On Sat, 2016-03-05 at 17:30 +0100, Andreas Metzler wrote: > Control: reopen 813598 > > On 2016-02-15 Nikos Mavrogiannopoulos <n...@gnutls.org> wrote: > > On Sun, Feb 14, 2016 at 3:14 PM, Andreas Metzler <ametz...@bebt.de> > > wrote: > > > this is http://bugs.debian.org/813598 reported by Steven > > > Chamberlain: > > > > gnutls28 tests/mini-loss-time fails about 20% of the time when I > > > try it > > > on kfreebsd-amd64. I think probably introduced by this commit: > > > https://gitlab.com/gnutls/gnutls/commit/e2a3ad31c487cbce997a08ddd > > > c55db639b60c024 > > > I've applied a similar fix in master. > > Well with 3.4.10 the timeout does not seem to make any difference. > Any of > > gnutls_dtls_set_timeouts(session, 1 * 1000, 29 * 1000); [Steven's > patch] > gnutls_dtls_set_timeouts(session, 1 * 1000, 30 * 1000); [3.4.10] > gnutls_dtls_set_timeouts(session, 1 * 1000, 29 * 1000); > > [e6dcb14dbbd3e9e40a1f193a7bf6657e82b88cb9] > *always* fails on kfreebsd-amd64.
Then the issue may have been different that what was initially thought. It would be nice to have some debugging output of the failed test (e.g., run with -v). Is that a 32-bit system? I'm wondering whether it relates to the undefined behavior fixes. regards, Nikos
