[pkg-gnupg-maint] Bug#844117: gnupg2 races and hangs

[Ian Jackson]

In #841143 I wrote:
> I'm afraid I don't have time now to investigate the gnupg2 source
> code.  For now I will arrange for my test suite to cause the migration
> to happen once for the whole test suite.

I did this.  gnupg2 in the dgit test suite is still causing trouble,
even so.  Just now I found that I had four separate gnupg2 processes
all stuck trying to make signatures.  I killed two gpg-agent processes
I found lying around, one of which was clearly too hold, and that
seemed to help.  That's with 2.1.16-3.

I also tried to look at the gnupg2 source code.  After reading it I
was not surprised the agent startup is racy, and I was very
discouraged from trying to debug it.  Is there a design document or
comment somewhere which I missed, which explains how the startup
protocol is supposed to work ?  It might be possible to compare the
code to such a specification.  Without such a specification it would
be almost impossible to write a correct race-free approach.

In #844117 you wrote:
> I haven't heard back from you about this, Ian.  Can you let me know
> whether this lockup is resolved for you?

I'm afraid I haven't had time to play around with caff.  I'm currently
trying to sort out a mess of corrupted git commits due to a dgit bug.
But I'm pretty sure there are still bugs here.

Ian.

-- 
Ian Jackson <ijackson at chiark.greenend.org.uk>   These opinions are my own.

If I emailed you from an address @fyvzl.net or @evade.org.uk, that is
a private address which bypasses my fierce spamfilter.