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

Jump to bottom

[FastISel][AArch64] Compare Instruction Miscompilation Fix #75993

Merged
merged 7 commits into from
Jan 3, 2024
Merged

[FastISel][AArch64] Compare Instruction Miscompilation Fix #75993

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
5657561
[FastISel][AArch64] compare instruction miscompilation fix
brendaso1 Dec 19, 2023
cdc10d8
Simplify test/CodeGen/AArch64/fastisel-shift-and-cmp.ll
brendaso1 Dec 25, 2023
9652999
removed extraneous function in unit test
brendaso1 Dec 25, 2023
44d4231
clean up unit test
brendaso1 Dec 26, 2023
a495c27
Removed left shift and compare merging logic because that cannot be r…
brendaso1 Jan 1, 2024
4457315
moved unit test to arm64-fast-isel-icmp.ll
brendaso1 Jan 2, 2024
7017aec
removed unnecessary changes to selectBranch
brendaso1 Jan 2, 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
8 changes: 6 additions & 2 deletions llvm/lib/Target/AArch64/AArch64FastISel.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1233,7 +1233,11 @@ unsigned AArch64FastISel::emitAddSub(bool UseAdd, MVT RetVT, const Value *LHS,
isValueAvailable(RHS)) {
if (const auto *SI = dyn_cast<BinaryOperator>(RHS))
if (const auto *C = dyn_cast<ConstantInt>(SI->getOperand(1)))
if ((SI->getOpcode() == Instruction::Shl) && (C->getZExtValue() < 4)) {
if ((SI->getOpcode() == Instruction::Shl) && (C->getZExtValue() < 4) &&
!NeedExtend) {
// We can only fold instructions that doesn't need extension because
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is it possible to reach this point now, if NeedExtend==false and ExtendType != AArch64_AM::InvalidShiftExtend?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I didn't add ExtendType != AArch64_AM::InvalidShiftExtend because there is already a check before this nested-if statement in line 1232. Do you think it's still better to add the check?

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah - I meant that if ExtendType != InvalidShift then does that mean that NeedExtend is always false? Is it ever possible to get to the emitAddSub_rx(..) line now?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ah Gotcha. You're right, if ExtendType != AArch64_AM::InvalidShiftExtend, NeedExtend is always true. So the logic would never be reached. Maybe it's a better idea to remove that part of the logic all together?

// in add/sub instruction, the instruction is first extended, then
// shifted
Register RHSReg = getRegForValue(SI->getOperand(0));
if (!RHSReg)
return 0;
Expand Down Expand Up @@ -2426,7 +2430,7 @@ bool AArch64FastISel::selectBranch(const Instruction *I) {
}

// Emit the cmp.
if (!emitCmp(CI->getOperand(0), CI->getOperand(1), CI->isUnsigned()))
if (!emitCmp(CI->getOperand(0), CI->getOperand(1), !CI->isSigned()))
return false;

// FCMP_UEQ and FCMP_ONE cannot be checked with a single branch
Expand Down
33 changes: 33 additions & 0 deletions llvm/test/CodeGen/AArch64/fastisel-shift-and-cmp.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,33 @@
; RUN: llc --global-isel=false -fast-isel -O0 -mtriple=aarch64-none-none < %s | FileCheck %s

; Check that the shl instruction did not get folded in together with
; the cmp instruction. It would create a miscompilation

@A = dso_local global [5 x i8] c"\C8\AA\C8\AA\AA"
@.str = private unnamed_addr constant [13 x i8] c"TRUE BRANCH\0A\00"
@.str.1 = private unnamed_addr constant [14 x i8] c"FALSE BRANCH\0A\00"

define dso_local i32 @main() {
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is it possible to simplify the test? I think it could

  • remove dso_local
  • remove the globals
  • remove the loads
  • remove the calls
  • remove --global-isel=false and -O0?

There are some existing tests in llvm/test/CodeGen/AArch64/arm64-fast-isel-icmp.ll, but they don't seem to include any with shift amounts at the moment.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

hmm in that case do you think it's more beneficial if I put this test in llvm/test/CodeGen/AArch64/arm64-fast-isel-icmp.ll?

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah that sounds like it might be a good place for it. Feel free to use the update_llc_test_checks script too to generate the check lines.


%tmp = load i8, ptr getelementptr inbounds ([5 x i8], ptr @A, i64 0, i64 1)
%tmp2 = load i8, ptr getelementptr inbounds ([5 x i8], ptr @A, i64 0, i64 2)
%op1 = xor i8 %tmp, -49
%op2 = mul i8 %op1, %op1
; CHECK-NOT: cmp [[REGS:.*]] #[[SHIFT_VAL:[0-9]+]]
%op3 = shl i8 %op2, 3
%tmp3 = icmp eq i8 %tmp2, %op3
br i1 %tmp3, label %_true, label %_false

_true:
%res = call i32 (ptr, ...) @printf(ptr noundef @.str)
br label %_ret

_false:
%res2 = call i32 (ptr, ...) @printf(ptr noundef @.str.1)
br label %_ret

_ret:
ret i32 0
}

declare i32 @printf(ptr noundef, ...)