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,
Nicolai
On 07.04.2016 04:15, Taras Prokopenko wrote:
---------- Forwarded message ----------
From: *Taras Prokopenko* <taras.prokopenko@xxxxxxxxx
<mailto:taras.prokopenko@xxxxxxxxx>>
Date: 2016-04-07 12:04 GMT+03:00
Subject: Please revert wrong kernel OOPS patch in -69 !
To: luis.henriques@xxxxxxxxxxxxx <mailto:luis.henriques@xxxxxxxxxxxxx>
Cc: jslaby@xxxxxxx <mailto:jslaby@xxxxxxx>
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: [<ffffffffc03b83ed>] 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?= <nicolai.haehnle@xxxxxxx>
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 Hähnle <nicolai.haehnle@xxxxxxx>
Reviewed-by: Christian König <christian.koenig@xxxxxxx> (v1 + fix)
Tested-by: Lutz Euler <lutz.euler@xxxxxxxxxx>
Cc: stable@xxxxxxxxxxxxxxx
---
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