Re: [PATCH 2/5] drm/msm: Remove abuse of drm_exec internals

["Thomas Hellström" <thomas.hellstrom@linux.intel.com>]

On Tue, 2026-03-31 at 12:08 -0700, Rob Clark wrote:
> On Tue, Mar 31, 2026 at 2:21 AM Thomas Hellström
> <thomas.hellstrom@linux.intel.com> wrote:
> > 
> > The code was reading drm_exec internal state to determine whether
> > the drm_exec structure had been initialized or not, and therefore
> > needed cleaning up, relying on undocumented behaviour.
> > 
> > Instead add a bool to struct msm_gem_submit to indicate whether
> > drm_exec cleaning up is needed.
> > 
> > Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
> 
> Reviewed-by: Rob Clark <rob.clark@oss.qualcomm.com>
> 
> This is pretty stand-alone, so I can pick it up for v7.1.  Or ack for
> landing it via drm-misc with the rest of the series if that is easier
> for you.  It shouldn't conflict with anything in flight.

Thanks Rob. Please pick it up and I'll exclude it from the next
iteration of the series.

Thanks,
Thomas

> 
> BR,
> -R
> 
> > ---
> >  drivers/gpu/drm/msm/msm_gem.h        | 1 +
> >  drivers/gpu/drm/msm/msm_gem_submit.c | 4 +++-
> >  2 files changed, 4 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/gpu/drm/msm/msm_gem.h
> > b/drivers/gpu/drm/msm/msm_gem.h
> > index cb32093fda47..762e546d25ef 100644
> > --- a/drivers/gpu/drm/msm/msm_gem.h
> > +++ b/drivers/gpu/drm/msm/msm_gem.h
> > @@ -452,6 +452,7 @@ struct msm_gem_submit {
> >         bool bos_pinned : 1;
> >         bool fault_dumped:1;/* Limit devcoredump dumping to one per
> > submit */
> >         bool in_rb : 1;     /* "sudo" mode, copy cmds into RB */
> > +       bool has_exec : 1;  /* @exec is initialized. */
> >         struct msm_ringbuffer *ring;
> >         unsigned int nr_cmds;
> >         unsigned int nr_bos;
> > diff --git a/drivers/gpu/drm/msm/msm_gem_submit.c
> > b/drivers/gpu/drm/msm/msm_gem_submit.c
> > index 75d9f3574370..26ea8a28be47 100644
> > --- a/drivers/gpu/drm/msm/msm_gem_submit.c
> > +++ b/drivers/gpu/drm/msm/msm_gem_submit.c
> > @@ -278,6 +278,7 @@ static int submit_lock_objects_vmbind(struct
> > msm_gem_submit *submit)
> >         int ret = 0;
> > 
> >         drm_exec_init(&submit->exec, flags, submit->nr_bos);
> > +       submit->has_exec = true;
> > 
> >         drm_exec_until_all_locked (&submit->exec) {
> >                 ret = drm_gpuvm_prepare_vm(submit->vm, exec, 1);
> > @@ -304,6 +305,7 @@ static int submit_lock_objects(struct
> > msm_gem_submit *submit)
> >                 return submit_lock_objects_vmbind(submit);
> > 
> >         drm_exec_init(&submit->exec, flags, submit->nr_bos);
> > +       submit->has_exec = true;
> > 
> >         drm_exec_until_all_locked (&submit->exec) {
> >                 ret = drm_exec_lock_obj(&submit->exec,
> > @@ -523,7 +525,7 @@ static void submit_cleanup(struct
> > msm_gem_submit *submit, bool error)
> >         if (error)
> >                 submit_unpin_objects(submit);
> > 
> > -       if (submit->exec.objects)
> > +       if (submit->has_exec)
> >                 drm_exec_fini(&submit->exec);
> > 
> >         /* if job wasn't enqueued to scheduler, early retirement:
> > */
> > --
> > 2.53.0
> >