[mapserver-users] mapcache_seed segfault when using threading [SEC=UNCLASSIFIED]

[Stephen Woodbridge]

On 10/22/2012 3:39 AM, Matthew Doyle wrote:
> Hi Thomas,
>
> Thanks for getting back to me.
>
> It seems we have reached an interesting situation.  After reconfiguring
> (with --enable-debug), reinstalling, executing gdb and seeding using 'run'
> with the same params, I can not reproduce the segfault no matter how many
> threads I throw at it, even all the way up to 64. However, outside of gdb I
> get the segfault as always.
>
> Perhaps we have some sort of race condition or timing issue. Some googling
> shows that this is not entirely uncommon. Some others in the same boat have
> suggested running with Valgrind.
> (http://stackoverflow.com/questions/7057651/cannot-reproduce-segfault-in-gdb)
>
> Could you please advise on the next step? Would the core dump be useful?

Hi Matt,

Get the core file and start  gdb -c corefile mapcache_seed and then do a 
bt that would be useful.

Also please check that you do not have an old version of the libraries 
or executable in the path. I recently had a mapcache_seed segv problem 
that went away when I pulled a recent copy and rebuilt.

If you start:

gdb mapcache_seed    - this looks in the path for the executable
gdb ./mapcache_seed  - this looks in the file system

So be careful with this, I have been tripped up on occasion by it.

-Steve