This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: patch for loop combination [Re: ppc]
- To: law at cygnus dot com
- Subject: Re: patch for loop combination [Re: ppc]
- From: Richard Henderson <rth at cygnus dot com>
- Date: Mon, 8 Feb 1999 21:18:45 -0800
- Cc: Joern Rennecke <amylaar at cygnus dot co dot uk>, meissner at cygnus dot com, egcs-patches at egcs dot cygnus dot com
- References: <19990207134650.A4396@cygnus.com> <15964.918524364@upchuck>
On Mon, Feb 08, 1999 at 06:39:24PM -0700, Jeffrey A Law wrote:
> Seems reasonable. We might as well kill the conditional initialization.
Ok.
> Though I thought there was some code which assumed that if that variable
> was initialized that the loop had a call.
Already caught it.
@@ -901,7 +898,8 @@ scan_loop (loop_start, end, loop_cont, u
Don't do this if P has a REG_RETVAL note or if we have
SMALL_REGISTER_CLASSES and SET_SRC is a hard register. */
- if (reg_single_usage && VARRAY_RTX (reg_single_usage, regno) != 0
+ if (loop_has_call
+ && VARRAY_RTX (reg_single_usage, regno) != 0
r~