Re: [Intel-gfx] [PATCH 1/4] drm/i915/gt: Prevent queuing retire workers on the virtual engine

2020-02-07 Thread Tvrtko Ursulin
On 06/02/2020 20:49, Chris Wilson wrote: Virtual engines are fleeting. They carry a reference count and may be freed when their last request is retired. This makes them unsuitable for the task of housing engine->retire.work so assert that it is not used. Tvrtko tracked down an instance where

Re: [Intel-gfx] [PATCH 1/4] drm/i915/gt: Prevent queuing retire workers on the virtual engine

2020-02-07 Thread Mika Kuoppala
Chris Wilson writes: > Quoting Mika Kuoppala (2020-02-07 09:13:22) >> Chris Wilson writes: >> >> > Virtual engines are fleeting. They carry a reference count and may be freed >> > when their last request is retired. This makes them unsuitable for the >> > task of housing engine->retire.work so

Re: [Intel-gfx] [PATCH 1/4] drm/i915/gt: Prevent queuing retire workers on the virtual engine

2020-02-07 Thread Chris Wilson
Quoting Mika Kuoppala (2020-02-07 09:13:22) > Chris Wilson writes: > > > Virtual engines are fleeting. They carry a reference count and may be freed > > when their last request is retired. This makes them unsuitable for the > > task of housing engine->retire.work so assert that it is not used. >

Re: [Intel-gfx] [PATCH 1/4] drm/i915/gt: Prevent queuing retire workers on the virtual engine

2020-02-07 Thread Mika Kuoppala
Chris Wilson writes: > Virtual engines are fleeting. They carry a reference count and may be freed > when their last request is retired. This makes them unsuitable for the > task of housing engine->retire.work so assert that it is not used. There is chicken and egg problem here that I fail to

[Intel-gfx] [PATCH 1/4] drm/i915/gt: Prevent queuing retire workers on the virtual engine

2020-02-06 Thread Chris Wilson
Virtual engines are fleeting. They carry a reference count and may be freed when their last request is retired. This makes them unsuitable for the task of housing engine->retire.work so assert that it is not used. Tvrtko tracked down an instance where we did indeed violate this rule. In