> > > Hmm, I have 64 128 512 in my /proc/sys/vm/freepages, do the
> > > recent MM changes require Thomases 128 256 1024 in order to
> > > perform as they were designed to?
>
> Please try the below patch, but not with such high numbers... If you
Apparently one more thing I have missunderstood in the kernel :(
> I saw a later posting mentioning that it didn't help :(
> Perhaps the patch below does?
Well, yes and no. I did some quick testing. With 2.0.31-pre6
64 128 512
64 96 128
8 64 96
won't produce the debug message on the 1st bonnie run.
128 256 1024
This one causes it even the first time. All the settings
I tested produced the debug message on the 2nd run. I didn't
bother to test it a 3rd time. I will now test it with 8 64 96
and your patch.
Well, with 2.0.31-pre6+Benjamin's patch
8 64 96
Produced the message on the first run.
With kernel defaults (64 96 128) the first two sequential runs
didn't produce any messages! But, the following two runs
both did. I didn't even bother to run simultaneous bonnies.
Does any of you hackers see a pattern here? I'm just a dumb
cs-student but I would say the more times I run bonnie the
frequent the messages get (I'm not doing anything besides
writing this message between runs), wouldn't that implicate
that there is a atomic/non-atomic buffer/cache/page leak
somewhere? Maybe I'm just over my head here. Well, tell me
what to test next I will give it a shot when I return from work.
All the tests were run with the ~same initial memory consumation
(2 console logins + X + 3 xterms + netscape + pine + 2 ssh) on a
p5-100, 32 MB RAM, 64 MB swap
-- Samuli Kaski, samkaski@cs.helsinki.fi Department of Computer Science, University of Helsinki, Finland.