This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [tree-ssa] PPC bootstraps still broken


In message <1057668748.16968.62.camel@frodo.toronto.redhat.com>, Diego Novillo 
writes:
 >On Tue, 2003-07-08 at 08:46, law@redhat.com wrote:
 >> In message <1057664080.16968.46.camel@frodo.toronto.redhat.com>, Diego Novi
 >llo 
 >> writes:
 >>  >
 >>  >I'm starting to suspect that there's something wrong with this machine
 >>  >or my local tree.  Could folks with access to PPC try a bootstrap of the
 >>  >branch to see if they can reproduce this failure?
 >> Why do you think that?  Are you getting failures again?
 >>
 >Because I seem to be the only one with bootstrapping problems.  I don't
 >get this failure on any of the other 6 architectures.  Toon didn't get
 >this failure on his ppc and Andrew Pinski says he gets bootstrap
 >failures with -fstrict-aliasing on pcc Darwin (he has been getting these
 >failures for a while now, IIRC).
 >
 >
 >> It sounds like there is a real problem to me and that it's extremely
 >> sensitive to anything changing.  That sounds like it precisely the kind
 >> of problem we want to nail down.
 >> 
 >Heh, no kidding.
 >
 >Could you try bootstrapping on your PPC box?
I'm on the road all week (leaving in about 20 minutes).  I can fire up a
bootstrap after I return.

However, I doubt that's going to be all that useful.  The bootstrap is clearly
failing on your end and from the way it sounds, it's unlikely I'll trip
the failure here.  This really needs to be debugged on your end.

So, what happens if you compile one of the failing .i files with the stage1
and stage2 compilers?  Do you get the same .s file or ones that differ?

Jeff



Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]