Quoting Tvrtko Ursulin (2019-12-17 17:21:28)
>
> On 16/12/2019 12:53, Chris Wilson wrote:
> > Quoting Tvrtko Ursulin (2019-12-16 12:07:01)
> >> diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
> >> index 0781b6326b8c..9fcbcb6d6f76 100644
> >> --- a/drivers/gpu/drm/i915/i915_drv.h
> >> +++ b/drivers/gpu/drm/i915/i915_drv.h
> >> @@ -224,6 +224,20 @@ struct drm_i915_file_private {
> >> /** ban_score: Accumulated score of all ctx bans and fast hangs. */
> >> atomic_t ban_score;
> >> unsigned long hang_timestamp;
> >> +
> >> + struct i915_drm_client {
> >> + unsigned int id;
> >> +
> >> + struct pid *pid;
> >> + char *name;
> >
> > Hmm. Should we scrap i915_gem_context.pid and just use the client.pid?
>
> Or maybe leave as it so I don't have to worry about ctx vs client
> lifetime. In other words places where we access ctx->pid and the client
> is maybe long gone. I don't want to ref count clients, or maybe I do..
> hmm.. keeping GPU load of a client which exited and left work running
> visible?
Yeah. If we don't and all the GPU time is being hogged by zombies, users
of the interface will not be impressed they can't identify those. Next
up, kill(client_id, SIGKILL).
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
