Re: problem upgrading the toolchain

Please don't file a duplicate bug about this problem as we're well aware
of the existing bug. The big problem is that we don't seem to be able to
reproduce the bug as everything seems fine in our testing using
glibc-2.3.5*.

At least 3 Gentoo developers (myself included) have tried to reproduce
it without any luck which makes it quite hard to fix.

So what I'd like is a bit more information about the machines this
problem happens on. Model of the box, glibc version, kernel version and
patchset and output from emerge --info.

Please add all this info to bug 100259 or send it to this list or me
directly at [email protected] and I'll see if I can find some pattern
to this. Hopefully this will give me some much needed clue about this
problem.

Regards,
Bryan Østergaard

On Tue, Dec 06, 2005 at 07:39:53PM -0800, Jeff Donsbach wrote:
> Thanks, Marc,
> 
> I guess it's a good thing I didn't build with
> USE="-nptlonly" then, huh? ;-)
> 
> Should I file a duplicate bug in bugzilla (if it would
> help)?
> 
> So, is there some kind of catch 22 going on here?
> Weren't there some reports of problems with
> linuxthreads in glibc on Alpha?
> 
> Jeff D
> 
> --- Marc Giger <[email protected]> wrote:
> 
> > Hi,
> > 
> > I reported the same problem to bugzilla...
> > http://bugs.gentoo.org/show_bug.cgi?id=100259#c36
> > (Perhaps it finds more attention here...)
> > 
> > It seems it has to do with the current state of ntpl
> > on
> > alphas.
> > 
> > If you set LD_ASSUME_KERNEL=2.4.1 and therefore
> > force
> > to use linuxthreads, emerge works again.
> > But I don't know if it really is nptl's fault. I can
> > manually
> > compile everything with the same parameters as
> > portage does. So a portage (sandbox?) fault is also
> > possible.
> > 
> > Marc
> > 
> > 
> > On Tue, 6 Dec 2005 10:45:57 -0800 (PST)
> > Jeff Donsbach <[email protected]> wrote:
> > 
> > > 
> > > Good point, Nils. Perhaps "5" is the configure.log
> > I
> > > included? I don't know.I'll try that tonight (the
> > > Miata is at home).
> > > 
> > > Jeff D
> > > 
> > > --- "Nils o. Janus" <[email protected]> wrote:
> > > 
> > > > Hi Jeff,
> > > > 
> > > > > configure:1892: checking whether the C
> > compiler
> > > > > (/usr/bin/gcc -mieee -mcpu=ev56 -pipe -O2 )
> > works
> > > > > configure:1908: /usr/bin/gcc -o conftest
> > -mieee
> > > > > -mcpu=ev56 -pipe -O2   conftest.c  1>&5
> > > > 						 ^^^^
> > > > what perplexes me is that stdout is redirecting
> > to
> > > > the filedescriptor 5
> > > > which is not a standardized one.
> > > > Does that filedescriptor ( /dev/fd/5 ) exist?
> > > > This may be the reason why compiling without
> > > > redirecting the output, like
> > > > you already tried, works fine.
> > > > 
> > > > Try passing the 1>&5 to the call of gcc too and
> > > > check wether it segfaults  
> > > > again.
> > > > 
> > > > On a DS20e, gcc just throws a "Bad file
> > descriptor"
> > > > error without  
> > > > segfaulting.
> > > > 
> > > > HTH,
> > > > 
> > > > greetings
> > > > Nils
> > > > -- 
> > > > [email protected] mailing list
> > > > 
> > > > 
> > > 
> > > 
> > > 
> > > 		
> > > __________________________________________ 
> > > Yahoo! DSL – Something to write home about. 
> > > Just $16.99/mo. or less. 
> > > dsl.yahoo.com 
> > > 
> > > -- 
> > > [email protected] mailing list
> > > 
> > 
> > -- 
> > [email protected] mailing list
> > 
> > 
> 
> 
> __________________________________________________
> Do You Yahoo!?
> Tired of spam?  Yahoo! Mail has the best spam protection around 
> http://mail.yahoo.com 
> -- 
> [email protected] mailing list
> 
> 
-- 
[email protected] mailing list