This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] fix target/44606, reload bug on SPE
On Sat, Oct 02, 2010 at 02:45:08AM +0200, Bernd Schmidt wrote:
> On 10/02/2010 01:16 AM, Bernd Schmidt wrote:
> > For the release branches it's conceivable to disable this optimization,
> > it shouldn't trigger that often anyway.
>
> Actually, maybe it's best to just do that everywhere. It doesn't seem
> to trigger in an i686-linux bootstrap, and it's something we can clean
> up in postreload if we really want to.
I'd be really surprised if it triggered in a bootstrap anywhere, given
that GCC's floating-point usage is practically zero.
I honestly can't think of many places it'd help: soft-float targets and
slightly odd targets like E500 are the only ones that come to mind. Any
target that permits SImode subregs of DFmode registers would benefit, I
suppose, but even then the stars have to align...
-Nathan