Skip to content
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

InstCombine: Increase threadlocal.address alignment if pointee is more aligned #88435

Merged
merged 2 commits into from
Apr 16, 2024
Merged

InstCombine: Increase threadlocal.address alignment if pointee is more aligned #88435

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
b871d1c
ValueTracking: Skip threadlocal.address in computeKnownBits
MatzeB Apr 11, 2024
821b5ed
Add aligment attribute in InstCombine instead
MatzeB Apr 12, 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
9 changes: 9 additions & 0 deletions llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3407,6 +3407,15 @@ Instruction *InstCombinerImpl::visitCallInst(CallInst &CI) {
return I;
break;
}
case Intrinsic::threadlocal_address: {
Align MinAlign = getKnownAlignment(II->getArgOperand(0), DL, II, &AC, &DT);
MaybeAlign Align = II->getRetAlign();
if (MinAlign > Align.valueOrOne()) {
II->addRetAttr(Attribute::getWithAlignment(II->getContext(), MinAlign));
return II;
}
break;
}
default: {
// Handle target specific intrinsics
std::optional<Instruction *> V = targetInstCombineIntrinsic(*II);
Expand Down
41 changes: 41 additions & 0 deletions llvm/test/Transforms/InstCombine/threadlocal_address.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,41 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 4
; RUN: opt -o - -S %s -passes=instcombine | FileCheck %s

@tlsvar_a4 = thread_local global i32 4, align 4

define void @func_increase_alignment() {
; CHECK-LABEL: define void @func_increase_alignment() {
; CHECK-NEXT: [[P:%.*]] = call align 4 ptr @llvm.threadlocal.address.p0(ptr @tlsvar_a4)
; CHECK-NEXT: store i32 42, ptr [[P]], align 2
; CHECK-NEXT: ret void
;
%p = call align 2 ptr @llvm.threadlocal.address(ptr @tlsvar_a4)
store i32 42, ptr %p, align 2
ret void
}

@tlsvar_a32 = thread_local global i32 5, align 32

define i1 @func_add_alignment() {
; CHECK-LABEL: define i1 @func_add_alignment() {
; CHECK-NEXT: ret i1 true
;
%p = call ptr @llvm.threadlocal.address(ptr @tlsvar_a32)
%p_int = ptrtoint ptr %p to i32
%lowbits = and i32 %p_int, 31
%zero = icmp eq i32 %lowbits, 0
ret i1 %zero
}

@tlsvar_a1 = thread_local global i8 6, align 1

define i1 @func_dont_reduce_alignment() {
; CHECK-LABEL: define i1 @func_dont_reduce_alignment() {
; CHECK-NEXT: ret i1 true
;
%p = call align 4 ptr @llvm.threadlocal.address(ptr @tlsvar_a1)
%p_int = ptrtoint ptr %p to i32
%lowbits = and i32 %p_int, 3
%zero = icmp eq i32 %lowbits, 0
ret i1 %zero
}
Loading