0; i < ARRAY_SIZE(flex_regs); i++)
  		reg_state[ctx_flexeu0 + i * 2 + 1] =
  			oa_config_flex_reg(stream->oa_config, flex_regs[i]);
 
@@ -2224,34 +2216,49 @@ static int gen8_configure_context(struct i915_gem_context *ctx,
  	return err;
  }
 
-static int gen12_emit_oar_config(struct intel_context *ce, bool enable)
+static int gen12_configure_oar_context(struct i915_perf_stream *stream, bool enable)
  {
-	struct i915_request *rq;
-	u32 *cs;
-	int err = 0;
-
-	rq = i915_request_create(ce);
-	if (IS_ERR(rq))

drivers/gpu/drm/i915/i915_perf.c static int gen8_configure_context(struct i915_gem_context *ctx, · line 2229

Quoted 10 lines
drivers/gpu/drm/i915/i915_perf.c +9 -1
-		return PTR_ERR(rq);
+	int err;
+	struct intel_context *ce = stream->pinned_ctx;
+	struct flex regs_context[] = {
+		{
+			GEN8_OACTXCONTROL,
+			stream->perf->ctx_oactxctrl_offset + 1,
+			enable ? GEN8_OA_COUNTER_RESUME : 0,
+		},
+	};

When do we configure the Flex registers?

Quoted 19 lines
drivers/gpu/drm/i915/i915_perf.c +12 -5
+	struct flex regs_lri[] = {
+		{
+			GEN12_OAR_OACONTROL,
+		},
+		{
+			RING_CONTEXT_CONTROL(ce->engine->mmio_base),
+		},
+	};
+	u32 format = stream->oa_buffer.format;
 
-	cs = intel_ring_begin(rq, 4);
-	if (IS_ERR(cs)) {
-		err = PTR_ERR(cs);
-		goto out;
-	}
+	regs_lri[0].value =
+		(format << GEN12_OAR_OACONTROL_COUNTER_FORMAT_SHIFT) |
+		(enable ? GEN12_OAR_OACONTROL_COUNTER_ENABLE : 0);
 

drivers/gpu/drm/i915/i915_perf.c static int gen8_configure_context(struct i915_gem_context *ctx, · line 2248

Quoted 10 lines
drivers/gpu/drm/i915/i915_perf.c +5 -5
-	*cs++ = MI_LOAD_REGISTER_IMM(1);
-	*cs++ = i915_mmio_reg_offset(RING_CONTEXT_CONTROL(ce->engine->mmio_base));
-	*cs++ = _MASKED_FIELD(GEN12_CTX_CTRL_OAR_CONTEXT_ENABLE,
-			      enable ? GEN12_CTX_CTRL_OAR_CONTEXT_ENABLE : 0);
-	*cs++ = MI_NOOP;
+	regs_lri[1].value =
+		_MASKED_FIELD(GEN12_CTX_CTRL_OAR_CONTEXT_ENABLE,
+			      enable ?
+			      GEN12_CTX_CTRL_OAR_CONTEXT_ENABLE :
+			      0);

Can't we put those values in the array above?

Quoted 164 lines
drivers/gpu/drm/i915/i915_perf.c +68 -49
-	intel_ring_advance(rq, cs);
+	/* Modify the context image of pinned context with regs_context*/
+	err = intel_context_lock_pinned(ce);
+	if (err)
+		return err;
 
-out:
-	i915_request_add(rq);
+	err = gen8_modify_context(ce, regs_context, ARRAY_SIZE(regs_context));
+	intel_context_unlock_pinned(ce);
+	if (err)
+		return err;
 
-	return err;
+	/* Apply regs_lri using LRI with pinned context */
+	return gen8_modify_self(ce, regs_lri, ARRAY_SIZE(regs_lri));
  }
 
  /*
@@ -2277,53 +2284,16 @@ static int gen12_emit_oar_config(struct intel_context *ce, bool enable)
   *   per-context OA state.
   *
   * Note: it's only the RCS/Render context that has any OA state.
+ * Note: the first flex register passed must always be R_PWR_CLK_STATE
   */
-static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
-				      const struct i915_oa_config *oa_config)
+static int oa_configure_all_contexts(struct i915_perf_stream *stream,
+				     struct flex *regs,
+				     size_t num_regs)
  {
  	struct drm_i915_private *i915 = stream->perf->i915;
-	/* The MMIO offsets for Flex EU registers aren't contiguous */
-	const u32 ctx_flexeu0 = stream->perf->ctx_flexeu0_offset;
-#define ctx_flexeuN(N) (ctx_flexeu0 + 2 * (N) + 1)
-	struct flex regs[] = {
-		{
-			GEN8_R_PWR_CLK_STATE,
-			CTX_R_PWR_CLK_STATE,
-		},
-		{
-			IS_GEN(i915, 12) ?
-			GEN12_OAR_OACONTROL : GEN8_OACTXCONTROL,
-			stream->perf->ctx_oactxctrl_offset + 1,
-		},
-		{ EU_PERF_CNTL0, ctx_flexeuN(0) },
-		{ EU_PERF_CNTL1, ctx_flexeuN(1) },
-		{ EU_PERF_CNTL2, ctx_flexeuN(2) },
-		{ EU_PERF_CNTL3, ctx_flexeuN(3) },
-		{ EU_PERF_CNTL4, ctx_flexeuN(4) },
-		{ EU_PERF_CNTL5, ctx_flexeuN(5) },
-		{ EU_PERF_CNTL6, ctx_flexeuN(6) },
-	};
-#undef ctx_flexeuN
  	struct intel_engine_cs *engine;
  	struct i915_gem_context *ctx, *cn;
-	size_t array_size = IS_GEN(i915, 12) ? 2 : ARRAY_SIZE(regs);
-	int i, err;
-
-	if (IS_GEN(i915, 12)) {
-		u32 format = stream->oa_buffer.format;
-
-		regs[1].value =
-			(format << GEN12_OAR_OACONTROL_COUNTER_FORMAT_SHIFT) |
-			(oa_config ? GEN12_OAR_OACONTROL_COUNTER_ENABLE : 0);
-	} else {
-		regs[1].value =
-			(stream->period_exponent << GEN8_OA_TIMER_PERIOD_SHIFT) |
-			(stream->periodic ? GEN8_OA_TIMER_ENABLE : 0) |
-			GEN8_OA_COUNTER_RESUME;
-	}
-
-	for (i = 2; !!ctx_flexeu0 && i < array_size; i++)
-		regs[i].value = oa_config_flex_reg(oa_config, regs[i].reg);
+	int err;
 
  	lockdep_assert_held(&stream->perf->lock);
 
@@ -2353,7 +2323,7 @@ static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
 
  		spin_unlock(&i915->gem.contexts.lock);
 
-		err = gen8_configure_context(ctx, regs, array_size);
+		err = gen8_configure_context(ctx, regs, num_regs);
  		if (err) {
  			i915_gem_context_put(ctx);
  			return err;
@@ -2378,7 +2348,7 @@ static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
 
  		regs[0].value = intel_sseu_make_rpcs(i915, &ce->sseu);
 
-		err = gen8_modify_self(ce, regs, array_size);
+		err = gen8_modify_self(ce, regs, num_regs);
  		if (err)
  			return err;
  	}
@@ -2386,6 +2356,56 @@ static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
  	return 0;
  }
 
+static int gen12_configure_all_contexts(struct i915_perf_stream *stream,
+					const struct i915_oa_config *oa_config)
+{
+	struct flex regs[] = {
+		{
+			GEN8_R_PWR_CLK_STATE,
+			CTX_R_PWR_CLK_STATE,
+		},
+	};
+
+	return oa_configure_all_contexts(stream, regs, ARRAY_SIZE(regs));
+}
+
+static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
+				      const struct i915_oa_config *oa_config)
+{
+	/* The MMIO offsets for Flex EU registers aren't contiguous */
+	const u32 ctx_flexeu0 = stream->perf->ctx_flexeu0_offset;
+#define ctx_flexeuN(N) (ctx_flexeu0 + 2 * (N) + 1)
+	struct flex regs[] = {
+		{
+			GEN8_R_PWR_CLK_STATE,
+			CTX_R_PWR_CLK_STATE,
+		},
+		{
+			GEN8_OACTXCONTROL,
+			stream->perf->ctx_oactxctrl_offset + 1,
+		},
+		{ EU_PERF_CNTL0, ctx_flexeuN(0) },
+		{ EU_PERF_CNTL1, ctx_flexeuN(1) },
+		{ EU_PERF_CNTL2, ctx_flexeuN(2) },
+		{ EU_PERF_CNTL3, ctx_flexeuN(3) },
+		{ EU_PERF_CNTL4, ctx_flexeuN(4) },
+		{ EU_PERF_CNTL5, ctx_flexeuN(5) },
+		{ EU_PERF_CNTL6, ctx_flexeuN(6) },
+	};
+#undef ctx_flexeuN
+	int i;
+
+	regs[1].value =
+		(stream->period_exponent << GEN8_OA_TIMER_PERIOD_SHIFT) |
+		(stream->periodic ? GEN8_OA_TIMER_ENABLE : 0) |
+		GEN8_OA_COUNTER_RESUME;
+
+	for (i = 2; i < ARRAY_SIZE(regs); i++)
+		regs[i].value = oa_config_flex_reg(oa_config, regs[i].reg);
+
+	return oa_configure_all_contexts(stream, regs, ARRAY_SIZE(regs));
+}
+
  static int gen8_enable_metric_set(struct i915_perf_stream *stream)
  {
  	struct intel_uncore *uncore = stream->uncore;
@@ -2469,7 +2489,7 @@ static int gen12_enable_metric_set(struct i915_perf_stream *stream)
  	 * to make sure all slices/subslices are ON before writing to NOA
  	 * registers.
  	 */
-	ret = lrc_configure_all_contexts(stream, oa_config);
+	ret = gen12_configure_all_contexts(stream, oa_config);
  	if (ret)
  		return ret;
 

drivers/gpu/drm/i915/i915_perf.c static int gen12_enable_metric_set(struct i915_perf_stream *stream) · line 2502

Quoted 6 lines
drivers/gpu/drm/i915/i915_perf.c +1 -2
@@ -2479,8 +2499,7 @@ static int gen12_enable_metric_set(struct i915_perf_stream *stream)
  	 * requested this.
  	 */
  	if (stream->ctx) {
-		ret = gen12_emit_oar_config(stream->pinned_ctx,
-					    oa_config != NULL);
+		ret = gen12_configure_oar_context(stream, oa_config != NULL);

I think you can assume oa_config is going to be != NULL in the enable_metric_set vfunc.

Quoted 20 lines
drivers/gpu/drm/i915/i915_perf.c +2 -2
  		if (ret)
  			return ret;
  	}
@@ -2514,11 +2533,11 @@ static void gen12_disable_metric_set(struct i915_perf_stream *stream)
  	struct intel_uncore *uncore = stream->uncore;
 
  	/* Reset all contexts' slices/subslices configurations. */
-	lrc_configure_all_contexts(stream, NULL);
+	gen12_configure_all_contexts(stream, NULL);
 
  	/* disable the context save/restore or OAR counters */
  	if (stream->ctx)
-		gen12_emit_oar_config(stream->pinned_ctx, false);
+		gen12_configure_oar_context(stream, false);
 
  	/* Make sure we disable noa to save power. */
  	intel_uncore_rmw(uncore, RPM_CONFIG1, GEN10_GT_NOA_ENABLE, 0);
@@ -2860,7 +2879,11 @@ void i915_oa_init_reg_state(const struct intel_context *ce,
  		return;
 

drivers/gpu/drm/i915/i915_perf.c void i915_oa_init_reg_state(const struct intel_context *ce, · line 2889

Quoted 10 lines
drivers/gpu/drm/i915/i915_perf.c +5 -1
  	stream = engine->i915->perf.exclusive_stream;
-	if (stream)
+	/*
+	 * For gen12, only CTX_R_PWR_CLK_STATE needs update, but the caller
+	 * is already doing that, so nothing to be done for gen12 here.
+	 */
+	if (stream && INTEL_GEN(stream->perf->i915) < 12)
  		gen8_update_reg_state_unlocked(ce, stream);
  }
 

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#4 lore raw

Re: [PATCH 2/2] drm/i915/perf: Configure OAR for specific context

Quoted 58 lines · Lionel Landwerlin
Quoted 57 lines · Umesh Nerlige Ramappa

Gen12 supports saving/restoring render counters per context. Apply OAR configuration only for the context that is passed in to perf.

v2:
- Fix OACTXCONTROL value to only stop/resume counters.
- Remove gen12_update_reg_state_unlocked as power state is already applied by the caller.

---
 drivers/gpu/drm/i915/i915_perf.c | 193 +++++++++++++++++--------------
 1 file changed, 108 insertions(+), 85 deletions(-)
drivers/gpu/drm/i915/i915_perf.c +10 -21
index 221c1090ae93..2f0be5fbef4b 100644
@@ -2078,20 +2078,12 @@ gen8_update_reg_state_unlocked(const struct intel_context *ce,
 	u32 *reg_state = ce->lrc_reg_state;
 	int i;
-	if (IS_GEN(stream->perf->i915, 12)) {
-		u32 format = stream->oa_buffer.format;
+	reg_state[ctx_oactxctrl + 1] =
+		(stream->period_exponent << GEN8_OA_TIMER_PERIOD_SHIFT) |
+		(stream->periodic ? GEN8_OA_TIMER_ENABLE : 0) |
+		GEN8_OA_COUNTER_RESUME;
-		reg_state[ctx_oactxctrl + 1] =
-			(format << GEN12_OAR_OACONTROL_COUNTER_FORMAT_SHIFT) |
-			(stream->oa_config ? GEN12_OAR_OACONTROL_COUNTER_ENABLE : 0);
-	} else {
-		reg_state[ctx_oactxctrl + 1] =
-			(stream->period_exponent << GEN8_OA_TIMER_PERIOD_SHIFT) |
-			(stream->periodic ? GEN8_OA_TIMER_ENABLE : 0) |
-			GEN8_OA_COUNTER_RESUME;
-	}
-
-	for (i = 0; !!ctx_flexeu0 && i < ARRAY_SIZE(flex_regs); i++)
+	for (i = 0; i < ARRAY_SIZE(flex_regs); i++)
 		reg_state[ctx_flexeu0 + i * 2 + 1] =
 			oa_config_flex_reg(stream->oa_config, flex_regs[i]);
@@ -2224,34 +2216,49 @@ static int gen8_configure_context(struct i915_gem_context *ctx,
 	return err;
 }
-static int gen12_emit_oar_config(struct intel_context *ce, bool enable)
+static int gen12_configure_oar_context(struct i915_perf_stream *stream, bool enable)
 {
-	struct i915_request *rq;
-	u32 *cs;
-	int err = 0;
-
-	rq = i915_request_create(ce);
-	if (IS_ERR(rq))
-		return PTR_ERR(rq);
+	int err;
+	struct intel_context *ce = stream->pinned_ctx;
+	struct flex regs_context[] = {
+		{
Quoted 8 lines

drivers/gpu/drm/i915/i915_perf.c static int gen8_configure_context(struct i915_gem_context *ctx, · line 2229

Quoted 5 lines
drivers/gpu/drm/i915/i915_perf.c +5 -0
+			GEN8_OACTXCONTROL,
+			stream->perf->ctx_oactxctrl_offset + 1,
+			enable ? GEN8_OA_COUNTER_RESUME : 0,
+		},
+	};

When do we configure the Flex registers?

I don't see flex registers in the context image dump or in the spec, so I guess we would only configure them if they are part of the metric set.

Will post a new version with below comments.

Thanks,
Umesh

Quoted 214 lines
Quoted 19 lines
drivers/gpu/drm/i915/i915_perf.c +12 -5
+	struct flex regs_lri[] = {
+		{
+			GEN12_OAR_OACONTROL,
+		},
+		{
+			RING_CONTEXT_CONTROL(ce->engine->mmio_base),
+		},
+	};
+	u32 format = stream->oa_buffer.format;
-	cs = intel_ring_begin(rq, 4);
-	if (IS_ERR(cs)) {
-		err = PTR_ERR(cs);
-		goto out;
-	}
+	regs_lri[0].value =
+		(format << GEN12_OAR_OACONTROL_COUNTER_FORMAT_SHIFT) |
+		(enable ? GEN12_OAR_OACONTROL_COUNTER_ENABLE : 0);

drivers/gpu/drm/i915/i915_perf.c static int gen8_configure_context(struct i915_gem_context *ctx, · line 2248

Quoted 10 lines
drivers/gpu/drm/i915/i915_perf.c +5 -5
-	*cs++ = MI_LOAD_REGISTER_IMM(1);
-	*cs++ = i915_mmio_reg_offset(RING_CONTEXT_CONTROL(ce->engine->mmio_base));
-	*cs++ = _MASKED_FIELD(GEN12_CTX_CTRL_OAR_CONTEXT_ENABLE,
-			      enable ? GEN12_CTX_CTRL_OAR_CONTEXT_ENABLE : 0);
-	*cs++ = MI_NOOP;
+	regs_lri[1].value =
+		_MASKED_FIELD(GEN12_CTX_CTRL_OAR_CONTEXT_ENABLE,
+			      enable ?
+			      GEN12_CTX_CTRL_OAR_CONTEXT_ENABLE :
+			      0);

Can't we put those values in the array above?

Quoted 152 lines
drivers/gpu/drm/i915/i915_perf.c +68 -49
-	intel_ring_advance(rq, cs);
+	/* Modify the context image of pinned context with regs_context*/
+	err = intel_context_lock_pinned(ce);
+	if (err)
+		return err;
-out:
-	i915_request_add(rq);
+	err = gen8_modify_context(ce, regs_context, ARRAY_SIZE(regs_context));
+	intel_context_unlock_pinned(ce);
+	if (err)
+		return err;
-	return err;
+	/* Apply regs_lri using LRI with pinned context */
+	return gen8_modify_self(ce, regs_lri, ARRAY_SIZE(regs_lri));
 }
 /*
@@ -2277,53 +2284,16 @@ static int gen12_emit_oar_config(struct intel_context *ce, bool enable)
  *   per-context OA state.
  *
  * Note: it's only the RCS/Render context that has any OA state.
+ * Note: the first flex register passed must always be R_PWR_CLK_STATE
  */
-static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
-				      const struct i915_oa_config *oa_config)
+static int oa_configure_all_contexts(struct i915_perf_stream *stream,
+				     struct flex *regs,
+				     size_t num_regs)
 {
 	struct drm_i915_private *i915 = stream->perf->i915;
-	/* The MMIO offsets for Flex EU registers aren't contiguous */
-	const u32 ctx_flexeu0 = stream->perf->ctx_flexeu0_offset;
-#define ctx_flexeuN(N) (ctx_flexeu0 + 2 * (N) + 1)
-	struct flex regs[] = {
-		{
-			GEN8_R_PWR_CLK_STATE,
-			CTX_R_PWR_CLK_STATE,
-		},
-		{
-			IS_GEN(i915, 12) ?
-			GEN12_OAR_OACONTROL : GEN8_OACTXCONTROL,
-			stream->perf->ctx_oactxctrl_offset + 1,
-		},
-		{ EU_PERF_CNTL0, ctx_flexeuN(0) },
-		{ EU_PERF_CNTL1, ctx_flexeuN(1) },
-		{ EU_PERF_CNTL2, ctx_flexeuN(2) },
-		{ EU_PERF_CNTL3, ctx_flexeuN(3) },
-		{ EU_PERF_CNTL4, ctx_flexeuN(4) },
-		{ EU_PERF_CNTL5, ctx_flexeuN(5) },
-		{ EU_PERF_CNTL6, ctx_flexeuN(6) },
-	};
-#undef ctx_flexeuN
 	struct intel_engine_cs *engine;
 	struct i915_gem_context *ctx, *cn;
-	size_t array_size = IS_GEN(i915, 12) ? 2 : ARRAY_SIZE(regs);
-	int i, err;
-
-	if (IS_GEN(i915, 12)) {
-		u32 format = stream->oa_buffer.format;
-
-		regs[1].value =
-			(format << GEN12_OAR_OACONTROL_COUNTER_FORMAT_SHIFT) |
-			(oa_config ? GEN12_OAR_OACONTROL_COUNTER_ENABLE : 0);
-	} else {
-		regs[1].value =
-			(stream->period_exponent << GEN8_OA_TIMER_PERIOD_SHIFT) |
-			(stream->periodic ? GEN8_OA_TIMER_ENABLE : 0) |
-			GEN8_OA_COUNTER_RESUME;
-	}
-
-	for (i = 2; !!ctx_flexeu0 && i < array_size; i++)
-		regs[i].value = oa_config_flex_reg(oa_config, regs[i].reg);
+	int err;
 	lockdep_assert_held(&stream->perf->lock);
@@ -2353,7 +2323,7 @@ static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
 		spin_unlock(&i915->gem.contexts.lock);
-		err = gen8_configure_context(ctx, regs, array_size);
+		err = gen8_configure_context(ctx, regs, num_regs);
 		if (err) {
 			i915_gem_context_put(ctx);
 			return err;
@@ -2378,7 +2348,7 @@ static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
 		regs[0].value = intel_sseu_make_rpcs(i915, &ce->sseu);
-		err = gen8_modify_self(ce, regs, array_size);
+		err = gen8_modify_self(ce, regs, num_regs);
 		if (err)
 			return err;
 	}
@@ -2386,6 +2356,56 @@ static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
 	return 0;
 }
+static int gen12_configure_all_contexts(struct i915_perf_stream *stream,
+					const struct i915_oa_config *oa_config)
+{
+	struct flex regs[] = {
+		{
+			GEN8_R_PWR_CLK_STATE,
+			CTX_R_PWR_CLK_STATE,
+		},
+	};
+
+	return oa_configure_all_contexts(stream, regs, ARRAY_SIZE(regs));
+}
+
+static int lrc_configure_all_contexts(struct i915_perf_stream *stream,
+				      const struct i915_oa_config *oa_config)
+{
+	/* The MMIO offsets for Flex EU registers aren't contiguous */
+	const u32 ctx_flexeu0 = stream->perf->ctx_flexeu0_offset;
+#define ctx_flexeuN(N) (ctx_flexeu0 + 2 * (N) + 1)
+	struct flex regs[] = {
+		{
+			GEN8_R_PWR_CLK_STATE,
+			CTX_R_PWR_CLK_STATE,
+		},
+		{
+			GEN8_OACTXCONTROL,
+			stream->perf->ctx_oactxctrl_offset + 1,
+		},
+		{ EU_PERF_CNTL0, ctx_flexeuN(0) },
+		{ EU_PERF_CNTL1, ctx_flexeuN(1) },
+		{ EU_PERF_CNTL2, ctx_flexeuN(2) },
+		{ EU_PERF_CNTL3, ctx_flexeuN(3) },
+		{ EU_PERF_CNTL4, ctx_flexeuN(4) },
+		{ EU_PERF_CNTL5, ctx_flexeuN(5) },
+		{ EU_PERF_CNTL6, ctx_flexeuN(6) },
+	};
+#undef ctx_flexeuN
+	int i;
+
+	regs[1].value =
+		(stream->period_exponent << GEN8_OA_TIMER_PERIOD_SHIFT) |
+		(stream->periodic ? GEN8_OA_TIMER_ENABLE : 0) |
+		GEN8_OA_COUNTER_RESUME;
+
+	for (i = 2; i < ARRAY_SIZE(regs); i++)
+		regs[i].value = oa_config_flex_reg(oa_config, regs[i].reg);
+
+	return oa_configure_all_contexts(stream, regs, ARRAY_SIZE(regs));
+}
+
 static int gen8_enable_metric_set(struct i915_perf_stream *stream)
 {
 	struct intel_uncore *uncore = stream->uncore;
@@ -2469,7 +2489,7 @@ static int gen12_enable_metric_set(struct i915_perf_stream *stream)
 	 * to make sure all slices/subslices are ON before writing to NOA
 	 * registers.
 	 */
-	ret = lrc_configure_all_contexts(stream, oa_config);
+	ret = gen12_configure_all_contexts(stream, oa_config);
 	if (ret)
 		return ret;

drivers/gpu/drm/i915/i915_perf.c static int gen12_enable_metric_set(struct i915_perf_stream *stream) · line 2502

Quoted 6 lines
drivers/gpu/drm/i915/i915_perf.c +1 -2
@@ -2479,8 +2499,7 @@ static int gen12_enable_metric_set(struct i915_perf_stream *stream)
 	 * requested this.
 	 */
 	if (stream->ctx) {
-		ret = gen12_emit_oar_config(stream->pinned_ctx,
-					    oa_config != NULL);
+		ret = gen12_configure_oar_context(stream, oa_config != NULL);

I think you can assume oa_config is going to be != NULL in the enable_metric_set vfunc.

Quoted 17 lines
drivers/gpu/drm/i915/i915_perf.c +2 -2
 		if (ret)
 			return ret;
 	}
@@ -2514,11 +2533,11 @@ static void gen12_disable_metric_set(struct i915_perf_stream *stream)
 	struct intel_uncore *uncore = stream->uncore;
 	/* Reset all contexts' slices/subslices configurations. */
-	lrc_configure_all_contexts(stream, NULL);
+	gen12_configure_all_contexts(stream, NULL);
 	/* disable the context save/restore or OAR counters */
 	if (stream->ctx)
-		gen12_emit_oar_config(stream->pinned_ctx, false);
+		gen12_configure_oar_context(stream, false);
 	/* Make sure we disable noa to save power. */
 	intel_uncore_rmw(uncore, RPM_CONFIG1, GEN10_GT_NOA_ENABLE, 0);
@@ -2860,7 +2879,11 @@ void i915_oa_init_reg_state(const struct intel_context *ce,
 		return;
 	stream = engine->i915->perf.exclusive_stream;
Quoted 10 lines

drivers/gpu/drm/i915/i915_perf.c void i915_oa_init_reg_state(const struct intel_context *ce, · line 2888

Quoted 8 lines
drivers/gpu/drm/i915/i915_perf.c +5 -1
-	if (stream)
+	/*
+	 * For gen12, only CTX_R_PWR_CLK_STATE needs update, but the caller
+	 * is already doing that, so nothing to be done for gen12 here.
+	 */
+	if (stream && INTEL_GEN(stream->perf->i915) < 12)
 		gen8_update_reg_state_unlocked(ce, stream);
 }

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#5 lore raw

✗ Fi.CI.CHECKPATCH: warning for series starting with [1/2] drm/i915/perf: Allow non-privileged access when OA buffer is not sampled

In reply to Umesh Nerlige Ramappa
== Series Details ==

== Series Details ==

Series: series starting with [1/2] drm/i915/perf: Allow non-privileged access when OA buffer is not sampled
URL : https://patchwork.freedesktop.org/series/69491/
State : warning

== Summary ==

$ dim checkpatch origin/drm-tip
61f9bbf85888 drm/i915/perf: Allow non-privileged access when OA buffer is not sampled
e58eb681f814 drm/i915/perf: Configure OAR for specific context
-:267: CHECK:COMPARISON_TO_NULL: Comparison to NULL could be written "oa_config"
#267: FILE: drivers/gpu/drm/i915/i915_perf.c:2502:
+		ret = gen12_configure_oar_context(stream, oa_config != NULL);

total: 0 errors, 0 warnings, 1 checks, 268 lines checked

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#6 lore raw

✗ Fi.CI.BAT: failure for series starting with [1/2] drm/i915/perf: Allow non-privileged access when OA buffer is not sampled

In reply to Umesh Nerlige Ramappa
== Series Details ==

== Series Details ==

Series: series starting with [1/2] drm/i915/perf: Allow non-privileged access when OA buffer is not sampled
URL : https://patchwork.freedesktop.org/series/69491/
State : failure

== Summary ==

CI Bug Log - changes from CI_DRM_7346 -> Patchwork_15267
====================================================

Summary
-------

**FAILURE**

Serious unknown changes coming with Patchwork_15267 absolutely need to be verified manually.

If you think the reported changes have nothing to do with the changes introduced in Patchwork_15267, please notify your bug team to allow them to document this new failure mode, which will reduce false positives in CI.

External URL: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_15267/index.html

Possible new issues
-------------------

Here are the unknown changes that may have been introduced in Patchwork_15267:

### IGT changes ###

#### Possible regressions ####

* igt@i915_selftest@live_sanitycheck:
- fi-skl-lmem: NOTRUN -> [DMESG-WARN][1]
[1]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_15267/fi-skl-lmem/igt@i915_selftest@live_sanitycheck.html

Known issues
------------

Here are the changes found in Patchwork_15267 that come from known issues:

### IGT changes ###

#### Issues hit ####

* igt@gem_exec_suspend@basic-s3:
- fi-icl-dsi: [PASS][2] -> [INCOMPLETE][3] ([fdo#107713])
[2]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_7346/fi-icl-dsi/igt@[email protected]
[3]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_15267/fi-icl-dsi/igt@[email protected]

#### Possible fixes ####

* igt@i915_module_load@reload-with-fault-injection:
- {fi-kbl-7560u}: [INCOMPLETE][4] ([fdo#109964] / [fdo#110343]) -> [PASS][5]
[4]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_7346/fi-kbl-7560u/igt@[email protected]
[5]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_15267/fi-kbl-7560u/igt@[email protected]

* igt@i915_pm_rpm@module-reload:
- fi-skl-lmem: [DMESG-WARN][6] ([fdo#112261]) -> [PASS][7]
[6]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_7346/fi-skl-lmem/igt@[email protected]
[7]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_15267/fi-skl-lmem/igt@[email protected]

* igt@i915_selftest@live_gem_contexts:
- fi-bsw-n3050: [INCOMPLETE][8] ([fdo# 111542]) -> [PASS][9]
[8]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_7346/fi-bsw-n3050/igt@i915_selftest@live_gem_contexts.html
[9]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_15267/fi-bsw-n3050/igt@i915_selftest@live_gem_contexts.html

* igt@kms_chamelium@hdmi-edid-read:
- fi-kbl-7500u: [FAIL][10] ([fdo#109483]) -> [PASS][11]
[10]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_7346/fi-kbl-7500u/igt@[email protected]
[11]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_15267/fi-kbl-7500u/igt@[email protected]

{name}: This element is suppressed. This means it is ignored when computing
the status of the difference (SUCCESS, WARNING, or FAILURE).

[fdo# 111542]: https://bugs.freedesktop.org/show_bug.cgi?id= 111542 [fdo#107713]: https://bugs.freedesktop.org/show_bug.cgi?id=107713
[fdo#109483]: https://bugs.freedesktop.org/show_bug.cgi?id=109483
[fdo#109964]: https://bugs.freedesktop.org/show_bug.cgi?id=109964
[fdo#110343]: https://bugs.freedesktop.org/show_bug.cgi?id=110343
[fdo#112261]: https://bugs.freedesktop.org/show_bug.cgi?id=112261

Participating hosts (49 -> 45)
------------------------------

Missing (4): fi-byt-clapper fi-byt-squawks fi-bsw-cyan fi-hsw-4200u

Build changes
-------------

* CI: CI-20190529 -> None
* Linux: CI_DRM_7346 -> Patchwork_15267

CI-20190529: 20190529
CI_DRM_7346: 78fd2497a0a373d6a68c047d759851e0a895f112 @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_5284: b28d801e4c010942d05821fada6dd9d2d67d6e4d @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
Patchwork_15267: e58eb681f814760556613c83fae52e3f1bd27e26 @ git://anongit.freedesktop.org/gfx-ci/linux

== Linux commits ==

e58eb681f814 drm/i915/perf: Configure OAR for specific context 61f9bbf85888 drm/i915/perf: Allow non-privileged access when OA buffer is not sampled

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_15267/index.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx