On 8/7/24 7:01 AM, Jin Ma wrote:
When rs1 is the immediate 0, the following ICE occurs:
error: unrecognizable insn:
(insn 8 5 12 2 (set (reg:RVVM1DI 134 [ <retval> ])
(if_then_else:RVVM1DI (unspec:RVVMF64BI [
(const_vector:RVVMF64BI repeat [
(const_int 1 [0x1])
])
(reg/v:DI 137 [ vl ])
(const_int 2 [0x2]) repeated x2
(const_int 0 [0])
(reg:SI 66 vl)
(reg:SI 67 vtype)
] UNSPEC_VPREDICATE)
(plus:RVVM1DI (mult:RVVM1DI (vec_duplicate:RVVM1DI (const_int 0
[0]))
(reg/v:RVVM1DI 136 [ vs2 ]))
(reg/v:RVVM1DI 135 [ vd ]))
(reg/v:RVVM1DI 135 [ vd ])))
gcc/ChangeLog:
* config/riscv/vector.md: Allow scalar operand to be 0.
gcc/testsuite/ChangeLog:
* gcc.target/riscv/rvv/base/bug-7.c: New test.
* gcc.target/riscv/rvv/base/bug-8.c: New test.
So I think we had a patch from Pan last year which tried to optimize
this case. While I don't think it makes sense to try and optimize this
case, we absolutely do need to avoid the ICE.
The lint phase of the pre-commit tester has several spaces vs tabs
errors flagged. Fix those and this is fine for the trunk. No need to
retest or wait for additional approvals.
https://github.com/ewlu/gcc-precommit-ci/issues/2021#issuecomment-2273493378
Jeff
