4 GCC regressions, 2 new, with your patch on 2002-07-18T09:56:17Z.

Jan Hubicka jh@suse.cz
Mon Jul 22 13:15:00 GMT 2002


> On Mon, Jul 22, 2002 at 03:39:57PM +0200, Jan Hubicka wrote:
> > The code is wrong as well I believe.  It should verify whether the code
> > in between earliest and jump change the operands used by to be inserted
> > instruction. This is bit tricky to do, I plan to do in separate patch.
> 
> This is already done.  The problem must be something else.

I see, then I will check deeper what is going on in ifcvt part.
Still I believe the patch to avoid splitting of flags setters and users
is profitable, especially for i386.
Once we will be able to do motion of instructions containing parallels,
like cfg branch does, we will lose opurtunities by trying to place it
where the flags are not dead.  I had the patch prepared for perfomrance
reasons before this problem had come out.

Honza
> 
> 
> r~



More information about the Gcc-patches mailing list