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
201 lines
8.0 KiB
LLVM
201 lines
8.0 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 6
|
|
; RUN: opt -passes='loop-mssa(licm)' < %s -S | FileCheck %s
|
|
|
|
;; It should hoist fn_write_inaccessible_mem
|
|
;; because there is no conflict between inaccessible memory
|
|
;; fn_read_inaccessible_mem is a nice side effect
|
|
define i32 @loop_alias(i64 %x, ptr %start) {
|
|
; CHECK-LABEL: define i32 @loop_alias(
|
|
; CHECK-SAME: i64 [[X:%.*]], ptr [[START:%.*]]) {
|
|
; CHECK-NEXT: [[ENTRY:.*]]:
|
|
; CHECK-NEXT: call void @fn_write_inaccessible_mem()
|
|
; CHECK-NEXT: br label %[[LOOP:.*]]
|
|
; CHECK: [[LOOP]]:
|
|
; CHECK-NEXT: [[PHI:%.*]] = phi ptr [ [[START]], %[[ENTRY]] ], [ [[GEP:%.*]], %[[LOOP]] ]
|
|
; CHECK-NEXT: [[LOAD:%.*]] = load i32, ptr [[PHI]], align 4
|
|
; CHECK-NEXT: [[VAL:%.*]] = call i32 @fn_args(i32 [[LOAD]])
|
|
; CHECK-NEXT: call void @fn_read_inaccessible_mem(i32 [[LOAD]])
|
|
; CHECK-NEXT: [[GEP]] = getelementptr inbounds nuw i32, ptr [[PHI]], i64 [[X]]
|
|
; CHECK-NEXT: [[ACC:%.*]] = add nuw nsw i32 [[VAL]], 1
|
|
; CHECK-NEXT: [[CMP:%.*]] = icmp ult i32 [[ACC]], 10
|
|
; CHECK-NEXT: br i1 [[CMP]], label %[[LOOP]], label %[[AFTER_LOOP:.*]]
|
|
; CHECK: [[AFTER_LOOP]]:
|
|
; CHECK-NEXT: [[ACC_LCSSA:%.*]] = phi i32 [ [[ACC]], %[[LOOP]] ]
|
|
; CHECK-NEXT: ret i32 [[ACC_LCSSA]]
|
|
;
|
|
entry:
|
|
br label %loop
|
|
loop:
|
|
%phi = phi ptr [ %start, %entry ], [ %gep, %loop ]
|
|
%load = load i32, ptr %phi
|
|
%val = call i32 @fn_args(i32 %load)
|
|
call void @fn_write_inaccessible_mem()
|
|
call void @fn_read_inaccessible_mem(i32 %load)
|
|
%gep = getelementptr inbounds nuw i32, ptr %phi, i64 %x
|
|
%acc = add nuw nsw i32 %val, 1
|
|
%cmp = icmp ult i32 %acc, 10
|
|
br i1 %cmp, label %loop, label %after_loop
|
|
after_loop:
|
|
ret i32 %acc
|
|
}
|
|
|
|
;; It should NOT hoist fn_write_inaccessible_mem
|
|
;; Because fn_read_inaccessible_mem_2 reads the same inaccessible memory
|
|
define i32 @ne_loop_alias(i64 %x, ptr %start) {
|
|
; CHECK-LABEL: define i32 @ne_loop_alias(
|
|
; CHECK-SAME: i64 [[X:%.*]], ptr [[START:%.*]]) {
|
|
; CHECK-NEXT: [[ENTRY:.*]]:
|
|
; CHECK-NEXT: br label %[[LOOP:.*]]
|
|
; CHECK: [[LOOP]]:
|
|
; CHECK-NEXT: [[PHI:%.*]] = phi ptr [ [[START]], %[[ENTRY]] ], [ [[GEP:%.*]], %[[LOOP]] ]
|
|
; CHECK-NEXT: [[LOAD:%.*]] = load i32, ptr [[PHI]], align 4
|
|
; CHECK-NEXT: [[VAL:%.*]] = call i32 @fn_read_inaccessible_mem_2(i32 [[LOAD]])
|
|
; CHECK-NEXT: call void @fn_write_inaccessible_mem()
|
|
; CHECK-NEXT: call void @fn_read_inaccessible_mem(i32 [[VAL]])
|
|
; CHECK-NEXT: [[GEP]] = getelementptr inbounds nuw i32, ptr [[PHI]], i64 [[X]]
|
|
; CHECK-NEXT: [[ACC:%.*]] = add nuw nsw i32 [[VAL]], 1
|
|
; CHECK-NEXT: [[CMP:%.*]] = icmp ult i32 [[ACC]], 10
|
|
; CHECK-NEXT: br i1 [[CMP]], label %[[LOOP]], label %[[AFTER_LOOP:.*]]
|
|
; CHECK: [[AFTER_LOOP]]:
|
|
; CHECK-NEXT: [[ACC_LCSSA:%.*]] = phi i32 [ [[ACC]], %[[LOOP]] ]
|
|
; CHECK-NEXT: ret i32 [[ACC_LCSSA]]
|
|
;
|
|
entry:
|
|
br label %loop
|
|
loop:
|
|
%phi = phi ptr [ %start, %entry ], [ %gep, %loop ]
|
|
%load = load i32, ptr %phi
|
|
%val = call i32 @fn_read_inaccessible_mem_2(i32 %load)
|
|
call void @fn_write_inaccessible_mem()
|
|
call void @fn_read_inaccessible_mem(i32 %val)
|
|
%gep = getelementptr inbounds nuw i32, ptr %phi, i64 %x
|
|
%acc = add nuw nsw i32 %val, 1
|
|
%cmp = icmp ult i32 %acc, 10
|
|
br i1 %cmp, label %loop, label %after_loop
|
|
after_loop:
|
|
ret i32 %acc
|
|
}
|
|
|
|
;; Should hoist fn_write_inaccessible_mem
|
|
;; It does not alias with store
|
|
define void @hoist_untill_I(ptr noalias %loc, ptr noalias %loc2){
|
|
; CHECK-LABEL: define void @hoist_untill_I(
|
|
; CHECK-SAME: ptr noalias [[LOC:%.*]], ptr noalias [[LOC2:%.*]]) {
|
|
; CHECK-NEXT: [[ENTRY:.*:]]
|
|
; CHECK-NEXT: [[VAL:%.*]] = load i32, ptr [[LOC2]], align 4
|
|
; CHECK-NEXT: store i32 [[VAL]], ptr [[LOC]], align 4
|
|
; CHECK-NEXT: call void @fn_write_inaccessible_mem()
|
|
; CHECK-NEXT: call void @fn_read_inaccessible_mem(i32 [[VAL]])
|
|
; CHECK-NEXT: br label %[[FOR_BODY:.*]]
|
|
; CHECK: [[FOR_COND_CLEANUP:.*:]]
|
|
; CHECK-NEXT: ret void
|
|
; CHECK: [[FOR_BODY]]:
|
|
; CHECK-NEXT: br label %[[FOR_BODY]]
|
|
;
|
|
entry:
|
|
br label %for.body
|
|
for.cond.cleanup: ; preds = %for.body
|
|
ret void
|
|
for.body:
|
|
%val = load i32, ptr %loc2
|
|
store i32 %val, ptr %loc
|
|
call void @fn_write_inaccessible_mem()
|
|
call void @fn_read_inaccessible_mem(i32 %val)
|
|
br label %for.body
|
|
}
|
|
|
|
;; Should NOT hoist fn_write_inaccessible_mem
|
|
;; because fn_read_inaccessible_mem reads the same memory location
|
|
define void @neg_hoist_untill_I(ptr noalias %loc, ptr noalias %loc2){
|
|
; CHECK-LABEL: define void @neg_hoist_untill_I(
|
|
; CHECK-SAME: ptr noalias [[LOC:%.*]], ptr noalias [[LOC2:%.*]]) {
|
|
; CHECK-NEXT: [[ENTRY:.*:]]
|
|
; CHECK-NEXT: [[VAL:%.*]] = load i32, ptr [[LOC2]], align 4
|
|
; CHECK-NEXT: store i32 [[VAL]], ptr [[LOC]], align 4
|
|
; CHECK-NEXT: br label %[[FOR_BODY:.*]]
|
|
; CHECK: [[FOR_BODY]]:
|
|
; CHECK-NEXT: call void @fn_read_inaccessible_mem(i32 0)
|
|
; CHECK-NEXT: call void @fn_write_inaccessible_mem()
|
|
; CHECK-NEXT: call void @fn_read_inaccessible_mem(i32 [[VAL]])
|
|
; CHECK-NEXT: br label %[[FOR_BODY]]
|
|
;
|
|
entry:
|
|
br label %for.body
|
|
for.body:
|
|
%val = load i32, ptr %loc2
|
|
store i32 %val, ptr %loc
|
|
call void @fn_read_inaccessible_mem(i32 0)
|
|
call void @fn_write_inaccessible_mem()
|
|
call void @fn_read_inaccessible_mem(i32 %val)
|
|
br label %for.body
|
|
}
|
|
|
|
|
|
;; Nothing should be hoisted from the loop because volatile
|
|
;; sets inaccessible memory to read write
|
|
define void @neg_volatile(ptr %loc, ptr %loc2) {
|
|
; CHECK-LABEL: define void @neg_volatile(
|
|
; CHECK-SAME: ptr [[LOC:%.*]], ptr [[LOC2:%.*]]) {
|
|
; CHECK-NEXT: [[ENTRY:.*:]]
|
|
; CHECK-NEXT: br label %[[LOOP:.*]]
|
|
; CHECK: [[LOOP]]:
|
|
; CHECK-NEXT: [[VAL:%.*]] = load i32, ptr [[LOC2]], align 4
|
|
; CHECK-NEXT: store volatile i32 0, ptr [[LOC]], align 4
|
|
; CHECK-NEXT: call void @fn_write_inaccessible_mem()
|
|
; CHECK-NEXT: call void @fn_read_inaccessible_mem(i32 [[VAL]])
|
|
; CHECK-NEXT: br label %[[LOOP]]
|
|
;
|
|
entry:
|
|
br label %loop
|
|
|
|
loop:
|
|
%val = load i32, ptr %loc2
|
|
store volatile i32 0, ptr %loc
|
|
call void @fn_write_inaccessible_mem()
|
|
call void @fn_read_inaccessible_mem(i32 %val)
|
|
br label %loop
|
|
}
|
|
|
|
define i32 @non_dominated_load_that_does_not_alias_inaccessible_mem(ptr %dst, i64 %x, ptr %start) {
|
|
; CHECK-LABEL: define i32 @non_dominated_load_that_does_not_alias_inaccessible_mem(
|
|
; CHECK-SAME: ptr [[DST:%.*]], i64 [[X:%.*]], ptr [[START:%.*]]) {
|
|
; CHECK-NEXT: [[ENTRY:.*]]:
|
|
; CHECK-NEXT: call void @fn_write_inaccessible_mem()
|
|
; CHECK-NEXT: br label %[[LOOP:.*]]
|
|
; CHECK: [[LOOP]]:
|
|
; CHECK-NEXT: [[PHI:%.*]] = phi ptr [ [[START]], %[[ENTRY]] ], [ [[GEP:%.*]], %[[LOOP]] ]
|
|
; CHECK-NEXT: [[LOAD:%.*]] = load i32, ptr [[PHI]], align 4
|
|
; CHECK-NEXT: [[VAL:%.*]] = call i32 @fn_args(i32 [[LOAD]])
|
|
; CHECK-NEXT: [[RES:%.*]] = call i32 @fn_read_inaccessible_mem_2(i32 [[LOAD]])
|
|
; CHECK-NEXT: store i32 [[RES]], ptr [[DST]], align 4
|
|
; CHECK-NEXT: [[GEP]] = getelementptr inbounds nuw i32, ptr [[PHI]], i64 [[X]]
|
|
; CHECK-NEXT: [[ACC:%.*]] = add nuw nsw i32 [[VAL]], 1
|
|
; CHECK-NEXT: [[CMP:%.*]] = icmp ult i32 [[ACC]], 10
|
|
; CHECK-NEXT: br i1 [[CMP]], label %[[LOOP]], label %[[AFTER_LOOP:.*]]
|
|
; CHECK: [[AFTER_LOOP]]:
|
|
; CHECK-NEXT: [[ACC_LCSSA:%.*]] = phi i32 [ [[ACC]], %[[LOOP]] ]
|
|
; CHECK-NEXT: ret i32 [[ACC_LCSSA]]
|
|
;
|
|
entry:
|
|
br label %loop
|
|
loop:
|
|
%phi = phi ptr [ %start, %entry ], [ %gep, %loop ]
|
|
%load = load i32, ptr %phi
|
|
%val = call i32 @fn_args(i32 %load)
|
|
call void @fn_write_inaccessible_mem()
|
|
%res = call i32 @fn_read_inaccessible_mem_2(i32 %load)
|
|
store i32 %res, ptr %dst
|
|
%gep = getelementptr inbounds nuw i32, ptr %phi, i64 %x
|
|
%acc = add nuw nsw i32 %val, 1
|
|
%cmp = icmp ult i32 %acc, 10
|
|
br i1 %cmp, label %loop, label %after_loop
|
|
after_loop:
|
|
ret i32 %acc
|
|
}
|
|
|
|
declare i32 @fn_args(i32) nounwind willreturn memory(argmem: read)
|
|
declare i32 @fn_read_inaccessible_mem_2(i32) nounwind willreturn memory(inaccessiblemem: read)
|
|
declare void @fn_write_inaccessible_mem() nounwind memory(inaccessiblemem: write)
|
|
declare void @fn_read_inaccessible_mem(i32) nounwind memory(inaccessiblemem: read)
|
|
declare void @fn_readwrite_inaccessible_mem() nounwind memory(inaccessiblemem: readwrite)
|