[Intel-gfx] [PATCH 08/19] drm/i915: Fix runtime pm handling in i915_gem_shrink

[Matthew Auld]

On Wed, 29 Sept 2021 at 09:37, Matthew Auld
<matthew.william.auld at gmail.com> wrote:
>
> On Mon, 30 Aug 2021 at 13:10, Maarten Lankhorst
> <maarten.lankhorst at linux.intel.com> wrote:
> >
> > We forgot to call intel_runtime_pm_put on error, fix it!
> >
> > Signed-off-by: Maarten Lankhorst <maarten.lankhorst at linux.intel.com>
> > Fixes: cf41a8f1dc1e ("drm/i915: Finally remove obj->mm.lock.")
> > Cc: Thomas Hellström <thomas.hellstrom at linux.intel.com>
> > Cc: Daniel Vetter <daniel.vetter at ffwll.ch>
> > Cc: <stable at vger.kernel.org> # v5.13+
>
> How does the err handling work? gem_shrink is meant to return the
> number of pages reclaimed which is an unsigned long, and yet we are
> also just returning the err here? Fortunately it looks like nothing is
> calling gem_shrinker with an actual ww context, so nothing will hit
> this yet? I think the interface needs to be reworked or something.

Can we just remove the ww context argument, or is that needed for
something in the future?

>
> > ---
> >  drivers/gpu/drm/i915/gem/i915_gem_shrinker.c | 7 +++++--
> >  1 file changed, 5 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/gem/i915_gem_shrinker.c b/drivers/gpu/drm/i915/gem/i915_gem_shrinker.c
> > index e382b7f2353b..5ab136ffdeb2 100644
> > --- a/drivers/gpu/drm/i915/gem/i915_gem_shrinker.c
> > +++ b/drivers/gpu/drm/i915/gem/i915_gem_shrinker.c
> > @@ -118,7 +118,7 @@ i915_gem_shrink(struct i915_gem_ww_ctx *ww,
> >         intel_wakeref_t wakeref = 0;
> >         unsigned long count = 0;
> >         unsigned long scanned = 0;
> > -       int err;
> > +       int err = 0;
> >
> >         /* CHV + VTD workaround use stop_machine(); need to trylock vm->mutex */
> >         bool trylock_vm = !ww && intel_vm_no_concurrent_access_wa(i915);
> > @@ -242,12 +242,15 @@ i915_gem_shrink(struct i915_gem_ww_ctx *ww,
> >                 list_splice_tail(&still_in_list, phase->list);
> >                 spin_unlock_irqrestore(&i915->mm.obj_lock, flags);
> >                 if (err)
> > -                       return err;
> > +                       break;
> >         }
> >
> >         if (shrink & I915_SHRINK_BOUND)
> >                 intel_runtime_pm_put(&i915->runtime_pm, wakeref);
> >
> > +       if (err)
> > +               return err;
> > +
> >         if (nr_scanned)
> >                 *nr_scanned += scanned;
> >         return count;
> > --
> > 2.32.0
> >