derek.holloway
7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
232 lines
6.3 KiB
LLVM
232 lines
6.3 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -verify-machineinstrs | FileCheck %s --check-prefix=X64
|
|
; RUN: llc < %s -mtriple=i686-unknown-unknown -verify-machineinstrs | FileCheck %s --check-prefix=X86
|
|
|
|
; Make sure that flags are properly preserved despite atomic optimizations.
|
|
|
|
define i32 @atomic_and_flags_1(ptr %p, i32 %a, i32 %b) {
|
|
; X64-LABEL: atomic_and_flags_1:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: cmpl %edx, %esi
|
|
; X64-NEXT: jne .LBB0_3
|
|
; X64-NEXT: # %bb.1: # %L1
|
|
; X64-NEXT: incb (%rdi)
|
|
; X64-NEXT: cmpl %edx, %esi
|
|
; X64-NEXT: jne .LBB0_2
|
|
; X64-NEXT: # %bb.4: # %L3
|
|
; X64-NEXT: movl $3, %eax
|
|
; X64-NEXT: retq
|
|
; X64-NEXT: .LBB0_3: # %L2
|
|
; X64-NEXT: movl $2, %eax
|
|
; X64-NEXT: retq
|
|
; X64-NEXT: .LBB0_2: # %L4
|
|
; X64-NEXT: movl $4, %eax
|
|
; X64-NEXT: retq
|
|
;
|
|
; X86-LABEL: atomic_and_flags_1:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-NEXT: cmpl %eax, %ecx
|
|
; X86-NEXT: jne .LBB0_3
|
|
; X86-NEXT: # %bb.1: # %L1
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-NEXT: incb (%edx)
|
|
; X86-NEXT: cmpl %eax, %ecx
|
|
; X86-NEXT: jne .LBB0_2
|
|
; X86-NEXT: # %bb.4: # %L3
|
|
; X86-NEXT: movl $3, %eax
|
|
; X86-NEXT: retl
|
|
; X86-NEXT: .LBB0_3: # %L2
|
|
; X86-NEXT: movl $2, %eax
|
|
; X86-NEXT: retl
|
|
; X86-NEXT: .LBB0_2: # %L4
|
|
; X86-NEXT: movl $4, %eax
|
|
; X86-NEXT: retl
|
|
; Generate flags value, and use it.
|
|
%cmp = icmp eq i32 %a, %b
|
|
br i1 %cmp, label %L1, label %L2
|
|
|
|
L1:
|
|
; The following pattern will get folded.
|
|
%1 = load atomic i8, ptr %p seq_cst, align 1
|
|
%2 = add i8 %1, 1 ; This forces the INC instruction to be generated.
|
|
store atomic i8 %2, ptr %p release, align 1
|
|
|
|
; Use the comparison result again. We need to rematerialize the comparison
|
|
; somehow. This test checks that cmpl gets emitted again, but any
|
|
; rematerialization would work (the optimizer used to clobber the flags with
|
|
; the add).
|
|
br i1 %cmp, label %L3, label %L4
|
|
|
|
L2:
|
|
ret i32 2
|
|
|
|
L3:
|
|
ret i32 3
|
|
|
|
L4:
|
|
ret i32 4
|
|
}
|
|
|
|
; Same as above, but using 2 as immediate to avoid the INC instruction.
|
|
define i32 @atomic_and_flags_2(ptr %p, i32 %a, i32 %b) {
|
|
; X64-LABEL: atomic_and_flags_2:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: cmpl %edx, %esi
|
|
; X64-NEXT: jne .LBB1_3
|
|
; X64-NEXT: # %bb.1: # %L1
|
|
; X64-NEXT: addb $2, (%rdi)
|
|
; X64-NEXT: cmpl %edx, %esi
|
|
; X64-NEXT: jne .LBB1_2
|
|
; X64-NEXT: # %bb.4: # %L3
|
|
; X64-NEXT: movl $3, %eax
|
|
; X64-NEXT: retq
|
|
; X64-NEXT: .LBB1_3: # %L2
|
|
; X64-NEXT: movl $2, %eax
|
|
; X64-NEXT: retq
|
|
; X64-NEXT: .LBB1_2: # %L4
|
|
; X64-NEXT: movl $4, %eax
|
|
; X64-NEXT: retq
|
|
;
|
|
; X86-LABEL: atomic_and_flags_2:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-NEXT: cmpl %eax, %ecx
|
|
; X86-NEXT: jne .LBB1_3
|
|
; X86-NEXT: # %bb.1: # %L1
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-NEXT: addb $2, (%edx)
|
|
; X86-NEXT: cmpl %eax, %ecx
|
|
; X86-NEXT: jne .LBB1_2
|
|
; X86-NEXT: # %bb.4: # %L3
|
|
; X86-NEXT: movl $3, %eax
|
|
; X86-NEXT: retl
|
|
; X86-NEXT: .LBB1_3: # %L2
|
|
; X86-NEXT: movl $2, %eax
|
|
; X86-NEXT: retl
|
|
; X86-NEXT: .LBB1_2: # %L4
|
|
; X86-NEXT: movl $4, %eax
|
|
; X86-NEXT: retl
|
|
%cmp = icmp eq i32 %a, %b
|
|
br i1 %cmp, label %L1, label %L2
|
|
L1:
|
|
%1 = load atomic i8, ptr %p seq_cst, align 1
|
|
%2 = add i8 %1, 2
|
|
store atomic i8 %2, ptr %p release, align 1
|
|
br i1 %cmp, label %L3, label %L4
|
|
L2:
|
|
ret i32 2
|
|
L3:
|
|
ret i32 3
|
|
L4:
|
|
ret i32 4
|
|
}
|
|
|
|
; PR20841 - ensure we don't reuse the ZF flag from XADD for compares with zero,
|
|
; the flags are set for the result of the add result (the value stored to memory),
|
|
; not the value returned by the atomicrmw add.
|
|
|
|
define zeroext i1 @xadd_cmp0_i64(ptr %x) nounwind {
|
|
; X64-LABEL: xadd_cmp0_i64:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movl $1, %eax
|
|
; X64-NEXT: lock xaddq %rax, (%rdi)
|
|
; X64-NEXT: testq %rax, %rax
|
|
; X64-NEXT: sete %al
|
|
; X64-NEXT: retq
|
|
;
|
|
; X86-LABEL: xadd_cmp0_i64:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: pushl %ebx
|
|
; X86-NEXT: pushl %esi
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %esi
|
|
; X86-NEXT: movl (%esi), %eax
|
|
; X86-NEXT: movl 4(%esi), %edx
|
|
; X86-NEXT: .p2align 4
|
|
; X86-NEXT: .LBB2_1: # %atomicrmw.start
|
|
; X86-NEXT: # =>This Inner Loop Header: Depth=1
|
|
; X86-NEXT: movl %eax, %ebx
|
|
; X86-NEXT: addl $1, %ebx
|
|
; X86-NEXT: movl %edx, %ecx
|
|
; X86-NEXT: adcl $0, %ecx
|
|
; X86-NEXT: lock cmpxchg8b (%esi)
|
|
; X86-NEXT: jne .LBB2_1
|
|
; X86-NEXT: # %bb.2: # %atomicrmw.end
|
|
; X86-NEXT: orl %edx, %eax
|
|
; X86-NEXT: sete %al
|
|
; X86-NEXT: popl %esi
|
|
; X86-NEXT: popl %ebx
|
|
; X86-NEXT: retl
|
|
%add = atomicrmw add ptr %x, i64 1 seq_cst
|
|
%cmp = icmp eq i64 %add, 0
|
|
ret i1 %cmp
|
|
}
|
|
|
|
define zeroext i1 @xadd_cmp0_i32(ptr %x) nounwind {
|
|
; X64-LABEL: xadd_cmp0_i32:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movl $1, %eax
|
|
; X64-NEXT: lock xaddl %eax, (%rdi)
|
|
; X64-NEXT: testl %eax, %eax
|
|
; X64-NEXT: setne %al
|
|
; X64-NEXT: retq
|
|
;
|
|
; X86-LABEL: xadd_cmp0_i32:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movl $1, %ecx
|
|
; X86-NEXT: lock xaddl %ecx, (%eax)
|
|
; X86-NEXT: testl %ecx, %ecx
|
|
; X86-NEXT: setne %al
|
|
; X86-NEXT: retl
|
|
%add = atomicrmw add ptr %x, i32 1 seq_cst
|
|
%cmp = icmp ne i32 %add, 0
|
|
ret i1 %cmp
|
|
}
|
|
|
|
define zeroext i1 @xadd_cmp0_i16(ptr %x) nounwind {
|
|
; X64-LABEL: xadd_cmp0_i16:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movw $1, %ax
|
|
; X64-NEXT: lock xaddw %ax, (%rdi)
|
|
; X64-NEXT: testw %ax, %ax
|
|
; X64-NEXT: sete %al
|
|
; X64-NEXT: retq
|
|
;
|
|
; X86-LABEL: xadd_cmp0_i16:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movw $1, %cx
|
|
; X86-NEXT: lock xaddw %cx, (%eax)
|
|
; X86-NEXT: testw %cx, %cx
|
|
; X86-NEXT: sete %al
|
|
; X86-NEXT: retl
|
|
%add = atomicrmw add ptr %x, i16 1 seq_cst
|
|
%cmp = icmp eq i16 %add, 0
|
|
ret i1 %cmp
|
|
}
|
|
|
|
define zeroext i1 @xadd_cmp0_i8(ptr %x) nounwind {
|
|
; X64-LABEL: xadd_cmp0_i8:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movb $1, %al
|
|
; X64-NEXT: lock xaddb %al, (%rdi)
|
|
; X64-NEXT: testb %al, %al
|
|
; X64-NEXT: setne %al
|
|
; X64-NEXT: retq
|
|
;
|
|
; X86-LABEL: xadd_cmp0_i8:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movb $1, %cl
|
|
; X86-NEXT: lock xaddb %cl, (%eax)
|
|
; X86-NEXT: testb %cl, %cl
|
|
; X86-NEXT: setne %al
|
|
; X86-NEXT: retl
|
|
%add = atomicrmw add ptr %x, i8 1 seq_cst
|
|
%cmp = icmp ne i8 %add, 0
|
|
ret i1 %cmp
|
|
}
|