This is the mail archive of the
mailing list for the GCC project.
Re: [patch] PR28621 (-ffast-math -Os segfault)
> > Maybe not. However unless you're planning on fixing PR27537 before we
> > release 4.2, I'd say that's a different problem. It's also always been
> > broken.
> > PR28621 is a regression that can effect users even if they don't use
> > MMX/SSE in their application.
> I'd like to see a solution which can be ported to 4.1.
I don't buy that argument. We have an attribute for exactly this purpose, so
we may as well use it. Uglifying the code for the sake of consistency with
previous releases doesn't seem like it gains us anything.
There is a patch that may be suitable for 4.1 in the PR, which you are free to
test and submit for that branch.