Re: 2.6.5-aa1: Badness in __remove_from_page_cache at mm/filemap.c:104 && page_remove_rmap at mm/objrmap.c:379

[Andrea Arcangeli]

On Sun, Apr 04, 2004 at 08:25:53PM +0200, Marcello Barnaba wrote:
> Hi Andrea,
> 
> I've just installed 2.6.5-aa1, and after some hours of uptime and
> testing, with everything working properly, I experienced some WARN_ON()
> triggered (my syslog became quite big :).
> 
> I've got both the nvidia and vmware modules installed.
> 
> The bug is fully reproducible, you just have to suspend a virtual
> machine from the vmware GUI, vmware-vmx starts using 100%CPU (just one
> in my 2-way SMP system), and the syslog is completely _filled_ with call
> traces:
> 
> vjt@nowhere:~$ grep Badness.in.page.remove.rmap /var/log/syslog | wc -l
> 8859
> vjt@nowhere:~$ grep Badness.in.__remove_from_pa /var/log/syslog | wc -l
> 8767
> 
> And here there are two of them (they are all identical).
> First occur the 8700 __remove_from_page_cache() warnings, and then the
> page_remove_rmap() ones.

that's an xfs bug found by Andrew, it started to be visible in -aa
because of the objrmap hardness checks from Dave. You can safely ignore
it, the xfs folks are already testing a fix. I didn't delete the WARN_ON
because it's a race condition we want to trap, but you can go ahead and
delete by hand mm/filemap.c:104 and mm/objrmap.c:379 if you're annoyed
by these warning messages. Kernel will work as stable as mainline
despite of the warnings.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/