[TAG] segmentation fault

[Pete Jewell]

J.Bakshi wrote:
> On Tue, 27 Sep 2005 08:25:19 -0700
> Mike Orr <mso at oz.net> wrote:
> 
> 
>>Run "df" and make sure you aren't running out of disk space.  Sometimes 
>>a full disk causes seemingly-unrelated errors.
> 
> 
> *df -H* shows
> 
> Filesystem             Size   Used  Avail Use% Mounted on
> /dev/hda2              21G   1.8G    19G   9%         /
> tmpfs                     65M      0    65M        0%     /dev/shm
> /dev/hda3               20G   936M    19G   5%     /data
> 
>>If the segfault happens at random times, the culprit is usually bad 
>>memory.  But if it happens consistently at a certain point, it's often a 
>>library mismatch.  Or Linux thinks you have more memory than you 
>>actually do, and it tried to put something at the top of memory and then 
>>retrieve it.  Run "free" and "dmesg | less" to verify how much memory 
>>Linux thinks you have.  If it guessed wrong, you'll have to set LILO/grub
> 
>  
> I have add on PCI so no memory sharing. I have 128MB RAM. BIOS shows 128MB at DDR1. POST shows 131072KB OK. but *free -m* shows as below
> 
>             total       used       free     shared    buffers     cached
> Mem:  123        117          6          0          9         39
> -/+ buffers/cache:           68         55
> Swap:   243         0        243
> 
> *memtest86+* shows *Memory 128MB*, but *memtest all* can't be executed as mlock fails due to page allocation problem though *memtest 100m * running fine.
> 
> I have also used *mem=128m* with grub but result is same as above.
> 
> the  swap in my PC  = 128MB x 2

What happens if you try increasing the amount of swap space?  Instead of
altering partitions (if there's no unused area on your HD), you could
try creating a swap file and using that.  For instance:

``
$ dd if=/dev/zero of=/swapfile bs=1024 count=524288
$ mkswap /swapfile
$ swapon /swapfile
''

This will create a 512mb swap file in the root directory, and use it.
You can confirm it's in use with the *free* command.  Then try the
processes that were more likely to trigger the segfault and see what
happens.

-- 
PeteJ