[Intel-gfx] [PATCH 11/14] drm/i915: Use uncached(WC) mapping for acessing the GuC log buffer

[akash.goel at intel.com]

From: Akash Goel <akash.goel at intel.com>

Host needs to sample the GuC log buffer on every flush interrupt from GuC.
To ensure that we always get the up-to-date data from log buffer, its
better to access the buffer through an uncached CPU mapping. Also the way
buffer is accessed from GuC & Host side, manually doing cache flush may
not be effective always if cached CPU mapping is used.
Since the data to be read is less than 50 KB, there would not be any
performance implications.

Signed-off-by: Akash Goel <akash.goel at intel.com>
---
 drivers/gpu/drm/i915/i915_guc_submission.c | 53 +++++++++++++++++++++++-------
 drivers/gpu/drm/i915/intel_guc.h           |  1 +
 2 files changed, 43 insertions(+), 11 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_guc_submission.c b/drivers/gpu/drm/i915/i915_guc_submission.c
index ec68a6d..4d76588 100644
--- a/drivers/gpu/drm/i915/i915_guc_submission.c
+++ b/drivers/gpu/drm/i915/i915_guc_submission.c
@@ -874,7 +874,7 @@ static void guc_read_update_log_buffer(struct drm_device *dev, bool capture_all)
 	u32 num_pages_to_copy;
 	int i;
 
-	if (!guc->log.obj)
+	if (!guc->log.obj || !guc->log.buf_addr)
 		return;
 
 	num_pages_to_copy = guc->log.obj->base.size / PAGE_SIZE;
@@ -884,8 +884,7 @@ static void guc_read_update_log_buffer(struct drm_device *dev, bool capture_all)
 	if (!capture_all)
 		num_pages_to_copy -= (GUC_LOG_CRASH_PAGES + 1);
 
-	log_buffer_state = src_ptr =
-		kmap_atomic(i915_gem_object_get_page(guc->log.obj, 0));
+	log_buffer_state = src_ptr = guc->log.buf_addr;
 
 	/* Get the pointer to local buffer to store the logs */
 	dst_ptr = log_buffer_copy_state = guc_get_write_buffer(guc);
@@ -906,8 +905,6 @@ static void guc_read_update_log_buffer(struct drm_device *dev, bool capture_all)
 			log_buffer_copy_state++;
 		}
 
-		/* FIXME: invalidate/flush for log buffer needed */
-
 		/* Update the read pointer in the shared log buffer */
 		log_buffer_state->read_ptr =
 			log_buffer_state->sampled_write_ptr;
@@ -917,14 +914,11 @@ static void guc_read_update_log_buffer(struct drm_device *dev, bool capture_all)
 		log_buffer_state++;
 	}
 
-	kunmap_atomic(src_ptr);
-
 	/* Now copy the actual logs */
 	for (i=1; (i < num_pages_to_copy) && dst_ptr; i++) {
 		dst_ptr += PAGE_SIZE;
-		src_ptr = kmap_atomic(i915_gem_object_get_page(guc->log.obj, i));
+		src_ptr += PAGE_SIZE;
 		memcpy(dst_ptr, src_ptr, PAGE_SIZE);
-		kunmap_atomic(src_ptr);
 	}
 }
 
@@ -1016,7 +1010,7 @@ static int guc_create_log_relay_file(struct intel_guc *guc)
 
 	/* Keep the size of sub buffers same as shared log buffer */
 	subbuf_size = guc->log.obj->base.size;
-	n_subbufs = i915.guc_log_buffer_nr
+	n_subbufs = i915.guc_log_buffer_nr;
 
 	guc_log_relay_chan = relay_open("guc_log", log_dir,
 			subbuf_size, n_subbufs, &relay_callbacks, dev);
@@ -1051,6 +1045,7 @@ static int guc_log_late_setup(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
 	struct intel_guc *guc = &dev_priv->guc;
+	void *vaddr;
 	int ret;
 
 	lockdep_assert_held(&dev->struct_mutex);
@@ -1061,9 +1056,26 @@ static int guc_log_late_setup(struct drm_device *dev)
 	if (WARN_ON(guc->log.relay_chan))
 		return -EINVAL;
 
+	/* If log_level was set as -1 at boot time, then vmalloc mapping would
+	 * not have been created for the log buffer, so create one now.
+	 */
+	if (!guc->log.buf_addr) {
+		vaddr = i915_gem_object_pin_map(guc->log.obj, true);
+		if (IS_ERR(vaddr)) {
+			ret = PTR_ERR(vaddr);
+			DRM_DEBUG_DRIVER("Couldn't map log buffer %d\n", ret);
+			goto err;
+		}
+		guc->log.buf_addr = vaddr;
+	}
+
+
 	ret = guc_create_log_relay_file(guc);
-	if (ret)
+	if (ret) {
+		guc->log.buf_addr = NULL;
+		i915_gem_object_unpin_map(guc->log.obj);
 		goto err;
+	}
 
 	return 0;
 err:
@@ -1098,6 +1110,25 @@ static void guc_create_log(struct intel_guc *guc)
 			return;
 		}
 
+		if (i915.guc_log_level >= 0) {
+			/* Create a WC (Uncached for read) mapping so that we
+			 * can directly get the data (up-to-date) from memory.
+			 * Also create the mapping now only to properly handle
+			 * the flush interrupts which can come before we create
+			 * a relay channel at the end of Driver load.
+			 */
+			void *vaddr = i915_gem_object_pin_map(obj, true);
+			if (IS_ERR(vaddr)) {
+				DRM_DEBUG_DRIVER("Couldn't map log buffer %ld\n",
+				PTR_ERR(vaddr));
+				gem_release_guc_obj(obj);
+				i915.guc_log_level = -1;
+				return;
+			}
+
+			guc->log.buf_addr = vaddr;
+		}
+
 		guc->log.obj = obj;
 	}
 
diff --git a/drivers/gpu/drm/i915/intel_guc.h b/drivers/gpu/drm/i915/intel_guc.h
index 3e90279..5e6accb 100644
--- a/drivers/gpu/drm/i915/intel_guc.h
+++ b/drivers/gpu/drm/i915/intel_guc.h
@@ -127,6 +127,7 @@ struct intel_guc_log {
 	struct drm_i915_gem_object *obj;
 	struct workqueue_struct *wq;
 	struct rchan *relay_chan;
+	void *buf_addr;
 };
 
 struct intel_guc {
-- 
1.9.2