From: Pan Li <[email protected]>
This patch would like to try to match the the unsigned
SAT_MUL form 5, aka below:
#define DEF_SAT_U_MUL_FMT_5(NT, WT) \
NT __attribute__((noinline)) \
sat_u_mul_##NT##_from_##WT##_fmt_5 (NT a, NT b) \
{ \
WT x = (WT)a * (WT)b; \
NT hi = x >> (sizeof(NT) * 8); \
NT lo = (NT)x; \
return lo | -!!hi; \
}
while WT is uint16_t, uint32_t and uint64_t.
and T is uint8_t, uint16_t, uint32_t.
gcc/ChangeLog:
* match.pd: Add pattern of mult and reuse the widen-mul
by for keyword.
Signed-off-by: Pan Li <[email protected]>
---
gcc/match.pd | 55 ++++++++++++++++++++++++++++------------------------
1 file changed, 30 insertions(+), 25 deletions(-)
diff --git a/gcc/match.pd b/gcc/match.pd
index 2e629fd31ce..b3fd26e18dd 100644
--- a/gcc/match.pd
+++ b/gcc/match.pd
@@ -3695,31 +3695,36 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
bool c2_is_type_precision_p = c2 == prec;
}
(if (widen_prec > prec && c2_is_type_precision_p && c4_is_max_p)))))
- (match (unsigned_integer_sat_mul @0 @1)
- /* SAT_U_MUL (X, Y) = {
- WT x = (WT)a * (WT)b;
- NT hi = x >> (sizeof(NT) * 8);
- NT lo = (NT)x;
- return lo | -!!hi;
- } while WT is uint128_t, T is uint8_t, uint16_t, uint32_t or uint64_t.
*/
- (convert1?
- (bit_ior (convert? (negate (convert (ne (convert (rshift @3 INTEGER_CST@2))
- integer_zerop))))
- (convert (widen_mult:c@3 (convert@4 @0)
- (convert@5 @1)))))
- (if (types_match (type, @0, @1))
- (with
- {
- unsigned widen_prec = TYPE_PRECISION (TREE_TYPE (@3));
- unsigned cvt4_prec = TYPE_PRECISION (TREE_TYPE (@4));
- unsigned cvt5_prec = TYPE_PRECISION (TREE_TYPE (@5));
- bool widen_mult_p = cvt4_prec == cvt5_prec && widen_prec == cvt5_prec *
2;
-
- unsigned c2 = tree_to_uhwi (@2);
- unsigned prec = TYPE_PRECISION (type);
- bool c2_is_type_precision_p = c2 == prec;
- }
- (if (widen_mult_p && c2_is_type_precision_p)))))
+ (for mult_op (mult widen_mult)
+ (match (unsigned_integer_sat_mul @0 @1)
+ /* SAT_U_MUL (X, Y) = {
+ WT x = (WT)a * (WT)b;
+ NT hi = x >> (sizeof(NT) * 8);
+ NT lo = (NT)x;
+ return lo | -!!hi;
+ } while WT is uint128_t, uint64_t, uint32_t, uint16_t,
+ and T is uint64_t, uint32_t, uint16_t, uint8_t. */
+ (convert1?
+ (bit_ior
+ (convert?
+ (negate
+ (convert (ne (convert2? (rshift @3 INTEGER_CST@2)) integer_zerop))))
+ (convert (mult_op:c@3 (convert@4 @0) (convert@5 @1)))))
+ (if (types_match (type, @0, @1))
+ (with
+ {
+ unsigned prec = TYPE_PRECISION (type);
+ unsigned widen_prec = TYPE_PRECISION (TREE_TYPE (@3));
+ unsigned cvt4_prec = TYPE_PRECISION (TREE_TYPE (@4));
+ unsigned cvt5_prec = TYPE_PRECISION (TREE_TYPE (@5));
+
+ bool widen_mult_p = mult_op == WIDEN_MULT_EXPR && cvt4_prec == cvt5_prec
+ && widen_prec == cvt5_prec * 2;
+ bool mult_p = mult_op == MULT_EXPR && cvt4_prec == cvt5_prec
+ && cvt4_prec == widen_prec && widen_prec > prec;
+ bool c2_is_type_precision_p = tree_to_uhwi (@2) == prec;
+ }
+ (if (c2_is_type_precision_p && (mult_p || widen_mult_p)))))))
)
/* The boundary condition for case 10: IMM = 1:
--
2.43.0
