> On 10 Jun, Mark Hemment wrote:
> > Yeek! try_to_free_page() is being called during an interrupt!
>
> Yep. Not a Good Thing!
Nope, and a silly mistake, too.
...
> > Perhaps someone with a tree who can apply the mmap-diff, can make a
> > patch for this change (and check my suggested mod). Hopefully the orignal
> > author will pick this up....
>
> I've made the change you suggested (can provide unified diffs if
> there's any desire). It crashed during my quicky stress test (ftp
> Xemacs-20 while simultaneously untarring and rm'ing four 2.1.x kernel
> trees) without any errors, but has otherwise been stable so far (~12
> hours).
;) Being the original author, yep that was close to the change I made
(see ftp://dot.superaje.com/pub/linux/blah-mm-2.1.42-2.diff for the
typo-thinko fixed version). That plus the smp-on-uni patch by Berhard
Kaindl has resulted in a >7 day uptime, helped by the fact that I'm no
longer using NFS -- on the other machine that's nfs mounting most things
I can lock it by running X. I still haven't been able to track it down
(very hard lock -- interrupts disabled, no output, spinning somewhere).
An NMI would be very useful...
-benjamin