✗ Fi.CI.BAT: warning for series starting with [1/3] drm/i915: Handle GEN6_PMINTRMSK for proper masking of RPS interrupts with GuC submission (rev2)

Chris Wilson chris at chris-wilson.co.uk
Fri Jan 13 20:29:40 UTC 2017


On Fri, Jan 13, 2017 at 07:53:26PM -0000, Patchwork wrote:
> == Series Details ==
> 
> Series: series starting with [1/3] drm/i915: Handle GEN6_PMINTRMSK for proper masking of RPS interrupts with GuC submission (rev2)
> URL   : https://patchwork.freedesktop.org/series/17958/
> State : warning
> 
> == Summary ==
> 
> Series 17958v2 Series without cover letter
> https://patchwork.freedesktop.org/api/1.0/series/17958/revisions/2/mbox/
> 
> Test gem_exec_nop:
>         Subgroup basic-parallel:
>                 pass       -> DMESG-WARN (fi-hsw-4770)

Huh.

We definitely call sw_fence_fini (to move the fence from the destroyed
to free state) in i915_fence_release before kmem_cache_free.  That should
go to a per-cpu freelist, so a subsequent alloc should be ordered (I
think). Doubtful that we will see this frequently enough to debug, but
my first instinct would be to throw a smp_wmb into i915_sw_fence_fini().
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre


More information about the Intel-gfx-trybot mailing list