diff mbox series

tree-optimization/110838 - less aggressively fold out-of-bound shifts

Message ID 20230804101537.C02F313904@imap2.suse-dmz.suse.de
State Unresolved
Headers
Series tree-optimization/110838 - less aggressively fold out-of-bound shifts |

Checks

Context Check Description
snail/gcc-patch-check warning Git am fail log

Commit Message

Richard Biener Aug. 4, 2023, 10:15 a.m. UTC 
  The following adjusts the shift simplification patterns to avoid
touching out-of-bound shift value arithmetic right shifts of
possibly negative values.  While simplifying those to zero isn't
wrong it's violating the principle of least surprise.

Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.

	PR tree-optimization/110838
	* match.pd (([rl]shift @0 out-of-bounds) -> zero): Restrict
	the arithmetic right-shift case to non-negative operands.
---
 gcc/match.pd | 4 ++++
 1 file changed, 4 insertions(+)
diff mbox series

Patch

diff --git a/gcc/match.pd b/gcc/match.pd
index 53e622bf28f..a1a82a5f954 100644
--- a/gcc/match.pd
+++ b/gcc/match.pd
@@ -1064,6 +1064,10 @@  DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
  (simplify
   (shift @0 uniform_integer_cst_p@1)
   (if ((GIMPLE || !sanitize_flags_p (SANITIZE_SHIFT_EXPONENT))
+       /* Leave arithmetic right shifts of possibly negative values alone.  */
+       && (TYPE_UNSIGNED (type)
+           || shift == LSHIFT_EXPR
+	   || tree_expr_nonnegative_p (@0))
        /* Use a signed compare to leave negative shift counts alone.  */
        && wi::ges_p (wi::to_wide (uniform_integer_cst_p (@1)),
 		     element_precision (type)))