Skip to content

[clang-tidy] Fix macro-repeated-side-effects false positive on assert - #235410

Open
yashaggarwal85d wants to merge 1 commit into
llvm:mainfrom
yashaggarwal85d:fix-macro-repeated-side-effects-assert
Open

yashaggarwal85d wants to merge 1 commit into
llvm:mainfrom
yashaggarwal85d:fix-macro-repeated-side-effects-assert

Conversation

@yashaggarwal85d

Copy link
Copy Markdown
Contributor

Since glibc 2.44, assert() is variadic and mentions __VA_ARGS__ twice, but only one copy is evaluated. bugprone-macro-repeated-side-effects counted both, so assert(++x == 1) got flagged.

There are two shapes. In C the extra copy is inside sizeof(__assert_single_arg(__VA_ARGS__)). In C++26 it's in the dead branch of void(1 ? 1 : bool(__VA_ARGS__)).

So the check now skips args inside sizeof/alignof/_Alignof (like it already does for __builtin_constant_p), and skips the false branch of (N ? a : b) when N is a non-zero integer literal, up to the closing paren or a comma. (0 ? a : b) and non-literal conditions are unchanged.

I checked it against the real glibc 2.44 assert.h with -std=c23 and -std=c++2c: warning before, none after. Added tests with glibc-style macros and a release note.

Fixes #233597

This fix was made with help from an AI tool. I reviewed and tested the change myself.

@github-actions

Copy link
Copy Markdown

Hello @yashaggarwal85d 👋

Thank you for submitting a Pull Request (PR) to the LLVM Project. Since this is your first PR, here are a few useful links covering our main contribution policies and review practices.

  • All contributions to LLVM must follow our LLVM AI Tool Use Policy. In particular, if you used AI while working on this PR, remember to add a note to the PR description.
  • The LLVM Code-Review Policy and Practices document contains practical information about the PR process, including how patches are reviewed and accepted, and who can review a PR.
  • Our LLVM Developer Policy describes our expectations for code quality, commit summaries and contains notes on our CI system.

Please reply to this message to confirm that you have read these policies, especially the LLVM AI Tool Use Policy, and that any AI tool usage has been noted in the PR description.


Frequently asked questions

How do I add reviewers?

This PR will be automatically labeled, and the relevant teams will be notified. For some parts of the project, reviewers may also be added automatically.

You can also add reviewers manually using the Reviewers section on this page. If you cannot use that section, it is probably because you do not have write permissions for the repository. In that case, you can request a review by tagging reviewers in a comment using @ followed by their GitHub username.

What if there are no comments?

If you have not received any comments on your PR after a week, you can request a review by pinging the PR with a comment such as “Ping”. The common courtesy ping rate is once a week. Please remember that you are asking for volunteer time from other developers.

Are any special GitHub settings required to contribute to LLVM?

We only require contributors to have a public email address associated with their GitHub commits, see this section of LLVM Developer Policy for details.


If you have questions, feel free to leave a comment on this PR, or ask on LLVM Discord or LLVM Discourse.

Thank you,
The LLVM Community

@llvmorg-github-actions

llvmorg-github-actions Bot commented Oct 25, 2026 •

Copy link
Copy Markdown

@llvm/pr-subscribers-clang-tools-extra

@llvm/pr-subscribers-clang-tidy

Author: Yash Aggarwal (yashaggarwal85d)

Changes

Since glibc 2.44, assert() is variadic and mentions __VA_ARGS__ twice, but only one copy is evaluated. bugprone-macro-repeated-side-effects counted both, so assert(++x == 1) got flagged.

There are two shapes. In C the extra copy is inside sizeof(__assert_single_arg(__VA_ARGS__)). In C++26 it's in the dead branch of void(1 ? 1 : bool(__VA_ARGS__)).

So the check now skips args inside sizeof/alignof/_Alignof (like it already does for __builtin_constant_p), and skips the false branch of (N ? a : b) when N is a non-zero integer literal, up to the closing paren or a comma. (0 ? a : b) and non-literal conditions are unchanged.

I checked it against the real glibc 2.44 assert.h with -std=c23 and -std=c++2c: warning before, none after. Added tests with glibc-style macros and a release note.

Fixes #233597

This fix was made with help from an AI tool. I reviewed and tested the change myself.


Full diff: https://github.com/llvm/llvm-project/pull/235410.diff

3 Files Affected:

  • (modified) clang-tools-extra/clang-tidy/bugprone/MacroRepeatedSideEffectsCheck.cpp (+49-5)
  • (modified) clang-tools-extra/docs/ReleaseNotes.md (+6)
  • (modified) clang-tools-extra/test/clang-tidy/checkers/bugprone/macro-repeated-side-effects.c (+25)
diff --git a/clang-tools-extra/clang-tidy/bugprone/MacroRepeatedSideEffectsCheck.cpp b/clang-tools-extra/clang-tidy/bugprone/MacroRepeatedSideEffectsCheck.cpp
index 4306cb11ce4bac..6297e675580559 100644
--- a/clang-tools-extra/clang-tidy/bugprone/MacroRepeatedSideEffectsCheck.cpp
+++ b/clang-tools-extra/clang-tidy/bugprone/MacroRepeatedSideEffectsCheck.cpp
@@ -12,6 +12,7 @@
 #include "clang/Lex/MacroArgs.h"
 #include "clang/Lex/PPCallbacks.h"
 #include "clang/Lex/Preprocessor.h"
+#include <optional>
 #include <stack>
 
 namespace clang::tidy::bugprone {
@@ -32,6 +33,8 @@ class MacroRepeatedPPCallbacks : public PPCallbacks {
   unsigned countArgumentExpansions(const MacroInfo *MI,
                                    const IdentifierInfo *Arg) const;
 
+  bool isNonZeroLiteral(const Token &T) const;
+
   bool hasSideEffects(const Token *ResultArgToks) const;
 };
 } // End of anonymous namespace.
@@ -84,10 +87,30 @@ unsigned MacroRepeatedPPCallbacks::countArgumentExpansions(
   // Has a __builtin_constant_p been found?
   bool FoundBuiltin = false;
   bool PrevTokenIsHash = false;
+  int ParenDepth = 0;
+  // Paren depth of the false branch of "(1 ? a : b)" while we are skipping it.
+  std::optional<int> DeadBranchDepth;
   // Count when "?" is reached. The "Current" will get this value when the ":"
-  // is reached.
-  std::stack<unsigned, SmallVector<unsigned, 8>> CountAtQuestion;
-  for (const auto &T : MI->tokens()) {
+  // is reached. Also remembers if the condition was a non-zero literal.
+  std::stack<std::pair<unsigned, bool>,
+             SmallVector<std::pair<unsigned, bool>, 8>>
+      CountAtQuestion;
+  const ArrayRef<Token> Tokens = MI->tokens();
+  for (const auto [I, T] : llvm::enumerate(Tokens)) {
+    if (T.is(tok::l_paren))
+      ParenDepth++;
+    else if (T.is(tok::r_paren))
+      ParenDepth--;
+
+    // The false branch of "(1 ? a : b)" is never evaluated. Skip it until the
+    // closing parenthesis or a comma.
+    if (DeadBranchDepth) {
+      if (ParenDepth > *DeadBranchDepth ||
+          (ParenDepth == *DeadBranchDepth && T.isNot(tok::comma)))
+        continue;
+      DeadBranchDepth.reset();
+    }
+
     // The result of __builtin_constant_p(x) is 0 if x is a macro argument
     // with side effects. If we see a __builtin_constant_p(x) followed by a
     // "?" "&&" or "||", then we need to reason about control flow to report
@@ -108,12 +131,19 @@ unsigned MacroRepeatedPPCallbacks::countArgumentExpansions(
 
     // Handling of ? and :.
     if (T.is(tok::question)) {
-      CountAtQuestion.push(Current);
+      CountAtQuestion.emplace(Current, I >= 2 &&
+                                           Tokens[I - 2].is(tok::l_paren) &&
+                                           isNonZeroLiteral(Tokens[I - 1]));
     } else if (T.is(tok::colon)) {
       if (CountAtQuestion.empty())
         return 0;
-      Current = CountAtQuestion.top();
+      bool SkipFalseBranch = false;
+      std::tie(Current, SkipFalseBranch) = CountAtQuestion.top();
       CountAtQuestion.pop();
+      if (SkipFalseBranch) {
+        DeadBranchDepth = ParenDepth;
+        continue;
+      }
     }
 
     // If current token is a parenthesis, skip it.
@@ -132,6 +162,12 @@ unsigned MacroRepeatedPPCallbacks::countArgumentExpansions(
     if (TII == nullptr)
       continue;
 
+    // Arguments of sizeof and alignof are not evaluated.
+    if (T.isOneOf(tok::kw_sizeof, tok::kw_alignof, tok::kw__Alignof)) {
+      SkipParen = true;
+      continue;
+    }
+
     // If a __builtin_constant_p is found within the macro definition, don't
     // count arguments inside the parentheses and remember that it has been
     // seen in case there are "?", "&&" or "||" operators later.
@@ -159,6 +195,14 @@ unsigned MacroRepeatedPPCallbacks::countArgumentExpansions(
   return Max;
 }
 
+bool MacroRepeatedPPCallbacks::isNonZeroLiteral(const Token &T) const {
+  if (T.isNot(tok::numeric_constant))
+    return false;
+  SmallString<16> Buffer;
+  uint64_t Value = 0;
+  return !PP.getSpelling(T, Buffer).getAsInteger(0, Value) && Value != 0;
+}
+
 bool MacroRepeatedPPCallbacks::hasSideEffects(
     const Token *ResultArgToks) const {
   for (; ResultArgToks->isNot(tok::eof); ++ResultArgToks)
diff --git a/clang-tools-extra/docs/ReleaseNotes.md b/clang-tools-extra/docs/ReleaseNotes.md
index 037df4cd0937c4..4337ce234db855 100644
--- a/clang-tools-extra/docs/ReleaseNotes.md
+++ b/clang-tools-extra/docs/ReleaseNotes.md
@@ -204,6 +204,12 @@ infrastructure are described first, followed by tool-specific sections.
   <clang-tidy/checks/bugprone/macro-parentheses>` by fixing invalid fixes for
   macro arguments used as names in alias declarations.
 
+- Improved {doc}`bugprone-macro-repeated-side-effects
+  <clang-tidy/checks/bugprone/macro-repeated-side-effects>` check by not
+  counting macro arguments that are never evaluated, such as inside `sizeof`
+  or in the false branch of `(1 ? a : b)`. This fixes false positives on
+  `assert` from glibc 2.44.
+
 - Fixed a crash in {doc}`bugprone-misplaced-operator-in-strlen-in-alloc
   <clang-tidy/checks/bugprone/misplaced-operator-in-strlen-in-alloc>` when
   checking an array new expression without a size expression.
diff --git a/clang-tools-extra/test/clang-tidy/checkers/bugprone/macro-repeated-side-effects.c b/clang-tools-extra/test/clang-tidy/checkers/bugprone/macro-repeated-side-effects.c
index 995d872dce76a0..6657719a49c45b 100644
--- a/clang-tools-extra/test/clang-tidy/checkers/bugprone/macro-repeated-side-effects.c
+++ b/clang-tools-extra/test/clang-tidy/checkers/bugprone/macro-repeated-side-effects.c
@@ -123,3 +123,28 @@ void log(const char *s, int v);
 void test_log(int a) {
   LOG(a++);
 }
+
+_Bool single_arg(_Bool);
+void fail(const char *);
+#define sizeofgood(x)   (sizeof(x) + _Alignof(x) + (x))
+#define sizeofbad(x)    (sizeof(x) + (x) + (x))
+#define assertc(...)    (((void)sizeof(single_arg(__VA_ARGS__)), __VA_ARGS__) ? (void)0 : fail(#__VA_ARGS__))
+#define assertcpp(...)  ((__VA_ARGS__) ? (void)(1 ? 1 : (_Bool)(__VA_ARGS__)) : fail(#__VA_ARGS__))
+#define deadbad1(x)     ((x) + (0 ? 1 : (x)))
+#define deadbad2(x)     ((x) + (1 ? 1 : (x), (x)))
+void unevaluated(int ret, int a) {
+  ret += sizeofgood(a++);
+  ret += sizeofbad(a++);
+  // CHECK-NOTES: :[[@LINE-1]]:20: warning: side effects in the 1st macro argument 'x'
+  // CHECK-NOTES: :[[@LINE-9]]:9: note: macro 'sizeofbad' defined here
+
+  assertc(++a == 1);
+  assertcpp(++a == 1);
+
+  ret += deadbad1(a++);
+  // CHECK-NOTES: :[[@LINE-1]]:19: warning: side effects in the 1st macro argument 'x'
+  // CHECK-NOTES: :[[@LINE-13]]:9: note: macro 'deadbad1' defined here
+  ret += deadbad2(a++);
+  // CHECK-NOTES: :[[@LINE-1]]:19: warning: side effects in the 1st macro argument 'x'
+  // CHECK-NOTES: :[[@LINE-15]]:9: note: macro 'deadbad2' defined here
+}

@yashaggarwal85d

Copy link
Copy Markdown
Contributor Author

I have read all the policies mentioned by the bot. Thanks

@github-actions

Copy link
Copy Markdown

⚠️ We detected that you are using a GitHub private e-mail address to contribute to the repo.
Please turn off Keep my email addresses private setting in your account.
See LLVM Developer Policy and LLVM Discourse for more information.

@yashaggarwal85d

Copy link
Copy Markdown
Contributor Author

⚠️ We detected that you are using a GitHub private e-mail address to contribute to the repo. Please turn off Keep my email addresses private setting in your account. See LLVM Developer Policy and LLVM Discourse for more information.

Fixed this!

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[clang-tidy] bugprone-macro-repeated-side-effects false positive with assert() in glibc 2.44

1 participant