Created attachment 2926 [details] Create mux socket for regress in temp directory In some setups, creating the socket under OBJ may result in a path that is too long for a Unix domain socket. The failure in https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-zesty/zesty/s390x/o/openssh/20170103_003152_0b07a@/log.gz was due to this. While of course it's possible to say that everyone must ensure that OBJ is short enough to avoid exceeding this limit, the limit is quite low at least on Linux (108 characters), and it would seem friendlier to create a temporary directory instead so that it's more likely to fit. Patch attached (which overlaps a bit with that in https://bugzilla.mindrot.org/show_bug.cgi?id=2659, but not terribly).
This has indeed fixed my remaining failures: with this and my other recent regress/ patches, I now have a full set of passing tests on Debian/amd64 and Ubuntu/{amd64,armhf,i386,ppc64el,s390x}.
Created attachment 2964 [details] create forwarding.sh CTL socket in /tmp I think Colin's solution is best, but we're too close to release now. This just makes forwarding.sh use the same location as multiplex.sh and should take the immediate pain away.
Move incomplete bugs to openssh-7.6 target since 7.5 shipped a while back. To calibrate expectations, there's little chance all of these are going to make 7.6.
remove 7.5 target
Move to OpenSSH 7.8 tracking bug
Comment on attachment 2926 [details] Create mux socket for regress in temp directory looks reasonable to me
I've committed a slightly-modified version of Colin's patch. The only substantive change was s/TMP/SSH_REGRESS_TMP/ because the "rm -rf $TMP" gave me the heebie-jeebies wrt namespace pollution. commit c59aca8adbdf7f5597084ad360a19bedb3f80970 Author: Damien Miller <djm@mindrot.org> Date: Fri Jul 20 14:53:42 2018 +1000 Create control sockets in clean temp directories Adds a regress/mkdtemp tool and uses it to create empty temp directories for tests needing control sockets. Patch from Colin Watson via bz#2660; ok dtucker
closing resolved bugs as of 8.6p1 release