在 2025/1/23 上午11:36, Xi Ruoyao 写道:
On Thu, 2025-01-23 at 11:21 +0800, Lulu Cheng wrote:
在 2025/1/22 下午9:26, Xi Ruoyao 写道:
The test case added in r15-7073 now triggers an ICE, indicating we need
the same fix as AArch64.
gcc/ChangeLog:
PR target/118501
* config/loongarch/loongarch.md (@xorsign<mode>3): Use
force_lowpart_subreg.
---
Bootstrapped and regtested on loongarch64-linux-gnu, ok for trunk?
LGTM!
Here we try to use force_lowpart_subreg when can_create_pseudo_p () is
true, right?
As this is a define_expand, can_create_pseudo_p should be just true (it
turns to false when the reload pass starts, and reload is far after
expand).
Um, that's not what I meant. What I meant is that in future code
implementation,
if can_create_pseudo_p is satisfied, we will not use lowpart_subreg but
use force_lowpart_subreg.
I'm pushing the patch into master now.
gcc/config/loongarch/loongarch.md | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/gcc/config/loongarch/loongarch.md
b/gcc/config/loongarch/loongarch.md
index 0325994ebd6..04a9a79d548 100644
--- a/gcc/config/loongarch/loongarch.md
+++ b/gcc/config/loongarch/loongarch.md
@@ -1343,8 +1343,8 @@ (define_expand "@xorsign<mode>3"
machine_mode lsx_mode
= <MODE>mode == SFmode ? V4SFmode : V2DFmode;
rtx tmp = gen_reg_rtx (lsx_mode);
- rtx op1 = lowpart_subreg (lsx_mode, operands[1], <MODE>mode);
- rtx op2 = lowpart_subreg (lsx_mode, operands[2], <MODE>mode);
+ rtx op1 = force_lowpart_subreg (lsx_mode, operands[1], <MODE>mode);
+ rtx op2 = force_lowpart_subreg (lsx_mode, operands[2], <MODE>mode);
emit_insn (gen_xorsign3 (lsx_mode, tmp, op1, op2));
emit_move_insn (operands[0],
lowpart_subreg (<MODE>mode, tmp, lsx_mode));
