LMCE keeps hard locking....

[colinjones]

Occasionally (sometimes once per day, other times it can go for several days) my core locks up completely, keyboard LEDs flashing. Have to reset the box completely...

I have noticed in dmesg some errors which are obviously related. They tend to happen some time before the lock up and obviously the kernel eventually falls over. But I have noticed the last few all seem to be complaining about Photo Screen Saver particularly.. Can anybody shed any light on what might be causing this?

[28100.604000] Bad page state in process 'Photo_Screen_Sa'
[28100.604000] page:c15cb7e0 flags:0x40000000 mapping:00000000 mapcount:-1 count:0
[28100.604000] Trying to fix it up, but a reboot is needed
[28100.604000] Backtrace:
[28100.604000]  [<c01628d3>] bad_page+0x63/0xa0
[28100.604000]  [<c01635f3>] get_page_from_freelist+0x343/0x3a0
[28100.604000]  [<c016369f>] __alloc_pages+0x4f/0x340
[28100.604000]  [<c0122786>] try_to_wake_up+0x46/0x480
[28100.604000]  [<c0172e3d>] anon_vma_prepare+0x1d/0xe0
[28100.604000]  [<c0165d66>] __pagevec_lru_add_active+0xb6/0xd0
[28100.604000]  [<c016d457>] __handle_mm_fault+0x8c7/0xb00
[28100.604000]  [<c017b123>] add_partial+0x13/0x40
[28100.604000]  [<c01032d6>] __switch_to+0x176/0x1d0
[28100.604000]  [<c02f1f8a>] schedule+0x2ca/0x890
[28100.604000]  [<c02f5b36>] do_page_fault+0x126/0x690
[28100.604000]  [<c02f5a10>] do_page_fault+0x0/0x690
[28100.604000]  [<c02f4292>] error_code+0x72/0x80
[28100.604000]  =======================

[hari]

ouch

[daballiemo]

Colin,

don't know about any software issues, but I presume you have run fsck, memcheck etc to make sure its not coming out of the hardware.

rgds

Han

[colinjones]

fsck just happens to have run yesterday, no issues. Haven't run any memory check yet, but I think it is indicative that it only seems to happen for the one process.. PSS... I guess that could either be a bug in the code generally, or perhaps a particular photo is triggering an arcane bug? esp. given the rescaling function that was recently added...

[hari]

do you run the nvidia binary driver? Revert to vesa to see if it keeps stable.

br, Hari

[Afkpuz]

2 possibilities: 

1.) Hardware compatibility
2.) Thermal



Check temps, then try swapping out parts.  Main things are mobo and video card that cause incompatibility.  I got hard lock up about the same frequency as you and putting a new motherboard in fixed it.

[colinjones]

Will have a think about using vesa - but I'm concerned it will be unusably slow (how slow is it, hari?)

Its not compatibility, the same hardware has been running very successfully for over a year. I have checked heat before, and it runs well below the critical temperatures, but I will take another look to make sure something hasn't changed - but again, I point out that it is the same process failing each time. Although PSS runs most of the time, and that statistically increases the likelihood of a "random" hardware issue effecting that process repeatedly, there are dozens of other processes that run all the time as well... wish I could find a clear explanation of that the error message means generally. If its just memory management getting pages mixed up/page tables corrupted then it is extremely unlikely to be hardware, but if it is the hardware memory mapping CAM or such like, then very likely to be hardware... hmmm

[Zaerc]

It could even be "bit-rot" on the HD (granted, not very likely) maybe try reinstalling the pluto-photo-screen-saver package and see if that makes any difference: sudo apt-get --reinstall pluto-photo-screen-saver (from memory).

[hari]

i vote for:

1.) bad hardware (most likely mainboard, maybe memory)
2.) binary module going havoc (less likely in that area of the kernel, but who knows)
3.) bug in the vm subsystem (very unlikely)

I'd say vesa is fast enough for a test :-)

br, Hari

[skeptic]

+1 vote for hardware. 

[Marie.O]

Put a fan into the case to blow fresh air onto your nvidia card.

[totallymaxed]

Will have a think about using vesa - but I'm concerned it will be unusably slow (how slow is it, hari?)

Its not compatibility, the same hardware has been running very successfully for over a year. I have checked heat before, and it runs well below the critical temperatures, but I will take another look to make sure something hasn't changed - but again, I point out that it is the same process failing each time. Although PSS runs most of the time, and that statistically increases the likelihood of a "random" hardware issue effecting that process repeatedly, there are dozens of other processes that run all the time as well... wish I could find a clear explanation of that the error message means generally. If its just memory management getting pages mixed up/page tables corrupted then it is extremely unlikely to be hardware, but if it is the hardware memory mapping CAM or such like, then very likely to be hardware... hmmm

I think the suggestion of using vesa mode was purely to allow a test of your hardware... not intended for permanent use etc.

Andrew

[colinjones]

Thanks for all the suggestions, guys! I will try the vesa thing first and run it for a few days (that should also help with heat on the video card as it won't be doing the graphics acc)

Thanks Zaerc for the tip on reinstalling PSS, that's exactly the type of thing I would like to eliminate..

[indulis]

If you are worried about the HD having problems you can also install the smart tools which allow monitoring and self-test of the disks.  I can't remember if they are installed/activated by default.

Code: [Select]

apt-get install smartmontools
More info here http://www.linuxjournal.com/article/6983

[colinjones]

All - before I can use apt-get install for either the smartmontools or to reinstall PSS I have to resolve a dependency issue.

I recently got a SqueezeBox Boom, and as a result of it being incompatible with the existing streamer, I had to go through a process of upgrading to SqueezeCentre. This has left my dependencies a bit screwed up. So now if I try to issue any apt-get install commands it comes back with:

The following packages have unmet dependencies:
  pluto-slim-server-streamer: PreDepends: slimserver but it is not going to be installed
E: Unmet dependencies. Try 'apt-get -f install' with no packages (or specify a solution).
linuxmce@dcerouter:/etc/pluto$ man apt-get

The dependencies indicate that is it still using slimserver when in fact it is really using squeezecenter (Zaerc did squeezecenter make it into the alpha1?)

I would like just to be able to override this check/failure for now and go ahead with the install/reinstall I want to do - how can I do this? I really don't want to 1) mess around with the dependencies or 2) accidentally install/uninstall a component at this stage because I can't remember how I got squeezecenter hooked into the system in the first place! So just getting it to ignore this problem temporarily would be great!