Bug#888769: gegl: FTBFS on machines with lots of cores

[James Cowgill]

Source: gegl
Version: 0.3.28-1
Severity: normal

Hi,

I started having a look at the mips build failure. To do this I tried to
run gegl on a mips machine with 64 cores. Unfortunately the testsuite
started randomly failing without out of memory errors.

Example:
> (gegl:8396): GLib-ERROR **: ../../../../glib/gmem.c:100: failed to allocate 262176 bytes
> FAIL clones.xml

Running gegl in gdb, I see that it runs out of memory trying to allocate
stack space for over 250 threads. Given the default stack size of 8MB,
256 threads would require 2GB of space and exceed the amount of virtual
memory available. 250 threads is close enough to cause problems.

I have not debugged this completely, but I do notice that there are a
number of calls to g_thread_pool_new in gegl/operation/*, but no calls
to g_thread_pool_free. Possibly these thread pools (and their threads)
are being leaked?

Thanks,
James

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <http://lists.alioth.debian.org/pipermail/pkg-gnome-maintainers/attachments/20180129/4deeebc8/attachment.sig>