Re: Boost build failure on hppa
On Sat, May 16, 2009 at 02:53:41PM +0000, Philipp Kern wrote:
> [ Please Cc me as I'm not subscribed to the list. ]
> On 2009-05-14, Steve M. Robbins <email@example.com> wrote:
> > On Wed, May 13, 2009 at 05:37:36PM +0200, Domenico Andreoli wrote:
> >> On 5/12/09, Steve M. Robbins <firstname.lastname@example.org> wrote:
> >> >
> >> > No other arch failed to copy these files, and the previous build (-5)
> >> > on hppa  exhibited no such trouble. The only difference between
> >> > -5 and -6 is a debian/control change to a conflicts specification. I
> >> > blame the build daemon. Can someone have a look?
> >> indeed it built successfully on my hppa box.
> > Thanks, Domenico.
> > Could the powers-that-be please reschedule boost1.38 on hppa?
> Please Cc email@example.com next time.
> I just discovered this mail by chance. I rescheduled it yesterday in order
> to get more build material for a "new" hppa buildds (paer)[*] and it built
> Sadly, in other news, paer suffers from random segfaults, too. This makes
> building quite unreliable, but at least the machine does not crash like
> our other buildds (peri and penalosa) do. It would be nice if someone
> could debug this, as it makes buildd maintainance fairly unpleasant.
FYI: Kyle Mcmartin, John David Anglin (hppa gcc maintainer), Carlos
O'Donnell, and some others have looked into this. As of earlier this
week, the problem was narrowed down to funtion pointers getting corrupted
somehow. I don't know what current status is though. Mostly writing this
so you know a few people are looking into it.