From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from youngberry.canonical.com ([91.189.89.112]:39980 "EHLO youngberry.canonical.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750993AbcDGNcA (ORCPT ); Thu, 7 Apr 2016 09:32:00 -0400 Date: Thu, 7 Apr 2016 14:31:57 +0100 From: Luis Henriques To: Nicolai =?iso-8859-1?Q?H=E4hnle?= Cc: Taras Prokopenko , christian.koenig@amd.com, stable@vger.kernel.org, Kamal Mostafa , Sasha Levin , Jiri Slaby , Greg Kroah-Hartman Subject: Re: Fwd: Please revert wrong kernel OOPS patch in -69 ! Message-ID: <20160407133157.GA2539@hercules> References: <57065B31.5010503@amd.com> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <57065B31.5010503@amd.com> Sender: stable-owner@vger.kernel.org List-ID: On Thu, Apr 07, 2016 at 08:05:53AM -0500, Nicolai H�hnle wrote: > Hmm, I was under the impression from an email I got that this was already > reverted from 3.16. Clearly, communication is hard. > > Let me re-iterate, to be sure: kernels 3.17 and older > > 1. need to revert the backport of upstream commit > f6ff4f67cdf8455d0a4226eeeaf5af17c37d05eb (I think most have done that > already) and > > 2. should apply the corrected backport which I am attaching yet again (seems > like when I sent this patch around two weeks ago, it wasn't picked up). > Thanks a lot for reiterating this, Nicolai. So, you're absolutely right: this *was* reverted in the 3.16 kernel, in version 3.16.7-ckt26 (and afaik from any other stable trees). Unfortunately, version 3.16.7-ckt25 was actually released *before* this regression was identified. I'm not sure which other stable kernels were actually released with this regression, but there was definitely a window of time where this could have happen. Cheers, -- Lu�s > Thanks, > Nicolai > > On 07.04.2016 04:15, Taras Prokopenko wrote: > > > >---------- Forwarded message ---------- > >From: *Taras Prokopenko* >> > >Date: 2016-04-07 12:04 GMT+03:00 > >Subject: Please revert wrong kernel OOPS patch in -69 ! > >To: luis.henriques@canonical.com > >Cc: jslaby@suse.cz > > > > > >Please test your changes! Lost 1/2 a day to figure out whats causing my > >pc to hang. > > > >OopsText: > > BUG: unable to handle kernel NULL pointer dereference at 0000000000000008 > > IP: [] radeon_fence_ref+0xd/0x50 [radeon] > > PGD 0 > > Oops: 0002 [#1] SMP > > > > > >https://lists.ubuntu.com/archives/kernel-team/2016-February/072148.html > > > >Damn > > > From ad94965f69c2681832f64473d28c23ae71b6e52f Mon Sep 17 00:00:00 2001 > From: =?UTF-8?q?Nicolai=20H=C3=A4hnle?= > Date: Tue, 15 Mar 2016 12:56:45 -0500 > Subject: [PATCH] drm/radeon: hold reference to fences in radeon_sa_bo_new > (3.17 and older) > MIME-Version: 1.0 > Content-Type: text/plain; charset=UTF-8 > Content-Transfer-Encoding: 8bit > > [Backport of upstream commit f6ff4f67cdf8455d0a4226eeeaf5af17c37d05eb, with > an additional NULL pointer guard that is required for kernels 3.17 and older. > > To be precise, any kernel that does *not* have commit 954605ca3 "drm/radeon: > use common fence implementation for fences, v4" requires this additional > NULL pointer guard.] > > 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 H??hnle > Reviewed-by: Christian K??nig (v1 + fix) > Tested-by: Lutz Euler > Cc: stable@vger.kernel.org > --- > drivers/gpu/drm/radeon/radeon_sa.c | 7 +++++++ > 1 file changed, 7 insertions(+) > > diff --git a/drivers/gpu/drm/radeon/radeon_sa.c b/drivers/gpu/drm/radeon/radeon_sa.c > index f0bac68..8962411 100644 > --- a/drivers/gpu/drm/radeon/radeon_sa.c > +++ b/drivers/gpu/drm/radeon/radeon_sa.c > @@ -349,8 +349,15 @@ int radeon_sa_bo_new(struct radeon_device *rdev, > /* see if we can skip over some allocations */ > } while (radeon_sa_bo_next_hole(sa_manager, fences, tries)); > > + for (i = 0; i < RADEON_NUM_RINGS; ++i) { > + if (fences[i]) > + radeon_fence_ref(fences[i]); > + } > + > spin_unlock(&sa_manager->wq.lock); > r = radeon_fence_wait_any(rdev, fences, false); > + for (i = 0; i < RADEON_NUM_RINGS; ++i) > + radeon_fence_unref(&fences[i]); > spin_lock(&sa_manager->wq.lock); > /* if we have nothing to wait for block */ > if (r == -ENOENT && block) { > -- > 2.5.0 >