diff mbox series

[v15,10/13] drm/i915/perf: execute OA configuration from command stream

Message ID 20190906093256.26198-11-lionel.g.landwerlin@intel.com (mailing list archive)
State New, archived
Headers show
Series drm/i915: Vulkan performance query support | expand

Commit Message

Lionel Landwerlin Sept. 6, 2019, 9:32 a.m. UTC
We haven't run into issues with programming the global OA/NOA
registers configuration from CPU so far, but HW engineers actually
recommend doing this from the command streamer. On TGL in particular
one of the clock domain in which some of that programming goes might
not be powered when we poke things from the CPU.

Since we have a command buffer prepared for the execbuffer side of
things, we can reuse that approach here too.

This also allows us to significantly reduce the amount of time we hold
the main lock.

v2: Drop the global lock as much as possible

v3: Take global lock to pin global

v4: Create i915 request in emit_oa_config() to avoid deadlocks (Lionel)

v5: Move locking to the stream (Lionel)

v6: Move active reconfiguration request into i915_perf_stream (Lionel)

v7: Pin VMA outside request creation (Chris)
    Lock VMA before move to active (Chris)

v8: Fix double free on stream->initial_oa_config_bo (Lionel)
    Don't allow interruption when waiting on active config request
    (Lionel)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
---
 drivers/gpu/drm/i915/i915_perf.c       | 170 ++++++++++++++++---------
 drivers/gpu/drm/i915/i915_perf_types.h |  13 +-
 2 files changed, 122 insertions(+), 61 deletions(-)
diff mbox series

Patch

diff --git a/drivers/gpu/drm/i915/i915_perf.c b/drivers/gpu/drm/i915/i915_perf.c
index f2b778d84b52..8e3532518139 100644
--- a/drivers/gpu/drm/i915/i915_perf.c
+++ b/drivers/gpu/drm/i915/i915_perf.c
@@ -1558,18 +1558,23 @@  free_oa_configs(struct i915_perf_stream *stream)
 static void i915_oa_stream_destroy(struct i915_perf_stream *stream)
 {
 	struct drm_i915_private *dev_priv = stream->dev_priv;
+	int err;
 
 	BUG_ON(stream != dev_priv->perf.exclusive_stream);
 
-	/*
-	 * Unset exclusive_stream first, it will be checked while disabling
-	 * the metric set on gen8+.
-	 */
 	mutex_lock(&dev_priv->drm.struct_mutex);
-	dev_priv->perf.exclusive_stream = NULL;
+	mutex_lock(&stream->config_mutex);
 	dev_priv->perf.ops.disable_metric_set(stream);
+	err = i915_active_request_retire(&stream->active_config_rq, 0,
+					 &stream->config_mutex);
+	mutex_unlock(&stream->config_mutex);
+	dev_priv->perf.exclusive_stream = NULL;
 	mutex_unlock(&dev_priv->drm.struct_mutex);
 
+	if (err)
+		DRM_ERROR("Failed to disable perf stream\n");
+
+
 	free_oa_buffer(stream);
 	free_noa_wait(stream);
 
@@ -1795,6 +1800,10 @@  static int alloc_noa_wait(struct i915_perf_stream *stream)
 		return PTR_ERR(bo);
 	}
 
+	ret = i915_mutex_lock_interruptible(&i915->drm);
+	if (ret)
+		goto err_unref;
+
 	/*
 	 * We pin in GGTT because we jump into this buffer now because
 	 * multiple OA config BOs will have a jump to this address and it
@@ -1802,10 +1811,13 @@  static int alloc_noa_wait(struct i915_perf_stream *stream)
 	 */
 	vma = i915_gem_object_ggtt_pin(bo, NULL, 0, 4096, 0);
 	if (IS_ERR(vma)) {
+		mutex_unlock(&i915->drm.struct_mutex);
 		ret = PTR_ERR(vma);
 		goto err_unref;
 	}
 
+	mutex_unlock(&i915->drm.struct_mutex);
+
 	batch = cs = i915_gem_object_pin_map(bo, I915_MAP_WB);
 	if (IS_ERR(batch)) {
 		ret = PTR_ERR(batch);
@@ -1939,7 +1951,9 @@  static int alloc_noa_wait(struct i915_perf_stream *stream)
 	return 0;
 
 err_unpin:
-	__i915_vma_unpin(vma);
+	mutex_lock(&i915->drm.struct_mutex);
+	i915_vma_unpin_and_release(&vma, 0);
+	mutex_unlock(&i915->drm.struct_mutex);
 
 err_unref:
 	i915_gem_object_put(bo);
@@ -1947,50 +1961,73 @@  static int alloc_noa_wait(struct i915_perf_stream *stream)
 	return ret;
 }
 
-static void config_oa_regs(struct drm_i915_private *dev_priv,
-			   const struct i915_oa_reg *regs,
-			   u32 n_regs)
+static int emit_oa_config(struct drm_i915_private *i915,
+			  struct i915_perf_stream *stream)
 {
-	u32 i;
+	struct i915_request *rq;
+	struct i915_vma *vma;
+	u32 *cs;
+	int err;
 
-	for (i = 0; i < n_regs; i++) {
-		const struct i915_oa_reg *reg = regs + i;
+	lockdep_assert_held(&stream->config_mutex);
+
+	vma = i915_vma_instance(stream->initial_oa_config_bo,
+				&stream->engine->gt->ggtt->vm, NULL);
+	if (unlikely(IS_ERR(vma)))
+		return PTR_ERR(vma);
 
-		I915_WRITE(reg->addr, reg->value);
+	err = i915_vma_pin(vma, 0, 0, PIN_GLOBAL);
+	if (err)
+		goto err_vma_unpin;
+
+	rq = i915_request_create(stream->engine->kernel_context);
+	if (IS_ERR(rq)) {
+		err = PTR_ERR(rq);
+		goto err_add_request;
 	}
-}
 
-static void delay_after_mux(void)
-{
-	/*
-	 * It apparently takes a fairly long time for a new MUX
-	 * configuration to be be applied after these register writes.
-	 * This delay duration was derived empirically based on the
-	 * render_basic config but hopefully it covers the maximum
-	 * configuration latency.
-	 *
-	 * As a fallback, the checks in _append_oa_reports() to skip
-	 * invalid OA reports do also seem to work to discard reports
-	 * generated before this config has completed - albeit not
-	 * silently.
-	 *
-	 * Unfortunately this is essentially a magic number, since we
-	 * don't currently know of a reliable mechanism for predicting
-	 * how long the MUX config will take to apply and besides
-	 * seeing invalid reports we don't know of a reliable way to
-	 * explicitly check that the MUX config has landed.
-	 *
-	 * It's even possible we've miss characterized the underlying
-	 * problem - it just seems like the simplest explanation why
-	 * a delay at this location would mitigate any invalid reports.
-	 */
-	usleep_range(15000, 20000);
+	err = i915_active_request_set(&stream->active_config_rq,
+				      rq);
+	if (err)
+		goto err_add_request;
+
+	i915_vma_lock(vma);
+	err = i915_request_await_object(rq, vma->obj, 0);
+	if (!err)
+		err = i915_vma_move_to_active(vma, rq, 0);
+	i915_vma_unlock(vma);
+	if (err)
+		goto err_add_request;
+
+	cs = intel_ring_begin(rq, INTEL_GEN(i915) >= 8 ? 4 : 2);
+	if (IS_ERR(cs)) {
+		err = PTR_ERR(cs);
+		goto err_add_request;
+	}
+
+	if (INTEL_GEN(i915) > 8) {
+		*cs++ = MI_BATCH_BUFFER_START_GEN8;
+		*cs++ = lower_32_bits(vma->node.start);
+		*cs++ = upper_32_bits(vma->node.start);
+		*cs++ = MI_NOOP;
+	} else {
+		*cs++ = MI_BATCH_BUFFER_START;
+		*cs++ = vma->node.start;
+	}
+
+	intel_ring_advance(rq, cs);
+
+err_add_request:
+	i915_request_add(rq);
+err_vma_unpin:
+	i915_vma_unpin(vma);
+
+	return err;
 }
 
 static int hsw_enable_metric_set(struct i915_perf_stream *stream)
 {
 	struct drm_i915_private *dev_priv = stream->dev_priv;
-	const struct i915_oa_config *oa_config = stream->oa_config;
 
 	/*
 	 * PRM:
@@ -2007,13 +2044,7 @@  static int hsw_enable_metric_set(struct i915_perf_stream *stream)
 	I915_WRITE(GEN6_UCGCTL1, (I915_READ(GEN6_UCGCTL1) |
 				  GEN6_CSUNIT_CLOCK_GATE_DISABLE));
 
-	config_oa_regs(dev_priv, oa_config->mux_regs, oa_config->mux_regs_len);
-	delay_after_mux();
-
-	config_oa_regs(dev_priv, oa_config->b_counter_regs,
-		       oa_config->b_counter_regs_len);
-
-	return 0;
+	return emit_oa_config(dev_priv, stream);
 }
 
 static void hsw_disable_metric_set(struct i915_perf_stream *stream)
@@ -2372,13 +2403,7 @@  static int gen8_enable_metric_set(struct i915_perf_stream *stream)
 	if (ret)
 		return ret;
 
-	config_oa_regs(dev_priv, oa_config->mux_regs, oa_config->mux_regs_len);
-	delay_after_mux();
-
-	config_oa_regs(dev_priv, oa_config->b_counter_regs,
-		       oa_config->b_counter_regs_len);
-
-	return 0;
+	return emit_oa_config(dev_priv, stream);
 }
 
 static void gen8_disable_metric_set(struct i915_perf_stream *stream)
@@ -2597,6 +2622,9 @@  static int i915_oa_stream_init(struct i915_perf_stream *stream,
 
 	stream->engine = props->engine;
 
+	INIT_ACTIVE_REQUEST(&stream->active_config_rq,
+			    &stream->config_mutex);
+
 	stream->sample_flags |= SAMPLE_OA_REPORT;
 	stream->sample_size += format_size;
 
@@ -2625,8 +2653,9 @@  static int i915_oa_stream_init(struct i915_perf_stream *stream,
 		goto err_noa_wait_alloc;
 	}
 
-	ret = i915_perf_get_oa_config(dev_priv, props->metrics_set,
-				      &stream->oa_config);
+	ret = i915_perf_get_oa_config_and_bo(stream, props->metrics_set,
+					     &stream->oa_config,
+					     &stream->initial_oa_config_bo);
 	if (ret) {
 		DRM_DEBUG("Invalid OA config id=%i\n", props->metrics_set);
 		goto err_config;
@@ -2658,16 +2687,31 @@  static int i915_oa_stream_init(struct i915_perf_stream *stream,
 	stream->ops = &i915_oa_stream_ops;
 	dev_priv->perf.exclusive_stream = stream;
 
+	mutex_lock(&stream->config_mutex);
 	ret = dev_priv->perf.ops.enable_metric_set(stream);
 	if (ret) {
 		DRM_DEBUG("Unable to enable metric set\n");
-		goto err_enable;
+		/*
+		 * Ignore the return value since we already have an error from
+		 * the enable vfunc.
+		 */
+		i915_active_request_retire(&stream->active_config_rq, 0,
+					   &stream->config_mutex);
+	} else {
+		ret = i915_active_request_retire(&stream->active_config_rq, 0,
+						 &stream->config_mutex);
 	}
 
-	DRM_DEBUG("opening stream oa config uuid=%s\n", stream->oa_config->uuid);
-
+	mutex_unlock(&stream->config_mutex);
 	mutex_unlock(&dev_priv->drm.struct_mutex);
 
+	i915_gem_object_put(stream->initial_oa_config_bo);
+	stream->initial_oa_config_bo = NULL;
+	if (ret)
+		goto err_enable;
+
+	DRM_DEBUG("opening stream oa config uuid=%s\n", stream->oa_config->uuid);
+
 	hrtimer_init(&stream->poll_check_timer,
 		     CLOCK_MONOTONIC, HRTIMER_MODE_REL);
 	stream->poll_check_timer.function = oa_poll_check_timer_cb;
@@ -2677,8 +2721,11 @@  static int i915_oa_stream_init(struct i915_perf_stream *stream,
 	return 0;
 
 err_enable:
-	dev_priv->perf.exclusive_stream = NULL;
+	mutex_lock(&dev_priv->drm.struct_mutex);
+	mutex_lock(&stream->config_mutex);
 	dev_priv->perf.ops.disable_metric_set(stream);
+	mutex_unlock(&stream->config_mutex);
+	dev_priv->perf.exclusive_stream = NULL;
 	mutex_unlock(&dev_priv->drm.struct_mutex);
 
 err_lock:
@@ -2690,6 +2737,9 @@  static int i915_oa_stream_init(struct i915_perf_stream *stream,
 
 	free_oa_configs(stream);
 
+	if (stream->initial_oa_config_bo)
+		i915_gem_object_put(stream->initial_oa_config_bo);
+
 err_config:
 	free_noa_wait(stream);
 
diff --git a/drivers/gpu/drm/i915/i915_perf_types.h b/drivers/gpu/drm/i915/i915_perf_types.h
index 695c6ad2fcd5..1ba511c12673 100644
--- a/drivers/gpu/drm/i915/i915_perf_types.h
+++ b/drivers/gpu/drm/i915/i915_perf_types.h
@@ -166,7 +166,8 @@  struct i915_perf_stream {
 	const struct i915_perf_stream_ops *ops;
 
 	/**
-	 * @active_config_mutex: Protects access to @oa_config & @oa_config_bos.
+	 * @active_config_mutex: Protects access to @active_config_rq,
+	 * @oa_config & @oa_config_bos.
 	 */
 	struct mutex config_mutex;
 
@@ -181,6 +182,16 @@  struct i915_perf_stream {
 	 */
 	struct list_head oa_config_bos;
 
+	/**
+	 * @active_config_rq: Last request reconfiguring the HW.
+	 */
+	struct i915_active_request active_config_rq;
+
+	/**
+	 * @initial_oa_config_bo: First OA configuration BO to be run.
+	 */
+	struct drm_i915_gem_object *initial_oa_config_bo;
+
 	/**
 	 * The OA context specific information.
 	 */