Re: [PATCH 3.13.y-ckt 078/138] drm/radeon: hold reference to fences in radeon_sa_bo_new

[Kamal Mostafa]

On Wed, 2016-03-09 at 18:42 -0500, Nicolai HÃhnle wrote:
> On 09.03.2016 18:13, Kamal Mostafa wrote:
> > 
> > 3.13.11-ckt36 -stable review patch.ÂÂIf anyone has any objections,
> > please let me know.
> Please drop the patch for now, it causes a NULL pointer dereference
> onÂ
> kernels <= 3.17. We will follow up with a correctly backported patch.
> 
> Thanks,
> Nicolai


Dropped from 3.13-stable. ÂThanks for the heads-up, Nicolai!

Â-Kamal


> > 
> > 
> > ---8<------------------------------------------------------------
> > 
> > From: =?UTF-8?q?Nicolai=20H=C3=A4hnle?= <nicolai.haehnle@xxxxxxx>
> > 
> > commit f6ff4f67cdf8455d0a4226eeeaf5af17c37d05eb upstream.
> > 
> > An arbitrary amount of time can pass between spin_unlock and
> > radeon_fence_wait_any, so we need to ensure that nobody frees the
> > fences from under us.
> > 
> > Based on the analogous fix for amdgpu.
> > 
> > Signed-off-by: Nicolai HÃhnle <nicolai.haehnle@xxxxxxx>
> > Reviewed-by: Christian KÃnig <christian.koenig@xxxxxxx>
> > Signed-off-by: Kamal Mostafa <kamal@xxxxxxxxxxxxx>
> > ---
> > Â drivers/gpu/drm/radeon/radeon_sa.c | 5 +++++
> > Â 1 file changed, 5 insertions(+)
> > 
> > diff --git a/drivers/gpu/drm/radeon/radeon_sa.c
> > b/drivers/gpu/drm/radeon/radeon_sa.c
> > index f0bac68..bb16684 100644
> > --- a/drivers/gpu/drm/radeon/radeon_sa.c
> > +++ b/drivers/gpu/drm/radeon/radeon_sa.c
> > @@ -349,8 +349,13 @@ int radeon_sa_bo_new(struct radeon_device
> > *rdev,
> > ÂÂ			/* see if we can skip over some
> > allocations */
> > ÂÂ		} while (radeon_sa_bo_next_hole(sa_manager,
> > fences, tries));
> > 
> > +		for (i = 0; i < RADEON_NUM_RINGS; ++i)
> > +			radeon_fence_ref(fences[i]);
> > +
> > ÂÂ		spin_unlock(&sa_manager->wq.lock);
> > ÂÂ		r = radeon_fence_wait_any(rdev, fences, false);
> > +		for (i = 0; i < RADEON_NUM_RINGS; ++i)
> > +			radeon_fence_unref(&fences[i]);
> > ÂÂ		spin_lock(&sa_manager->wq.lock);
> > ÂÂ		/* if we have nothing to wait for block */
> > ÂÂ		if (r == -ENOENT && block) {
> >