[PATCH] drm/i915: remember to call i915_sw_fence_fini

Ville Syrjälä ville.syrjala at linux.intel.com
Fri Sep 24 15:35:58 UTC 2021


On Fri, Sep 24, 2021 at 03:46:46PM +0100, Matthew Auld wrote:
> Seems to fix some object-debug splat which appeared while debugging
> something unrelated.
> 
> v2: s/guc_blocked/guc_state.blocked/
> 
> Signed-off-by: Matthew Auld <matthew.auld at intel.com>
> Cc: Ville Syrjälä <ville.syrjala at linux.intel.com>
> Cc: Matthew Brost <matthew.brost at intel.com>

Seems to help.
Tested-by: Ville Syrjälä <ville.syrjala at linux.intel.com>

Bonus points if someone can figure out how this getting past CI in its
current state. I can trigger it very easily by pretty much running any
igt. Though for some reason it doesn't always trigger on the first run
of an igt after boot, sometimes takes a few runs. Dunno if the checks
are  a bit lazy or something.

> ---
>  drivers/gpu/drm/i915/gt/intel_context.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/gpu/drm/i915/gt/intel_context.c b/drivers/gpu/drm/i915/gt/intel_context.c
> index ff637147b1a9..e9a0cad5c34d 100644
> --- a/drivers/gpu/drm/i915/gt/intel_context.c
> +++ b/drivers/gpu/drm/i915/gt/intel_context.c
> @@ -419,6 +419,7 @@ void intel_context_fini(struct intel_context *ce)
>  
>  	mutex_destroy(&ce->pin_mutex);
>  	i915_active_fini(&ce->active);
> +	i915_sw_fence_fini(&ce->guc_state.blocked);
>  }
>  
>  void i915_context_module_exit(void)
> -- 
> 2.26.3

-- 
Ville Syrjälä
Intel


More information about the dri-devel mailing list