On Wed, 9 Nov 2005, Robert Watson wrote: >> I have now had another live-lock after upgrading: >> >> http://www.nostrum.com/hang/hang.RELENG_6-trace-2005-11-09-0.txt >> Any other suggestions or pointers on how to identify this livelock? > > This looks like much the same issue in the UNIX domain sockets. I have > been looking at executing unp_gc in a deferred context, and have an > initial patch which I need to test some before I send to you. > Hopefully it will be ready for you to try out in a day or two. Last night I successfully sent this patch to the wrong person, as I'm chasing a number of different bugs currently. While it won't help with his quota-related problems, using a combination of patches (attached) I'm now able to run a set of file descriptor passing edge case regression tests successfully. I've committed the regression test to src/tools/regression/sockets/unix_passfd. This test should only be run once the patches are applied, needless to say. If you could try out the patches and let me know if things improve, that would be great. Thanks, Robert N M Watson
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:47 UTC