[pkg-gnupg-maint] Bug#789246: FTBFS amd64

Daniel Kahn Gillmor dkg at fifthhorseman.net
Sun Sep 13 21:08:52 UTC 2015


On Wed 2015-07-08 09:42:14 -0400, Daniel Kahn Gillmor wrote:
> On Tue 2015-06-30 18:37:05 -0400, Daniel Kahn Gillmor wrote:
>> looking at the test that it seems to hang on (agent/t-protect), i tried
>> running it on my own machine and got the attached strace.  (it completes
>> quickly)
>>
>> we checked with Q_, who looked at the buildd, and it's definitely
>> hanging running agent/t-protect: the process is in R (running) state.
>>
>> agent/t-protect.c is pretty simple, and it's not hanging on other
>> architectures.  I'm perplexed as to why it's not terminating on that
>> platform.
>
> We managed to get a copy of t-protect from the affected system
> (x86-csail-01) and it runs fine without any delay or hang on other
> systems.  So i suspect the problem is with x86-csail-01 itself.
>
> 2.1.6-1 built fine yesterday on babin, so i'm closing this ticket.

2.1.8-1 is again failing in the same place on x86-csail-01:

  https://buildd.debian.org/status/fetch.php?pkg=gnupg2&arch=amd64&ver=2.1.8-1&stamp=1441933875

I think this is a recurring problem with the GnuPG upstream sources,
gcrypt, and x86-csail-01.  Kurt Roeckx helped a bit with diagnosis
earlier on https://bugs.debian.org/789246 , and Niibe suggested some
possibilities having to do with gcrypt, but i've seen no further fixes
or attempts to address the problem on the machine.

I don't think i have any special access to that buildd -- what can we do
to get this sorted out?

In the meantime, is it possible to give gnupg2 2.1.8-1 back to a
different amd64 buildd (brahms and x86-grnet-01 both have worked in the
past) so that the package can propagate through the archive?

      --dkg



More information about the pkg-gnupg-maint mailing list