[inline-asm] Fixed issue no. 164973 #167316

AnSaki57 · 2025-11-10T14:08:38Z

Made modifications to CGStmt.cpp and added a clang CodeGen test named "asm-srcloc-split-literal.c".

…pp and adding a clang CodeGen test named "asm-srcloc-split-literal.c".

github-actions · 2025-11-10T14:08:57Z

Thank you for submitting a Pull Request (PR) to the LLVM Project!

This PR will be automatically labeled and the relevant teams will be notified.

If you wish to, you can add reviewers by using the "Reviewers" section on this page.

If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using @ followed by their GitHub username.

If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers.

If you have further questions, they may be answered by the LLVM GitHub User Guide.

You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums.

llvmbot · 2025-11-10T14:09:27Z

@llvm/pr-subscribers-clang-codegen

@llvm/pr-subscribers-clang

Author: Anakala Sashikiran (AnSaki57)

Changes

Made modifications to CGStmt.cpp and added a clang CodeGen test named "asm-srcloc-split-literal.c".

Full diff: https://github.com/llvm/llvm-project/pull/167316.diff

2 Files Affected:

(modified) clang/lib/CodeGen/CGStmt.cpp (+34-9)
(added) clang/test/CodeGen/asm-srcloc-split-literal.c (+23)

diff --git a/clang/lib/CodeGen/CGStmt.cpp b/clang/lib/CodeGen/CGStmt.cpp
index 36be3295950b8..5a062b74b0607 100644
--- a/clang/lib/CodeGen/CGStmt.cpp
+++ b/clang/lib/CodeGen/CGStmt.cpp
@@ -2572,24 +2572,49 @@ CodeGenFunction::EmitAsmInput(const TargetInfo::ConstraintInfo &Info,
 static llvm::MDNode *getAsmSrcLocInfo(const StringLiteral *Str,
                                       CodeGenFunction &CGF) {
   SmallVector<llvm::Metadata *, 8> Locs;
+
+  // We need these to find the correct location for the first line.
+  StringRef StrVal = Str->getString();
+  const SourceManager &SM = CGF.CGM.getContext().getSourceManager();
+  const LangOptions &LangOpts = CGF.CGM.getLangOpts();
+  unsigned StartToken = 0;
+  unsigned ByteOffset = 0;
+
   // Add the location of the first line to the MDNode.
+
+  // Find the offset of the first character that isn't horizontal whitespace.
+  size_t FirstLocOffset = StrVal.find_first_not_of(" \t\v\f");
+
+  // If the string is empty or all-whitespace, default to offset 0.
+  if (FirstLocOffset == StringRef::npos)
+    FirstLocOffset = 0;
+
+  SourceLocation FirstLineLoc = Str->getLocationOfByte(
+    FirstLocOffset, SM, LangOpts, CGF.getTarget(), &StartToken, &ByteOffset);
+
   Locs.push_back(llvm::ConstantAsMetadata::get(llvm::ConstantInt::get(
-      CGF.Int64Ty, Str->getBeginLoc().getRawEncoding())));
-  StringRef StrVal = Str->getString();
-  if (!StrVal.empty()) {
-    const SourceManager &SM = CGF.CGM.getContext().getSourceManager();
-    const LangOptions &LangOpts = CGF.CGM.getLangOpts();
-    unsigned StartToken = 0;
-    unsigned ByteOffset = 0;
+    CGF.Int64Ty, FirstLineLoc.getRawEncoding())));
 
+  if (!StrVal.empty()) {
     // Add the location of the start of each subsequent line of the asm to the
     // MDNode.
     for (unsigned i = 0, e = StrVal.size() - 1; i != e; ++i) {
       if (StrVal[i] != '\n') continue;
+
+      // The next line starts at byte offset i + 1.
+      // Find the first non-horizontal-whitespace at or after this offset.
+      size_t NextLineOffset = StrVal.find_first_not_of(" \t\v\f", i + 1);
+
+      // If the rest of the string is empty or all-whitespace,
+      // just use the location right after the newline (i + 1).
+      if (NextLineOffset == StringRef::npos)
+        NextLineOffset = i + 1;
+
       SourceLocation LineLoc = Str->getLocationOfByte(
-          i + 1, SM, LangOpts, CGF.getTarget(), &StartToken, &ByteOffset);
+        NextLineOffset, SM, LangOpts, CGF.getTarget(), &StartToken, &ByteOffset);
+
       Locs.push_back(llvm::ConstantAsMetadata::get(
-          llvm::ConstantInt::get(CGF.Int64Ty, LineLoc.getRawEncoding())));
+        llvm::ConstantInt::get(CGF.Int64Ty, LineLoc.getRawEncoding())));
     }
   }
 
diff --git a/clang/test/CodeGen/asm-srcloc-split-literal.c b/clang/test/CodeGen/asm-srcloc-split-literal.c
new file mode 100644
index 0000000000000..a1d898264e6bd
--- /dev/null
+++ b/clang/test/CodeGen/asm-srcloc-split-literal.c
@@ -0,0 +1,23 @@
+/// Test that inline asm source location corresponds to the actual
+/// instruction line, not the first line of the asm block.
+///
+/// RUN: not %clang_cc1 -triple x86_64-pc-linux-gnu -emit-obj %s 2>&1 | FileCheck %s
+
+// #include <stdint.h>
+// #include <string.h>
+
+void *memset(void *dest, int c, int n)__attribute__((naked));
+void *memset(void *dest, int c, int n) {
+    __asm__(
+        "\t"            // <-- line with only a tab
+        "xchg %eax, %eax\n" // <-- A valid instruction
+        "\t"            // <-- line with only a tab
+        "mov rdi, 1\n"  // <-- An invalid instruction
+    );
+}
+
+int main() { return 0; }
+
+// CHECK: error: unknown use of instruction mnemonic
+// CHECK-NEXT: mov rdi, 1
+

github-actions · 2025-11-10T18:45:12Z

⚠️ C/C++ code formatter, clang-format found issues in your code. ⚠️

You can test this locally with the following command:

git-clang-format --diff origin/main HEAD --extensions c,cpp -- clang/test/CodeGen/asm-srcloc-split-literal.c clang/lib/CodeGen/CGStmt.cpp --diff_from_common_commit

⚠️
The reproduction instructions above might return results for more than one PR
in a stack if you are using a stacked PR workflow. You can limit the results by
changing origin/main to the base branch/commit you want to compare against.
⚠️

View the diff from clang-format here.

diff --git a/clang/lib/CodeGen/CGStmt.cpp b/clang/lib/CodeGen/CGStmt.cpp
index 5a062b74b..94a5c338e 100644
--- a/clang/lib/CodeGen/CGStmt.cpp
+++ b/clang/lib/CodeGen/CGStmt.cpp
@@ -2590,10 +2590,10 @@ static llvm::MDNode *getAsmSrcLocInfo(const StringLiteral *Str,
     FirstLocOffset = 0;
 
   SourceLocation FirstLineLoc = Str->getLocationOfByte(
-    FirstLocOffset, SM, LangOpts, CGF.getTarget(), &StartToken, &ByteOffset);
+      FirstLocOffset, SM, LangOpts, CGF.getTarget(), &StartToken, &ByteOffset);
 
-  Locs.push_back(llvm::ConstantAsMetadata::get(llvm::ConstantInt::get(
-    CGF.Int64Ty, FirstLineLoc.getRawEncoding())));
+  Locs.push_back(llvm::ConstantAsMetadata::get(
+      llvm::ConstantInt::get(CGF.Int64Ty, FirstLineLoc.getRawEncoding())));
 
   if (!StrVal.empty()) {
     // Add the location of the start of each subsequent line of the asm to the
@@ -2610,11 +2610,12 @@ static llvm::MDNode *getAsmSrcLocInfo(const StringLiteral *Str,
       if (NextLineOffset == StringRef::npos)
         NextLineOffset = i + 1;
 
-      SourceLocation LineLoc = Str->getLocationOfByte(
-        NextLineOffset, SM, LangOpts, CGF.getTarget(), &StartToken, &ByteOffset);
+      SourceLocation LineLoc =
+          Str->getLocationOfByte(NextLineOffset, SM, LangOpts, CGF.getTarget(),
+                                 &StartToken, &ByteOffset);
 
       Locs.push_back(llvm::ConstantAsMetadata::get(
-        llvm::ConstantInt::get(CGF.Int64Ty, LineLoc.getRawEncoding())));
+          llvm::ConstantInt::get(CGF.Int64Ty, LineLoc.getRawEncoding())));
     }
   }

efriedma-quic

Please fix the title to explain what you're fixing. It's good to put something like "Fixes #164973" somewhere, but not the title/first line.

efriedma-quic · 2025-11-10T18:45:32Z

clang/test/CodeGen/asm-srcloc-split-literal.c

+int main() { return 0; }
+
+// CHECK: error: unknown use of instruction mnemonic
+// CHECK-NEXT: mov rdi, 1


The CHECK lines here don't seem to actually check the line numbers.

efriedma-quic · 2025-11-10T18:46:05Z

clang/test/CodeGen/asm-srcloc-split-literal.c

+/// Test that inline asm source location corresponds to the actual
+/// instruction line, not the first line of the asm block.
+///
+/// RUN: not %clang_cc1 -triple x86_64-pc-linux-gnu -emit-obj %s 2>&1 | FileCheck %s


Please add -o %t or something like that so we don't accidentally get output files in weird places.

Needs // REQUIRES: x86-registered-target somewhere because this uses the x86 asm parser.

efriedma-quic · 2025-11-10T18:46:16Z

clang/test/CodeGen/asm-srcloc-split-literal.c

+/// RUN: not %clang_cc1 -triple x86_64-pc-linux-gnu -emit-obj %s 2>&1 | FileCheck %s
+
+// #include <stdint.h>
+// #include <string.h>


Unnecessary comments.

efriedma-quic · 2025-11-10T18:49:31Z

clang/lib/CodeGen/CGStmt.cpp

+
+      // The next line starts at byte offset i + 1.
+      // Find the first non-horizontal-whitespace at or after this offset.
+      size_t NextLineOffset = StrVal.find_first_not_of(" \t\v\f", i + 1);


Can you rearrange the loop so you don't have to duplicate the code for the first line?

[inline-asm] Fixed issue no. 164973, making modifications to CGStmt.c…

d7b4416

…pp and adding a clang CodeGen test named "asm-srcloc-split-literal.c".

llvmbot added clang Clang issues not falling into any other category clang:codegen IR generation bugs: mangling, exceptions, etc. labels Nov 10, 2025

efriedma-quic reviewed Nov 10, 2025

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[inline-asm] Fixed issue no. 164973 #167316

[inline-asm] Fixed issue no. 164973 #167316

AnSaki57 commented Nov 10, 2025

Uh oh!

github-actions bot commented Nov 10, 2025

Uh oh!

llvmbot commented Nov 10, 2025 •

edited

Loading

Uh oh!

github-actions bot commented Nov 10, 2025

Uh oh!

efriedma-quic left a comment

Uh oh!

efriedma-quic Nov 10, 2025

Uh oh!

efriedma-quic Nov 10, 2025

Uh oh!

efriedma-quic Nov 10, 2025

Uh oh!

efriedma-quic Nov 10, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

[inline-asm] Fixed issue no. 164973 #167316

Are you sure you want to change the base?

[inline-asm] Fixed issue no. 164973 #167316

Conversation

AnSaki57 commented Nov 10, 2025

Uh oh!

github-actions bot commented Nov 10, 2025

Uh oh!

llvmbot commented Nov 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions bot commented Nov 10, 2025

Uh oh!

efriedma-quic left a comment

Choose a reason for hiding this comment

Uh oh!

efriedma-quic Nov 10, 2025

Choose a reason for hiding this comment

Uh oh!

efriedma-quic Nov 10, 2025

Choose a reason for hiding this comment

Uh oh!

efriedma-quic Nov 10, 2025

Choose a reason for hiding this comment

Uh oh!

efriedma-quic Nov 10, 2025

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

llvmbot commented Nov 10, 2025 •

edited

Loading