Re: optimization/7147: ifcvt.c problem (regression)

[Franz Sirl <Franz.Sirl-kernel@lauterbach.com>]

The following reply was made to PR optimization/7147; it has been noted by GNATS.

From: Franz Sirl <Franz.Sirl-kernel@lauterbach.com>
To: Richard Henderson <rth@redhat.com>
Cc: gcc-gnats@gcc.gnu.org,gcc-patches@gcc.gnu.org,
 Mark Mitchell <mark@codesourcery.com>
Subject: Re: optimization/7147: ifcvt.c problem (regression)
Date: Thu, 18 Jul 2002 15:05:16 +0200

 At 23:27 16.07.2002, Richard Henderson wrote:
 >This looks like fallout from
 >
 >2002-05-03  Richard Henderson  <rth@redhat.com>
 >
 >         PR opt/6534
 >         * ifcvt.c (noce_try_store_flag, noce_try_store_flag_constants,
 >         noce_try_store_flag_inc, noce_try_store_flag_mask, noce_try_cmove,
 >         noce_try_cmove_arith, noce_try_minmax, noce_try_abs): Insert new
 >         code before JUMP, not EARLIEST.
 >
 >I'm testing the following on x86.  Give it a whirl on ppc as well?
 >
 >
 >r~
 >
 >
 >         * ifcvt.c (noce_get_condition): Make certain that the condition
 >         is valid at JUMP.
 
 That fixes the small executable testcase I posted and inspection of the 
 assembly shows that the original mozilla source code is fine too now. 
 Bootstrap and regression testing on powerpc-linux-gnu without problems.
 
 Franz.