Skip to contentSkip to conversation

drm/i915: i915_gem_object_sync must handle NULL

PATCH Applied Ben Widawsky 3 participants 7 messages Apr 11, 2012

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/3 drm/i915: i915_gem_object_sync must handle NULL

When I extracted the synchronization code for implementing semaphorified pageflips (74f5f6e0), I neglected the non pipelined case which also calls this code. The modesetting code wants to make sure the object has finished rendering to the frame before configuring the scanout (ie. non-pipelined case).

As a result of a follow on discussion on IRC, I've decided to add a comment about the function itself which received much inspiration from Chris as well. So really, this patch was ghost-written by Chris :).

---
 drivers/gpu/drm/i915/i915_gem.c |   14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_gem.c +13 -1
index 1bfb0d2..9fcdc9a 100644
@@ -1953,6 +1953,18 @@ i915_gem_object_wait_rendering(struct drm_i915_gem_object *obj)
 	return 0;
 }
 
+/**
+ * i915_gem_object_sync - sync an object to a ring.
+ *
+ * @obj: object which may be in use on another ring.
+ * @to: ring we wish to use the object on. May be NULL.
+ *
+ * This code is meant to abstract object synchronization with the GPU.
+ * Calling with NULL implies synchronizing the object with the CPU
+ * rather than a particular GPU ring.
+ *
+ * Returns 0 if successful, else propagates up the lower layer error.
+ */
 int
 i915_gem_object_sync(struct drm_i915_gem_object *obj,
 		     struct intel_ring_buffer *to)
@@ -1964,7 +1976,7 @@ i915_gem_object_sync(struct drm_i915_gem_object *obj,
 	if (from == NULL || to == from)
 		return 0;
 
-	if (!i915_semaphore_is_enabled(obj->base.dev))
+	if (to == NULL || !i915_semaphore_is_enabled(obj->base.dev))
 		return i915_gem_object_wait_rendering(obj);
 
 	idx = intel_ring_sync_index(from, to);
Signature · 1 line
1.7.10
#1 lore raw

PATCH 2/3 drm/i915: fix for when semaphore updates fail

This fixes a long standing issue where emitting the semaphore updates may have failed, but we've already updated our internal data structure.

---
 drivers/gpu/drm/i915/i915_gem.c |    6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_gem.c +4 -2
index 9fcdc9a..0115b12 100644
@@ -2001,10 +2001,12 @@ i915_gem_object_sync(struct drm_i915_gem_object *obj,
 		seqno = request->seqno;
 	}
 
-	from->sync_seqno[idx] = seqno;
 
-	return to->sync_to(to, from, seqno - 1);
+	ret = to->sync_to(to, from, seqno - 1);
+	if (!ret)
+		from->sync_seqno[idx] = seqno;
 
+	return ret;
 }
 
 static void i915_gem_object_finish_gtt(struct drm_i915_gem_object *obj)
Signature · 1 line
1.7.10
Reviewed-by on v1 2/3
#2 lore raw

Re: [PATCH 2/3] drm/i915: fix for when semaphore updates fail

Quoted 5 lines · Ben Widawsky

This fixes a long standing issue where emitting the semaphore updates may have failed, but we've already updated our internal data structure.

I would have written it as ret == 0, but since I didn't write this,

Fortunately the impact of the bug would have been render corruption and not a GPU hang.
-Chris

Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre
#3 lore raw

PATCH 3/3 drm/i915: hide (seqno-1) in ringbuffer code

In reply to Ben Widawsky

Waiting for seqno-1 in our object synchronization code is an implementation detail given how we've decided to do the waits within the rest of our code.

---
 drivers/gpu/drm/i915/i915_gem.c         |    2 +-
 drivers/gpu/drm/i915/intel_ringbuffer.c |    6 ++++++
 2 files changed, 7 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_gem.c +1 -1
index 0115b12..71934dd 100644
@@ -2002,7 +2002,7 @@ i915_gem_object_sync(struct drm_i915_gem_object *obj,
 	}
 
 
-	ret = to->sync_to(to, from, seqno - 1);
+	ret = to->sync_to(to, from, seqno);
 	if (!ret)
 		from->sync_seqno[idx] = seqno;
 
drivers/gpu/drm/i915/intel_ringbuffer.c +6 -0
index dfdb613..467b331 100644
@@ -482,6 +482,12 @@ intel_ring_sync(struct intel_ring_buffer *waiter,
 		  MI_SEMAPHORE_COMPARE |
 		  MI_SEMAPHORE_REGISTER;
 
+	/* Throughout all of the GEM code, seqno passed implies our current
+	 * seqno is >= the last seqno executed. However for hardware the
+	 * comparison is strictly greater than.
+	 */
+	seqno -= 1;
+
 	ret = intel_ring_begin(waiter, 4);
 	if (ret)
 		return ret;
Signature · 1 line
1.7.10
Reviewed-by on v1 3/3
#4 lore raw

Re: [PATCH 3/3] drm/i915: hide (seqno-1) in ringbuffer code

Quoted 6 lines · Ben Widawsky

Waiting for seqno-1 in our object synchronization code is an implementation detail given how we've decided to do the waits within the rest of our code.

Thanks for the comment, I had forgotten about that hw detail.

-Chris

Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre
Reviewed-by Tested-by on v1 1/3
#5 lore raw
In reply to Ben Widawsky
Quoted 6 lines · Ben Widawsky

When I extracted the synchronization code for implementing semaphorified pageflips (74f5f6e0), I neglected the non pipelined case which also calls this code. The modesetting code wants to make sure the object has finished rendering to the frame before configuring the scanout (ie. non-pipelined case).

Quoted 6 lines

As a result of a follow on discussion on IRC, I've decided to add a comment about the function itself which received much inspiration from Chris as well. So really, this patch was ghost-written by Chris :).

-Chris

Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre
#6 lore raw
Quoted 7 lines · Chris Wilson
Quoted 6 lines · Ben Widawsky

When I extracted the synchronization code for implementing semaphorified pageflips (74f5f6e0), I neglected the non pipelined case which also calls this code. The modesetting code wants to make sure the object has finished rendering to the frame before configuring the scanout (ie. non-pipelined case).

Quoted 8 lines
Quoted 6 lines

As a result of a follow on discussion on IRC, I've decided to add a comment about the function itself which received much inspiration from Chris as well. So really, this patch was ghost-written by Chris :).

I've picked up all 3 patches for -next, thanks for them. -Daniel

Signature · 3 lines
Daniel Vetter
Mail: [email protected]
Mobile: +41 (0)79 365 57 48