Bug#558360: Debian Bug report logs - #558360
"devnull@lordb.de" <devnull@lordb.de> writes:
> Sorry for the delay. Today I had some time to try your approach. I've
> attached the log to this email (I hope this works).
> I've no real clue on how to read this file but I think the two last
> lines describe the problem:
> read(5, "Package: ghostscript\nVersion: 1:"..., 4096) = 791
> --- SIGPIPE (Broken pipe) @ 0 (0) ---
I'm afraid your problem continues to make absolutely no sense to me (and
I've never been able to duplicate it). The trace that you included seems
to indicate that Lintian is dying with a SIGPIPE, which is a signal
delivered when a process tries to write to a pipe that's been closed on
the other end. However, you didn't use strace -f, just a regular strace,
so you're only seeing the parent process. The SIGPIPE appears to have
happened during get_deb_info() on the *.deb file that you're checking. In
that situation, I don't see where the parent Lintian process would ever be
writing to a pipe, nor does your strace show any sign of a write.
For the next release, I will have Lintian globally ignore SIGPIPE, which
is generally the correct programming method anyway since it's easier to
deal with write failures than signals. Maybe this will somehow solve your
problem, but I don't understand how these circumstances could exist.
--
Russ Allbery (rra@debian.org) <http://www.eyrie.org/~eagle/>
Reply to: