Moritz Mühlenhoff <[email protected]>: > On Sun, Nov 10, 2013 at 04:39:25PM +1300, Jan Larres wrote: >> Leaving out the patch didn't change anything, but I did some experiments >> with older versions and discovered that 1.0.0~pre18-1.3 worked fine, but >> from 1.0.1-1 on it's broken. > > Did you assess this by rebuilding 1.0.0~pre18-1.3 in current unstable or > by running 1.0.0~pre18-1.3 in a stable system?
I used a binary package from snapshot.debian.org in my current unstable
system. And just to make sure I just built 1.0.0~pre18-1.3 and 1.0.1-1
from source on the same system, and again the former works and the
latter is broken.
> I had a look at the diff between 1.0.0~pre18 and 1.0.1 and the only
> changes (from src/sltcp.c) are the logging code (which is present in
> your log) and some commented SSL_CTX_* calls.
I also had a look at the code and didn't see anything else immediately
suspicious, so I tried commenting out the two
gnutls_global_set_log_function() and gnutls_global_set_log_level() calls
and discovered that the latter was the culprit: setting the level to
anything higher than 1 triggered the bug, probably because the lowest
level at which gnutls seems to start printing out messages in this case
is 2. A quick look at the logging and print_error() functions in slrn
doesn't reveal anything obvious, but there must be some problem between
gnutls and slrn's error printing functionality.
Cheers,
Jan
--
-[ OpenPGP key ID: 00A0FD5F ]-
Why should I fear death? If I am, death is not. If death is, I am not. Why
should I fear that which cannot exist when I do?
-- Epicurus
signature.asc
Description: Digital signature
