On Thu, 14 Dec 2023 at 19:26, Jon Hunter <jonathanh@xxxxxxxxxx> wrote:
>
>
>
> On 08/12/2023 10:46, Thierry Reding wrote:
> > From: Thierry Reding <treding@xxxxxxxxxx>
> >
> > Commit 12c9b05da918 ("drm/nouveau/imem: support allocations not
> > preserved across suspend") uses container_of() to cast from struct
> > nvkm_memory to struct nvkm_instobj, assuming that all instance objects
> > are derived from struct nvkm_instobj. For the gk20a family that's not
> > the case and they are derived from struct nvkm_memory instead. This
> > causes some subtle data corruption (nvkm_instobj.preserve ends up
> > mapping to gk20a_instobj.vaddr) that causes a NULL pointer dereference
> > in gk20a_instobj_acquire_iommu() (and possibly elsewhere) and also
> > prevents suspend/resume from working.
> >
> > Fix this by making struct gk20a_instobj derive from struct nvkm_instobj
> > instead.
> >
> > Fixes: 12c9b05da918 ("drm/nouveau/imem: support allocations not preserved across suspend")
> > Reported-by: Jonathan Hunter <jonathanh@xxxxxxxxxx>
> > Signed-off-by: Thierry Reding <treding@xxxxxxxxxx>
I've applied this to drm-fixes.
Dave.
