Messages in this thread | | | Date | Sat, 3 Apr 2004 13:35:51 -0500 | From | Chris Shoemaker <> | Subject | Re: [2.6.4] Bad swap file entry, then kernel BUG at mm/shmem.c:475 |
| |
On Fri, Apr 02, 2004 at 08:38:09AM +0300, Denis Vlasenko wrote: > On Friday 02 April 2004 02:27, Chris Shoemaker wrote: > > and I haven't found evidence of off-by-bit errors. However, 2 days > > running memtest86 with all tests, gave no errors. (BTW, does anyone > > have a suggestion of a more rigorous general hardware stress-test for > > detecting flakiness?) If my hardware is bad it's pretty intermittent > > cpuburn.
Thanks for the tip. I ran burnBX, burnP5, and burnMMX overnight. No problems at all. But at least I'll know about cpuburn next time I want to test a CPU.
> > Also try to slightly underclock your hardware, > downgrade DMA mode, etc (although it seems you use SCSI...).
Actually, the SCSI drives aren't active. root and swap are on a 13GB Fujitsu IDE drive. That got me thinking though... it seems maybe there is a common thread to my oopses. I think they tend to happen during disk i/o. Either I'm loading a program, or closing X or even during boot-up ( 3 times oopsed during boot now. ) -- all i/o activities.
My disk is using udma2 mode, according to 'hdparm -I'. I guess it's getting set that way from the BIOS, which is set to Auto for the PIO/DMA mode. I'll follow your suggestion and try to force it lower in the BIOS, but first I checked more carefully over the CONFIG options related to IDE and found some things that maybe shouldn't have been set:
CONFIG_IDEDISK_MULTI_MODE=y CONFIG_IDEDISK_STROKE=y CONFIG_BLK_DEV_IDETAPE=m CONFIG_IDE_TASK_IOCTL=y CONFIG_IDE_TASKFILE_IO=y CONFIG_BLK_DEV_IDEPNP=y CONFIG_IDEPCI_SHARE_IRQ=y CONFIG_BLK_DEV_OFFBOARD=y CONFIG_BLK_DEV_GENERIC=y CONFIG_IDEDMA_ONLYDISK=y
As far as I can tell, I don't really need any of those. So, I disabled all of the above. I'm running with this new config for a few hours now. If it oopses, I'll try the lower dma mode.
-Chris
> > vda - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |