intrepid hang with screen corruption during boot with 2.6.27-2-generic on x61

[Martin Pool]

I have a Thinkpad X61s running intrepid, and I just upgraded it from the version of about 3 weeks ago to yesterday's current version, including a kernel upgrade to 2.6.27-2-generic. It now reliably hangs during startup, with the screen oddly corrupted, just after "loading files needed to boot". This still happens if I choose a recovery boot.

Choosing the previous kernel, 2.6.26-5-generic, works fine.

[Martin Pool (mbp)]

I can't get the machine up far enough on the faulty kernel to save the log, but here is the dmesg from a boot of the working kernel.

[Thierry Carrez (ttx)]

I think I have the same garbled screen problem on a Dell XPS M1330, from your screenshot it hangs at about the same point...

However, I sometimes (half of the time) manage to boot it (?)
From what happens just before the screen is garbled, I can narrow the issue to ACPI, around

[ 24.684203] ACPI: AC Adapter [AC] (on-line)
[ 24.762166] input: Lid Switch as /class/input/input2
[ 24.792634] ACPI: Lid Switch [LID]
[ 24.792716] input: Power Button (CM) as /class/input/input3
[ 24.856150] ACPI: Power Button (CM) [PBTN]
[ 24.856255] input: Sleep Button (CM) as /class/input/input4
[ 24.904143] ACPI: Sleep Button (CM) [SBTN]
[ 24.934974] ACPI: Battery Slot [BAT0] (battery present)
[ 24.935057] ACPI: WMI: Mapper loaded
[ 25.206679] acpi device:32: registered as cooling_device2
[ 25.207411] input: Video Bus as /class/input/input5

it seems to log two or three lines after the input4 register, then screen is garbled and you can't really tell anything :)
Attached is a complete dmesg.log from when 2.6.27 boots correctly.

[Stefan Bader (smb)]

Do combinations of "highres=off nohz=off" or "nopaictimer", "hpet=disabled" help or change something?

[Thierry Carrez (ttx)]

You mean "noacpitimer" or "nopaictimer" ?

Tests so far:
"noacpitimer" -> FAIL (1/2)
"highres=off nohz=off hpet=disabled" -> BOOT OK (3/3)
"hpet=disabled" -> FAIL (1/2)
"highres=off nohz=off" -> BOOT OK (4/4)

fyi my Dell XPS M1330 is shipping an Intel Core2Duo T7250.

[Thierry Carrez (ttx)]

Hm. Test results update:

"noacpitimer" -> FAIL (1/2)
"hpet=disabled" -> FAIL (1/2)
"highres=off nohz=off" -> FAIL (4/5)
"highres=off nohz=off hpet=disabled" -> BOOT OK (5/5)
"noacpitimer hpet=disabled" -> NON-TESTED (0/0)

[Stefan Bader (smb)]

There is (will be) an updated 2.6.27-3.3smb4 at http://people.ubuntu.com/~smb/bug254668 which has some changes that are currently discussed upstream. I am aware that there still seems to be a piece missing but overall there is an improvement. Maybe you could give that a try.

[Alfredo (frederick-jansen)]

Stefan, this kernel fixes the problem for me. Thanks

[Thierry Carrez (ttx)]

So far testing is OK with this kernel. Note that it also fixes suspend problems for me (bug 263862).

[Thierry Carrez (ttx)]

I finally reproduced the boot failure with 2.6.27-3.3smb4, same symptoms. So I suppose my issue is slightly different than Alfredo's. I keep it because the suspend problem was more of a blocker to me. Looking forward a rc5-based kernel that would fix both issues ;)

[Stefan Bader (smb)]

There are i386 and amd64 versions of 2.6.27-3.3smb5 at http://people.ubuntu.com/~smb/bug254668. These have some more changes from upstream regarding hpet code.

[Thierry Carrez (ttx)]

I still have the screen-garbling locks at boot with smb5. Occurs about 1/3 of the time. Can't find a reproduction logic.

[Martin Pool (mbp)]

With 2.6.27-2-generic I'm now getting hangs with corruption about 1/4
of the time; about 1/4 of the time it will spontaneously reboot
instead of hanging, and the rest of the time it boots ok. I haven't
tested Stefan's kernels yet.

--
Martin <http://launchpad.net/~mbp/>

[Martin Pool (mbp)]

Current intrepid has worked ok for me the last several times. I'm not sure if it's fixed or just intermittent.

[Thierry Carrez (ttx)]

Same here, OK so far with 2.6.27-3.4

[Martin Pool (mbp)]

It did happen again this morning, so it's still present but intermittent.

[ozonehole (taibei)]

I'm having the same problem with a Dell Inspiron 1420N (that laptop that Dell ships with Linux pre-installed). I was able to boot OK with the Alpha-6 live CD, and it booted OK after install. As of today (Sept 28) I did a full update-manager, so it should be the latest kernel. After the update, the machine crashes every time at bootup, with screen corruption. I can't get to my logs or anything else, as I can't boot.

[Leandro Pereira de Lima e Silva (leandro-limaesilva)]

I don't have the screen problem, but it hangs on boot too. Pressing the alt key a few times makes it wake up.

Pressing alt after a few seconds also gives me the message "clocksource tsc unstable".

I'm on an HP DV6736 (Turion 64 X2, 2ghz).

I'll try Stefan's kernel now.

[Alexander Butenko (avb)]

i have same problem with screen garbage on boot with 2.6.27-4-generic kernel.

[Alexander Butenko (avb)]

by the way, i have lenovo thinkpad X61 with

[James Ward (jamesward)]

I have also been getting random hangs on boot with all 2.6.27 kernels. It hangs about 3 of 4 boots. When I've done a non-silent boot there were no errors and each crash seemed to happen at different times during boot.

[Thierry Carrez (ttx)]

It occurred again for me this morning after upgrading to 2.6.27-5.

I don't think it ever occurred while I was using the the 2.6.27-4 series, but since it's so random it's difficult to draw conclusions.

[Alexander Butenko (avb)]

same issue with -7

[Alexander Butenko (avb)]

-6, sorry

[James Ward (jamesward)]

I haven't had a hang in the past day with about 5 boots. So maybe this is fixed? I'm not sure what changed but it wasn't my kernel. Can others having this problem update to the latest packages and see if this is resolved for them?

[Arnau Alcázar (arnau)]

I've got the same problem in Dell XPS M1330 with intel x3100.

[Martin Pool (mbp)]

It was still happening on my X61s as of last Friday.

--
Martin <http://launchpad.net/~mbp/>

[Thierry Carrez (ttx)]

For me (Dell XPS 1330) it happens more than ever with 2.6.27-7.10...
4 hits on 4 consecutive tries.

[Thierry Carrez (ttx)]

OK so far with 2.6.27-7.11... that's really random.

This *might* be a duplicate from bug 263059, except that there is no screen corruption in their case.
Are we all using iwl3945 as the wireless chipset ?

[J. Bruce Fields (bfields-fieldses)]

2.6.27-7-generic, Dell 1420n, I reliably get a corrupted splash screen and boot hang.

Switching to the first virtual console (alt-ctl-F1) early in the boot works around the problem for me.

[Alexander Butenko (avb)]

im confirming what Thierry Carrez was saying. Yesterday i made a research of this problem and figure out that screen corruption happends during loading of iwl3945 driver.

[Martin Pool (mbp)]

On Fri, Oct 17, 2008 at 4:43 AM, J. Bruce Fields <email address hidden> wrote:
> Switching to the first virtual console (alt-ctl-F1) early in the boot
> works around the problem for me.

This doesn't reliably fix it for me. Because the problem is
intermittent it's hard to make a definitive statement.

--
Martin <http://launchpad.net/~mbp/>

[Colin Watson (cjwatson)]

With the mention of iwl3945, this sounds very much like bug 263059, which was fixed in linux 2.6.27-7.12 by disabling CONFIG_DYNAMIC_FTRACE. Could folks please upgrade and confirm or deny?

[Thierry Carrez (ttx)]

OK so far for me with >=2.6.27-7.12 . Dell XPS 1330 with iwl3945.

[Chris Jones (cmsj)]

If this is actually the same as bug 282700 then I believe it is fixed by the dynamic ftrace disablement. I have screenshots on the other bug of what it looked like on my system. I marked it Fix Released because after 31 reboot cycles I couldn't reproduce it and it hasn't happened since then. Previously I was able to reproduce it at around 15 cycles.
I didn't dupe it to this bug because after talking to Martin I wasn't entirely convinced it was the same bug, so I erred on the side of caution.

[J. Bruce Fields (bfields-fieldses)]

Following up on https://bugs.launchpad.net/ubuntu/+source/linux/+bug/263782/comments/32. I can no longer reproduce the problem. I don't know when exactly it stopped, but I'm on 2.6.27-7.13 now. I do use iwl3945, and the corruption in http://launchpadlibrarian.net/18498919/snowcrash.jpg looks like the same thing I was seeing.

I don't think I saw corruption like that in http://launchpadlibrarian.net/17218490/IMG_4804.JPG, so rather than confirmation that this bug is a dupe of 282700, this may just mean that my comment should have been to 282700 rather than this bug....

[Ramaddan (ramaddan)]

Have the same problem after upgrading to Intrepid from Hardy with linux-image-2.6.27-7-generic and linux-image-2.6.27-7-server

I do not even get any image at boot after the first few times it came as corrupted lines.

I need to switch to ctrl-alt-F1 and input my encryption password for it to continue.

[Leann Ogasawara (leannogasawara)]

Hi Martin,

Since you are the original bug reporter, can you comment if this is resolved with a 2.6.27-7.12 or newer kernel?

@Ramaddan, which specific 2.6.27-7 kernel are you using? 'cat /proc/version_signature' should show you. Thanks.

[Martin Pool (mbp)]

On Sun, Oct 26, 2008 at 3:51 PM, Leann Ogasawara <email address hidden> wrote:
> Hi Martin,
>
> Since you are the original bug reporter, can you comment if this is
> resolved with a 2.6.27-7.12 or newer kernel?

2.6.27-7.12 booted successfully on the one time I've tried it so far.

--
Martin <http://launchpad.net/~mbp/>

[Arnau Alcázar (arnau)]

I didn't get corrupted boot screen again.

[Ramaddan (ramaddan)]

Hi Leann,

> @Ramaddan, which specific 2.6.27-7 kernel are you using? 'cat /proc/version_signature' should show you. Thanks.

This is what I got:
Ubuntu 2.6.27-7.14-generic

The thing is, it stopped going corrupt after I used StartUp-Manager to change the resolution to 800x600, and I also changed the color depth to 16bits.

So I'm not sure if it is stopped because of that, or an update that I did not notice.

[Leann Ogasawara (leannogasawara)]

Based on the previous comments I'm marking this "Fix Released". Thanks.

[Launchpad Janitor (janitor)]

Per a decision made by the Ubuntu Kernel Team, bugs will longer be assigned to the ubuntu-kernel-team in Launchpad as part of the bug triage process. The ubuntu-kernel-team is being unassigned from this bug report. Refer to https://wiki.ubuntu.com/KernelTeamBugPolicies for more information. Thanks.