Hi Matt,

On Mon, Mar 18, 2024 at 02:50:00PM +0100, Nirmoy Das wrote:

Add a warn for NULL job when sync->type is
DRM_XE_SYNC_TYPE_USER_FENCE. This should be a programming
error and should never happen so warn and let the kernel crash
if that ever happens.

Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Nirmoy Das <nirmoy.das@intel.com>
---
 drivers/gpu/drm/xe/xe_sync.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/gpu/drm/xe/xe_sync.c b/drivers/gpu/drm/xe/xe_sync.c
index 02c9577fe418..2b45950aa0af 100644
--- a/drivers/gpu/drm/xe/xe_sync.c
+++ b/drivers/gpu/drm/xe/xe_sync.c
@@ -255,6 +255,7 @@ void xe_sync_entry_signal(struct xe_sync_entry *sync, struct xe_sched_job *job,
 			dma_fence_put(fence);
 		}
 	} else if (sync->type == DRM_XE_SYNC_TYPE_USER_FENCE) {
+		XE_WARN_ON(job);

For a warn on you this inverted - a WARN_ON(true) will pop.

But I think we should actually use an xe_assert here which has this
opposite behavior - pops on xe_assert(xe, false).

So the I think it should be:

xe_assert(xe, job);

xe_assert needed a xe device struct which I can't find a better way to retrieve than to pass it on as new argument

I think that is bit too much change for it. I will resend with  WARN_ON(!job).

Regards,

Nirmoy

Matt

 		job->user_fence.used = true;
 		job->user_fence.addr = sync->addr;
 		job->user_fence.value = sync->timeline_value;
-- 
2.42.0