Re: pkg 1.4 freeze please test test test!

On 29 Oct, Baptiste Daroussin wrote:
> Hi all,
> 
> We are starting the release process of pkg 1.4, we want to have a better release
> process than with every single previous version of pkg. For that we will need
> you help!
> 
> pkg-devel has been updated to the latest version of pkg as of alpha2.
> 
> Changes you can expect in pkg 1.4 are the following:
> - Loads of bug fixes

I kind of doubt that I'll have time to test it, but I've stumbled across
an interesting test case for package building with pkg-1.3.8_3.

When I tried to build a multimedia/2mandvd package with
poudriere (either bulk or testport) in a FreeBSD 10 amd64 host and jail,
pkg-static segfaults.  Portsmon also sees this failure, which also
seems to be affecting head/amd64 as well:
<http://portsmon.freebsd.org/portoverview.py?category=multimedia&portname=2mandvd&wildcard=>

If I run poudriere jail -i to keep the jail around, I don't see any
leftover core files, I'm guessing because pkg-static's cwd is in the r/o
/usr/ports tree.  If I then cd /usr/ports/multimedia/2mandvd in the
jail and run:
	make clean
	make stage
	make package
pkg-static doesn't segfault, but it never exits either.  I left it
running for a couple of days and it was still stuck at 100% CPU.  If
I truss -p the process, I don't get any output, which means it's not
doing any syscalls.

I tried attaching gdb to the process, but got some strange error
messages that I didn't understand.  I then ran pkg-static under gdb with
the same command line arguments and it still looped forever.  I
interrupted it to get a stack trace, but that wasn't helpful because the
executable was stripped.

If I run pkg instead of pkg-static, it seems to work properly.

I was hoping to gather some more information to file a bug report, but
haven't had time to work on this in the last week.