[Intel-gfx] [PATCH 2/2] drm/i915: Prevent concurrent tiling/framebuffer modifications
Chris Wilson
chris at chris-wilson.co.uk
Wed Mar 1 15:47:55 UTC 2017
On Wed, Mar 01, 2017 at 03:09:36PM +0200, Ville Syrjälä wrote:
> On Tue, Feb 28, 2017 at 04:22:33PM +0000, Chris Wilson wrote:
> > Reintroduce a lock around tiling vs framebuffer creation to prevent
> > modification of the obj->tiling_and_stride whilst the framebuffer is
> > being created. Rather than use struct_mutex once again, use the
> > per-object lock - this will also be required in future to prevent
> > changing the tiling whilst submitting rendering.
> >
> > Fixes: 24dbf51a5517 ("drm/i915: struct_mutex is not required for allocating the framebuffer")
> > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > Cc: Ville Syrjälä <ville.syrjala at linux.intel.com>
> > ---
> > drivers/gpu/drm/i915/i915_gem_object.h | 18 +++++++++++++++++-
> > drivers/gpu/drm/i915/i915_gem_shrinker.c | 2 +-
> > drivers/gpu/drm/i915/i915_gem_tiling.c | 9 ++++++++-
> > drivers/gpu/drm/i915/intel_display.c | 25 ++++++++++++++++---------
> > 4 files changed, 42 insertions(+), 12 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/i915_gem_object.h b/drivers/gpu/drm/i915/i915_gem_object.h
> > index ad1bc0b1a0c2..8c02c8ec2a3b 100644
> > --- a/drivers/gpu/drm/i915/i915_gem_object.h
> > +++ b/drivers/gpu/drm/i915/i915_gem_object.h
> > @@ -169,7 +169,7 @@ struct drm_i915_gem_object {
> > struct reservation_object *resv;
> >
> > /** References from framebuffers, locks out tiling changes. */
> > - atomic_t framebuffer_references;
> > + unsigned int framebuffer_references;
> >
> > /** Record of address bit 17 of each page at last unbind. */
> > unsigned long *bit_17;
> > @@ -263,6 +263,16 @@ extern void drm_gem_object_unreference(struct drm_gem_object *);
> > __deprecated
> > extern void drm_gem_object_unreference_unlocked(struct drm_gem_object *);
> >
> > +static inline void i915_gem_object_lock(struct drm_i915_gem_object *obj)
> > +{
> > + reservation_object_lock(obj->resv, NULL);
> > +}
> > +
> > +static inline void i915_gem_object_unlock(struct drm_i915_gem_object *obj)
> > +{
> > + reservation_object_unlock(obj->resv);
> > +}
> > +
> > static inline bool
> > i915_gem_object_has_struct_page(const struct drm_i915_gem_object *obj)
> > {
> > @@ -303,6 +313,12 @@ i915_gem_object_clear_active_reference(struct drm_i915_gem_object *obj)
> >
> > void __i915_gem_object_release_unless_active(struct drm_i915_gem_object *obj);
> >
> > +static inline bool
> > +i915_gem_object_is_framebuffer(const struct drm_i915_gem_object *obj)
> > +{
> > + return READ_ONCE(obj->framebuffer_references);
> > +}
> > +
> > static inline unsigned int
> > i915_gem_object_get_tiling(struct drm_i915_gem_object *obj)
> > {
> > diff --git a/drivers/gpu/drm/i915/i915_gem_shrinker.c b/drivers/gpu/drm/i915/i915_gem_shrinker.c
> > index 7e3bb48e043e..630697001b38 100644
> > --- a/drivers/gpu/drm/i915/i915_gem_shrinker.c
> > +++ b/drivers/gpu/drm/i915/i915_gem_shrinker.c
> > @@ -210,7 +210,7 @@ i915_gem_shrink(struct drm_i915_private *dev_priv,
> >
> > if (!(flags & I915_SHRINK_ACTIVE) &&
> > (i915_gem_object_is_active(obj) ||
> > - atomic_read(&obj->framebuffer_references)))
> > + i915_gem_object_is_framebuffer(obj)))
> > continue;
> >
> > if (!can_release_pages(obj))
> > diff --git a/drivers/gpu/drm/i915/i915_gem_tiling.c b/drivers/gpu/drm/i915/i915_gem_tiling.c
> > index c1d669e32f41..ad5e05f6b836 100644
> > --- a/drivers/gpu/drm/i915/i915_gem_tiling.c
> > +++ b/drivers/gpu/drm/i915/i915_gem_tiling.c
> > @@ -238,7 +238,7 @@ i915_gem_object_set_tiling(struct drm_i915_gem_object *obj,
> > if ((tiling | stride) == obj->tiling_and_stride)
> > return 0;
> >
> > - if (atomic_read(&obj->framebuffer_references))
> > + if (i915_gem_object_is_framebuffer(obj))
> > return -EBUSY;
> >
> > /* We need to rebind the object if its current allocation
> > @@ -258,6 +258,12 @@ i915_gem_object_set_tiling(struct drm_i915_gem_object *obj,
> > if (err)
> > return err;
> >
> > + i915_gem_object_lock(obj);
> > + if (i915_gem_object_is_framebuffer(obj)) {
> > + i915_gem_object_unlock(obj);
> > + return -EBUSY;
> > + }
> > +
> > /* If the memory has unknown (i.e. varying) swizzling, we pin the
> > * pages to prevent them being swapped out and causing corruption
> > * due to the change in swizzling.
> > @@ -294,6 +300,7 @@ i915_gem_object_set_tiling(struct drm_i915_gem_object *obj,
> > }
> >
> > obj->tiling_and_stride = tiling | stride;
> > + i915_gem_object_unlock(obj);
> >
> > /* Force the fence to be reacquired for GTT access */
> > i915_gem_release_mmap(obj);
> > diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
> > index 77936ddd860a..62a1e628e399 100644
> > --- a/drivers/gpu/drm/i915/intel_display.c
> > +++ b/drivers/gpu/drm/i915/intel_display.c
> > @@ -14255,7 +14255,10 @@ static void intel_user_framebuffer_destroy(struct drm_framebuffer *fb)
> >
> > drm_framebuffer_cleanup(fb);
> >
> > - WARN_ON(atomic_dec_return(&intel_fb->obj->framebuffer_references) < 0);
> > + i915_gem_object_lock(intel_fb->obj);
> > + WARN_ON(!intel_fb->obj->framebuffer_references--);
> > + i915_gem_object_unlock(intel_fb->obj);
> > +
> > i915_gem_object_put(intel_fb->obj);
> >
> > kfree(intel_fb);
> > @@ -14332,12 +14335,16 @@ static int intel_framebuffer_init(struct intel_framebuffer *intel_fb,
> > struct drm_mode_fb_cmd2 *mode_cmd)
> > {
> > struct drm_i915_private *dev_priv = to_i915(obj->base.dev);
> > - unsigned int tiling = i915_gem_object_get_tiling(obj);
> > - u32 pitch_limit, stride_alignment;
> > struct drm_format_name_buf format_name;
> > + u32 pitch_limit, stride_alignment;
> > + unsigned int tiling, stride;
> > int ret = -EINVAL;
> >
> > - atomic_inc(&obj->framebuffer_references);
> > + i915_gem_object_lock(obj);
> > + obj->framebuffer_references++;
> > + tiling = i915_gem_object_get_tiling(obj);
> > + stride = i915_gem_object_get_stride(obj);
> > + i915_gem_object_unlock(obj);
>
> I can't say I'm really up to date on the object locking stuff, but
> from the display POV this looks all right to me. So good enough for me
> :)
It'll do for the moment. The whole fence and vma pinning needs a
dramatic overhaul to allow greater concurrency, and a mix of per-object,
per-GTT and per-global resource mutexes. :|
> Reviewed-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
Thanks.
-Chris
--
Chris Wilson, Intel Open Source Technology Centre
More information about the Intel-gfx
mailing list