how to verify gpg-signed commits

James Henstridge james.henstridge at gmail.com
Wed Sep 20 14:36:04 BST 2006


On 19/09/06, John Arbash Meinel <john at arbash-meinel.com> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> James Henstridge wrote:
> > On 15/09/06, John Arbash Meinel <john at arbash-meinel.com> wrote:
> >> > There is only one crasher bug I've run into with the current version
> >> > (to do with listing/reading keys), and that turned out to be a
> >> > double-free bug in gpgme itself.  That bug was fixed in 1.1.0 though.
> >> >
> >> > If you can provide details of the crash, I can take a look at it.
> >>
>
> ...
>
> >> Judging by the error, I don't know that these tests are 100% meant to
> >> pass. (It looks like gpgme-1.1.0 is giving incomplete data that
> >> gpgme-1.1.2-19 is returning).
> >
> > The above test failures look harmless, but they should be fixed.  I'll
> > look at fixing those problems (probably by loosening the affected
> > tests).
> >
> >
> >> I haven't been able to test on the Mac yet, because darwinports decided
> >> to be uncooperative when I wanted to update everything.
> >
> > I'd expect that you see similar test failures to those above, or no
> > errors at all.  I would be interested if you are still seeing segfault
> > failures though.
> >
> > James.
> >
>
> Well, I seem to be running with gpgme-11.6.1. According to port search
> it is gpgme-1.1.2, and I'm still getting the segfault. (Bus Error)
>
> It seems to be crashing in the initialization routines:
> 0   libSystem.B.dylib   0x9000197c pthread_mutex_lock + 36
> 1   <<00000000>>        0x00000001 0 + 1
> 2   org.python.python   0x100bc3cc PyThread_release_lock + 40
> (thread_pthread.h:440)
> 3   _gpgme.so           0x000bdb34 init_gpgme + 3312
> 4   libgpgme.11.dylib   0x0020717c _gpgme_data_outbound_handler + 52
> (data.c:246)
> 5   libgpgme.11.dylib   0x002085ac _gpgme_run_io_cb + 236 (wait.c:204)
> 6   libgpgme.11.dylib   0x00208fd8 _gpgme_wait_on_condition + 268
> (wait-private.c:109)
> 7   _gpgme.so           0x000bfbb4 init_gpgme + 11632
>
>
> I can give you more of the traceback if you like.

That'd be useful if you can get debug symbols in _gpgme.so.  The above
looks like it is just giving offsets from the one exposed symbol in
the library.

>
> Now, it is possible that the problem is actually when binding to
> pthread. I'm not sure why, considering I just compiled all of the gpgme
> libraries (this is a fresh darwinports install). And
> /usr/lib/libpthread.dylib is available (symlinked to
> /usr/lib/libSystem.B.dylib).
>
> I would have guessed that if pthread wasn't working, other things would
> be failing, but I could be wrong about that.
>
> Do you know a way to make it choose gpgme-pth instead of gpgme-pthread?

At the moment it is linking to the plain "libgpgme".  It should
probably be linking with libgpgme-pthread though, or whatever thread
implementation Python uses.  Does Python use pth on OSX?

James.




More information about the bazaar mailing list