[Intel-gfx] [PATCH] [RFC] mm/shrinker: Add a shrinker flag to always shrink a bit

Daniel Vetter daniel.vetter at ffwll.ch
Wed Sep 18 23:57:04 PDT 2013

On Wed, Sep 18, 2013 at 10:38 PM, Dave Chinner <david at fromorbit.com> wrote:
> No, that's wrong. ->count_objects should never ass SHRINK_STOP.
> Indeed, it should always return a count of objects in the cache,
> regardless of the context.
> SHRINK_STOP is for ->scan_objects to tell the shrinker it can make
> any progress due to the context it is called in. This allows the
> shirnker to defer the work to another call in a different context.
> However, if ->count-objects doesn't return a count, the work that
> was supposed to be done cannot be deferred, and that is what
> ->count_objects should always return the number of objects in the
> cache.

So we should rework the locking in the drm/i915 shrinker to be able to
always count objects? Thus far no one screamed yet that we're not
really able to do that in all call contexts ...

So should I revert 81e49f or will the early return 0; completely upset
the core shrinker logic?
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch

More information about the dri-devel mailing list