thoughts of looking at android fences

Message ID 5277777E.5060904@canonical.com
State New
Headers show

Commit Message

Maarten Lankhorst Nov. 4, 2013, 10:31 a.m.
op 02-11-13 22:36, Colin Cross schreef:
> On Wed, Oct 30, 2013 at 5:17 AM, Maarten Lankhorst
> <maarten.lankhorst@canonical.com> wrote:
>> op 24-10-13 14:13, Maarten Lankhorst schreef:
>>> So I actually tried to implement it now. I killed all the deprecated members and assumed a linear timeline.
>>> This means that syncpoints can only be added at the end, not in between. In particular it means sw_sync
>>> might be slightly broken.
>>>
>>> I only tested it with a simple program I wrote called ufence.c, it's in drivers/staging/android/ufence.c in the following tree:
>>>
>>> http://cgit.freedesktop.org/~mlankhorst/linux
>>>
>>> the "rfc: convert android to fence api" has all the changes from my dma-fence proposal to what android would need,
>>> it also converts the userspace fence api to use the dma-fence api.
>>>
>>> sync_pt is implemented as fence too. This meant not having to convert all of android right away, though I did make some changes.
>>> I killed the deprecated members and made all the fence calls forward to the sync_timeline_ops. dup and compare are no longer used.
>>>
>>> I haven't given this a spin on a full android kernel, only with the components that are in mainline kernel under staging and my dumb test program.
>>>
>>> ~Maarten
>>>
>>> PS: The nomenclature is very confusing. I want to rename dma-fence to syncpoint, but I want some feedback from the android devs first. :)
>>>
>> Come on, any feedback? I want to move the discussion forward.
>>
>> ~Maarten
> I experimented with it a little on a device that uses sync and came
> across a few bugs:
> 1.  sync_timeline_signal needs to call __fence_signal on all signaled
> points on the timeline, not just the first
> 2.  fence_add_callback doesn't always initialize cb.node
> 3.  sync_fence_wait should take ms
> 4.  sync_print_pt status printing was incorrect
> 5.  there is a deadlock:
>    sync_print_obj takes obj->child_list_lock
>    sync_print_pt
>    fence_is_signaled
>    fence_signal takes fence->lock == obj->child_list_lock
> 6.  freeing a timeline before all the fences holding points on that
> timeline have timed out results in a crash
>
> With the attached patch to fix these issues, our libsync and sync_test
> give the same results as with our sync code.  I haven't tested against
> the full Android framework yet.
>
> The compare op and timeline ordering is critical to the efficiency of
> sync points on Android.  The compare op is used when merging fences to
> drop all but the latest point on the same timeline.  This is necessary
> for example when the same buffer is submitted to the display on
> multiple frames, like when there is a live wallpaper in the background
> updating at 60 fps and a static screen of widgets on top of it.  The
> static widget buffer is submitted on every frame, returning a new
> fence each time.  The compositor merges the new fence with the fence
> for the previous buffer, and because they are on the same timeline it
> merges down to a single point.  I experimented with disabling the
> merge optimization on a Nexus 10, and found that leaving the screen on
> running a live wallpaper eventually resulted in 100k outstanding sync
> points.

Well, here I did the same for dma-fence, can you take a look?

---

Comments

Maarten Lankhorst Nov. 8, 2013, 10:43 a.m. | #1
op 07-11-13 22:11, Rom Lemarchand schreef:
> Hi Maarten, I tested your changes and needed the attached patch: behavior
> now seems equivalent as android sync. I haven't tested performance.
>
> The issue resolved by this patch happens when i_b < b->num_fences and
> i_a >= a->num_fences (or vice versa). Then, pt_a is invalid and so
> dereferencing pt_a->context causes a crash.

Oops, thinko. :) Originally I had it correct by doing this:

+       /*
+        * Assume sync_fence a and b are both ordered and have no
+        * duplicates with the same context.
+        *
+        * If a sync_fence can only be created with sync_fence_merge
+        * and sync_fence_create, this is a reasonable assumption.
+        */
+       for (i = i_a = i_b = 0; i_a < a->num_fences && i_b < b->num_fences; ) {
+               struct fence *pt_a = a->cbs[i_a].sync_pt;
+               struct fence *pt_b = b->cbs[i_b].sync_pt;
+
+               if (pt_a->context < pt_b->context) {
+                       sync_fence_add_pt(fence, &i, pt_a);
+
+                       i_a++;
+               } else if (pt_a->context > pt_b->context) {
+                       sync_fence_add_pt(fence, &i, pt_b);
+
+                       i_b++;
+               } else {
+                       if (pt_a->seqno - pt_b->seqno <= INT_MAX)
+                               sync_fence_add_pt(fence, &i, pt_a);
+                       else
+                               sync_fence_add_pt(fence, &i, pt_b);
+
+                       i_a++;
+                       i_b++;
+               }
+        }
+
+        /* Add remaining fences from a or b*/
+        for (; i_a < a->num_fences; i_a++)
+               sync_fence_add_pt(fence, &i, a->cbs[i_a].sync_pt);
+
+        for (; i_b < b->num_fences; i_b++)
+               sync_fence_add_pt(fence, &i, b->cbs[i_b].sync_pt);

Then I thought I could clean it up by merging it, but that ended up being
more unreadable and crashing... so I guess I'll revert back to this version. :)
Maarten Lankhorst Nov. 8, 2013, 11:43 a.m. | #2
op 07-11-13 22:11, Rom Lemarchand schreef:
> Hi Maarten, I tested your changes and needed the attached patch: behavior
> now seems equivalent as android sync. I haven't tested performance.
>
> The issue resolved by this patch happens when i_b < b->num_fences and i_a
>> = a->num_fences (or vice versa). Then, pt_a is invalid and so
> dereferencing pt_a->context causes a crash.
>
Yeah, I pushed my original fix. I intended to keep android userspace behavior the same, and I tried to keep the kernelspace the api same as much as I could. If peformance is the same, or not noticeably worse, would there be any objections on the android side about renaming dma-fence to syncpoint, and getting it in mainline?

~Maarten
Rom Lemarchand Nov. 12, 2013, 1:53 a.m. | #3
I ran some benchmarks and things seem to be running about the same.
No one on our graphics team seemed concerned about the change.

The only concern I heard was about the increased complexity of the new sync
code as opposed to the old sync framework which tried to keep things
straightforward.


On Fri, Nov 8, 2013 at 3:43 AM, Maarten Lankhorst <
maarten.lankhorst@canonical.com> wrote:

> op 07-11-13 22:11, Rom Lemarchand schreef:
> > Hi Maarten, I tested your changes and needed the attached patch: behavior
> > now seems equivalent as android sync. I haven't tested performance.
> >
> > The issue resolved by this patch happens when i_b < b->num_fences and i_a
> >> = a->num_fences (or vice versa). Then, pt_a is invalid and so
> > dereferencing pt_a->context causes a crash.
> >
> Yeah, I pushed my original fix. I intended to keep android userspace
> behavior the same, and I tried to keep the kernelspace the api same as much
> as I could. If peformance is the same, or not noticeably worse, would there
> be any objections on the android side about renaming dma-fence to
> syncpoint, and getting it in mainline?
>
> ~Maarten
>

Patch hide | download patch | download mbox

diff --git a/drivers/staging/android/sync.c b/drivers/staging/android/sync.c
index 2c7fd3f2ab23..d1d89f1f8553 100644
--- a/drivers/staging/android/sync.c
+++ b/drivers/staging/android/sync.c
@@ -232,39 +232,62 @@  void sync_fence_install(struct sync_fence *fence, int fd)
 }
 EXPORT_SYMBOL(sync_fence_install);
 
+static void sync_fence_add_pt(struct sync_fence *fence, int *i, struct fence *pt) {
+	fence->cbs[*i].sync_pt = pt;
+	fence->cbs[*i].fence = fence;
+
+	if (!fence_add_callback(pt, &fence->cbs[*i].cb, fence_check_cb_func)) {
+		fence_get(pt);
+		(*i)++;
+	}
+}
+
 struct sync_fence *sync_fence_merge(const char *name,
 				    struct sync_fence *a, struct sync_fence *b)
 {
 	int num_fences = a->num_fences + b->num_fences;
 	struct sync_fence *fence;
-	int i;
+	int i, i_a, i_b;
 
 	fence = sync_fence_alloc(offsetof(struct sync_fence, cbs[num_fences]), name);
 	if (fence == NULL)
 		return NULL;
 
-	fence->num_fences = num_fences;
 	atomic_set(&fence->status, num_fences);
 
-	for (i = 0; i < a->num_fences; ++i) {
-		struct fence *pt = a->cbs[i].sync_pt;
-
-		fence_get(pt);
-		fence->cbs[i].sync_pt = pt;
-		fence->cbs[i].fence = fence;
-		if (fence_add_callback(pt, &fence->cbs[i].cb, fence_check_cb_func))
-			atomic_dec(&fence->status);
+	/*
+	 * Assume sync_fence a and b are both ordered and have no
+	 * duplicates with the same context.
+	 *
+	 * If a sync_fence can only be created with sync_fence_merge
+	 * and sync_fence_create, this is a reasonable assumption.
+	 */
+	for (i = i_a = i_b = 0; i_a < a->num_fences || i_b < b->num_fences; ) {
+		struct fence *pt_a = i_a < a->num_fences ? a->cbs[i_a].sync_pt : NULL;
+		struct fence *pt_b = i_b < b->num_fences ? b->cbs[i_b].sync_pt : NULL;
+
+		if (!pt_b || pt_a->context < pt_b->context) {
+			sync_fence_add_pt(fence, &i, pt_a);
+
+			i_a++;
+		} else if (!pt_a || pt_a->context > pt_b->context) {
+			sync_fence_add_pt(fence, &i, pt_b);
+
+			i_b++;
+		} else {
+			if (pt_a->seqno - pt_b->seqno <= INT_MAX)
+				sync_fence_add_pt(fence, &i, pt_a);
+			else
+				sync_fence_add_pt(fence, &i, pt_b);
+
+			i_a++;
+			i_b++;
+		}
 	}
 
-	for (i = 0; i < b->num_fences; ++i) {
-		struct fence *pt = b->cbs[i].sync_pt;
-
-		fence_get(pt);
-		fence->cbs[a->num_fences + i].sync_pt = pt;
-		fence->cbs[a->num_fences + i].fence = fence;
-		if (fence_add_callback(pt, &fence->cbs[a->num_fences + i].cb, fence_check_cb_func))
-			atomic_dec(&fence->status);
-	}
+	if (num_fences > i)
+		atomic_sub(num_fences - i, &fence->status);
+	fence->num_fences = i;
 
 	sync_fence_debug_add(fence);
 	return fence;