On Fri, Oct 21, 2016 at 11:41:01AM +0100, Daniel P. Berrange wrote: > On Fri, Oct 21, 2016 at 11:35:38AM +0100, Peter Maydell wrote: > > On 21 October 2016 at 11:28, Daniel P. Berrange <berra...@redhat.com> wrote: > > > On Thu, Oct 20, 2016 at 05:11:56PM +0100, Peter Maydell wrote: > > >> Unfortunately the new test seems to fail on OSX: > > >> > > >> GTESTER tests/test-io-channel-socket > > >> ** > > >> ERROR:/Users/pm215/src/qemu-for-merges/tests/test-io-channel-socket.c:386:void > > >> test_io_channel_unix(_Bool): assertion failed: > > >> (g_file_test(TEST_SOCKET, G_FILE_TEST_EXISTS) == FALSE) > > >> GTester: last random seed: R02S300c198252e54fe12ff5d64603150e68 > > >> ** > > >> ERROR:/Users/pm215/src/qemu-for-merges/tests/test-io-channel-socket.c:386:void > > >> test_io_channel_unix(_Bool): assertion failed: > > >> (g_file_test(TEST_SOCKET, G_FILE_TEST_EXISTS) == FALSE) > > >> GTester: last random seed: R02S7ade3405366ffc33ad5cf50619671f53 > > > > > > Oh interesting - I've just seen my Travis CI build check also failed > > > over night on OS-X. > > > > > > I wonder if OS-X automatically deletes UNIX non-abstract sockets paths > > > on the filesystem when the socket FD is closed ?!?! > > > > The assertion is the other way round, isn't it? It's trying to > > assert that the path doesn't exist, but in fact it does > > (and it's still lying around in my working tree after the > > test failure). > > Oooooh, I'm looking at the wrong test case. I was assuming it was the > newly added test case that failed, whereas it is the original test > case we've regressd on.
Ok, the problem is that SO_ACCEPTCONN is not working on OS-X. Googling it seems this is likely broken on some versions of BSD too, so I'll rework the patch to avoid that function. Regards, Daniel -- |: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org :| |: http://entangle-photo.org -o- http://search.cpan.org/~danberr/ :|
