Skip to content

[InstCombine] Preserve the nsw/nuw flags for (X | Op01C) + Op1C --> X + (Op01C + Op1C) #94586

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 8 commits into from
Jun 8, 2024
Merged

[InstCombine] Preserve the nsw/nuw flags for (X | Op01C) + Op1C --> X + (Op01C + Op1C) #94586

Show file tree
Hide file tree
Changes from 4 commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
170cfa2
[InstCombine] Preserve the nsw flag for (X | Op01C) + Op1C --> X + (O…
csstormq Jun 6, 2024
8b6eca1
update test case using the update_test_checks.py script
csstormq Jun 6, 2024
3cf7a70
fixup! Preserve the nsw flag iff the sum of Op01C and Op1C will not o…
csstormq Jun 7, 2024
40037ff
Add more test cases
csstormq Jun 7, 2024
d2f4bee
Resolve conversations
csstormq Jun 7, 2024
8997ab0
Remove added new lines
csstormq Jun 7, 2024
045f3ba
Resolve conversations
csstormq Jun 7, 2024
1050bde
Fix test to not always return poison
csstormq Jun 7, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 11 additions & 2 deletions llvm/lib/Transforms/InstCombine/InstCombineAddSub.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -905,8 +905,17 @@ Instruction *InstCombinerImpl::foldAddWithConstant(BinaryOperator &Add) {

// (X | Op01C) + Op1C --> X + (Op01C + Op1C) iff the `or` is actually an `add`
Constant *Op01C;
if (match(Op0, m_DisjointOr(m_Value(X), m_ImmConstant(Op01C))))
return BinaryOperator::CreateAdd(X, ConstantExpr::getAdd(Op01C, Op1C));
if (match(Op0, m_DisjointOr(m_Value(X), m_ImmConstant(Op01C)))) {
BinaryOperator *NewAdd =
BinaryOperator::CreateAdd(X, ConstantExpr::getAdd(Op01C, Op1C));
// Preserve the nsw flag iff the sum of Op01C and Op1C will not overflow
// so that there is a chance to make some other transformations.
// For some cases, sdiv can be converted to udiv when the newly created add
// carrying the nsw flag is one of its operands.
if (willNotOverflowSignedAdd(Op01C, Op1C, Add))
NewAdd->setHasNoSignedWrap(Add.hasNoSignedWrap());
return NewAdd;
}

// (X | C2) + C --> (X | C2) ^ C2 iff (C2 == -C)
const APInt *C2;
Expand Down
20 changes: 20 additions & 0 deletions llvm/test/Transforms/InstCombine/add.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4137,5 +4137,25 @@ define i32 @fold_zext_nneg_add_const_fail2(i8 %x) {
ret i32 %r
}

define i8 @fold_add_constant_preserve_nsw(i8 %x) {
; CHECK-LABEL: @fold_add_constant_preserve_nsw(
; CHECK-NEXT: [[ADD:%.*]] = add nsw i8 [[X:%.*]], -120
; CHECK-NEXT: ret i8 [[ADD]]
;
%or = or disjoint i8 %x, -128
%add = add nsw i8 %or, 8
ret i8 %add
}

define i8 @fold_add_constant_no_nsw(i8 %x) {
; CHECK-LABEL: @fold_add_constant_no_nsw(
; CHECK-NEXT: [[ADD:%.*]] = add i8 [[X:%.*]], 120
; CHECK-NEXT: ret i8 [[ADD]]
;
%or = or disjoint i8 %x, -128
%add = add nsw i8 %or, -8
ret i8 %add
}

declare void @llvm.assume(i1)
declare void @fake_func(i32)
2 changes: 1 addition & 1 deletion llvm/test/Transforms/InstCombine/sadd-with-overflow.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -125,7 +125,7 @@ define { i32, i1 } @fold_sub_simple(i32 %x) {

define { i32, i1 } @fold_with_distjoin_or(i32 %x) {
; CHECK-LABEL: @fold_with_distjoin_or(
; CHECK-NEXT: [[B:%.*]] = add i32 [[X:%.*]], 6
; CHECK-NEXT: [[B:%.*]] = add nsw i32 [[X:%.*]], 6
; CHECK-NEXT: [[TMP1:%.*]] = insertvalue { i32, i1 } { i32 poison, i1 false }, i32 [[B]], 0
; CHECK-NEXT: ret { i32, i1 } [[TMP1]]
;
Expand Down
17 changes: 17 additions & 0 deletions llvm/test/Transforms/InstCombine/sdiv-simplify.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,17 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
; RUN: opt < %s -passes=instcombine -S | FileCheck %s

define i32 @sdiv_to_udiv(i32 %arg0, i32 %arg1) {
; CHECK-LABEL: define i32 @sdiv_to_udiv(
; CHECK-SAME: i32 [[ARG0:%.*]], i32 [[ARG1:%.*]]) {
; CHECK-NEXT: [[T0:%.*]] = shl nuw nsw i32 [[ARG0]], 8
; CHECK-NEXT: [[T2:%.*]] = add nuw nsw i32 [[T0]], 6242049
; CHECK-NEXT: [[T3:%.*]] = udiv i32 [[T2]], 192
; CHECK-NEXT: ret i32 [[T3]]
;
%t0 = shl nuw nsw i32 %arg0, 8
%t1 = or disjoint i32 %t0, 1
%t2 = add nuw nsw i32 %t1, 6242048
%t3 = sdiv i32 %t2, 192
ret i32 %t3
}
18 changes: 9 additions & 9 deletions llvm/test/Transforms/PhaseOrdering/AArch64/matrix-extract-insert.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -182,11 +182,11 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK: vector.body.1:
; CHECK-NEXT: [[INDEX_1:%.*]] = phi i64 [ 0, [[VECTOR_PH_1]] ], [ [[INDEX_NEXT_1:%.*]], [[VECTOR_BODY_1]] ]
; CHECK-NEXT: [[TMP33:%.*]] = add nuw nsw i64 [[INDEX_1]], 15
; CHECK-NEXT: [[TMP34:%.*]] = add i64 [[INDEX_1]], 16
; CHECK-NEXT: [[TMP34:%.*]] = add nsw i64 [[INDEX_1]], 16
; CHECK-NEXT: [[TMP35:%.*]] = insertelement <2 x i64> poison, i64 [[TMP33]], i64 0
; CHECK-NEXT: [[TMP36:%.*]] = insertelement <2 x i64> [[TMP35]], i64 [[TMP34]], i64 1
; CHECK-NEXT: [[TMP37:%.*]] = add i64 [[INDEX_1]], 17
; CHECK-NEXT: [[TMP38:%.*]] = add i64 [[INDEX_1]], 18
; CHECK-NEXT: [[TMP37:%.*]] = add nsw i64 [[INDEX_1]], 17
; CHECK-NEXT: [[TMP38:%.*]] = add nsw i64 [[INDEX_1]], 18
; CHECK-NEXT: [[TMP39:%.*]] = insertelement <2 x i64> poison, i64 [[TMP37]], i64 0
; CHECK-NEXT: [[TMP40:%.*]] = insertelement <2 x i64> [[TMP39]], i64 [[TMP38]], i64 1
; CHECK-NEXT: [[TMP41:%.*]] = icmp ult <2 x i64> [[TMP36]], <i64 225, i64 225>
Expand Down Expand Up @@ -259,11 +259,11 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK: vector.body.2:
; CHECK-NEXT: [[INDEX_2:%.*]] = phi i64 [ 0, [[VECTOR_PH_2]] ], [ [[INDEX_NEXT_2:%.*]], [[VECTOR_BODY_2]] ]
; CHECK-NEXT: [[TMP64:%.*]] = add nuw nsw i64 [[INDEX_2]], 30
; CHECK-NEXT: [[TMP65:%.*]] = add i64 [[INDEX_2]], 31
; CHECK-NEXT: [[TMP65:%.*]] = add nsw i64 [[INDEX_2]], 31
; CHECK-NEXT: [[TMP66:%.*]] = insertelement <2 x i64> poison, i64 [[TMP64]], i64 0
; CHECK-NEXT: [[TMP67:%.*]] = insertelement <2 x i64> [[TMP66]], i64 [[TMP65]], i64 1
; CHECK-NEXT: [[TMP68:%.*]] = add i64 [[INDEX_2]], 32
; CHECK-NEXT: [[TMP69:%.*]] = add i64 [[INDEX_2]], 33
; CHECK-NEXT: [[TMP68:%.*]] = add nsw i64 [[INDEX_2]], 32
; CHECK-NEXT: [[TMP69:%.*]] = add nsw i64 [[INDEX_2]], 33
; CHECK-NEXT: [[TMP70:%.*]] = insertelement <2 x i64> poison, i64 [[TMP68]], i64 0
; CHECK-NEXT: [[TMP71:%.*]] = insertelement <2 x i64> [[TMP70]], i64 [[TMP69]], i64 1
; CHECK-NEXT: [[TMP72:%.*]] = icmp ult <2 x i64> [[TMP67]], <i64 225, i64 225>
Expand Down Expand Up @@ -336,11 +336,11 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK: vector.body.3:
; CHECK-NEXT: [[INDEX_3:%.*]] = phi i64 [ 0, [[VECTOR_PH_3]] ], [ [[INDEX_NEXT_3:%.*]], [[VECTOR_BODY_3]] ]
; CHECK-NEXT: [[TMP95:%.*]] = add nuw nsw i64 [[INDEX_3]], 45
; CHECK-NEXT: [[TMP96:%.*]] = add i64 [[INDEX_3]], 46
; CHECK-NEXT: [[TMP96:%.*]] = add nsw i64 [[INDEX_3]], 46
; CHECK-NEXT: [[TMP97:%.*]] = insertelement <2 x i64> poison, i64 [[TMP95]], i64 0
; CHECK-NEXT: [[TMP98:%.*]] = insertelement <2 x i64> [[TMP97]], i64 [[TMP96]], i64 1
; CHECK-NEXT: [[TMP99:%.*]] = add i64 [[INDEX_3]], 47
; CHECK-NEXT: [[TMP100:%.*]] = add i64 [[INDEX_3]], 48
; CHECK-NEXT: [[TMP99:%.*]] = add nsw i64 [[INDEX_3]], 47
; CHECK-NEXT: [[TMP100:%.*]] = add nsw i64 [[INDEX_3]], 48
; CHECK-NEXT: [[TMP101:%.*]] = insertelement <2 x i64> poison, i64 [[TMP99]], i64 0
; CHECK-NEXT: [[TMP102:%.*]] = insertelement <2 x i64> [[TMP101]], i64 [[TMP100]], i64 1
; CHECK-NEXT: [[TMP103:%.*]] = icmp ult <2 x i64> [[TMP98]], <i64 225, i64 225>
Expand Down
Loading