This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] rtlopt merge part 1 -- loop analysis (resent)
Hello,
> > Index: flow.c
> > ===================================================================
> > RCS file: /cvs/gcc/gcc/gcc/flow.c,v
> > retrieving revision 1.547
> > diff -c -3 -p -r1.547 flow.c
> > *** flow.c 17 Jan 2003 14:24:20 -0000 1.547
> > --- flow.c 24 Jan 2003 00:28:26 -0000
> > *************** init_propagate_block_info (bb, live, loc
> > *** 2000,2012 ****
> > rtx mem = SET_DEST (set);
> > rtx canon_mem = canon_rtx (mem);
> >
> > - /* This optimization is performed by faking a store to the
> > - memory at the end of the block. This doesn't work for
> > - unchanging memories because multiple stores to unchanging
> > - memory is illegal and alias analysis doesn't consider it. */
> > - if (RTX_UNCHANGING_P (canon_mem))
> > - continue;
>
> This bit shouldn't have been part of this patch.
>
> I also happen to think it's wrong.
Sorry; this must have leaked in from my attempt to fix PR8967
(http://gcc.gnu.org/ml/gcc-patches/2002-12/msg01164.html); by the way --
why do you think it is wrong?
> The rest of this looks fine.
Should I take it as an approval?
Zdenek