On Aug 19 11:19, Damien Miller wrote: > On Mon, 18 Aug 2014, Corinna Vinschen wrote: > > > I tested from CVS HEAD and there's a bug in serverloop.c. On systems > > not defining NO_IPPORT_RESERVED_CONCEPT, a stray "||" leads to a syntax > > error. Here's a patch: > > > > Index: serverloop.c > > applied - thanks. > > > Also, I can't run the testsuite on Cygwin anymore: > [...] > > -L. -Lopenbsd-compat/ -fstack-protector-all -lssh -lopenbsd-compat -lssh -lopenbsd-compat -lcrypto -lz /usr/lib/textreadmode.o -lcrypt > > make: *** No rule to make target 'regress/unittests/sshbuf/tests.o', needed by 'regress/unittests/sshbuf/test_sshbuf.exe'. Stop. > > > > This is using GNU make. I'm not sure what's missing. Is that because > > I'm not building in the source dir, by any chance? > > I don't think so - it works on Ubuntu with GNU Make and building outside > the source directory. There is probably some other make trickery that > I've messed up. Maybe it was the fact that I called `cvs up' without the -d flag...? Duh. Sorry about that. So, now I can run the testsuite and I'm having a problem with the "forwarding" test. The older version of this test worked fine, the latest from CVS sometimes fails: run test forwarding.sh ... failed copy /bin/ls.exe cmp: EOF on /home/corinna/src/openssh/64bin/regress/copy corrupted copy of /bin/ls.exe This is the last test in forwarding.sh, "transfer over chained unix domain socket forwards and check result" Inspecting the log files I see this a couple of times in failed-ssh.log: debug1: Local connections to LOCALHOST:3301 forwarded to remote address 127.0.0.1:4242 debug3: channel_setup_fwd_listener_tcpip: type 2 wildcard 0 addr NULL debug3: sock_set_v6only: set socket 5 IPV6_V6ONLY debug1: Local forwarding listening on ::1 port 3301. bind: Address already in use debug1: Local forwarding listening on 127.0.0.1 port 3301. bind: Address already in use channel_setup_fwd_listener_tcpip: cannot listen to port: 3301 Could not request local forwarding. failed-sshd.log shows a similar error. Running the test another time, it runs to completion. Running it again, it fails. It fails more often than succeeding, though. This may very well be a bug in Cygwin or a shortcoming of the underlying Windows, but I'm not sure what to look for. I attached the failed-foo.log files from a failed run to this mail, maybe somebody can sched some light on this problem? Other than this problem, the testsuite succeeds on Cygwin. Thanks, Corinna -- Corinna Vinschen Cygwin Maintainer Red Hat
Attachment:
failed-regress.tar.bz2
Description: BZip2 compressed data
Attachment:
pgpJUjRuoKehe.pgp
Description: PGP signature
_______________________________________________ openssh-unix-dev mailing list openssh-unix-dev@xxxxxxxxxxx https://lists.mindrot.org/mailman/listinfo/openssh-unix-dev