Re: kern/40027 (pagedaemon loops on memory shortage)

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Wed, Dec 03, 2008 at 02:05:03PM +0000, Andrew Doran wrote:
> The following reply was made to PR kern/40027; it has been noted by GNATS.
> 
> From: Andrew Doran <ad%NetBSD.org@localhost>
> To: gnats-bugs%netbsd.org@localhost
> Cc: 
> Subject: Re: kern/40027 (pagedaemon loops on memory shortage)
> Date: Wed, 3 Dec 2008 14:04:57 +0000
> 
>  With this patch against -current and the test program, a dual CPU machine
>  with 128MB RAM stalls during heavy paging activity, but eventually the
>  memory hog process is killed and it recovers.
>  
>  Memory: 3728K Act, 2156K Inact, 500K Wired, 4484K Exec, 124K File, 95M Free

I still see the problem here:
load averages:  1.17,  0.86,  0.41;               up 0+00:07:05        16:16:49
42 processes: 1 runnable, 40 sleeping, 1 on CPU
CPU states:  0.0% user,  0.0% nice, 97.1% system,  0.0% interrupt,  2.9% idle
Memory: 304M Act, 149M Inact, 5824K Wired, 8924K Exec, 21M File, 20K Free
Swap: 256M Total, 256M Used, 8K Free

  PID USERNAME PRI NICE   SIZE   RES STATE      TIME   WCPU    CPU COMMAND
    0 root     126    0     0K 9568K pgdaemon   3:38  0.88%  0.88% [system]
10573 root      81    0  5904K   22M RUN        0:11  0.00%  0.00% cc1
18891 root      85    0    76K    4K wait       0:02  0.00%  0.00% <pbulk-build
3229 bouyer    43    0    96K  844K CPU        0:00  0.00%  0.00% top

Did you try writing large files while running the memory hog program ?
I found that producing dirty pages to the file cache while running short
or memory makes the problem much more likely to happen.

-- 
Manuel Bouyer, LIP6, Universite Paris VI.           
Manuel.Bouyer%lip6.fr@localhost
     NetBSD: 26 ans d'experience feront toujours la difference
--