Is current -git working for anybody else?

Rob Landley rob at landley.net
Mon Nov 23 22:44:20 UTC 2009


On Sunday 22 November 2009 21:56:57 Mike Frysinger wrote:
> On Sunday 22 November 2009 22:22:23 Austin Foxley wrote:
> > On 11/11/2009 02:40 PM, Rob Landley wrote:
> > > I bisected the problem to:
> > >
> > > landley at driftwood:~/uclibc/git$ git bisect bad
> > > e0ac4efbdb498319f03a2a95d75d061ab6c68491 is first bad commit
> > > commit e0ac4efbdb498319f03a2a95d75d061ab6c68491
> > > Author: Mike Frysinger<vapier at gentoo.org>
> > > Date:   Thu Oct 22 01:05:28 2009 -0400
> > >
> > >      libc: add hidden calls to pthread cleanup funcs
> > >
> > >      A lot of libc code calls the pthread cleanup funcs implicitly (for
> > > stdio) which currently goes through the PLT.  Since we already have
> > > forwarding symbols for these funcs, it's safe to declare the internal
> > > libc usage hidden as a loaded libpthread will have the real symbols
> > > found.
> > >
> > >      Signed-off-by: Mike Frysinger<vapier at gentoo.org>
> > >
> > > I confirmed that reversing that patch fixes the problem.
> >
> > I can confirm that crash now as well, I'm going to revert this commit
> > until another fix can be made.
> >
> > Thanks for the bisecting work!
>
> how exactly did you confirm ?  dynamic x86_64 binaries seem to be running
> fine for me.

And my original post started:

>I built for i686, using a .config that works fine for 0.9.30.1 (plus one
> extra symbol to un-deprecate stuff busybox defconfig won't build without),
> and no program actually works linked against the result:

i686 != x86_64

> -mike

Rob
-- 
Latency is more important than throughput. It's that simple. - Linus Torvalds


More information about the uClibc mailing list