[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: coredump in scm_ithrow
From: |
Gary Houston |
Subject: |
Re: coredump in scm_ithrow |
Date: |
17 Sep 2000 09:27:33 -0000 |
> From: Marius Vollmer <address@hidden>
> Date: 10 Sep 2000 15:43:03 +0200
> Dirk Herrmann <address@hidden> writes:
>
> > Is the problem really platform specific, or does it occur on
> > several/most/all platforms? If so, then we could rather check for
> > the compiler version than to try to create a test case that on some
> > platforms works and on others doesn't.
>
> I think detecting the GCSE bug can be arbitrarily hairy, and we should
> probably just put in the barrier unconditionally. That is, even on
> GCCs that don't have the GCSE bug, we will inhibit some optimizations,
> but I think these optimizations aren't important enough to care so
> strong about them.
It seems like the inhibited optimisation would be the one that
shouldn't have been done, so I wouldn't worry too much about
performance. Most likely it would be possible to avoid the bug by
reorganising the code, e.g., move the abort into a separate procedure.
I've removed the probably-broken BROKEN_GCSE test from configure.