[Intel-gfx] [PATCH] drm/i915: Clear stop-engine for a pardoned reset

[Chris Wilson]

Quoting Mika Kuoppala (2018-08-15 09:52:18)
> Chris Wilson <chris at chris-wilson.co.uk> writes:
> 
> > If we pardon a per-engine reset, we may leave the STOP_RING bit asserted
> > in RING_MI_MODE resulting in the engine hanging. Unconditionally clear
> > it on the per-engine exit path as we know that either we skipped the
> > reset and so need the cancellation, or the reset was successful and the
> > cancellation is a no-op, or there was an error and we will follow up
> > with a full-reset or wedging (both of which will stop the engines again
> > as required).
> >
> > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > Cc: Mika Kuoppala <mika.kuoppala at linux.intel.com>
> > ---
> >  drivers/gpu/drm/i915/i915_drv.c         |  1 +
> >  drivers/gpu/drm/i915/intel_engine_cs.c  | 10 ++++++++++
> >  drivers/gpu/drm/i915/intel_ringbuffer.h |  1 +
> >  3 files changed, 12 insertions(+)
> >
> > diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
> > index 9dce55182c3a..41111f2a9c39 100644
> > --- a/drivers/gpu/drm/i915/i915_drv.c
> > +++ b/drivers/gpu/drm/i915/i915_drv.c
> > @@ -2079,6 +2079,7 @@ int i915_reset_engine(struct intel_engine_cs *engine, const char *msg)
> >               goto out;
> >  
> >  out:
> > +     intel_engine_cancel_stop_cs(engine);
> >       i915_gem_reset_finish_engine(engine);
> 
> Should we just lift the whole stop/start dance into
> gem_reset_prepare|finish_engine()s?

No, because it is also used by wedging where we do want the asymmetry.
Been there, done that, have the gem_eio bruises.
-Chris