Skip to content

[GlobalOpt] Fix global SRA incorrect alignment on some elements #115328

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 4 commits into from
Nov 18, 2024
Merged

[GlobalOpt] Fix global SRA incorrect alignment on some elements #115328

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
2d1cd65
[GlobalOpt] Fix global SRA incorrect use of original alignment on som…
brunodf-snps Nov 7, 2024
f5151ec
Checking for ABI type align is unnecessary
brunodf-snps Nov 7, 2024
0a59028
[GlobalOpt] Add test case for issue #115282
brunodf-snps Nov 7, 2024
f281c80
Update test case after fix
brunodf-snps Nov 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
7 changes: 4 additions & 3 deletions llvm/lib/Transforms/IPO/GlobalOpt.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -575,15 +575,16 @@ static GlobalVariable *SRAGlobal(GlobalVariable *GV, const DataLayout &DL) {
*GV->getParent(), Ty, false, GlobalVariable::InternalLinkage,
Initializer, GV->getName() + "." + Twine(NameSuffix++), GV,
GV->getThreadLocalMode(), GV->getAddressSpace());
// Start out by copying attributes from the original, including alignment.
NGV->copyAttributesFrom(GV);
NewGlobals.insert({OffsetForTy, NGV});

// Calculate the known alignment of the field. If the original aggregate
// had 256 byte alignment for example, something might depend on that:
// had 256 byte alignment for example, then the element at a given offset
// may also have a known alignment, and something might depend on that:
// propagate info to each field.
Align NewAlign = commonAlignment(StartAlignment, OffsetForTy);
if (NewAlign > DL.getABITypeAlign(Ty))
NGV->setAlignment(NewAlign);
NGV->setAlignment(NewAlign);

// Copy over the debug info for the variable.
transferSRADebugInfo(GV, NGV, OffsetForTy * 8,
Expand Down
2 changes: 1 addition & 1 deletion llvm/test/DebugInfo/X86/global-sra-struct-fit-segment.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,7 @@
%struct.BSS1 = type <{ [12 x i8] }>

;CHECK: @.BSS1.0 = internal unnamed_addr global i32 0, align 32, !dbg ![[GVE1:.*]]
;CHECK: @.BSS1.1 = internal unnamed_addr global i32 0, align 32, !dbg ![[GVE2:.*]], !dbg ![[GVE4:.*]]
;CHECK: @.BSS1.1 = internal unnamed_addr global i32 0, align 4, !dbg ![[GVE2:.*]], !dbg ![[GVE4:.*]]
;CHECK: @.BSS1.2 = internal unnamed_addr global i32 0, align 8, !dbg ![[GVE3:.*]]

@.BSS1 = internal global %struct.BSS1 zeroinitializer, align 32, !dbg !0, !dbg !7, !dbg !10, !dbg !27, !dbg !29
Expand Down
2 changes: 1 addition & 1 deletion llvm/test/DebugInfo/X86/global-sra-struct-part-overlap-segment.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,7 +26,7 @@

@.BSS3 = internal unnamed_addr global %struct.BSS3 zeroinitializer, align 32, !dbg !0, !dbg !7, !dbg !29
;CHECK: @.BSS3.0 = internal unnamed_addr global double 0.000000e+00, align 32, !dbg ![[GVE1:.*]], !dbg ![[GVE2:.*]]
;CHECK: @.BSS3.1 = internal unnamed_addr global double 0.000000e+00, align 32, !dbg ![[GVE3:.*]], !dbg ![[GVE4:.*]], !dbg ![[GVE6:.*]]
;CHECK: @.BSS3.1 = internal unnamed_addr global double 0.000000e+00, align 8, !dbg ![[GVE3:.*]], !dbg ![[GVE4:.*]], !dbg ![[GVE6:.*]]
;CHECK: @.BSS3.2 = internal unnamed_addr global double 0.000000e+00, align 16, !dbg ![[GVE5:.*]]

@.C363_mymod_bar_ = internal constant [2 x i8] c"IF"
Expand Down
8 changes: 4 additions & 4 deletions llvm/test/Transforms/GlobalOpt/globalsra-align.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,9 +15,9 @@ target datalayout = "p:16:32:64" ; 16-bit pointers with 32-bit ABI alignment and

;.
; CHECK: @[[A_4:[a-zA-Z0-9_$"\\.-]+]] = internal unnamed_addr externally_initialized global ptr null, align 8
; CHECK: @[[A_5:[a-zA-Z0-9_$"\\.-]+]] = internal unnamed_addr externally_initialized global ptr null, align 16
; CHECK: @[[A_5:[a-zA-Z0-9_$"\\.-]+]] = internal unnamed_addr externally_initialized global ptr null, align 8
; CHECK: @[[A_6:[a-zA-Z0-9_$"\\.-]+]] = internal unnamed_addr externally_initialized global ptr null, align 16
; CHECK: @[[A_7:[a-zA-Z0-9_$"\\.-]+]] = internal unnamed_addr externally_initialized global ptr null, align 16
; CHECK: @[[A_7:[a-zA-Z0-9_$"\\.-]+]] = internal unnamed_addr externally_initialized global ptr null, align 8
;.
define ptr @reduce_align_0() {
; CHECK-LABEL: @reduce_align_0(
Expand All @@ -31,7 +31,7 @@ define ptr @reduce_align_0() {

define ptr @reduce_align_1() {
; CHECK-LABEL: @reduce_align_1(
; CHECK-NEXT: [[X:%.*]] = load ptr, ptr @a.5, align 16
; CHECK-NEXT: [[X:%.*]] = load ptr, ptr @a.5, align 8
; CHECK-NEXT: ret ptr [[X]]
;
%x = load ptr, ptr getelementptr inbounds ([3 x [7 x ptr]], ptr @a, i64 0, i64 2, i64 1), align 4
Expand All @@ -51,7 +51,7 @@ define ptr @reduce_align_2() {

define ptr @reduce_align_3() {
; CHECK-LABEL: @reduce_align_3(
; CHECK-NEXT: [[X:%.*]] = load ptr, ptr @a.7, align 16
; CHECK-NEXT: [[X:%.*]] = load ptr, ptr @a.7, align 8
; CHECK-NEXT: ret ptr [[X]]
;
%x = load ptr, ptr getelementptr inbounds ([3 x [7 x ptr]], ptr @a, i64 0, i64 2, i64 3), align 4
Expand Down
51 changes: 51 additions & 0 deletions llvm/test/Transforms/GlobalOpt/pr115282.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,51 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals all --version 5
; RUN: opt < %s -passes=globalopt -S | FileCheck %s

@params = internal global [4 x i32] zeroinitializer, align 32

;.
; CHECK: @params.0 = internal unnamed_addr global i32 0, align 32
; CHECK: @params.1 = internal unnamed_addr global i32 0, align 4
; CHECK: @params.2 = internal unnamed_addr global i32 0, align 8
; CHECK: @params.3 = internal unnamed_addr global i32 0, align 4
;.
define void @set(i32 %a, i32 %b, i32 %c, i32 %d) {
; CHECK-LABEL: define void @set(
; CHECK-SAME: i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]], i32 [[D:%.*]]) local_unnamed_addr {
; CHECK-NEXT: store i32 [[A]], ptr @params.0, align 32
; CHECK-NEXT: store i32 [[B]], ptr @params.1, align 4
; CHECK-NEXT: store i32 [[C]], ptr @params.2, align 8
; CHECK-NEXT: store i32 [[D]], ptr @params.3, align 4
; CHECK-NEXT: ret void
;
store i32 %a, ptr @params
store i32 %b, ptr getelementptr ([4 x i32], ptr @params, i32 0, i32 1)
store i32 %c, ptr getelementptr ([4 x i32], ptr @params, i32 0, i32 2)
store i32 %d, ptr getelementptr ([4 x i32], ptr @params, i32 0, i32 3)
ret void
}

%S = type { i32, i32, i32, i32 }

define %S @get() {
; CHECK-LABEL: define %S @get() local_unnamed_addr {
; CHECK-NEXT: [[A:%.*]] = load i32, ptr @params.0, align 32
; CHECK-NEXT: [[SA:%.*]] = insertvalue [[S:%.*]] undef, i32 [[A]], 0
; CHECK-NEXT: [[B:%.*]] = load i32, ptr @params.1, align 4
; CHECK-NEXT: [[SB:%.*]] = insertvalue [[S]] [[SA]], i32 [[B]], 1
; CHECK-NEXT: [[C:%.*]] = load i32, ptr @params.2, align 8
; CHECK-NEXT: [[SC:%.*]] = insertvalue [[S]] [[SB]], i32 [[C]], 2
; CHECK-NEXT: [[D:%.*]] = load i32, ptr @params.3, align 4
; CHECK-NEXT: [[SD:%.*]] = insertvalue [[S]] [[SC]], i32 [[D]], 3
; CHECK-NEXT: ret [[S]] [[SD]]
;
%a = load i32, ptr @params
%sa = insertvalue %S undef, i32 %a, 0
%b = load i32, ptr getelementptr ([4 x i32], ptr @params, i32 0, i32 1)
%sb = insertvalue %S %sa, i32 %b, 1
%c = load i32, ptr getelementptr ([4 x i32], ptr @params, i32 0, i32 2)
%sc = insertvalue %S %sb, i32 %c, 2
%d = load i32, ptr getelementptr ([4 x i32], ptr @params, i32 0, i32 3)
%sd = insertvalue %S %sc, i32 %d, 3
ret %S %sd
}
Loading