I'm not sure if this is a regression of the fix to issue #156, or if it's a new bug. For now, I've written up the description in a new comment on issue #156 (see #156 (comment)), because it seems very likely that this is a regression of the older bug. But if that's wrong, then I guess there's a shiny new seg fault instead.
Hope this helps, and please let me know if there's some better procedure I should have followed.
Okay, I rebuilt against the latest libminiupnpc (commit 05cc5daf14fe of 6 Sep 2014, from https://github.com/miniupnp/miniupnp.git) and the seg fault is gone. This is with the exact same bitcoin sources (commit 93193c8) with which I was able to reproduce the bug every time yesterday.
So I strongly suspect this was just a bug in the Debian-packaged version of libminiupnpc-dev in the 'testing' distro. Closing this ticket for now; if it reoccurs, I'll re-open.