[LLD] [COFF] Generate a codeview build id signature for MinGW even when not creating a PDB

[mstorsjo]

Summary

GNU ld, which doesn't generate PDBs, can optionally generate a build id by passing the --build-id option. LLD's MinGW frontend knows about this option but ignores it, as I had incorrectly assumed that LLD already generated build IDs even in those cases.

If debug info is requested and no PDB path is set, generate a build id signature as a hash of the binary itself. This allows associating a binary to a minidump, even if debug info isn't written in PDB form by the linker.

I guess the main question is if we should add an lld specific option for this to the lld-link inteface, which we pass from the MinGW frontend when --build-id is received, or if we just should always generate it for the "mingw && debug && !pdb" case.

[ruiu]

COFF/Writer.cpp

1574

I think Config->MinGW && Config->Debug && Config->PDBPath.empty() is fine, but it seems this code needs a bit more explanation, because we create build id even if GenerateBuildId is false (if we are creating a PDB file.)

[mstorsjo]

Added a comment explaining why the added code is there, renamed the variable to GenerateSyntheticBuildId to differentiate it a bit further from the normal build id generated in a PDB file.

[mstorsjo]

Missed renaming the variable in one place in the previous version of the patch.

[ruiu]

Could you add a test?

[ruiu]

Oh sorry you have already added tests. For some reason they weren't displayed at first, but I can now see them. LGTM.

COFF/Writer.cpp

   // "timestamp" in the COFF file header, and the ones in the coff debug
   // directory.  Now we can hash the file and write that hash to the various
   // timestamp fields in the file.
   StringRef OutputFileData(
       reinterpret_cast<const char *>(Buffer->getBufferStart()),
       Buffer->getBufferSize());
 
   uint32_t Timestamp = Config->Timestamp;
+  uint64_t Hash = 0;
+  bool GenerateSyntheticBuildId =
+      Config->MinGW && Config->Debug && Config->PDBPath.empty();
+
+  if (Config->Repro || GenerateBuildId)
+    Hash = xxHash64(OutputFileData);
+
   if (Config->Repro)
-    Timestamp = static_cast<uint32_t>(xxHash64(OutputFileData));
+    Timestamp = static_cast<uint32_t>(Hash);
+
+  if (GenerateSyntheticBuildId) {
+    // For MinGW builds without a PDB file, we still generate a build id
+    // to allow associating a crash dump to the executable.
+    BuildId->BuildId->PDB70.CVSignature = OMF::Signature::PDB70;
+    BuildId->BuildId->PDB70.Age = 1;
+    memcpy(BuildId->BuildId->PDB70.Signature, &Hash, 8);
+    // xxhash only gives us 8 bytes, so put some fixed data in the other half.
+    memcpy(&BuildId->BuildId->PDB70.Signature[8], "LLD PDB.", 8);
+  }
 
   if (DebugDirectory)
     DebugDirectory->setTimeDateStamp(Timestamp);
 
   uint8_t *Buf = Buffer->getBufferStart();
   Buf += DOSStubSize + sizeof(PEMagic);
   object::coff_file_header *CoffHeader =
       reinterpret_cast<coff_file_header *>(Buf);

test/COFF/rsds.test

 # RUN: rm -f %t.dll %t.pdb
 # RUN: lld-link /Brepro /dll /out:%t.dll /entry:DllMain %t.obj
 # RUN: llvm-readobj -coff-debug-directory %t.dll | FileCheck --check-prefix REPRO %s
 
 # RUN: rm -f %t.dll %t.pdb
 # RUN: lld-link /Brepro /debug /dll /out:%t.dll /entry:DllMain %t.obj
 # RUN: llvm-readobj -coff-debug-directory %t.dll | FileCheck --check-prefix REPRODEBUG %s
 
+# RUN: rm -f %t.dll %t.pdb
+# RUN: lld-link /lldmingw /debug:dwarf /dll /out:%t.dll /entry:DllMain %t.obj
+# RUN: llvm-readobj -coff-debug-directory %t.dll | FileCheck --check-prefix MINGW %s
+
 # CHECK: File: [[FILE:.*]].dll
 # CHECK: DebugDirectory [
 # CHECK:   DebugEntry {
 # CHECK:     Characteristics: 0x0
 # CHECK:     TimeDateStamp: 
 # CHECK:     MajorVersion: 0x0
 # CHECK:     MinorVersion: 0x0
 # CHECK:     Type: CodeView (0x2)

 # REPRODEBUG:     MajorVersion: 0x0
 # REPRODEBUG:     MinorVersion: 0x0
 # REPRODEBUG:     Type: Repro (0x10)
 # REPRODEBUG:     SizeOfData: 0x0
 # REPRODEBUG:     AddressOfRawData: 0x0
 # REPRODEBUG:     PointerToRawData: 0x0
 # REPRODEBUG:   }
 # REPRODEBUG: ]
+
+# MINGW: File: {{.*}}.dll
+# MINGW: DebugDirectory [
+# MINGW:   DebugEntry {
+# MINGW:     Characteristics: 0x0
+# MINGW:     TimeDateStamp:
+# MINGW:     MajorVersion: 0x0
+# MINGW:     MinorVersion: 0x0
+# MINGW:     Type: CodeView (0x2)
+# MINGW:     SizeOfData: 0x{{[^0]}}
+# MINGW:     AddressOfRawData: 0x{{[^0]}}
+# MINGW:     PointerToRawData: 0x{{[^0]}}
+# MINGW:     PDBInfo {
+# MINGW:       PDBSignature: 0x53445352
+# MINGW:       PDBGUID: [[GUID:\(([A-Za-z0-9]{2} ?){16}\)]]
+# MINGW:       PDBAge: 1
+# MINGW:       PDBFileName:
+# MINGW:     }
+# MINGW:   }
+# MINGW: ]
 --- !COFF
 header:
   Machine:         IMAGE_FILE_MACHINE_I386
   Characteristics: [  ]
 sections:
   - Name:            .text
     Characteristics: [ IMAGE_SCN_CNT_CODE, IMAGE_SCN_MEM_EXECUTE, IMAGE_SCN_MEM_READ ]
     Alignment:       4