Hi Linus,
a few people reported an oops that looks to be fixed in drm-next already, so I've pulled the patch back.
Dave.
The following changes since commit 90a5a895cc8b284ac522757a01de15e36710c2b9:
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net (2015-03-23 10:16:13 -0700)
are available in the git repository at:
git://people.freedesktop.org/~airlied/linux drm-fixes
for you to fetch changes up to 8218c3f4df3bb1c637c17552405039a6dd3c1ee1:
drm: Fixup racy refcounting in plane_force_disable (2015-03-24 12:04:35 +1000)
---------------------------------------------------------------- Daniel Vetter (1): drm: Fixup racy refcounting in plane_force_disable
drivers/gpu/drm/drm_crtc.c | 13 +------------ 1 file changed, 1 insertion(+), 12 deletions(-)
drm/i915: Don't try to reference the fb in get_initial_plane_config()
On Mon, Mar 23, 2015 at 7:11 PM, Dave Airlie airlied@linux.ie wrote:
a few people reported an oops that looks to be fixed in drm-next already, so I've pulled the patch back.
Hmm. At least from Josh's report, he also needs
drm/i915: Don't try to reference the fb in get_initial_plane_config()
in addition to the commit you have.
Should I perhaps wait for that one too? Daniel?
Linus
On 25 March 2015 at 09:48, Linus Torvalds torvalds@linux-foundation.org wrote:
drm/i915: Don't try to reference the fb in get_initial_plane_config()On Mon, Mar 23, 2015 at 7:11 PM, Dave Airlie airlied@linux.ie wrote:
a few people reported an oops that looks to be fixed in drm-next already, so I've pulled the patch back.
Hmm. At least from Josh's report, he also needs
drm/i915: Don't try to reference the fb in get_initial_plane_config()in addition to the commit you have.
Should I perhaps wait for that one too? Daniel?
Yes, Daniel indicated to Jani this morning we should pull this, so I'll sidestep the middle man here for expediency.
updated pull:
The following changes since commit 90a5a895cc8b284ac522757a01de15e36710c2b9:
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net (2015-03-23 10:16:13 -0700)
are available in the git repository at:
git://people.freedesktop.org/~airlied/linux drm-fixes
for you to fetch changes up to 59a58cb34d3fe73e6c899cc5d9a87428ca662925:
drm/i915: Don't try to reference the fb in get_initial_plane_config() (2015-03-25 10:01:46 +1000)
---------------------------------------------------------------- Damien Lespiau (1): drm/i915: Don't try to reference the fb in get_initial_plane_config()
Daniel Vetter (1): drm: Fixup racy refcounting in plane_force_disable
drivers/gpu/drm/drm_crtc.c | 13 +------------ drivers/gpu/drm/i915/intel_display.c | 7 +++---- 2 files changed, 4 insertions(+), 16 deletions(-)
Dave.
On Wed, Mar 25, 2015 at 10:04:09AM +1000, Dave Airlie wrote:
On 25 March 2015 at 09:48, Linus Torvalds torvalds@linux-foundation.org wrote:
drm/i915: Don't try to reference the fb in get_initial_plane_config()On Mon, Mar 23, 2015 at 7:11 PM, Dave Airlie airlied@linux.ie wrote:
a few people reported an oops that looks to be fixed in drm-next already, so I've pulled the patch back.
Hmm. At least from Josh's report, he also needs
drm/i915: Don't try to reference the fb in get_initial_plane_config()in addition to the commit you have.
Should I perhaps wait for that one too? Daniel?
Yes, Daniel indicated to Jani this morning we should pull this, so I'll sidestep the middle man here for expediency.
There's going to be one more (waiting for confirmation since I can't repro here). But with these I think we have at least the oopses under control, the remaining one is just paranoid state self-checks. I hope ;-)
And since this is all a bit unpleasant for -rc5 I've thought a bit about what needs to be improved: - I'll be more broken record about correctly referencing commits that have introduced an issue. Some of the fixes in -next didn't have these references, so when we cherry-picked the first one it was hard to find the follow-up fixups.
- I'm stitching together a short scriptlet for cherry-picking to make sure we don't forget to look for all the fixup commits next time around.
- I'll start a drm-fixes branch to make sure drm core fixes don't get lost when I "temporarily" stash them in my drm-misc -next branch.
Cheers, Daniel
updated pull:
The following changes since commit 90a5a895cc8b284ac522757a01de15e36710c2b9:
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net (2015-03-23 10:16:13 -0700)
are available in the git repository at:
git://people.freedesktop.org/~airlied/linux drm-fixes
for you to fetch changes up to 59a58cb34d3fe73e6c899cc5d9a87428ca662925:
drm/i915: Don't try to reference the fb in get_initial_plane_config() (2015-03-25 10:01:46 +1000)
Damien Lespiau (1): drm/i915: Don't try to reference the fb in get_initial_plane_config()
Daniel Vetter (1): drm: Fixup racy refcounting in plane_force_disable
drivers/gpu/drm/drm_crtc.c | 13 +------------ drivers/gpu/drm/i915/intel_display.c | 7 +++---- 2 files changed, 4 insertions(+), 16 deletions(-)
Dave.
On Tue, Mar 24, 2015 at 5:04 PM, Dave Airlie airlied@gmail.com wrote:
Yes, Daniel indicated to Jani this morning we should pull this, so I'll sidestep the middle man here for expediency.
I'm going to wait a bit more with this, since clearly things are still in flux, and these two commits don't actually fix everything at all.
There's apparently at least one more fix necessary. Please holler/re-send when this has cleared up. I don't want to pull some half-way fix at this point, only to pull a fixup to the fixup tomorrow or whatever.
Linus
On Wed, Mar 25, 2015 at 3:43 PM, Linus Torvalds torvalds@linux-foundation.org wrote:
I'm going to wait a bit more with this, since clearly things are still in flux, and these two commits don't actually fix everything at all.
There's apparently at least one more fix necessary.
Indeed. I can see the problem myself.
With current git I get two warnings:
WARNING: CPU: 3 PID: 325 at include/linux/kref.h:47 drm_framebuffer_reference+0x72/0x80 [drm]() WARNING: CPU: 4 PID: 107 at drivers/gpu/drm/drm_atomic.c:482 drm_atomic_check_only+0x32d/0x4f0 [drm]()
and testing the pull it gets rid of the first one, but no tthe second one.
So that urgent fix is very much still lacking.
Linus
On 26 March 2015 at 13:04, Linus Torvalds torvalds@linux-foundation.org wrote:
On Wed, Mar 25, 2015 at 3:43 PM, Linus Torvalds torvalds@linux-foundation.org wrote:
I'm going to wait a bit more with this, since clearly things are still in flux, and these two commits don't actually fix everything at all.
There's apparently at least one more fix necessary.
Indeed. I can see the problem myself.
With current git I get two warnings:
WARNING: CPU: 3 PID: 325 at include/linux/kref.h:47 drm_framebuffer_reference+0x72/0x80 [drm]() WARNING: CPU: 4 PID: 107 at drivers/gpu/drm/drm_atomic.c:482 drm_atomic_check_only+0x32d/0x4f0 [drm]()
and testing the pull it gets rid of the first one, but no tthe second one.
So that urgent fix is very much still lacking.
I've pushed a drm-fixes-staging branch that backport's Daniel's drm-next fix from 9 hours ago,
However it isn't tested yet, so if you want to give it a whirl grab it.
Hopefully when Daniel comes on line he can provide assurance that my fix is a correct backport.
Dave.
On Thu, Mar 26, 2015 at 4:29 AM, Dave Airlie airlied@gmail.com wrote:
I've pushed a drm-fixes-staging branch that backport's Daniel's drm-next fix from 9 hours ago,
However it isn't tested yet, so if you want to give it a whirl grab it.
Hopefully when Daniel comes on line he can provide assurance that my fix is a correct backport.
Your missing the backport of
commit fb9981aa675eb7b398849915364916fd98833cfa Author: Damien Lespiau damien.lespiau@intel.com Date: Thu Feb 5 19:24:25 2015 +0000
drm/i915: Fix atomic state when reusing the firmware fb
My manual backport is here
http://www.mail-archive.com/linux-kernel@vger.kernel.org/msg854137.html
The backport of the other patch you have in -staging looks fine. -Daniel
On Thu, 26 Mar 2015, Daniel Vetter daniel@ffwll.ch wrote:
On Thu, Mar 26, 2015 at 4:29 AM, Dave Airlie airlied@gmail.com wrote:
I've pushed a drm-fixes-staging branch that backport's Daniel's drm-next fix from 9 hours ago,
However it isn't tested yet, so if you want to give it a whirl grab it.
Hopefully when Daniel comes on line he can provide assurance that my fix is a correct backport.
Your missing the backport of
commit fb9981aa675eb7b398849915364916fd98833cfa Author: Damien Lespiau damien.lespiau@intel.com Date: Thu Feb 5 19:24:25 2015 +0000
drm/i915: Fix atomic state when reusing the firmware fbMy manual backport is here
http://www.mail-archive.com/linux-kernel@vger.kernel.org/msg854137.html
The backport of the other patch you have in -staging looks fine.
The drm-intel-fixes pull request I just sent [1] covers both of these. It's on top of drm-fixes, i.e. superseeds -staging.
BR, Jani.
[1] http://mid.gmane.org/87egocrmyc.fsf@intel.com
dri-devel@lists.freedesktop.org
-
Daniel Vetter -
Dave Airlie -
Dave Airlie -
Jani Nikula -
Linus Torvalds