From a56122dea468bc9fc808adee44d788903ad9ff76 Mon Sep 17 00:00:00 2001
From: Alexandre Oliva <oliva@adacore.com>
Date: Thu, 23 Jan 2025 23:23:13 -0300
Subject: [PATCH] [ifcombine] improve reverse checking and operand swapping
Don't reject an ifcombine field-merging opportunity just because the
left-hand operands aren't both reversed, if the second compare needs
to be swapped for operands to match.
Also mention that reversep does NOT affect the turning of range tests
into bit tests.
for gcc/ChangeLog
* gimple-fold.cc (fold_truth_andor_for_ifcombine): Document
reversep's absence of effects on range tests. Don't reject
reversep mismatches before trying compare swapping.
---
gcc/gimple-fold.cc | 10 ++++++----
1 file changed, 6 insertions(+), 4 deletions(-)
diff --git a/gcc/gimple-fold.cc b/gcc/gimple-fold.cc
index 01c4d076af26..5f2fe56ea383 100644
--- a/gcc/gimple-fold.cc
+++ b/gcc/gimple-fold.cc
@@ -8095,8 +8095,9 @@ fold_truth_andor_for_ifcombine (enum tree_code code, tree truth_type,
return 0;
}
- /* Prepare to turn compares of signed quantities with zero into
- sign-bit tests. */
+ /* Prepare to turn compares of signed quantities with zero into sign-bit
+ tests. We need not worry about *_reversep here for these compare
+ rewrites: loads will have already been reversed before compares. */
bool lsignbit = false, rsignbit = false;
if ((lcode == LT_EXPR || lcode == GE_EXPR)
&& integer_zerop (lr_arg)
@@ -8203,10 +8204,11 @@ fold_truth_andor_for_ifcombine (enum tree_code code, tree truth_type,
the rhs's. If one is a load and the other isn't, we have to be
conservative and avoid the optimization, otherwise we could get
SRAed fields wrong. */
- if (volatilep || ll_reversep != rl_reversep)
+ if (volatilep)
return 0;
- if (! operand_equal_p (ll_inner, rl_inner, 0))
+ if (ll_reversep != rl_reversep
+ || ! operand_equal_p (ll_inner, rl_inner, 0))
{
/* Try swapping the operands. */
if (ll_reversep != rr_reversep
--
GitLab