Re: [PATCH v2] drm/panthor: fix for dma-fence safe access rules

[Chia-I Wu]

On Fri, Dec 5, 2025 at 4:43 AM Boris Brezillon
<boris.brezillon@xxxxxxxxxxxxx> wrote:
>
> On Thu,  4 Dec 2025 09:45:45 -0800
> Chia-I Wu <olvaffe@xxxxxxxxx> wrote:
>
> > Commit 506aa8b02a8d6 ("dma-fence: Add safe access helpers and document
> > the rules") details the dma-fence safe access rules. The most common
> > culprit is that drm_sched_fence_get_timeline_name may race with
> > group_free_queue.
> >
> > Signed-off-by: Chia-I Wu <olvaffe@xxxxxxxxx>
> > Reviewed-by: Boris Brezillon <boris.brezillon@xxxxxxxxxxxxx>
> > Reviewed-by: Liviu Dudau <liviu.dudau@xxxxxxx>
> > Reviewed-by: Steven Price <steven.price@xxxxxxx>
> >
> > ---
> > v2: collect R-bs and drop misleading Fixes: tag
>
> Should we at least
>
> Cc: stable@xxxxxxxxxxxxxxx
>
> so this gets considered for backporting?
This can be considered a follow-up to commit 506aa8b02a8d6. I think

  Cc: stable@xxxxxxxxxxxxxxx # v6.17+

is appropriate. (Not sure if I should send v3 or leave it to
maintainers, because I was once told to never add the line and always
leave it to maintainers).

To help make the decision, commit 506aa8b02a8d6 itself narrowed the
race window greatly. This follow-up closes the remaining case where
panthor signals the fence and destroys the group while the consumer
holds the rcu read lock.


>
> > ---
> >  drivers/gpu/drm/panthor/panthor_sched.c | 4 ++++
> >  1 file changed, 4 insertions(+)
> >
> > diff --git a/drivers/gpu/drm/panthor/panthor_sched.c b/drivers/gpu/drm/panthor/panthor_sched.c
> > index 33b9ef537e359..a8b1347e4da71 100644
> > --- a/drivers/gpu/drm/panthor/panthor_sched.c
> > +++ b/drivers/gpu/drm/panthor/panthor_sched.c
> > @@ -23,6 +23,7 @@
> >  #include <linux/module.h>
> >  #include <linux/platform_device.h>
> >  #include <linux/pm_runtime.h>
> > +#include <linux/rcupdate.h>
> >
> >  #include "panthor_devfreq.h"
> >  #include "panthor_device.h"
> > @@ -923,6 +924,9 @@ static void group_release_work(struct work_struct *work)
> >                                                  release_work);
> >       u32 i;
> >
> > +     /* dma-fences may still be accessing group->queues under rcu lock. */
> > +     synchronize_rcu();
> > +
> >       for (i = 0; i < group->queue_count; i++)
> >               group_free_queue(group, group->queues[i]);
> >
>