Nouveau - Apr 2025 - [PATCH v2 0/2] drm/nouveau: Don't set signaled fences' error codes

Changes in v2:
  - Only fix the issue by checking for a fence being signaled in
    nouveau_fence_context_kill(), before setting the fence's error.
    (Christian, Danilo)
  - Drop cleanup patches. Instead, idiomaticize for-each-loops.

Was called "Fix & improve nouveau_fence_done()" before.

I've tested this with KASAN & kmemleak.

P.

Philipp Stanner (2):
  drm/nouveau: Fix WARN_ON in nouveau_fence_context_kill()
  drm/nouveau: nouveau_fence: Standardize list iterations

 drivers/gpu/drm/nouveau/nouveau_fence.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

-- 
2.48.1

Nouveau is mostly designed in a way that it's expected that fences only
ever get signaled through nouveau_fence_signal(). However, in at least
one other place, nouveau_fence_done(), can signal fences, too. If that
happens (race) a signaled fence remains in the pending list for a while,
until it gets removed by nouveau_fence_update().

Should nouveau_fence_context_kill() run in the meantime, this would be
a bug because the function would attempt to set an error code on an
already signaled fence.

Have nouveau_fence_context_kill() check for a fence being signaled.

Cc: <stable at vger.kernel.org> # v5.10+
Fixes: ea13e5abf807 ("drm/nouveau: signal pending fences when channel has
been killed")
Suggested-by: Christian K?nig <christian.koenig at amd.com>
Signed-off-by: Philipp Stanner <phasta at kernel.org>
---
 drivers/gpu/drm/nouveau/nouveau_fence.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_fence.c
b/drivers/gpu/drm/nouveau/nouveau_fence.c
index 7622587f149e..6ded8c2b6d3b 100644
--- a/drivers/gpu/drm/nouveau/nouveau_fence.c
+++ b/drivers/gpu/drm/nouveau/nouveau_fence.c
@@ -90,7 +90,7 @@ nouveau_fence_context_kill(struct nouveau_fence_chan *fctx,
int error)
 	while (!list_empty(&fctx->pending)) {
 		fence = list_entry(fctx->pending.next, typeof(*fence), head);
 
-		if (error)
+		if (error && !dma_fence_is_signaled_locked(&fence->base))
 			dma_fence_set_error(&fence->base, error);
 
 		if (nouveau_fence_signal(fence))
-- 
2.48.1

nouveau_fence.c iterates over lists in a non-canonical way. Since the
operations done are just basic for-each-loops, they should be written in
the standard form.

Use for_each_safe() instead of the custom loop iterations.

Signed-off-by: Philipp Stanner <phasta at kernel.org>
---
 drivers/gpu/drm/nouveau/nouveau_fence.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_fence.c
b/drivers/gpu/drm/nouveau/nouveau_fence.c
index 6ded8c2b6d3b..60d961b43488 100644
--- a/drivers/gpu/drm/nouveau/nouveau_fence.c
+++ b/drivers/gpu/drm/nouveau/nouveau_fence.c
@@ -84,11 +84,12 @@ void
 nouveau_fence_context_kill(struct nouveau_fence_chan *fctx, int error)
 {
 	struct nouveau_fence *fence;
+	struct list_head *pos, *tmp;
 	unsigned long flags;
 
 	spin_lock_irqsave(&fctx->lock, flags);
-	while (!list_empty(&fctx->pending)) {
-		fence = list_entry(fctx->pending.next, typeof(*fence), head);
+	list_for_each_safe(pos, tmp, &fctx->pending) {
+		fence = list_entry(pos, struct nouveau_fence, head);
 
 		if (error && !dma_fence_is_signaled_locked(&fence->base))
 			dma_fence_set_error(&fence->base, error);
@@ -131,11 +132,12 @@ static int
 nouveau_fence_update(struct nouveau_channel *chan, struct nouveau_fence_chan
*fctx)
 {
 	struct nouveau_fence *fence;
+	struct list_head *pos, *tmp;
 	int drop = 0;
 	u32 seq = fctx->read(chan);
 
-	while (!list_empty(&fctx->pending)) {
-		fence = list_entry(fctx->pending.next, typeof(*fence), head);
+	list_for_each_safe(pos, tmp, &fctx->pending) {
+		fence = list_entry(pos, struct nouveau_fence, head);
 
 		if ((int)(seq - fence->base.seqno) < 0)
 			break;
-- 
2.48.1