Not strictly a bug maybe, although I could never finish the test, as I simply run out of diskspace.
The test called "bigcore" will not react if "kill -9"ed and does not stop when interrupting the sandbox with ctrl+c (sandbox cleans, closes yet "bigcore" keeps running).
Somehow like mount, while it is (endlessly) waiting for a device. No zombie, but unkillable.
I feel like at least this test is to be disabled, as it is really uncool to compile something, dont watch it, and come back and see the disk being full (as the test runs as root by default) with all of it's consequences.
are becoming huge...
The only way I could stop it was rebooting, which isnt nice either...
Steps to Reproduce:
Created attachment 115653 [details]
if you cant `kill -9` the process, then that usually means kernel/hardware problem
ive def completed the gdb tests w/only ~5gigs space available
Well - how to check this? I don't have any problems that would account for this, and I run the latest stable gentoo-sources (2.6.19-r5)
The process creates alot of system load, which might incidate that it's a problem with the kernel.
I'll try it again with the latest gentoo-sources available soon.
What was/is your ulimit -f set to?
Since it works without problems if I set it to 100MB for instance.
ulimit -c doesnt seem to affect this.
Maybe I'll have a closer look at what this test actually tries to do.
bigcore.c contains the following part:
/* Try to expand all the resource limits beyond the point of sanity
- we're after the biggest possible core file. */
print_string ("Maximize resource limits ...\n");
maximize_rlimit (RLIMIT_CORE, "core");
maximize_rlimit (RLIMIT_DATA, "data");
maximize_rlimit (RLIMIT_STACK, "stack");
maximize_rlimit (RLIMIT_AS, "stack");
Therefore, it is not strictly a bug, as what I expected.
I feel like it's a stupid thing anyway as it is unkillable...
sorry, forgot closing
(maybe will file a bug at gdb)
I'm having the exact same problems, except my entire system partition is 20GB, so my file is only using about 14GB of space.
If I remember correctly, root isn't affected by the resource limitations anyway, so if root privileges aren't dropped, this test seem quite stupid.
I would be thankful if you could give me a link to the corresponding gdb bug report.