Skip to content

Conversation

demoitem
Copy link
Contributor

@demoitem demoitem commented Oct 9, 2025

Add create file with checksum to the C interface

@llvmbot
Copy link
Member

llvmbot commented Oct 9, 2025

@llvm/pr-subscribers-debuginfo

@llvm/pr-subscribers-llvm-ir

Author: peter mckinna (demoitem)

Changes

Add create file with checksum to the C interface


Full diff: https://github.com/llvm/llvm-project/pull/162592.diff

4 Files Affected:

  • (modified) llvm/include/llvm-c/DebugInfo.h (+24)
  • (modified) llvm/lib/IR/DebugInfo.cpp (+33)
  • (modified) llvm/test/Bindings/llvm-c/debug_info_new_format.ll (+54-53)
  • (modified) llvm/tools/llvm-c-test/debuginfo.c (+4-1)
diff --git a/llvm/include/llvm-c/DebugInfo.h b/llvm/include/llvm-c/DebugInfo.h
index 2ecd69a187d4d..96ebec42fcdb4 100644
--- a/llvm/include/llvm-c/DebugInfo.h
+++ b/llvm/include/llvm-c/DebugInfo.h
@@ -203,6 +203,11 @@ enum {
 };
 typedef unsigned LLVMMetadataKind;
 
+/**
+ * The kind of checksum to emit.
+ */
+typedef enum { CSK_MD5, CSK_SHA1, CSK_SHA256 } LLVMChecksumKind;
+
 /**
  * An LLVM DWARF type encoding.
  */
@@ -326,6 +331,25 @@ LLVM_C_ABI LLVMMetadataRef LLVMDIBuilderCreateFile(LLVMDIBuilderRef Builder,
                                                    const char *Directory,
                                                    size_t DirectoryLen);
 
+/**
+ * Create a file descriptor to hold debugging information for a file.
+ * \param Builder      The \c DIBuilder.
+ * \param Filename     File name.
+ * \param FilenameLen  The length of the C string passed to \c Filename.
+ * \param Directory    Directory.
+ * \param DirectoryLen The length of the C string passed to \c Directory.
+ * \param ChecksumKind The kind of checksum. eg MD5, SHA256
+ * \param Checksum     The checksum.
+ * \param ChecksumLen  The length of the checksum.
+ * \param Souce        The embedded source.
+ * \param SourceLen    The length of the source.
+ */
+LLVM_C_ABI LLVMMetadataRef LLVMDIBuilderCreateFileWithCheckSum(
+    LLVMDIBuilderRef Builder, const char *Filename, size_t FilenameLen,
+    const char *Directory, size_t DirectoryLen, LLVMChecksumKind ChecksumKind,
+    const char *Checksum, size_t ChecksumLen, const char *Source,
+    size_t SourceLen);
+
 /**
  * Creates a new descriptor for a module with the specified parent scope.
  * \param Builder         The \c DIBuilder.
diff --git a/llvm/lib/IR/DebugInfo.cpp b/llvm/lib/IR/DebugInfo.cpp
index f9ded507f8328..2b767f19a3923 100644
--- a/llvm/lib/IR/DebugInfo.cpp
+++ b/llvm/lib/IR/DebugInfo.cpp
@@ -1095,6 +1095,39 @@ LLVMDIBuilderCreateFile(LLVMDIBuilderRef Builder, const char *Filename,
                                           StringRef(Directory, DirectoryLen)));
 }
 
+static llvm::DIFile::ChecksumKind
+map_from_llvmChecksumKind(LLVMChecksumKind CSKind) {
+  switch (CSKind) {
+  case LLVMChecksumKind::CSK_MD5:
+    return llvm::DIFile::CSK_MD5;
+  case LLVMChecksumKind::CSK_SHA1:
+    return llvm::DIFile::CSK_SHA1;
+  case LLVMChecksumKind::CSK_SHA256:
+    return llvm::DIFile::CSK_SHA256;
+  default:
+    llvm_unreachable("Unhandled Checksum Kind");
+  }
+}
+
+LLVMMetadataRef LLVMDIBuilderCreateFileWithCheckSum(
+    LLVMDIBuilderRef Builder, const char *Filename, size_t FilenameLen,
+    const char *Directory, size_t DirectoryLen, LLVMChecksumKind ChecksumKind,
+    const char *Checksum, size_t ChecksumLen, const char *Source,
+    size_t SourceLen) {
+  StringRef ChkSum = StringRef(Checksum, ChecksumLen);
+  std::optional<llvm::DIFile::ChecksumInfo<StringRef>> CSInfo;
+  auto CSK = map_from_llvmChecksumKind(ChecksumKind);
+  CSInfo.emplace(CSK, ChkSum);
+  std::optional<StringRef> Src;
+  if (SourceLen > 0)
+    Src = StringRef(Source, SourceLen);
+  else
+    Src = std::nullopt;
+  return wrap(unwrap(Builder)->createFile(StringRef(Filename, FilenameLen),
+                                          StringRef(Directory, DirectoryLen),
+                                          CSInfo, Src));
+}
+
 LLVMMetadataRef
 LLVMDIBuilderCreateModule(LLVMDIBuilderRef Builder, LLVMMetadataRef ParentScope,
                           const char *Name, size_t NameLen,
diff --git a/llvm/test/Bindings/llvm-c/debug_info_new_format.ll b/llvm/test/Bindings/llvm-c/debug_info_new_format.ll
index 83b37da759b5c..75e5fa01b14a0 100644
--- a/llvm/test/Bindings/llvm-c/debug_info_new_format.ll
+++ b/llvm/test/Bindings/llvm-c/debug_info_new_format.ll
@@ -3,37 +3,37 @@
 
 ; CHECK: ; ModuleID = 'debuginfo.c'
 ; CHECK-NEXT: source_filename = "debuginfo.c"
- 
-; CHECK:      define i64 @foo(i64 %0, i64 %1, <10 x i64> %2) !dbg !44 {
+
+; CHECK:      define i64 @foo(i64 %0, i64 %1, <10 x i64> %2) !dbg !45 {
 ; CHECK-NEXT: entry:
-; CHECK-NEXT:     #dbg_declare(i64 0, !49, !DIExpression(), !58)
-; CHECK-NEXT:     #dbg_declare(i64 0, !50, !DIExpression(), !58)
-; CHECK-NEXT:     #dbg_declare(i64 0, !51, !DIExpression(), !58)
-; CHECK-NEXT:     #dbg_label(!59, !58)
+; CHECK-NEXT:     #dbg_declare(i64 0, !50, !DIExpression(), !59)
+; CHECK-NEXT:     #dbg_declare(i64 0, !51, !DIExpression(), !59)
+; CHECK-NEXT:     #dbg_declare(i64 0, !52, !DIExpression(), !59)
+; CHECK-NEXT:     #dbg_label(!60, !59)
 ; CHECK-NEXT:   br label %vars
-; CHECK-NEXT:     #dbg_label(!60, !58)
+; CHECK-NEXT:     #dbg_label(!61, !59)
 ; CHECK-NEXT:   br label %vars
  
 ; CHECK:      vars:                                             ; preds = %entry, %entry
 ; CHECK-NEXT:   %p1 = phi i64 [ 0, %entry ]
 ; CHECK-NEXT:   %p2 = phi i64 [ 0, %entry ]
-; CHECK-NEXT:     #dbg_value(i64 0, !42, !DIExpression(DW_OP_constu, 0, DW_OP_stack_value), !61)
-; CHECK-NEXT:     #dbg_value(i64 1, !52, !DIExpression(DW_OP_constu, 1, DW_OP_stack_value), !61)
+; CHECK-NEXT:     #dbg_value(i64 0, !43, !DIExpression(DW_OP_constu, 0, DW_OP_stack_value), !62)
+; CHECK-NEXT:     #dbg_value(i64 1, !53, !DIExpression(DW_OP_constu, 1, DW_OP_stack_value), !62)
 ; CHECK-NEXT:   %a = add i64 %p1, %p2
 ; CHECK-NEXT:   ret i64 0
 ; CHECK-NEXT: }
  
 ; CHECK:      !llvm.dbg.cu = !{!0}
-; CHECK-NEXT: !FooType = !{!33}
+; CHECK-NEXT: !FooType = !{!34}
 ; CHECK-NEXT: !EnumTest = !{!3}
 ; CHECK-NEXT: !LargeEnumTest = !{!11}
-; CHECK-NEXT: !SubrangeType = !{!36}
-; CHECK-NEXT: !SetType1 = !{!37}
-; CHECK-NEXT: !SetType2 = !{!38}
-; CHECK-NEXT: !DynType = !{!39}
-; CHECK-NEXT: !ClassType = !{!54}
+; CHECK-NEXT: !SubrangeType = !{!37}
+; CHECK-NEXT: !SetType1 = !{!38}
+; CHECK-NEXT: !SetType2 = !{!39}
+; CHECK-NEXT: !DynType = !{!40}
+; CHECK-NEXT: !ClassType = !{!55}
  
-; CHECK:      !0 = distinct !DICompileUnit(language: DW_LANG_C, file: !1, producer: "llvm-c-test", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !2, globals: !16, imports: !24, macros: !28, splitDebugInlining: false, sysroot: "/")
+; CHECK:      !0 = distinct !DICompileUnit(language: DW_LANG_C, file: !1, producer: "llvm-c-test", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !2, globals: !16, imports: !24, macros: !29, splitDebugInlining: false, sysroot: "/")
 ; CHECK-NEXT: !1 = !DIFile(filename: "debuginfo.c", directory: ".")
 ; CHECK-NEXT: !2 = !{!3, !11}
 ; CHECK-NEXT: !3 = !DICompositeType(tag: DW_TAG_enumeration_type, name: "EnumTest", scope: !4, file: !1, baseType: !6, size: 64, elements: !7)
@@ -57,41 +57,42 @@
 ; CHECK-NEXT: !21 = !DIGlobalVariableExpression(var: !22, expr: !DIExpression(DW_OP_constu, 0, DW_OP_stack_value))
 ; CHECK-NEXT: !22 = distinct !DIGlobalVariable(name: "global", scope: !5, file: !1, line: 1, type: !23, isLocal: true, isDefinition: true)
 ; CHECK-NEXT: !23 = !DIDerivedType(tag: DW_TAG_typedef, name: "int64_t", scope: !1, file: !1, line: 42, baseType: !6)
-; CHECK-NEXT: !24 = !{!25, !27}
-; CHECK-NEXT: !25 = !DIImportedEntity(tag: DW_TAG_imported_module, scope: !5, entity: !26, file: !1, line: 42)
+; CHECK-NEXT: !24 = !{!25, !28}
+; CHECK-NEXT: !25 = !DIImportedEntity(tag: DW_TAG_imported_module, scope: !5, entity: !26, file: !27, line: 42)
 ; CHECK-NEXT: !26 = !DIModule(scope: null, name: "llvm-c-test-import", includePath: "/test/include/llvm-c-test-import.h")
-; CHECK-NEXT: !27 = !DIImportedEntity(tag: DW_TAG_imported_module, scope: !5, entity: !25, file: !1, line: 42)
-; CHECK-NEXT: !28 = !{!29}
-; CHECK-NEXT: !29 = !DIMacroFile(file: !1, nodes: !30)
-; CHECK-NEXT: !30 = !{!31, !32}
-; CHECK-NEXT: !31 = !DIMacro(type: DW_MACINFO_define, name: "SIMPLE_DEFINE")
-; CHECK-NEXT: !32 = !DIMacro(type: DW_MACINFO_define, name: "VALUE_DEFINE", value: "1")
-; CHECK-NEXT: !33 = !DIDerivedType(tag: DW_TAG_pointer_type, baseType: !34, size: 192, dwarfAddressSpace: 0)
-; CHECK-NEXT: !34 = !DICompositeType(tag: DW_TAG_structure_type, name: "MyStruct", scope: !4, file: !1, size: 192, elements: !35, runtimeLang: DW_LANG_C89, identifier: "MyStruct")
-; CHECK-NEXT: !35 = !{!6, !6, !6}
-; CHECK-NEXT: !36 = !DISubrangeType(name: "foo", scope: !1, file: !1, line: 42, size: 64, baseType: !6, lowerBound: i64 0, upperBound: i64 1, stride: i64 8, bias: i64 4)
-; CHECK-NEXT: !37 = !DIDerivedType(tag: DW_TAG_set_type, name: "enumset", scope: !1, file: !1, line: 42, baseType: !3, size: 64)
-; CHECK-NEXT: !38 = !DIDerivedType(tag: DW_TAG_set_type, name: "subrangeset", scope: !1, file: !1, line: 42, baseType: !36, size: 64)
-; CHECK-NEXT: !39 = !DICompositeType(tag: DW_TAG_array_type, name: "foo", scope: !1, file: !1, line: 42, baseType: !6, size: 640, elements: !40, dataLocation: !DIExpression(), associated: !42, rank: !DIExpression())
-; CHECK-NEXT: !40 = !{!41}
-; CHECK-NEXT: !41 = !DISubrange(count: 10, lowerBound: 0)
-; CHECK-NEXT: !42 = !DILocalVariable(name: "d", scope: !43, file: !1, line: 43, type: !6)
-; CHECK-NEXT: !43 = distinct !DILexicalBlock(scope: !44, file: !1, line: 42)
-; CHECK-NEXT: !44 = distinct !DISubprogram(name: "foo", linkageName: "foo", scope: !1, file: !1, line: 42, type: !45, scopeLine: 42, spFlags: DISPFlagLocalToUnit | DISPFlagDefinition, unit: !0, retainedNodes: !48)
-; CHECK-NEXT: !45 = !DISubroutineType(types: !46)
-; CHECK-NEXT: !46 = !{!6, !6, !47}
-; CHECK-NEXT: !47 = !DICompositeType(tag: DW_TAG_array_type, baseType: !6, size: 640, flags: DIFlagVector, elements: !40)
-; CHECK-NEXT: !48 = !{!49, !50, !51, !42, !52, !53}
-; CHECK-NEXT: !49 = !DILocalVariable(name: "a", arg: 1, scope: !44, file: !1, line: 42, type: !6)
-; CHECK-NEXT: !50 = !DILocalVariable(name: "b", arg: 2, scope: !44, file: !1, line: 42, type: !6)
-; CHECK-NEXT: !51 = !DILocalVariable(name: "c", arg: 3, scope: !44, file: !1, line: 42, type: !47)
-; CHECK-NEXT: !52 = !DILocalVariable(name: "e", scope: !43, file: !1, line: 44, type: !6)
-; CHECK-NEXT: !53 = !DILabel(scope: !44, name: "label3", file: !1, line: 42)
-; CHECK-NEXT: !54 = !DICompositeType(tag: DW_TAG_class_type, name: "Class", scope: !4, file: !1, size: 192, flags: DIFlagFwdDecl, elements: !55, identifier: "FooClass")
-; CHECK-NEXT: !55 = !{!56}
-; CHECK-NEXT: !56 = !{!6, !6, !57}
-; CHECK-NEXT: !57 = !DIBasicType(name: "Int32", size: 32)
-; CHECK-NEXT: !58 = !DILocation(line: 42, scope: !44)
-; CHECK-NEXT: !59 = !DILabel(scope: !44, name: "label1", file: !1, line: 42)
-; CHECK-NEXT: !60 = !DILabel(scope: !44, name: "label2", file: !1, line: 42)
-; CHECK-NEXT: !61 = !DILocation(line: 43, scope: !44)
+; CHECK-NEXT: !27 = !DIFile(filename: "debuginfo.c", directory: ".", checksumkind: CSK_MD5, checksum: "1234", source: "source")
+; CHECK-NEXT: !28 = !DIImportedEntity(tag: DW_TAG_imported_module, scope: !5, entity: !25, file: !1, line: 42)
+; CHECK-NEXT: !29 = !{!30}
+; CHECK-NEXT: !30 = !DIMacroFile(file: !1, nodes: !31)
+; CHECK-NEXT: !31 = !{!32, !33}
+; CHECK-NEXT: !32 = !DIMacro(type: DW_MACINFO_define, name: "SIMPLE_DEFINE")
+; CHECK-NEXT: !33 = !DIMacro(type: DW_MACINFO_define, name: "VALUE_DEFINE", value: "1")
+; CHECK-NEXT: !34 = !DIDerivedType(tag: DW_TAG_pointer_type, baseType: !35, size: 192, dwarfAddressSpace: 0)
+; CHECK-NEXT: !35 = !DICompositeType(tag: DW_TAG_structure_type, name: "MyStruct", scope: !4, file: !1, size: 192, elements: !36, runtimeLang: DW_LANG_C89, identifier: "MyStruct")
+; CHECK-NEXT: !36 = !{!6, !6, !6}
+; CHECK-NEXT: !37 = !DISubrangeType(name: "foo", scope: !1, file: !1, line: 42, size: 64, baseType: !6, lowerBound: i64 0, upperBound: i64 1, stride: i64 8, bias: i64 4)
+; CHECK-NEXT: !38 = !DIDerivedType(tag: DW_TAG_set_type, name: "enumset", scope: !1, file: !1, line: 42, baseType: !3, size: 64)
+; CHECK-NEXT: !39 = !DIDerivedType(tag: DW_TAG_set_type, name: "subrangeset", scope: !1, file: !1, line: 42, baseType: !37, size: 64)
+; CHECK-NEXT: !40 = !DICompositeType(tag: DW_TAG_array_type, name: "foo", scope: !1, file: !1, line: 42, baseType: !6, size: 640, elements: !41, dataLocation: !DIExpression(), associated: !43, rank: !DIExpression())
+; CHECK-NEXT: !41 = !{!42}
+; CHECK-NEXT: !42 = !DISubrange(count: 10, lowerBound: 0)
+; CHECK-NEXT: !43 = !DILocalVariable(name: "d", scope: !44, file: !1, line: 43, type: !6)
+; CHECK-NEXT: !44 = distinct !DILexicalBlock(scope: !45, file: !1, line: 42)
+; CHECK-NEXT: !45 = distinct !DISubprogram(name: "foo", linkageName: "foo", scope: !1, file: !1, line: 42, type: !46, scopeLine: 42, spFlags: DISPFlagLocalToUnit | DISPFlagDefinition, unit: !0, retainedNodes: !49)
+; CHECK-NEXT: !46 = !DISubroutineType(types: !47)
+; CHECK-NEXT: !47 = !{!6, !6, !48}
+; CHECK-NEXT: !48 = !DICompositeType(tag: DW_TAG_array_type, baseType: !6, size: 640, flags: DIFlagVector, elements: !41)
+; CHECK-NEXT: !49 = !{!50, !51, !52, !43, !53, !54}
+; CHECK-NEXT: !50 = !DILocalVariable(name: "a", arg: 1, scope: !45, file: !1, line: 42, type: !6)
+; CHECK-NEXT: !51 = !DILocalVariable(name: "b", arg: 2, scope: !45, file: !1, line: 42, type: !6)
+; CHECK-NEXT: !52 = !DILocalVariable(name: "c", arg: 3, scope: !45, file: !1, line: 42, type: !48)
+; CHECK-NEXT: !53 = !DILocalVariable(name: "e", scope: !44, file: !1, line: 44, type: !6)
+; CHECK-NEXT: !54 = !DILabel(scope: !45, name: "label3", file: !1, line: 42)
+; CHECK-NEXT: !55 = !DICompositeType(tag: DW_TAG_class_type, name: "Class", scope: !4, file: !1, size: 192, flags: DIFlagFwdDecl, elements: !56, identifier: "FooClass")
+; CHECK-NEXT: !56 = !{!57}
+; CHECK-NEXT: !57 = !{!6, !6, !58}
+; CHECK-NEXT: !58 = !DIBasicType(name: "Int32", size: 32)
+; CHECK-NEXT: !59 = !DILocation(line: 42, scope: !45)
+; CHECK-NEXT: !60 = !DILabel(scope: !45, name: "label1", file: !1, line: 42)
+; CHECK-NEXT: !61 = !DILabel(scope: !45, name: "label2", file: !1, line: 42)
+; CHECK-NEXT: !62 = !DILocation(line: 43, scope: !45)
diff --git a/llvm/tools/llvm-c-test/debuginfo.c b/llvm/tools/llvm-c-test/debuginfo.c
index 0f09c74a476bb..23fa38f4091d3 100644
--- a/llvm/tools/llvm-c-test/debuginfo.c
+++ b/llvm/tools/llvm-c-test/debuginfo.c
@@ -43,6 +43,9 @@ int llvm_test_dibuilder(void) {
   LLVMMetadataRef File = LLVMDIBuilderCreateFile(DIB, Filename,
     strlen(Filename), ".", 1);
 
+  LLVMMetadataRef FileCS = LLVMDIBuilderCreateFileWithCheckSum(
+      DIB, Filename, strlen(Filename), ".", 1, CSK_MD5, "1234", 4, "source", 6);
+
   LLVMMetadataRef CompileUnit = LLVMDIBuilderCreateCompileUnit(
       DIB, LLVMDWARFSourceLanguageC, File, "llvm-c-test", 11, 0, NULL, 0, 0,
       NULL, 0, LLVMDWARFEmissionFull, 0, 0, 0, "/", 1, "", 0);
@@ -61,7 +64,7 @@ int llvm_test_dibuilder(void) {
                               "/test/include/llvm-c-test-import.h", 34,
                               "", 0);
   LLVMMetadataRef ImportedModule = LLVMDIBuilderCreateImportedModuleFromModule(
-      DIB, Module, OtherModule, File, 42, NULL, 0);
+      DIB, Module, OtherModule, FileCS, 42, NULL, 0);
   LLVMDIBuilderCreateImportedModuleFromAlias(DIB, Module, ImportedModule, File,
                                              42, NULL, 0);
 

Copy link
Contributor

@OCHyams OCHyams left a comment

Choose a reason for hiding this comment

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

LGTM with a few nits, test looks good

llvm_unreachable("Unhandled Checksum Kind");
}

LLVMMetadataRef LLVMDIBuilderCreateFileWithCheckSum(
Copy link
Contributor

Choose a reason for hiding this comment

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

nit: I think checksum should be one word (so the S should be lower case in the function name).

Comment on lines +1123 to +1124
else
Src = std::nullopt;
Copy link
Contributor

Choose a reason for hiding this comment

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

I think we can drop the else block as I believe a default initialized std::optional is empty

const char *Checksum, size_t ChecksumLen, const char *Source,
size_t SourceLen) {
StringRef ChkSum = StringRef(Checksum, ChecksumLen);
std::optional<llvm::DIFile::ChecksumInfo<StringRef>> CSInfo;
Copy link
Contributor

Choose a reason for hiding this comment

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

Can we drop the optional part here, since it looks like we're expecting to always have a value. I think you can pass the llvm::DIFile::ChecksumInfo<StringRef> directly to the optional<> param?

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.

3 participants