I am running Debian 64bit and have had intermittent problems with booting. Initially (Aug 2010) I was running 2.6.32-15 kernel from Debian Squeeze. Please see http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=591622 for details of my but report to Debian. I am reporting the bug here, following the advice I got there. The problem was that about half the time the system would not boot unless I removed power and battery to force a reset. The fan would spin up and disk light go on and off briefly, but no messages on the display. Following recommendation from that bug report, I upgraded to kernel 2.6.35-trunk-amd64. The failures continued but a little less often. Recently I installed updates, including update of ACPI packages (I don't recall the details and don't know how to look them up - sorry). Now the problem is much worse. I have to remove all power at most, but not all, reboots. If I disable ACPI, I don't have any problem with reboots. No doubt you will need more information to isolate the fault. I would be pleased to provide any information if you tell me how to find it.
does the problem still exist if you boot with boot option "init=/bin/bash"?
From the debian bug report Aug 7th > The system reboots reliably if it hasn't gone to sleep, > but after having gone to sleep a subsequent boot fails > with the same behavior as the older kernel exhibited in > all cases. Is this still true? Does the system boot reliably if there is no system suspend involved? ie, if you apply power, boot/shutdown/boot/reboot a number of time w/o any suspends, does it work reliably? Is there any previous version of Linux that worked reliably on this machine?
Hi Len, Thanks for the follow-up. Yes, this is still a problem for me. In the mean time, I have installed 32bit lenny and it exhibits similar problems. Currently I am using it (32bit) with "acpi=off". Initially I had thought the problem only occurred after a suspend, but now I suspect it happens even without a suspend having occurred. Rarely, I experienced the boot problem after the system running for only a few minutes and other times after several hours of continuous use. In both these cases I don't think the system was ever idle long enough for a suspend to have happened. Thus is seems to be an intermittent problem that is more likely if the system has been running longer. I don't know that I have ever had a successful boot after a suspend. If I boot/restart/restart/restart it seems to work OK - another reason I thought it was suspend that was causing the problem initially. I am trying to remember if any previous version worked with ACPI enabled. If there was it was Ubuntu 9.04 or 9.10. I was using these for a while but I think I disabled ACPI there also. I regret I don't have notes and have since overwritten the installation with Debian, so can't confirm. Everything I have tried has given me boot problems and eventually I got everything booting successfully, but I think it may have been by disabling ACPI in every case. Regards, Ian Goodacre On Thu, 2010-09-30 at 02:25 +0000, bugzilla-daemon@bugzilla.kernel.org wrote: > https://bugzilla.kernel.org/show_bug.cgi?id=18772 > > > Len Brown <lenb@kernel.org> changed: > > What |Removed |Added > ---------------------------------------------------------------------------- > CC| |lenb@kernel.org > > > > > --- Comment #2 from Len Brown <lenb@kernel.org> 2010-09-30 02:25:52 --- > >From the debian bug report Aug 7th > > > The system reboots reliably if it hasn't gone to sleep, > > but after having gone to sleep a subsequent boot fails > > with the same behavior as the older kernel exhibited in > > all cases. > > Is this still true? > Does the system boot reliably if there is no system suspend involved? > ie, if you apply power, boot/shutdown/boot/reboot a number of time w/o > any suspends, does it work reliably? > > Is there any previous version of Linux that worked > reliably on this machine? >
(In reply to comment #1) > does the problem still exist if you boot with boot option "init=/bin/bash"? I tried three times, once leaving the system idle for 20 minutes and the other times rebooting after a few minutes. I don't know how to shut down cleanly from this state. The shutdown and init commands fail, I expect because the init process isn't running. If I merely exit the shell, I get a kernel panic then have to force the power off, which is what I did all three times. Boot was successful after each such shutdown.
Reply-To: neiljamieso@gmail.com I have a similar problem having installed Debian Squeeze. I did not have this problem with Lenny. It seems to be related to other problems reported with this device which also relate to acpi. The failed restarts generally follow flooding of the log files with Apr 22 20:41:55 debian kernel: [ 7841.500055] ACPI Exception: AE_TIME, Returned by Handler for [EmbeddedControl] (20090903/evregion-424) Apr 22 20:41:55 debian kernel: [ 7841.500070] ACPI Error (psparse-0537): Method parse/execution failed [\_SB_.AMW0.WMCA] (Node f6c178e8), AE_TIME Apr 22 20:41:55 debian kernel: [ 7842.000062] ACPI Exception: AE_TIME, Returned by Handler for [EmbeddedControl] (20090903/evregion-424) Apr 22 20:41:55 debian kernel: [ 7842.000079] ACPI Error (psparse-0537): Method parse/execution failed [\_SB_.AMW0.WMCA] (Node f6c178e8), AE_TIME Apr 22 20:41:57 debian kernel: [ 7843.500046] ACPI Exception: AE_TIME, Returned by Handler for [EmbeddedControl] (20090903/evregion-424) Apr 22 20:41:57 debian kernel: [ 7843.500064] ACPI Error (psparse-0537): Method parse/execution failed [\_SB_.AMW0.WMCA] (Node f6c178e8), AE_TIME Apr 22 20:41:57 debian kernel: [ 7844.000070] ACPI Exception: AE_TIME, Returned by Handler for [EmbeddedControl] (20090903/evregion-424) Apr 22 20:41:57 debian kernel: [ 7844.000088] ACPI Error (psparse-0537): Method parse/execution failed [\_SB_.AMW0.WMCA] (Node f6c178e8), AE_TIME Apr 22 20:41:59 debian kernel: [ 7845.500060] ACPI Exception: AE_TIME, Returned by Handler for [EmbeddedControl] (20090903/evregion-424) Apr 22 20:41:59 debian kernel: [ 7845.500078] ACPI Error (psparse-0537): Method parse/execution failed [\_SB_.AMW0.WMCA] (Node f6c178e8), AE_TIME Apr 22 20:41:59 debian kernel: [ 7846.000067] ACPI Exception: AE_TIME, Returned by Handler for [EmbeddedControl] (20090903/evregion-424) Apr 22 20:41:59 debian kernel: [ 7846.000089] ACPI Error (psparse-0537): Method parse/execution failed [\_SB_.AMW0.WMCA] (Node f6c178e8), AE_TIME This is associated with marked performance degradation (as you'd expect I guess). I have send a bug report to Debian, the number of which I can't access in Gmail while writing another email! It had 5735Z in the topic. The frequency (but not the severity of failures) is reduced since I followed advice to add " i915.powersave=0 pci=noacpi" to the GRUB_CMDLINE_LINUX_DEFAULT line in /etc/default/grub. Cheers, neil
I'm not running 64bit any more, but have exactly the same problem with 32bit Debian Squeeze (Linux debian 2.6.32-5-686). I also see reduced frequency of boot failure with "i915.powersave=0 pci=noacpi", but still about half of boots fail. On the other hand, I can hibernate and restart with very low failure rate, though still get failures and have to pull the power and battery occasionally.
some ACPI setting messed up the BIOS? please check if there is any BIOS update. what if you boot into console mode with boot option "nomodeset" and without ACPI video driver? what if you boot with AC only? or Battery only?
Bug closed as there is no response from the bug reporter. Please feel free to reopen it if the problem still exists in the latest upstream kernel.