12b4374cd5
Also fix and extend the current implementation: - AtomicRMWOp::Clr != AtomicRmwOp::And, as the input needs to be inverted first. - Inputs to the cmp for the RMWLoop case are sign-extended when needed. - Lower Xchg to Swp. - Lower Sub to Add with a negated input. - Added more runtests. Copyright (c) 2022, Arm Limited.
41 lines
1017 B
Plaintext
41 lines
1017 B
Plaintext
test run
|
|
target s390x
|
|
|
|
; We can't test that these instructions are right regarding atomicity, but we can
|
|
; test if they perform their operation correctly
|
|
|
|
function %atomic_cas_i64(i64, i64, i64) -> i64 {
|
|
ss0 = explicit_slot 8
|
|
|
|
block0(v0: i64, v1: i64, v2: i64):
|
|
v3 = stack_addr.i64 ss0
|
|
store.i64 little v0, v3
|
|
|
|
v4 = atomic_cas.i64 little v3, v1, v2
|
|
|
|
v5 = load.i64 little v3
|
|
return v5
|
|
}
|
|
; run: %atomic_cas_i64(0, 0, 2) == 2
|
|
; run: %atomic_cas_i64(1, 0, 2) == 1
|
|
; run: %atomic_cas_i64(0, 1, 2) == 0
|
|
; run: %atomic_cas_i64(0, 0, 0xC0FFEEEE_DECAFFFF) == 0xC0FFEEEE_DECAFFFF
|
|
|
|
function %atomic_cas_i32(i32, i32, i32) -> i32 {
|
|
ss0 = explicit_slot 4
|
|
|
|
block0(v0: i32, v1: i32, v2: i32):
|
|
v3 = stack_addr.i64 ss0
|
|
store.i32 little v0, v3
|
|
|
|
v4 = atomic_cas.i32 little v3, v1, v2
|
|
|
|
v5 = load.i32 little v3
|
|
return v5
|
|
}
|
|
; run: %atomic_cas_i32(0, 0, 2) == 2
|
|
; run: %atomic_cas_i32(1, 0, 2) == 1
|
|
; run: %atomic_cas_i32(0, 1, 2) == 0
|
|
; run: %atomic_cas_i32(0, 0, 0xC0FFEEEE) == 0xC0FFEEEE
|
|
|