Bugzilla – Bug 669556
11.4 m6 often crashing on boot
Last modified: 2011-04-20 09:08:42 UTC
Perhaps 3 times out of 4, with default boot options, m6 crashes during boot on my Samsung NP-N310 netbook. I suspect i915 might be the problem as it doesn't seem to happen with the failsafe option where mode setting is disabled. Unfortunately there's no serial port, so I can't get proper details, but I'll take a photo of the screen and upload shortly, as well as hwinfo output.
Created attachment 412281 [details] photo of screen after crash on boot
When this happens, there's a wait after loading i915, followed by the crash.
Created attachment 412282 [details] output of hwinfo
booting like this: kernel /vmlinuz-2.6.37-20-default root=/dev/disk/by-id/ata-SAMSUNG_HM160HI_S1WWJH0S445408-part2 resume=/dev/disk/by-id/ata-SAMSUNG_HM160HI_S1WWJH0S445408-part3 splash=silent edd=off I had switched off edd in the hope that it might improve matters.
If you suspect modeset, can you boot with nomodeset?
Yes - after playing some more I'm fairly sure about this. With nomodeset, it boots reliably. Without, sometimes it boots OK, and sometimes it crashes (ratio of crashes to good boots maybe 3:1). And the crash happens following a long wait after loading i915, just at the point where on a good boot the video mode would switch.
OpenSUSE 11.4 Milestone 6 was not booting at all for me (tested on MSI S271 notebook, and Acer 6935G) Couple of Factory releases after MS6 - were crashing in kernel during boot. Downloaded today: openSUSE-KDE-LiveCD-i686-Build1056-Media.iso and tried it once more. Kernel is crashing, I get into continuous loop of core dumps on console. OpenSUSE 11.3 works fine on this notebook. I also have install of 11.4 Milestone 4/5 on this notebook. It boots fine with kernel 2.6.37-rc5-16, but doesn't boot with 2.6.37-30 (milestone 5)
Selecting "Kernel Safe Options" with MSI S271 notebook and Build1056 -> -> kernel crash pid: 1428, comm: clicfs Not tainted 2.6.37-20-default EIP: 0060 [<c032bb9e>] EFLAGS: 00010246 CPU: 1 EIP is at _mark_inode_dirty+0xfe/0x1c0 Then Stack and Call Trace follows
This problem is also has been discussed in bug: https://bugzilla.novell.com/show_bug.cgi?id=668141
with 'nomodeset' - I also get crash But in different place now EXT3-fs (sdb4): I/O error while writing superblock EXT3-fs (sdb4): error in ext3_reserve_inode_write: IO failure EXT3-fs (sdb4): I/O error while writing superblock BUG: unable to handle kernel NULL pointer dereference at 00000010 IP: [<c032bb9e>] __mark_inode_dirty+0xfe/0x1c0 *pde = 00000000 Ooops: 0000 [#1] SMP last sysfs file: /sys/devices/pci0000:00/0000:00:14.2/uevent ...
Created attachment 413358 [details] hwinfo for MSI S271 noebook (OpenSUSE 11.4 Factory) Booted with kernel: Linux linux-msi-s271.site 2.6.37-rc5-16-default #1 SMP 2010-12-09 16:04:57 +0100 i686 athlon i386 GNU/Linux Ths is last bootable kernel from recent (MS4-MS6) milestones
Created attachment 413362 [details] dmesg for MSI S271 notebook (boot with kernel 2.6.37-RC5) dmesg for MSI S271 notebook (OpenSUSE 11.4 Factory) boot :~> uname -a Linux linux-msi-s271.site 2.6.37-rc5-16-default #1 SMP 2010-12-09 16:04:57 +0100 i686 athlon i386 GNU/Linux system boots with this kernel without problems networking and graphics working fine.
Can you attach the full trace? Just __mark_inode_dirty isn't enough to debug it. A full capture would be ideal, but a photo would work as well.
(In reply to comment #13) > Can you attach the full trace? Just __mark_inode_dirty isn't enough to debug > it. A full capture would be ideal, but a photo would work as well. Jeff, pls advise how I can get full trace - as OpenSUSE doesn't boot, I can't reach command prompt.
You quoted the beginning of it in comment #10. Boot without splash=silent or quiet. Ideally a serial console could be used to capture it but failing that, a photo will do.
Created attachment 413552 [details] OpenSUSE 11.4 MS6 boot - core dump (photo) OpenSUSE 11.4 MS6 boot - core dump (photo) System: MSI S271 notebook Boot parameters: nomodeset
Created attachment 413555 [details] OpenSUSE 11.4 MS6 boot - core dump on Acer 6935G - photo OpenSUSE 11.4 MS6 boot - core dump on Acer 6935G (notebook) Acer 6935G has Intel CPU (T9400)/ ans Intel chipset. So problem with boot is not related to CPU or chipset. (on previous photo - MSI S271 has AMD CPU and AMD chipset)
Created attachment 413556 [details] Hardware info - Acer 6935G (notebook) Hardware info - Acer 6935G (notebook) - for second photo (core dump) taken from OpenSUSE 11.3
Comment on attachment 413556 [details] Hardware info - Acer 6935G (notebook) "Intel(R) Core(TM)2 Duo CPU T9400 @ 2.53GHz"
Created attachment 414288 [details] dmesg from Rawhide (20110214.15) boot, kernel 2.6.38 To make sure that problem is not related to recent kernel (2.6.37.x and up), I downloaded Fedora Rawhide 20110214.15 and booted from USB stick on MSI S271 notebook. Fedora Rawhide has kernel 2.6.38-0.rc4.git0.2.fc15.i686 I got up to graphical screen, with flashing window says that I have crash in 'mutter' (filed bug report: https://bugzilla.redhat.com/show_bug.cgi?id=677810 ) But boot process was going through. Hope this information would be helpful. Looking forward to test new snapshot with updated kernel! :)
By using rawhide, you're taking the fault out of the equation. The crash involved clicfs, which is a FUSE file system. Please open a new report for this since it's different than the original issue.
Jeff, thanks for explanation. Would problem with clicfs/FUSE file system be fixed in OpenSUSE soon? My understanding that OpenSUSE has RC1 state. But it's even not possible to boot system right now, on two computers which I have for tests like this.
I successfully booted OpenSUSE 11.4 RC2 (x86) KDE LiveCD on Acer 6935G.
On MSI S271: OpenSUSE 11.4 RC2 doesn't crash But it also doens't boot to Login ... squashfs: version 4.0 fuse init (API vrsion 7.15) kjournald starting. Commit interval 15 sec. EXT3-fs (sdb4): mounted filesystem with ordered data mode .. /etc/initscript: line 77: /etc/config/ulimit: unptup/output error Id "2" respawning too fast. disabled for 5 minutes INIT: Id "3" respawning too fast. disabled for 5 minutes Not sure if I should make separate bug report, or it's already known issue and can be fixed ASAP?
This report now has three separate bugs in it and is going to be tough to unravel. If any of them still exist, please open new reports.
I still see the behaviour I originally reported with the final 11.4 on the same hardware. I'll open a new report as suggested.
https://bugzilla.novell.com/show_bug.cgi?id=688714