Hi All, Both sides of the VEC_PERM_EXPR need to be a MULT but the check was accidentally checking if both sides are a mul.
The FMS case would be handled by the validate_multiplication but this makes the requirement more explicit and we exit earlier. Bootstrapped Regtested on aarch64-none-linux-gnu and no issues. Ok for master? Thanks, Tamar gcc/ChangeLog: PR tree-optimizaion/103311 * tree-vect-slp-patterns.c (complex_mul_pattern::matches, complex_fms_pattern::matches): Check for multiplications. gcc/testsuite/ChangeLog: PR tree-optimizaion/103311 * gcc.target/aarch64/pr103311.c: New test. --- inline copy of patch -- diff --git a/gcc/testsuite/gcc.target/aarch64/pr103311.c b/gcc/testsuite/gcc.target/aarch64/pr103311.c new file mode 100644 index 0000000000000000000000000000000000000000..7bcc7db79be8f1b4dc1f95c6cf10f5292191a892 --- /dev/null +++ b/gcc/testsuite/gcc.target/aarch64/pr103311.c @@ -0,0 +1,16 @@ +/* { dg-do compile } */ +/* { dg-require-effective-target arm_v8_3a_complex_neon_ok } */ +/* { dg-add-options arm_v8_3a_complex_neon } */ +/* { dg-additional-options "-std=c99" } */ + +_Complex double y; + +void +cbknu (_Complex double f) +{ + const _Complex double cone = 1.0e0; + + f = f * cone; + y = f * cone; +} + diff --git a/gcc/tree-vect-slp-patterns.c b/gcc/tree-vect-slp-patterns.c index 53fbe5185f543298b27d084a93c30a1c161d3d32..d916fc9cef9a86a0d797caaf2b2f253685e9b7dd 100644 --- a/gcc/tree-vect-slp-patterns.c +++ b/gcc/tree-vect-slp-patterns.c @@ -949,7 +949,7 @@ complex_mul_pattern::matches (complex_operation_t op, bool mul0 = vect_match_expression_p (l0node[0], MULT_EXPR); bool mul1 = vect_match_expression_p (l0node[1], MULT_EXPR); - if (!mul0 && !mul1) + if (!mul0 || !mul1) return IFN_LAST; /* Now operand2+4 may lead to another expression. */ @@ -1204,7 +1204,9 @@ complex_fms_pattern::matches (complex_operation_t op, /* If these nodes don't have any children then they're not ones we're interested in. */ - if (left_op.length () != 2 || right_op.length () != 2) + if (left_op.length () != 2 + || right_op.length () != 2 + || !vect_match_expression_p (l0node[1], MULT_EXPR)) return IFN_LAST; bool is_neg = vect_normalize_conj_loc (left_op); --
diff --git a/gcc/testsuite/gcc.target/aarch64/pr103311.c b/gcc/testsuite/gcc.target/aarch64/pr103311.c new file mode 100644 index 0000000000000000000000000000000000000000..7bcc7db79be8f1b4dc1f95c6cf10f5292191a892 --- /dev/null +++ b/gcc/testsuite/gcc.target/aarch64/pr103311.c @@ -0,0 +1,16 @@ +/* { dg-do compile } */ +/* { dg-require-effective-target arm_v8_3a_complex_neon_ok } */ +/* { dg-add-options arm_v8_3a_complex_neon } */ +/* { dg-additional-options "-std=c99" } */ + +_Complex double y; + +void +cbknu (_Complex double f) +{ + const _Complex double cone = 1.0e0; + + f = f * cone; + y = f * cone; +} + diff --git a/gcc/tree-vect-slp-patterns.c b/gcc/tree-vect-slp-patterns.c index 53fbe5185f543298b27d084a93c30a1c161d3d32..d916fc9cef9a86a0d797caaf2b2f253685e9b7dd 100644 --- a/gcc/tree-vect-slp-patterns.c +++ b/gcc/tree-vect-slp-patterns.c @@ -949,7 +949,7 @@ complex_mul_pattern::matches (complex_operation_t op, bool mul0 = vect_match_expression_p (l0node[0], MULT_EXPR); bool mul1 = vect_match_expression_p (l0node[1], MULT_EXPR); - if (!mul0 && !mul1) + if (!mul0 || !mul1) return IFN_LAST; /* Now operand2+4 may lead to another expression. */ @@ -1204,7 +1204,9 @@ complex_fms_pattern::matches (complex_operation_t op, /* If these nodes don't have any children then they're not ones we're interested in. */ - if (left_op.length () != 2 || right_op.length () != 2) + if (left_op.length () != 2 + || right_op.length () != 2 + || !vect_match_expression_p (l0node[1], MULT_EXPR)) return IFN_LAST; bool is_neg = vect_normalize_conj_loc (left_op);