On Wed, 2011-06-22 at 16:44 +0200, Jan Zelený wrote:
Nack. In monitor_config_file() you have
ret = monitor_config_file_fallback(mem_ctx, ctx, file, fn);
But you never check this return value. You always return EOK.
Thanks for catching this, new patch attached.
Also, please describe how you tested this bug. I assume you did so
exhausting the inotify watches like we discussed. Could you attach the
reproducer you wrote to the original BZ
so that QE can use it
to verify the issue?
Actually I didn't use the program we discussed. I wrote it, but it was useless
since it hit a watch limit before it managed to exhaust all watches in the
system. What I did was started sssd and setting the limit in proc to 1. I'll
write it to the BZ.