Skip to contentSkip to conversation

recent 2.6.26 kernel hangs at suspend

DISCUSSION Jeff Chua 4 participants 8 messages May 19, 2008 – May 20, 2008

lore
Share & tools
lore

Jeff Chua

author
#0 lore raw

I can simulate this and it's definitely a bug, but I could only bisect up to a point and the kernel would not boot after tat.

The bug is that my X60s notebook would not suspend-to-ram on the 2nd try.

Boot up -> suspend to ram -> ok -> resume -> suspend to ram -> hang!

The hang would only happen while in X.

If I don't enter X, it'll not hang on suspend to ram.

I tried git bisect but can only get to this commit 4a27214d7be31e122db4102166f49ec15958e8e9.

The next commit b82287587ef9917afbea5fcbf7aa63424b6f3719 would not boot.

How can I get pass the "bad" commit. I've tried "skip" or "next", but still won't boot.

And it doesn't seem like it's X as 2.6.25 kernel can suspend and resume many times in X without problem.

Here's how I bisected ...

git-bisect start
# bad: [b8291ad07a7f3b5b990900f0001198ac23ba893e] Linux 2.6.26-rc3
git-bisect bad b8291ad07a7f3b5b990900f0001198ac23ba893e
# good: [4b119e21d0c66c22e8ca03df05d9de623d0eb50f] Linux 2.6.25
git-bisect good 4b119e21d0c66c22e8ca03df05d9de623d0eb50f
# good: [c328d54cd4ad120d76284e46dcca6c6cf996154a] Merge
git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/v4l-dvb
git-bisect good c328d54cd4ad120d76284e46dcca6c6cf996154a
# bad: [25a025863e024f6b86b48137b10b4960c50351b0] Merge branch
'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
git-bisect bad 25a025863e024f6b86b48137b10b4960c50351b0
# bad: [62d9f0dbc92d7e398fde53fc6021338393522e68] KVM: add ioctls to save/store mpstate
git-bisect bad 62d9f0dbc92d7e398fde53fc6021338393522e68
# good: [ce1d5b23a8d1e19866ab82bdec0dc41fde5273d8] Merge branch
'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
git-bisect good ce1d5b23a8d1e19866ab82bdec0dc41fde5273d8
# good: [a2826190aa157a1d29bef70ca81f8b51a9b36d29] IDE: Coding Style
fixes to drivers/ide/pci/it8213.c
git-bisect good a2826190aa157a1d29bef70ca81f8b51a9b36d29
# bad: [9b79ed952bd7344d40152f8a560ad8a0d93f3886] Merge branch
'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/x86/linux-2.6-generic-bitops-v3
git-bisect bad 9b79ed952bd7344d40152f8a560ad8a0d93f3886
# bad: [4a27214d7be31e122db4102166f49ec15958e8e9] Merge branch
'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/x86/linux-2.6-x86-fixes
git-bisect bad 4a27214d7be31e122db4102166f49ec15958e8e9
# bad: [b82287587ef9917afbea5fcbf7aa63424b6f3719] Merge branch
'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/x86/linux-2.6-x86-misc
git-bisect bad b82287587ef9917afbea5fcbf7aa63424b6f3719

Thanks,
Jeff.

#1 lore raw
Quoted 3 lines · Jeff Chua

How can I get pass the "bad" commit. I've tried "skip" or "next", but still won't boot.

Try

git bisect visualize

and pick any commit that is active (avoiding the endpoints) and looks like a good candidate to test. Then do "git reset --hard <that-commit>" and try again..

Of course, when there are two overlapping bugs, and the other hides the bug you are trying to chase down, there *will* be cases where bisect simply doesn't work, if you are unlucky. Bisecting is wonderful, but it's not going so solve world hunger, or even be able to help with _all_ problems.

Linus

author
#2 lore raw
Quoted 5 lines · Linus Torvalds

git bisect visualize

and pick any commit that is active (avoiding the endpoints) and looks like a good candidate to test. Then do "git reset --hard <that-commit>" and try

Trying that now.

Thanks,
Jeff.

#3 lore raw
In reply to Jeff Chua
Quoted 8 lines · Jeff Chua

I can simulate this and it's definitely a bug, but I could only bisect up to a point and the kernel would not boot after tat.

The bug is that my X60s notebook would not suspend-to-ram on the 2nd try.

Boot up -> suspend to ram -> ok -> resume -> suspend to ram -> hang!

The hang would only happen while in X.

Does this notebook have intel graphics?
If yes, does anything change with the i915 driver loaded vs not loaded?

-Len

author
#4 lore raw
Quoted 3 lines · Len Brown

On Monday 19 May 2008, Jeff Chua wrote:

Does this notebook have intel graphics?

Yes.

(II) intel(0): Integrated Graphics Chipset: Intel(R) 945GM

Quoted 1 line

If yes, does anything change with the i915 driver loaded vs not loaded?

I'll try that. Currently, everything compliled into kernel.

Thanks,
Jeff.

#5 lore raw
Quoted 5 lines · Jeff Chua
Quoted 3 lines · Len Brown

On Monday 19 May 2008, Jeff Chua wrote:

Does this notebook have intel graphics?

Quoted 7 lines

Yes.

(II) intel(0): Integrated Graphics Chipset: Intel(R) 945GM

Quoted 1 line

If yes, does anything change with the i915 driver loaded vs not loaded?

I'll try that. Currently, everything compliled into kernel.

Another thing to check out is if VT switching is working. On the Toshiba U305, which also has Intel graphics, it broke and made suspend/resume look broken:

http://bugzilla.kernel.org/show_bug.cgi?id=10620

#6 lore raw
In reply to Jeff Chua

Hi!

Quoted 10 lines

I can simulate this and it's definitely a bug, but I could only bisect up to a point and the kernel would not boot after tat.

The bug is that my X60s notebook would not suspend-to-ram on the 2nd try.

Boot up -> suspend to ram -> ok -> resume -> suspend to ram -> hang!

The hang would only happen while in X.

If I don't enter X, it'll not hang on suspend to ram.

Ok, so we have X in the mix.

Can you try if it happens if you disable Intel framebuffer driver?
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
author
#7 lore raw
Quoted 5 lines · Pavel Machek
Quoted 1 line

If I don't enter X, it'll not hang on suspend to ram.

Ok, so we have X in the mix.

Can you try if it happens if you disable Intel framebuffer driver?

The whole problem went away after the latest git pull.

Looks like the commit below fixes the problem of failing on 2nd suspense.

I've tested whole day and it's suspending-resuming after many cycles.

Sorry, couldn't get to email whole day. Traveling. Slow train.

Thank you all for your help!!!
Jeff.

commit 860da5e578c25d1ab4528c0d1ad13f9969e3490f
Merge: 1bf9947... e948e99...
Author: Linus Torvalds <[email protected]>
Date: Mon May 19 13:30:40 2008 -0700

Merge branch 'drm-patches' of git://git.kernel.org/pub/scm/linux/kernel/git/
airlied/drm-2.6

* 'drm-patches' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm
-2.6:
drm/i915: save and restore dsparb and d_state registers. drm/i915: fix off by one in VGA save/restore of AR & CR regs.
drm: disable tasklets not IRQs when taking the drm lock spinlock Revert "drm/vbl rework: rework how the drm deals with vblank."