Re: [PATCH v2 4/4] drm/gem: Make the GEM LRU lock part of drm_device

[Boris Brezillon]

On Mon, 11 May 2026 17:27:13 +0100
Liviu Dudau <liviu.dudau@xxxxxxx> wrote:

> On Mon, May 11, 2026 at 09:16:49AM -0700, Rob Clark wrote:
> > On Mon, May 11, 2026 at 8:44 AM Boris Brezillon
> > <boris.brezillon@xxxxxxxxxxxxx> wrote:  
> > >
> > > On Mon, 11 May 2026 06:18:44 -0700
> > > Rob Clark <rob.clark@xxxxxxxxxxxxxxxx> wrote:
> > >  
> > > > On Fri, May 8, 2026 at 3:41 AM Boris Brezillon
> > > > <boris.brezillon@xxxxxxxxxxxxx> wrote:  
> > > > >
> > > > > Recently, a few races have been discovered in the GEM LRU logic, all
> > > > > of them caused by the fact the LRU lock is accessed through
> > > > > gem->lru->lock, and that lock itself also protects changes to
> > > > > gem->lru, leading to situations where gem->lru needs to first be
> > > > > accessed without the lock held, to then get the lru to access the lock
> > > > > through and finally take the lock and do the expected operation.
> > > > >
> > > > > Currently, the two drivers making use of this API declare device-wide
> > > > > locks, and there's no clue that we will ever have a driver that wants
> > > > > different pools of LRUs protected by different locks under the same
> > > > > drm_device. So we're better off moving this lock to drm_device and
> > > > > always locking it through obj->dev->gem_lru_mutex, or directly through
> > > > > dev->gem_lru_mutex.
> > > > >
> > > > > If anyone ever needs more fine-grained locking, this can be revisited
> > > > > to pass some drm_gem_lru_pool object represent the pool of LRUs under
> > > > > a specific lock, but for now, the per-device lock seems to be enough.
> > > > >
> > > > > Signed-off-by: Boris Brezillon <boris.brezillon@xxxxxxxxxxxxx>  
> > > >
> > > > Reviewed-by: Rob Clark <rob.clark@xxxxxxxxxxxxxxxx>  
> > >
> > > If every one is happy with patch 4, I'm tempted to drop patch 1 and 3,
> > > since they are irrelevant if the locking is fixed the way it is in this
> > > patch. Question is, do we care about patch 2 being backported to stable
> > > kernels then?  
> > 
> > Or backport patch 4?  It's a bit bigger, but mostly mechanical.. with
> > only issue being backporting across addition of panthor shrinker, I
> > guess?  
> 
> I'm agreeing with Rob here, but it makes a mess of the backport. Maybe split
> patch 4 into the drm_gem.c and msm changes and make another one for panthor,
> as that is still in linux-next?

We can't really do that if we want to keep things bisectable, but I can
prepare a version against drm-misc-fixes which wouldn't contain any of
the panthor changes, and then ask drm-misc maintainers to backmerge the
next -rc into drm-misc-next and fix things up in panthor as part of the
backmerge.