[Intel-gfx] [PATCH v2] drm/i915/gt: move files more files into debugfs

[Chris Wilson]

Quoting Tvrtko Ursulin (2020-03-20 12:01:14)
> 
> 
> On 20/03/2020 11:45, Andi Shyti wrote:
> > Hi Chris,
> > 
> > On Fri, Mar 20, 2020 at 11:11:46AM +0000, Chris Wilson wrote:
> >> Quoting Andi Shyti (2020-03-20 03:49:01)
> >>> From: Andi Shyti <andi.shyti at intel.com>
> >>>
> >>> The following interfaces:
> >>>
> >>> i915_wedged
> >>> i915_forcewake_user
> >>> i915_gem_interrupt
> >>> i915_sseu_status
> >>>
> >>> are dependent on gt values. Put them inside gt/ and drop the
> >>> "i915_" prefix name. This would be the new structure:
> >>>
> >>>    gt
> >>>    |
> >>>    +-- forcewake_user
> >>>    |
> >>>    +-- interrupt_info_show
> >>
> >> Please tell me you didn't actually include _show :)
> > 
> > You know me, everything can happen!
> > I did overlook indeed, but I had to check if I actually did
> > include _show. Thanks for spotting it.
> > 
> >>>    |
> >>>    +-- sseu_status
> >>>    |
> >>>    +-- wedge
> >>
> >> The world will rejoice when it's stopped being called wedged.
> >> Well Mika will at any rate.
> > 
> > well, I did keep the original name.
> > 
> >> 'echo rcs0 > reset' maybe? Yeah. Wait, but rcs0 is uabi name, so top
> >> level.
> >>
> >> Oh well, I definitely do not think copying a mistake is a good idea.
> > 
> > OK, I'll call it reset
> 
> Wedge is wedge and reset is reset, or is it not?

i915_wedged is reset :)

Hysterical raisons.

But my main question is what do you feed into a gt/reset? Currently we
have a random bitmask to reset a group of engines. Should we just go and
put reset into sysfs/engine/ ?
-Chris