2001-02-16 09:34:49

by Ketil Froyn

[permalink] [raw]
Subject: out of memory?

Hi.

I'm getting lots of this on a computer:
VM: do_try_to_free_pages failed for myprog.pl
VM: do_try_to_free_pages failed for myprog.pl
VM: do_try_to_free_pages failed for myprog.pl
....
VM: do_try_to_free_pages failed for kupdate (just saw this once)

myprog is basically making lots of directories, and is not using lots of
memory. Here's the info from 'ps auxw':

USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
root 994 22.6 0.3 2568 364 tty1 R Feb15 230:40 perl -w myprog.pl

It doesn't seem like I'm out of memory:
$ cat /proc/meminfo
total: used: free: shared: buffers: cached:
Mem: 97730560 96149504 1581056 2125824 5447680 5115904
Swap: 68083712 35213312 32870400
MemTotal: 95440 kB
MemFree: 1544 kB
MemShared: 2076 kB
Buffers: 5320 kB
Cached: 4996 kB
SwapTotal: 66488 kB
SwapFree: 32100 kB
$ uname -a
Linux localhost.localdomain 2.2.18RAID #6 Wed Feb 14 19:15:49 CET 2001 \
i586 unknown

The kernel is compiled with the rh-7.0 kgcc (egcs-2.91.66), and I've
patched it to get raid 0.90 and reiserfs 3.5.29.

What's going on? How bad is this?

Ketil Froyn


2001-02-16 10:12:58

by Alan

[permalink] [raw]
Subject: Re: out of memory?

> VM: do_try_to_free_pages failed for myprog.pl
> VM: do_try_to_free_pages failed for myprog.pl
> VM: do_try_to_free_pages failed for kupdate (just saw this once)
>
> The kernel is compiled with the rh-7.0 kgcc (egcs-2.91.66), and I've
> patched it to get raid 0.90 and reiserfs 3.5.29.
> What's going on? How bad is this?

The VM logs cases where it actually ends up having trouble getting free
memory and sits around before it eventually gets life going again. Its mostly
a debugging aid and in itself harmless.

2001-02-16 20:40:31

by George

[permalink] [raw]
Subject: Re: out of memory?

On Fri, 16 Feb 2001, Alan Cox wrote:

>> VM: do_try_to_free_pages failed for myprog.pl
>> VM: do_try_to_free_pages failed for myprog.pl
>> VM: do_try_to_free_pages failed for kupdate (just saw this once)
>>
>> The kernel is compiled with the rh-7.0 kgcc (egcs-2.91.66), and I've
>> patched it to get raid 0.90 and reiserfs 3.5.29.
>> What's going on? How bad is this?
>
>The VM logs cases where it actually ends up having trouble getting free
>memory and sits around before it eventually gets life going again. Its mostly
>a debugging aid and in itself harmless.

I've had 2.2.18 go into an infinite kernel loop with that message when
running out of memory. ALT+SysRq would respond and print what it tries to
do, but never actually accomplish it. (i.e., it'd say "Syncing disks." but
never actually do it.) Apparently userspace never ran again because sockets
would be accepted still but never progress.

-George Greer