[gcc r14-6099] RISC-V: Robostify the W43, W86, W87 constraint enabled attribute
Pan Li
panli@gcc.gnu.org
Mon Dec 4 01:25:53 GMT 2023
https://gcc.gnu.org/g:d3544cea63d0a642b6357a7be55986f5562beaa0
commit r14-6099-gd3544cea63d0a642b6357a7be55986f5562beaa0
Author: Juzhe-Zhong <juzhe.zhong@rivai.ai>
Date: Mon Dec 4 06:47:41 2023 +0800
RISC-V: Robostify the W43, W86, W87 constraint enabled attribute
Committed as it is obvious fix.
gcc/ChangeLog:
* config/riscv/riscv.md: Rostify the constraints.
Diff:
---
gcc/config/riscv/riscv.md | 19 +++++++++++++++++--
1 file changed, 17 insertions(+), 2 deletions(-)
diff --git a/gcc/config/riscv/riscv.md b/gcc/config/riscv/riscv.md
index 4c6f63677df..ff521454876 100644
--- a/gcc/config/riscv/riscv.md
+++ b/gcc/config/riscv/riscv.md
@@ -515,13 +515,28 @@
(match_test "riscv_get_v_regno_alignment (GET_MODE (operands[0])) != 2"))
(const_string "no")
- (and (eq_attr "group_overlap" "W42,W43")
+ (and (eq_attr "group_overlap" "W42")
(match_test "riscv_get_v_regno_alignment (GET_MODE (operands[0])) != 4"))
(const_string "no")
- (and (eq_attr "group_overlap" "W84,W86,W87")
+ (and (eq_attr "group_overlap" "W84")
(match_test "riscv_get_v_regno_alignment (GET_MODE (operands[0])) != 8"))
(const_string "no")
+
+ ;; According to RVV ISA:
+ ;; The destination EEW is greater than the source EEW, the source EMUL is at least 1,
+ ;; and the overlap is in the highest-numbered part of the destination register group
+ ;; (e.g., when LMUL=8, vzext.vf4 v0, v6 is legal, but a source of v0, v2, or v4 is not).
+ ;; So the source operand should have LMUL >= 1.
+ (and (eq_attr "group_overlap" "W43")
+ (match_test "riscv_get_v_regno_alignment (GET_MODE (operands[0])) != 4
+ && riscv_get_v_regno_alignment (GET_MODE (operands[3])) >= 1"))
+ (const_string "no")
+
+ (and (eq_attr "group_overlap" "W86,W87")
+ (match_test "riscv_get_v_regno_alignment (GET_MODE (operands[0])) != 8
+ && riscv_get_v_regno_alignment (GET_MODE (operands[3])) >= 1"))
+ (const_string "no")
]
(const_string "yes")))
More information about the Gcc-cvs
mailing list