Skip to content

[Sanitizer] remove array-bounds-pseudofn #149430

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 1 commit into from
Jul 18, 2025
Merged

[Sanitizer] remove array-bounds-pseudofn #149430

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
23 changes: 7 additions & 16 deletions clang/lib/CodeGen/CGDebugInfo.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -58,13 +58,6 @@
using namespace clang;
using namespace clang::CodeGen;

// TODO: consider deprecating ClArrayBoundsPseudoFn; functionality is subsumed
// by -fsanitize-annotate-debug-info
static llvm::cl::opt<bool> ClArrayBoundsPseudoFn(
"array-bounds-pseudofn", llvm::cl::Hidden, llvm::cl::Optional,
llvm::cl::desc("Emit debug info that places array-bounds instrumentation "
"in an inline function called __ubsan_check_array_bounds."));

static uint32_t getTypeAlignIfRequired(const Type *Ty, const ASTContext &Ctx) {
auto TI = Ctx.getTypeInfo(Ty);
if (TI.isAlignRequired())
Expand Down Expand Up @@ -6482,7 +6475,11 @@ llvm::DILocation *CodeGenFunction::SanitizerAnnotateDebugInfo(
SanitizerHandler Handler) {
llvm::DILocation *CheckDebugLoc = Builder.getCurrentDebugLocation();
auto *DI = getDebugInfo();
if (!DI)
if (!DI || !CheckDebugLoc)
return CheckDebugLoc;
Copy link
Contributor

Choose a reason for hiding this comment

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

Do all the extra early-exits speed up or slow down the codegen?

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 would expect a non-null check to be cheaper than constructing a string for the Label

Copy link
Contributor

Choose a reason for hiding this comment

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

The performance tradeoff depends on how often the check fails (if CheckDebugLoc is rarely null, then adding the check here may slow down codegen overall).

Copy link
Contributor Author

Choose a reason for hiding this comment

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

We need to check it somewhere?

const auto &AnnotateDebugInfo =
CGM.getCodeGenOpts().SanitizeAnnotateDebugInfo;
if (AnnotateDebugInfo.empty())
return CheckDebugLoc;

std::string Label;
Expand All @@ -6491,14 +6488,8 @@ llvm::DILocation *CodeGenFunction::SanitizerAnnotateDebugInfo(
else
Label = SanitizerHandlerToCheckLabel(Handler);

for (auto Ord : Ordinals) {
// TODO: deprecate ClArrayBoundsPseudoFn
if (((ClArrayBoundsPseudoFn && Ord == SanitizerKind::SO_ArrayBounds) ||
CGM.getCodeGenOpts().SanitizeAnnotateDebugInfo.has(Ord)) &&
CheckDebugLoc) {
return DI->CreateSyntheticInlineAt(CheckDebugLoc, Label);
}
}
if (any_of(Ordinals, [&](auto Ord) { return AnnotateDebugInfo.has(Ord); }))
Copy link
Contributor

Choose a reason for hiding this comment

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

Maybe !any_of could be combined with the AnnotateDebugInfo.empty() check earlier? That would avoid the need for return CheckDebugLoc; in line 6494.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Doesn't that kind of obfuscate the code? Right now, we have preconditions / early exit first, then the actual logic "if any Ord matches"

Copy link
Contributor

Choose a reason for hiding this comment

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

Arguably, "if no Ord matches" is a precondition / early exit; and, like your earlier checks, avoids the cost of constructing a string for the Label. But up to you, YOLO.

return DI->CreateSyntheticInlineAt(CheckDebugLoc, Label);

return CheckDebugLoc;
}
Expand Down
Loading