* [PATCH v4] drm: Fix drm_pending_vblank_event leak in error path for out_fence_ptr
@ 2026-08-26 11:46 Thadeu Lima de Souza Cascardo
2026-09-11 13:58 ` Melissa Wen
2026-09-11 14:43 ` Markus Elfring
0 siblings, 2 replies; 4+ messages in thread
From: Thadeu Lima de Souza Cascardo @ 2026-08-26 11:46 UTC (permalink / raw)
To: Maarten Lankhorst, Maxime Ripard, Thomas Zimmermann,
David Airlie, Simona Vetter
Cc: dri-devel, linux-kernel, Melissa Wen, kernel-dev, sashiko-bot,
Thadeu Lima de Souza Cascardo
When an out_fence_ptr is provided but DRM_MODE_PAGE_FLIP_EVENT is not
set, a drm_pending_vblank_event will be allocated. If later, there is an
allocation failure or another failure at setup_out_fence(), that event
will not have base.fence set and it will not be released at
complete_signaling().
Release the event and set crtc_state->event to NULL just like in the
DRM_MODE_PAGE_FLIP_EVENT case when there is a failure at
drm_event_reserve_init(). That is, prepare_signaling() releases the
event and there is nothing to be done at complete_signaling(). Use
drm_event_cancel_free() as that will also undo drm_event_reserve_init()
in case it has been called.
Reported-by: sashiko-bot@kernel.org
Closes: https://sashiko.dev/#/patchset/20260727-drm_crtc_atomic_commit_leak-v1-1-23d9948a9d7c@igalia.com?part=1
Fixes: 92c715fca907 ("drm/atomic: Fix double free in drm_atomic_state_default_clear")
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@igalia.com>
---
Changes in v4:
- Rename label to err_free_event.
- Use crtc_state->event instead of a local variable e.
- Link to v3: https://patch.msgid.link/20260817-drm_pending_vblank_event_leak-v3-1-7582b24447d0@igalia.com
Changes in v3:
- Use a label for the common exit pattern.
- Link to v2: https://patch.msgid.link/20260729-drm_pending_vblank_event_leak-v2-1-a5074aae07df@igalia.com
Changes in v2:
- Fix UAF when DRM_MODE_PAGE_FLIP_EVENT is used.
- Link to v1: https://patch.msgid.link/20260728-drm_pending_vblank_event_leak-v1-1-08429b920b16@igalia.com
---
drivers/gpu/drm/drm_atomic_uapi.c | 13 ++++++++++---
1 file changed, 10 insertions(+), 3 deletions(-)
diff --git a/drivers/gpu/drm/drm_atomic_uapi.c b/drivers/gpu/drm/drm_atomic_uapi.c
index ae7667d1072d..8c0eea2fed4a 100644
--- a/drivers/gpu/drm/drm_atomic_uapi.c
+++ b/drivers/gpu/drm/drm_atomic_uapi.c
@@ -1458,10 +1458,12 @@ static int prepare_signaling(struct drm_device *dev,
struct dma_fence *fence;
struct drm_out_fence_state *f;
+ ret = -ENOMEM;
+
f = krealloc(*fence_state, sizeof(**fence_state) *
(*num_fences + 1), GFP_KERNEL);
if (!f)
- return -ENOMEM;
+ goto err_free_event;
memset(&f[*num_fences], 0, sizeof(*f));
@@ -1470,12 +1472,12 @@ static int prepare_signaling(struct drm_device *dev,
fence = drm_crtc_create_fence(crtc);
if (!fence)
- return -ENOMEM;
+ goto err_free_event;
ret = setup_out_fence(&f[(*num_fences)++], fence);
if (ret) {
dma_fence_put(fence);
- return ret;
+ goto err_free_event;
}
crtc_state->event->base.fence = fence;
@@ -1531,6 +1533,11 @@ static int prepare_signaling(struct drm_device *dev,
}
return 0;
+
+err_free_event:
+ drm_event_cancel_free(dev, &crtc_state->event->base);
+ crtc_state->event = NULL;
+ return ret;
}
static void complete_signaling(struct drm_device *dev,
---
base-commit: 4d4be202165e832d74849b4a68e289a2a377039c
change-id: 20260728-drm_pending_vblank_event_leak-a36cedb296ba
Best regards,
--
Thadeu Lima de Souza Cascardo <cascardo@igalia.com>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH v4] drm: Fix drm_pending_vblank_event leak in error path for out_fence_ptr
2026-08-26 11:46 [PATCH v4] drm: Fix drm_pending_vblank_event leak in error path for out_fence_ptr Thadeu Lima de Souza Cascardo
@ 2026-09-11 13:58 ` Melissa Wen
2026-09-11 14:26 ` Melissa Wen
2026-09-11 14:43 ` Markus Elfring
1 sibling, 1 reply; 4+ messages in thread
From: Melissa Wen @ 2026-09-11 13:58 UTC (permalink / raw)
To: Thadeu Lima de Souza Cascardo, Maarten Lankhorst, Maxime Ripard,
Thomas Zimmermann, David Airlie, Simona Vetter
Cc: dri-devel, linux-kernel, kernel-dev, sashiko-bot
On 26/08/2026 13:46, Thadeu Lima de Souza Cascardo wrote:
> When an out_fence_ptr is provided but DRM_MODE_PAGE_FLIP_EVENT is not
> set, a drm_pending_vblank_event will be allocated. If later, there is an
> allocation failure or another failure at setup_out_fence(), that event
> will not have base.fence set and it will not be released at
> complete_signaling().
>
> Release the event and set crtc_state->event to NULL just like in the
> DRM_MODE_PAGE_FLIP_EVENT case when there is a failure at
> drm_event_reserve_init(). That is, prepare_signaling() releases the
> event and there is nothing to be done at complete_signaling(). Use
> drm_event_cancel_free() as that will also undo drm_event_reserve_init()
> in case it has been called.
LGTM. Thanks!
Reviewed-by: Melissa Wen <mwen@igalia.com>
>
> Reported-by: sashiko-bot@kernel.org
> Closes: https://sashiko.dev/#/patchset/20260727-drm_crtc_atomic_commit_leak-v1-1-23d9948a9d7c@igalia.com?part=1
> Fixes: 92c715fca907 ("drm/atomic: Fix double free in drm_atomic_state_default_clear")
> Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@igalia.com>
> ---
> Changes in v4:
> - Rename label to err_free_event.
> - Use crtc_state->event instead of a local variable e.
> - Link to v3: https://patch.msgid.link/20260817-drm_pending_vblank_event_leak-v3-1-7582b24447d0@igalia.com
>
> Changes in v3:
> - Use a label for the common exit pattern.
> - Link to v2: https://patch.msgid.link/20260729-drm_pending_vblank_event_leak-v2-1-a5074aae07df@igalia.com
>
> Changes in v2:
> - Fix UAF when DRM_MODE_PAGE_FLIP_EVENT is used.
> - Link to v1: https://patch.msgid.link/20260728-drm_pending_vblank_event_leak-v1-1-08429b920b16@igalia.com
> ---
> drivers/gpu/drm/drm_atomic_uapi.c | 13 ++++++++++---
> 1 file changed, 10 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/gpu/drm/drm_atomic_uapi.c b/drivers/gpu/drm/drm_atomic_uapi.c
> index ae7667d1072d..8c0eea2fed4a 100644
> --- a/drivers/gpu/drm/drm_atomic_uapi.c
> +++ b/drivers/gpu/drm/drm_atomic_uapi.c
> @@ -1458,10 +1458,12 @@ static int prepare_signaling(struct drm_device *dev,
> struct dma_fence *fence;
> struct drm_out_fence_state *f;
>
> + ret = -ENOMEM;
> +
> f = krealloc(*fence_state, sizeof(**fence_state) *
> (*num_fences + 1), GFP_KERNEL);
> if (!f)
> - return -ENOMEM;
> + goto err_free_event;
>
> memset(&f[*num_fences], 0, sizeof(*f));
>
> @@ -1470,12 +1472,12 @@ static int prepare_signaling(struct drm_device *dev,
>
> fence = drm_crtc_create_fence(crtc);
> if (!fence)
> - return -ENOMEM;
> + goto err_free_event;
>
> ret = setup_out_fence(&f[(*num_fences)++], fence);
> if (ret) {
> dma_fence_put(fence);
> - return ret;
> + goto err_free_event;
> }
>
> crtc_state->event->base.fence = fence;
> @@ -1531,6 +1533,11 @@ static int prepare_signaling(struct drm_device *dev,
> }
>
> return 0;
> +
> +err_free_event:
> + drm_event_cancel_free(dev, &crtc_state->event->base);
> + crtc_state->event = NULL;
> + return ret;
> }
>
> static void complete_signaling(struct drm_device *dev,
>
> ---
> base-commit: 4d4be202165e832d74849b4a68e289a2a377039c
> change-id: 20260728-drm_pending_vblank_event_leak-a36cedb296ba
>
> Best regards,
> --
> Thadeu Lima de Souza Cascardo <cascardo@igalia.com>
>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH v4] drm: Fix drm_pending_vblank_event leak in error path for out_fence_ptr
2026-09-11 13:58 ` Melissa Wen
@ 2026-09-11 14:26 ` Melissa Wen
0 siblings, 0 replies; 4+ messages in thread
From: Melissa Wen @ 2026-09-11 14:26 UTC (permalink / raw)
To: Thadeu Lima de Souza Cascardo, Maarten Lankhorst, Maxime Ripard,
Thomas Zimmermann, David Airlie, Simona Vetter
Cc: dri-devel, linux-kernel, kernel-dev, sashiko-bot
On 11/09/2026 15:58, Melissa Wen wrote:
>
>
> On 26/08/2026 13:46, Thadeu Lima de Souza Cascardo wrote:
>> When an out_fence_ptr is provided but DRM_MODE_PAGE_FLIP_EVENT is not
>> set, a drm_pending_vblank_event will be allocated. If later, there is an
>> allocation failure or another failure at setup_out_fence(), that event
>> will not have base.fence set and it will not be released at
>> complete_signaling().
>>
>> Release the event and set crtc_state->event to NULL just like in the
>> DRM_MODE_PAGE_FLIP_EVENT case when there is a failure at
>> drm_event_reserve_init(). That is, prepare_signaling() releases the
>> event and there is nothing to be done at complete_signaling(). Use
>> drm_event_cancel_free() as that will also undo drm_event_reserve_init()
>> in case it has been called.
>
> LGTM. Thanks!
>
> Reviewed-by: Melissa Wen <mwen@igalia.com>
Applied to drm-misc-fixes:
9eb1a393c89a ("drm: Fix drm_pending_vblank_event leak in error path for
out_fence_ptr")
>
>>
>> Reported-by: sashiko-bot@kernel.org
>> Closes:
>> https://sashiko.dev/#/patchset/20260727-drm_crtc_atomic_commit_leak-v1-1-23d9948a9d7c@igalia.com?part=1
>> Fixes: 92c715fca907 ("drm/atomic: Fix double free in
>> drm_atomic_state_default_clear")
>> Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@igalia.com>
>> ---
>> Changes in v4:
>> - Rename label to err_free_event.
>> - Use crtc_state->event instead of a local variable e.
>> - Link to v3:
>> https://patch.msgid.link/20260817-drm_pending_vblank_event_leak-v3-1-7582b24447d0@igalia.com
>>
>> Changes in v3:
>> - Use a label for the common exit pattern.
>> - Link to v2:
>> https://patch.msgid.link/20260729-drm_pending_vblank_event_leak-v2-1-a5074aae07df@igalia.com
>>
>> Changes in v2:
>> - Fix UAF when DRM_MODE_PAGE_FLIP_EVENT is used.
>> - Link to v1:
>> https://patch.msgid.link/20260728-drm_pending_vblank_event_leak-v1-1-08429b920b16@igalia.com
>> ---
>> drivers/gpu/drm/drm_atomic_uapi.c | 13 ++++++++++---
>> 1 file changed, 10 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/drm_atomic_uapi.c
>> b/drivers/gpu/drm/drm_atomic_uapi.c
>> index ae7667d1072d..8c0eea2fed4a 100644
>> --- a/drivers/gpu/drm/drm_atomic_uapi.c
>> +++ b/drivers/gpu/drm/drm_atomic_uapi.c
>> @@ -1458,10 +1458,12 @@ static int prepare_signaling(struct
>> drm_device *dev,
>> struct dma_fence *fence;
>> struct drm_out_fence_state *f;
>> + ret = -ENOMEM;
>> +
>> f = krealloc(*fence_state, sizeof(**fence_state) *
>> (*num_fences + 1), GFP_KERNEL);
>> if (!f)
>> - return -ENOMEM;
>> + goto err_free_event;
>> memset(&f[*num_fences], 0, sizeof(*f));
>> @@ -1470,12 +1472,12 @@ static int prepare_signaling(struct
>> drm_device *dev,
>> fence = drm_crtc_create_fence(crtc);
>> if (!fence)
>> - return -ENOMEM;
>> + goto err_free_event;
>> ret = setup_out_fence(&f[(*num_fences)++], fence);
>> if (ret) {
>> dma_fence_put(fence);
>> - return ret;
>> + goto err_free_event;
>> }
>> crtc_state->event->base.fence = fence;
>> @@ -1531,6 +1533,11 @@ static int prepare_signaling(struct drm_device
>> *dev,
>> }
>> return 0;
>> +
>> +err_free_event:
>> + drm_event_cancel_free(dev, &crtc_state->event->base);
>> + crtc_state->event = NULL;
>> + return ret;
>> }
>> static void complete_signaling(struct drm_device *dev,
>>
>> ---
>> base-commit: 4d4be202165e832d74849b4a68e289a2a377039c
>> change-id: 20260728-drm_pending_vblank_event_leak-a36cedb296ba
>>
>> Best regards,
>> --
>> Thadeu Lima de Souza Cascardo <cascardo@igalia.com>
>>
>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH v4] drm: Fix drm_pending_vblank_event leak in error path for out_fence_ptr
2026-08-26 11:46 [PATCH v4] drm: Fix drm_pending_vblank_event leak in error path for out_fence_ptr Thadeu Lima de Souza Cascardo
2026-09-11 13:58 ` Melissa Wen
@ 2026-09-11 14:43 ` Markus Elfring
1 sibling, 0 replies; 4+ messages in thread
From: Markus Elfring @ 2026-09-11 14:43 UTC (permalink / raw)
To: Thadeu Lima de Souza Cascardo, dri-devel, kernel-dev,
David Airlie, Maarten Lankhorst, Maxime Ripard, Melissa Wen,
Simona Vetter, Thomas Zimmermann
Cc: sashiko-bot, LKML, kernel-janitors
> +++ b/drivers/gpu/drm/drm_atomic_uapi.c
> @@ -1458,10 +1458,12 @@ static int prepare_signaling(struct drm_device *dev,
> struct dma_fence *fence;
> struct drm_out_fence_state *f;
>
> + ret = -ENOMEM;
> +
How do you think about to refine the source code a bit more?
I imagine that this assignment statement can be moved behind an additional label
like e_nomem.
> f = krealloc(*fence_state, sizeof(**fence_state) *
> (*num_fences + 1), GFP_KERNEL);
> if (!f)
> - return -ENOMEM;
> + goto err_free_event;
>
> memset(&f[*num_fences], 0, sizeof(*f));
>
…
> @@ -1531,6 +1533,11 @@ static int prepare_signaling(struct drm_device *dev,
> }
>
> return 0;
> +
> +err_free_event:
> + drm_event_cancel_free(dev, &crtc_state->event->base);
+err_reset_event:
Would you like to avoid a bit of duplicate source code from an other if branch?
https://elixir.bootlin.com/linux/v7.3-rc1/source/drivers/gpu/drm/drm_atomic_uapi.c#L1457-L1461
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/coding-style.rst?h=v7.3-rc2#n572
> + crtc_state->event = NULL;
> + return ret;
> }
Regards,
Markus
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2026-09-11 14:44 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-08-26 11:46 [PATCH v4] drm: Fix drm_pending_vblank_event leak in error path for out_fence_ptr Thadeu Lima de Souza Cascardo
2026-09-11 13:58 ` Melissa Wen
2026-09-11 14:26 ` Melissa Wen
2026-09-11 14:43 ` Markus Elfring
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®