[PATCH v2] drm/sched: Mark scheduler work queues with WQ_MEM_RECLAIM

Matthew Brost matthew.brost at intel.com
Thu Oct 24 15:47:14 UTC 2024


On Thu, Oct 24, 2024 at 05:35:47PM +0200, Philipp Stanner wrote:
> On Wed, 2024-10-23 at 16:59 -0700, Matthew Brost wrote:
> > drm_gpu_scheduler.submit_wq is used to submit jobs, jobs are in the
> > path
> > of dma-fences, and dma-fences are in the path of reclaim. Mark
> > scheduler
> > work queue with WQ_MEM_RECLAIM to ensure forward progress during
> > reclaim; without WQ_MEM_RECLAIM, work queues cannot make forward
> > progress during reclaim.
> > 
> > v2:
> >  - Fixes tags (Philipp)
> >  - Reword commit message (Philipp)
> > 
> > Cc: Luben Tuikov <ltuikov89 at gmail.com>
> > Cc: Danilo Krummrich <dakr at kernel.org>
> > Cc: Philipp Stanner <pstanner at redhat.com>
> > Cc: stable at vger.kernel.org
> > Fixes: 34f50cc6441b ("drm/sched: Use drm sched lockdep map for
> > submit_wq")
> > Fixes: a6149f039369 ("drm/sched: Convert drm scheduler to use a work
> > queue rather than kthread")
> > Signed-off-by: Matthew Brost <matthew.brost at intel.com>
> > ---
> >  drivers/gpu/drm/scheduler/sched_main.c | 5 +++--
> >  1 file changed, 3 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/scheduler/sched_main.c
> > b/drivers/gpu/drm/scheduler/sched_main.c
> > index 540231e6bac6..df0a5abb1400 100644
> > --- a/drivers/gpu/drm/scheduler/sched_main.c
> > +++ b/drivers/gpu/drm/scheduler/sched_main.c
> > @@ -1283,10 +1283,11 @@ int drm_sched_init(struct drm_gpu_scheduler
> > *sched,
> >  		sched->own_submit_wq = false;
> >  	} else {
> >  #ifdef CONFIG_LOCKDEP
> > -		sched->submit_wq =
> > alloc_ordered_workqueue_lockdep_map(name, 0,
> > +		sched->submit_wq =
> > alloc_ordered_workqueue_lockdep_map(name,
> > +								    
> >    WQ_MEM_RECLAIM,
> >  								    
> >    &drm_sched_lockdep_map);
> >  #else
> > -		sched->submit_wq = alloc_ordered_workqueue(name, 0);
> > +		sched->submit_wq = alloc_ordered_workqueue(name,
> > WQ_MEM_RECLAIM);
> >  #endif
> >  		if (!sched->submit_wq)
> >  			return -ENOMEM;
> 
> 
> Cool, thx – looks good from my POV.
> 

Can I get a RB?

> Since you now sent this patch as a single one, what would be the
> preferred merge plan for this? Your XE-Series doesn't depend on this
> IIUC, so should we take this patch here separately into drm-misc-next?
> 

Merge this one to drm-misc and we will backport into drm-xe-next.

Matt

> 
> Regards,
> P.
> 


More information about the dri-devel mailing list