s right, but the last change 
which tried to fix it but did not. It sounded more logical to continue 
the chain on fixes but dunno. Could replace it with the original one 
just as well.
the report as Reference: 
https://lore.kernel.org/dri-devel/aNFbXq8OeYl3QSdm@stanley.mountain/

?
I was mulling it over but then thought the kerneldoc for 
drm_sched_job_add_dependency() is good enough since it explains the 
function always consumes the reference and dma_resv_for_each_fence() 
does not hold one over the iteration. I can add it.
Given the relative size of the diff versus the function itself it did 
not look like a big deal to reduce to one return statement while 
touching it, but I can keep two returns just as well, no problem.

Regards,

Tvrtko[PATCH] drm/sched: Fix potential double free in drm_sched_job_add_resv_dependenciesTvrtko Ursulin <tvrtko.ursulin@igalia.com> undefinedphasta@kernel.org, dri-devel@lists.freedesktop.org undefined undefined undefined undefined undefined undefined undefined undefined undefined undefined undefined„dƒ•e