Reuse code from CGDebugInfo::getOrCreateFile() when creating the file for the DICompileUnit

[aprantl]

Summary

This addresses post-commit feedback for D55085. Without this patch, a main source file with an absolute paths may appear in different DIFiles, once with the absolute path and once with the common prefix between the absolute path and the current working directory.

[uabelho]

If I change what I think is two typos, then the patch compiles for me and solves the issue I had.

The remaining calls to DIBuilder::createFile, should they be changed to createFile too or should they be left as is for some reason?
If they should be left, perhaps it would be good to add some comments explaining why we should use DIBuilder::createFile directly in some place and not in others?

Thanks!

lib/CodeGen/CGDebugInfo.cpp

432 ↗

(On Diff #177544)

I suppose you mean to use the new
createFile
method here?

613 ↗

(On Diff #177544)

I suppose you mean to use the new
createFile
method here?

[aprantl]

I also removed the other references to DIBuilder::createFile() in r348866.

[uabelho]

Great, thanks!

[labath]

I am sorry about the late notice Adrian, but it looks like this patch completely breaks DWO file lookup. I think the search for the common prefix of compilation directory and file path is somehow broken. For example this is what I get when compiling a file from /tmp with PWD=/home/pavel (so, these paths only share / as a prefix):

~ $ /tmp/clang.bad  -g -gsplit-dwarf  -c -o foo.o /tmp/foo.cpp 
~ $ llvm-dwarfdump foo.o 
foo.o:	file format ELF64-x86-64

.debug_info contents:
0x00000000: Compile Unit: length = 0x0000001c version = 0x0004 abbr_offset = 0x0000 addr_size = 0x08 (next unit at 0x00000020)

0x0000000b: DW_TAG_compile_unit
              DW_AT_stmt_list	(0x00000000)
              DW_AT_GNU_dwo_name	("foo.dwo")
              DW_AT_GNU_pubnames	(true)
              DW_AT_GNU_dwo_id	(0x441cbc2aa7bff7c5)
              DW_AT_GNU_addr_base	(0x00000000)

Notice that DW_AT_comp_dir is completely missing (so the debugger has no chance of finding the dwo file). If I revert this (and r348866) then I get correct behavior:

~ $ /tmp/clang.good  -g -gsplit-dwarf  -c -o foo.o /tmp/foo.cpp 
~ $ llvm-dwarfdump foo.o 
foo.o:	file format ELF64-x86-64

.debug_info contents:
0x00000000: Compile Unit: length = 0x00000020 version = 0x0004 abbr_offset = 0x0000 addr_size = 0x08 (next unit at 0x00000024)

0x0000000b: DW_TAG_compile_unit
              DW_AT_stmt_list	(0x00000000)
              DW_AT_GNU_dwo_name	("foo.dwo")
              DW_AT_comp_dir	("/home/pavel")
              DW_AT_GNU_pubnames	(true)
              DW_AT_GNU_dwo_id	(0x441cbc2aa7bff7c5)
              DW_AT_GNU_addr_base	(0x00000000)

If I run the same command from /tmp then I get the correct compilation directory, no matter which clang I use:

/tmp $ /tmp/clang.bad  -g -gsplit-dwarf  -c -o foo.o /tmp/foo.cpp 
/tmp $ llvm-dwarfdump foo.o 
foo.o:	file format ELF64-x86-64

.debug_info contents:
0x00000000: Compile Unit: length = 0x00000020 version = 0x0004 abbr_offset = 0x0000 addr_size = 0x08 (next unit at 0x00000024)

0x0000000b: DW_TAG_compile_unit
              DW_AT_stmt_list	(0x00000000)
              DW_AT_GNU_dwo_name	("foo.dwo")
              DW_AT_comp_dir	("/tmp")
              DW_AT_GNU_pubnames	(true)
              DW_AT_GNU_dwo_id	(0xfc9fc3a6ed2d6bbb)
              DW_AT_GNU_addr_base	(0x00000000)

Could you please check this out?

Thank you.

[aprantl]

I believe I fixed this in r349065. (The DIFile used by the CU is special and distinct from the main source file. Its directory part specifies what becomes the DW_AT_comp_dir (the compilation directory), even if the source file was specified with an absolute path.)

cfe/trunk/lib/CodeGen/CGDebugInfo.h

private:
 
   /// Compute the file checksum debug info for input file ID.
   Optional<llvm::DIFile::ChecksumKind>
   computeChecksum(FileID FID, SmallString<32> &Checksum) const;
 
   /// Get the source of the given file ID.
   Optional<StringRef> getSource(const SourceManager &SM, FileID FID);
 
-  /// Get the file debug info descriptor for the input location.
+  /// Convenience function to get the file debug info descriptor for the input
+  /// location.
   llvm::DIFile *getOrCreateFile(SourceLocation Loc);
 
+  /// Create a file debug info descriptor for a source file.
+  llvm::DIFile *
+  createFile(StringRef FileName,
+             Optional<llvm::DIFile::ChecksumInfo<StringRef>> CSInfo,
+             Optional<StringRef> Source);
+
   /// Get the file info for main compile unit.
   llvm::DIFile *getOrCreateMainFile();
 
   /// Get the type from the cache or create a new type if necessary.
   llvm::DIType *getOrCreateType(QualType Ty, llvm::DIFile *Fg);
 
   /// Get a reference to a clang module.  If \p CreateSkeletonCU is true,
   /// this also creates a split dwarf skeleton compile unit.

cfe/trunk/lib/CodeGen/CGDebugInfo.cpp

llvm::DIFile *CGDebugInfo::getOrCreateFile(SourceLocation Loc) {
   }
 
   SmallString<32> Checksum;
   Optional<llvm::DIFile::ChecksumKind> CSKind =
       computeChecksum(SM.getFileID(Loc), Checksum);
   Optional<llvm::DIFile::ChecksumInfo<StringRef>> CSInfo;
   if (CSKind)
     CSInfo.emplace(*CSKind, Checksum);
+  return createFile(FileName, CSInfo, getSource(SM, SM.getFileID(Loc)));
+}
 
+llvm::DIFile *
+CGDebugInfo::createFile(StringRef FileName,
+                        Optional<llvm::DIFile::ChecksumInfo<StringRef>> CSInfo,
+                        Optional<StringRef> Source) {
   StringRef Dir;
   StringRef File;
   std::string RemappedFile = remapDIPath(FileName);
   std::string CurDir = remapDIPath(getCurrentDirname());
   SmallString<128> DirBuf;
   SmallString<128> FileBuf;
   if (llvm::sys::path::is_absolute(RemappedFile)) {
     // Strip the common prefix (if it is more than just "/") from current
if (std::distance(llvm::sys::path::begin(CurDir), CurDirIt) == 1) {
         llvm::sys::path::append(FileBuf, *FileIt);
       Dir = DirBuf;
       File = FileBuf;
     }
   } else {
     Dir = CurDir;
     File = RemappedFile;
   }
-  llvm::DIFile *F =
-      DBuilder.createFile(File, Dir, CSInfo,
-                          getSource(SM, SM.getFileID(Loc)));
-
+  llvm::DIFile *F = DBuilder.createFile(File, Dir, CSInfo, Source);
   DIFileCache[FileName.data()].reset(F);
   return F;
 }
 
 llvm::DIFile *CGDebugInfo::getOrCreateMainFile() {
-  return DBuilder.createFile(
+  return createFile(
       remapDIPath(TheCU->getFilename()), remapDIPath(TheCU->getDirectory()),
       TheCU->getFile()->getChecksum(),
       CGM.getCodeGenOpts().EmbedSource ? TheCU->getSource() : None);
 }
 
 std::string CGDebugInfo::remapDIPath(StringRef Path) const {
   for (const auto &Entry : DebugPrefixMap)
     if (Path.startswith(Entry.first))
void CGDebugInfo::CreateCompileUnit() {
   if (CSKind)
     CSInfo.emplace(*CSKind, Checksum);
 
   // Create new compile unit.
   // FIXME - Eliminate TheCU.
   auto &CGOpts = CGM.getCodeGenOpts();
   TheCU = DBuilder.createCompileUnit(
       LangTag,
-      DBuilder.createFile(remapDIPath(MainFileName),
-                          remapDIPath(getCurrentDirname()), CSInfo,
-                          getSource(SM, SM.getMainFileID())),
+      createFile(MainFileName, CSInfo, getSource(SM, SM.getMainFileID())),
       CGOpts.EmitVersionIdentMetadata ? Producer : "",
       LO.Optimize || CGOpts.PrepareForLTO || CGOpts.PrepareForThinLTO,
       CGOpts.DwarfDebugFlags, RuntimeVers,
       (CGOpts.getSplitDwarfMode() != CodeGenOptions::NoFission)
           ? ""
           : CGOpts.SplitDwarfFile,
       EmissionKind, 0 /* DWOid */, CGOpts.SplitDwarfInlining,
       CGOpts.DebugInfoForProfiling,

cfe/trunk/test/CodeGen/debug-info-compilation-dir.c

-// RUN: %clang_cc1 -fdebug-compilation-dir /nonsense -emit-llvm -debug-info-kind=limited %s -o - | FileCheck -check-prefix=CHECK-NONSENSE %s
+// RUN: mkdir -p %t.dir && cd %t.dir
+// RUN: cp %s rel.c
+// RUN: %clang_cc1 -fdebug-compilation-dir /nonsense -emit-llvm -debug-info-kind=limited rel.c -o - | FileCheck -check-prefix=CHECK-NONSENSE %s
 // CHECK-NONSENSE: nonsense
 
 // RUN: %clang_cc1 -emit-llvm -debug-info-kind=limited %s -o - | FileCheck -check-prefix=CHECK-DIR %s
 // CHECK-DIR: CodeGen
 

cfe/trunk/test/PCH/debug-info-pch-path.c

 //
 // CHECK-REL-NODIR: !DICompileUnit
 // CHECK-REL-NODIR-SAME:           file: ![[C:[0-9]+]]
 // CHECK-REL-NODIR-NOT: dwoId
 // CHECK-REL-NODIR: ![[C]] = !DIFile({{.*}}directory: "[[DIR:.*]]"
 // CHECK-REL-NODIR: !DICompileUnit(
 // CHECK-REL-NODIR-SAME:           file: ![[PCH:[0-9]+]]
 // CHECK-REL-NODIR-SAME:           splitDebugFilename: "prefix.pch"
-// CHECK-REL-NODIR: ![[PCH]] = !DIFile({{.*}}directory: "[[DIR]]"
+// CHECK-REL-NODIR: ![[PCH]] = !DIFile({{.*}}directory: "[[DIR]]
 
 // ---------------------------------------------------------------------
 // Relative PCH in a subdirectory.
 // ---------------------------------------------------------------------
 //
 // RUN: mkdir pchdir
 // RUN: %clang_cc1 -fmodule-format=obj -emit-pch                \
 // RUN:     -triple %itanium_abi_triple                         \

 
 // CHECK-ABS: !DICompileUnit
 // CHECK-ABS-SAME:           file: ![[C:[0-9]+]]
 // CHECK-ABS-NOT: dwoId
 // CHECK-ABS: ![[C]] = !DIFile({{.*}}directory: "[[DIR:.*]]"
 // CHECK-ABS: !DICompileUnit(
 // CHECK-ABS-SAME:           file: ![[PCH:[0-9]+]]
 // CHECK-ABS-SAME:           splitDebugFilename: "prefix.pch"
-// CHECK-ABS: ![[PCH]] = !DIFile({{.*}}directory: "[[DIR]]"
+// CHECK-ABS: ![[PCH]] = !DIFile({{.*}}directory: "[[DIR]]