https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126724
Bug ID: 126724
Summary: The check for SIDE EFFECTS on match patterns should be
using expr_no_side_effects_p instead
Product: gcc
Version: 17.0
Status: UNCONFIRMED
Keywords: internal-improvement, missed-optimization, wrong-code
Severity: normal
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: pinskia at gcc dot gnu.org
Target Milestone: ---
So the patterns like:
(simplify
(plus:c @0 (mult:c (minus @1 @0) zero_one_valued_p@2))
(if (INTEGRAL_TYPE_P (type)
&& (GIMPLE || !TREE_SIDE_EFFECTS (@1)))
(cond (convert:boolean_type_node @2) @1 @0)))
Checks TREE_SIDE_EFFECTS but those should be expr_no_side_effects_p instead.
The reason is TREE_SIDE_EFFECTS does not handle trapping instructions.
So in this case we start with an unconditional trapping instruction and make it
conditional.
Note there are 3 uses of TREE_SIDE_EFFECTS which is not needed at all:
```
/* (X ^ -(X < 0)) + (X < 0) -> abs (X) */
(simplify
(plus:c (bit_xor:c @0 (negate (convert@1 (lt @0 integer_zerop)))) @1)
(if (INTEGRAL_TYPE_P (TREE_TYPE (@0))
&& !TYPE_UNSIGNED (TREE_TYPE (@0))
&& !TYPE_SATURATING (TREE_TYPE (@0))
&& (GIMPLE || !TREE_SIDE_EFFECTS (@0)))
(abs @0)))
...
/* popcount(rotate(X Y)) is popcount(X). */
(for popcount (POPCOUNT)
(for rot (lrotate rrotate)
(simplify
(popcount (convert?@0 (rot:s@1 @2 @3)))
(if (INTEGRAL_TYPE_P (TREE_TYPE (@0))
&& INTEGRAL_TYPE_P (TREE_TYPE (@1))
&& (GIMPLE || !TREE_SIDE_EFFECTS (@3)))
(with { tree type0 = TREE_TYPE (@0);
tree type1 = TREE_TYPE (@1);
unsigned int prec0 = TYPE_PRECISION (type0);
unsigned int prec1 = TYPE_PRECISION (type1); }
(if (prec0 == prec1 || (prec0 > prec1 && TYPE_UNSIGNED (type1)))
(popcount (convert:type0 @2))))))))
/* parity(rotate(X Y)) is parity(X). */
(for parity (PARITY)
(for rot (lrotate rrotate)
(simplify
(parity (convert?@0 (rot:s@1 @2 @3)))
(if (INTEGRAL_TYPE_P (TREE_TYPE (@0))
&& INTEGRAL_TYPE_P (TREE_TYPE (@1))
&& (GIMPLE || !TREE_SIDE_EFFECTS (@3))
&& TYPE_PRECISION (TREE_TYPE (@0))
>= TYPE_PRECISION (TREE_TYPE (@1)))
(with { tree type0 = TREE_TYPE (@0); }
(parity (convert:type0 @2)))))))
```
The last 2 seems like they had them because of the removal of @3 but genmatch
has code already to handle that case correctly.So they can be safely removed.
