DebugInfo: Support debug_loc under fission

Implement debug_loc.dwo, as well as llvm-dwarfdump support for dumping
this section.

Outlined in the DWARF5 spec and http://gcc.gnu.org/wiki/DebugFission the
debug_loc.dwo section has more variation than the standard debug_loc,
allowing 3 different forms of entry (plus the end of list entry). GCC
seems to, and Clang certainly, only use one form, so I've just
implemented dumping support for that for now.

It wasn't immediately obvious that there was a good refactoring to share
the implementation of dumping support between debug_loc and
debug_loc.dwo, so they're separate for now - ideas welcome or I may come
back to it at some point.

As per a comment in the code, we could choose different forms that may
reduce the number of debug_addr entries we emit, but that will require
further study.

llvm-svn: 204697

llvm/include/llvm/DebugInfo/DIContext.h

@@ -109,6 +109,7 @@ enum DIDumpType {
   DIDT_Line,
   DIDT_LineDwo,
   DIDT_Loc,
+  DIDT_LocDwo,
   DIDT_Ranges,
   DIDT_Pubnames,
   DIDT_Pubtypes,

llvm/include/llvm/Support/Dwarf.h

@@ -771,6 +771,14 @@ enum Constants {
   DW_EH_PE_indirect = 0x80
 };
 
+enum LocationListEntry : unsigned char {
+  DW_LLE_end_of_list_entry,
+  DW_LLE_base_address_selection_entry,
+  DW_LLE_start_end_entry,
+  DW_LLE_start_length_entry,
+  DW_LLE_offset_pair_entry
+};
+
 enum ApplePropertyAttributes {
   // Apple Objective-C Property Attributes
   DW_APPLE_PROPERTY_readonly = 0x01,

llvm/lib/CodeGen/AsmPrinter/DIE.cpp

@@ -559,13 +559,15 @@ unsigned DIELocList::SizeOf(AsmPrinter *AP, dwarf::Form Form) const {
 /// EmitValue - Emit label value.
 ///
 void DIELocList::EmitValue(AsmPrinter *AP, dwarf::Form Form) const {
+  DwarfDebug *DD = AP->getDwarfDebug();
   MCSymbol *Label = AP->GetTempSymbol("debug_loc", Index);
-  MCSymbol *DwarfDebugLocSectionSym = AP->getDwarfDebug()->getDebugLocSym();
 
-  if (AP->MAI->doesDwarfUseRelocationsAcrossSections())
-    AP->EmitSectionOffset(Label, DwarfDebugLocSectionSym);
+  if (DD->useSplitDwarf())
+    AP->EmitLabelDifference(Label, DD->getDebugLocDWOSym(), 4);
+  else if (AP->MAI->doesDwarfUseRelocationsAcrossSections())
+    AP->EmitSectionOffset(Label, DD->getDebugLocSym());
   else
-    AP->EmitLabelDifference(Label, DwarfDebugLocSectionSym, 4);
+    AP->EmitLabelDifference(Label, DD->getDebugLocSym(), 4);
 }
 
 #ifndef NDEBUG

llvm/lib/CodeGen/AsmPrinter/DwarfDebug.cpp

@@ -184,7 +184,7 @@ DwarfDebug::DwarfDebug(AsmPrinter *A, Module *M)
 
   DwarfInfoSectionSym = DwarfAbbrevSectionSym = DwarfStrSectionSym = 0;
   DwarfDebugRangeSectionSym = DwarfDebugLocSectionSym = DwarfLineSectionSym = 0;
-  DwarfAddrSectionSym = 0;
+  DwarfAddrSectionSym = DwarfDebugLocDWOSectionSym = 0;
   DwarfAbbrevDWOSectionSym = DwarfStrDWOSectionSym = 0;
   FunctionBeginSym = FunctionEndSym = 0;
   CurFn = 0;
@@ -1865,7 +1865,6 @@ void DwarfDebug::emitSectionLabels() {
 
   DwarfLineSectionSym =
       emitSectionSym(Asm, TLOF.getDwarfLineSection(), "section_line");
-  emitSectionSym(Asm, TLOF.getDwarfLocSection());
   if (GenerateGnuPubSections) {
     DwarfGnuPubNamesSectionSym =
         emitSectionSym(Asm, TLOF.getDwarfGnuPubNamesSection());
@@ -1883,12 +1882,13 @@ void DwarfDebug::emitSectionLabels() {
         emitSectionSym(Asm, TLOF.getDwarfStrDWOSection(), "skel_string");
     DwarfAddrSectionSym =
         emitSectionSym(Asm, TLOF.getDwarfAddrSection(), "addr_sec");
-  }
+    DwarfDebugLocDWOSectionSym =
+        emitSectionSym(Asm, TLOF.getDwarfLocDWOSection(), "skel_loc");
+  } else
+    DwarfDebugLocSectionSym =
+        emitSectionSym(Asm, TLOF.getDwarfLocSection(), "section_debug_loc");
   DwarfDebugRangeSectionSym =
       emitSectionSym(Asm, TLOF.getDwarfRangesSection(), "debug_range");
-
-  DwarfDebugLocSectionSym =
-      emitSectionSym(Asm, TLOF.getDwarfLocSection(), "section_debug_loc");
 }
 
 // Recursively emits a debug information entry.
@@ -2379,7 +2379,8 @@ void DwarfDebug::emitDebugLocEntry(ByteStreamer &Streamer,
 void DwarfDebug::emitDebugLoc() {
   // Start the dwarf loc section.
   Asm->OutStreamer.SwitchSection(
-      Asm->getObjFileLowering().getDwarfLocSection());
+      useSplitDwarf() ? Asm->getObjFileLowering().getDwarfLocDWOSection()
+                      : Asm->getObjFileLowering().getDwarfLocSection());
   unsigned char Size = Asm->getDataLayout().getPointerSize();
   unsigned index = 0;
   for (const auto &DebugLoc : DotDebugLocEntries) {
@@ -2389,7 +2390,16 @@ void DwarfDebug::emitDebugLoc() {
       // compile unit. This is a hard coded 0 for low_pc when we're emitting
       // ranges, or the DW_AT_low_pc on the compile unit otherwise.
       const DwarfCompileUnit *CU = Entry.getCU();
-      if (CU->getRanges().size() == 1) {
+      if (useSplitDwarf()) {
+        // Just always use start_length for now - at least that's one address
+        // rather than two. We could get fancier and try to, say, reuse an
+        // address we know we've emitted elsewhere (the start of the function?
+        // The start of the CU or CU subrange that encloses this range?)
+        Asm->EmitInt8(dwarf::DW_LLE_start_length_entry);
+        unsigned idx = InfoHolder.getAddrPoolIndex(Entry.getBeginSym());
+        Asm->EmitULEB128(idx);
+        Asm->EmitLabelDifference(Entry.getEndSym(), Entry.getBeginSym(), 4);
+      } else if (CU->getRanges().size() == 1) {
         // Grab the begin symbol from the first range as our base.
         const MCSymbol *Base = CU->getRanges()[0].getStart();
         Asm->EmitLabelDifference(Entry.getBeginSym(), Base, Size);
@@ -2409,8 +2419,12 @@ void DwarfDebug::emitDebugLoc() {
       // Close the range.
       Asm->OutStreamer.EmitLabel(end);
     }
-    Asm->OutStreamer.EmitIntValue(0, Size);
-    Asm->OutStreamer.EmitIntValue(0, Size);
+    if (useSplitDwarf())
+      Asm->EmitInt8(dwarf::DW_LLE_end_of_list_entry);
+    else {
+      Asm->OutStreamer.EmitIntValue(0, Size);
+      Asm->OutStreamer.EmitIntValue(0, Size);
+    }
     ++index;
   }
 }

llvm/lib/CodeGen/AsmPrinter/DwarfDebug.h

@@ -423,7 +423,7 @@ class DwarfDebug : public AsmPrinterHandler {
   MCSymbol *DwarfDebugLocSectionSym, *DwarfLineSectionSym, *DwarfAddrSectionSym;
   MCSymbol *FunctionBeginSym, *FunctionEndSym;
   MCSymbol *DwarfInfoDWOSectionSym, *DwarfAbbrevDWOSectionSym;
-  MCSymbol *DwarfStrDWOSectionSym;
+  MCSymbol *DwarfStrDWOSectionSym, *DwarfDebugLocDWOSectionSym;
   MCSymbol *DwarfGnuPubNamesSectionSym, *DwarfGnuPubTypesSectionSym;
 
   // As an optimization, there is no need to emit an entry in the directory
@@ -756,6 +756,9 @@ public:
   /// Returns the section symbol for the .debug_loc section.
   MCSymbol *getDebugLocSym() const { return DwarfDebugLocSectionSym; }
 
+  /// Returns the section symbol for the .debug_loc section.
+  MCSymbol *getDebugLocDWOSym() const { return DwarfDebugLocDWOSectionSym; }
+
   /// Returns the previous section that was emitted into.
   const MCSection *getPrevSection() const { return PrevSection; }
 

llvm/lib/DebugInfo/DWARFContext.cpp

@@ -96,6 +96,11 @@ void DWARFContext::dump(raw_ostream &OS, DIDumpType DumpType) {
     getDebugLoc()->dump(OS);
   }
 
+  if (DumpType == DIDT_All || DumpType == DIDT_LocDwo) {
+    OS << "\n.debug_loc.dwo contents:\n";
+    getDebugLocDWO()->dump(OS);
+  }
+
   if (DumpType == DIDT_All || DumpType == DIDT_Frames) {
     OS << "\n.debug_frame contents:\n";
     getDebugFrame()->dump(OS);
@@ -237,6 +242,16 @@ const DWARFDebugLoc *DWARFContext::getDebugLoc() {
   return Loc.get();
 }
 
+const DWARFDebugLocDWO *DWARFContext::getDebugLocDWO() {
+  if (LocDWO)
+    return LocDWO.get();
+
+  DataExtractor LocData(getLocDWOSection().Data, isLittleEndian(), 0);
+  LocDWO.reset(new DWARFDebugLocDWO());
+  LocDWO->parse(LocData);
+  return LocDWO.get();
+}
+
 const DWARFDebugAranges *DWARFContext::getDebugAranges() {
   if (Aranges)
     return Aranges.get();
@@ -648,6 +663,7 @@ DWARFContextInMemory::DWARFContextInMemory(object::ObjectFile *Obj)
             .Case("debug_gnu_pubtypes", &GnuPubTypesSection)
             .Case("debug_info.dwo", &InfoDWOSection.Data)
             .Case("debug_abbrev.dwo", &AbbrevDWOSection)
+            .Case("debug_loc.dwo", &LocDWOSection.Data)
             .Case("debug_line.dwo", &LineDWOSection.Data)
             .Case("debug_str.dwo", &StringDWOSection)
             .Case("debug_str_offsets.dwo", &StringOffsetDWOSection)

llvm/lib/DebugInfo/DWARFContext.h

@@ -42,6 +42,7 @@ class DWARFContext : public DIContext {
   CUVector DWOCUs;
   TUVector DWOTUs;
   std::unique_ptr<DWARFDebugAbbrev> AbbrevDWO;
+  std::unique_ptr<DWARFDebugLocDWO> LocDWO;
 
   DWARFContext(DWARFContext &) LLVM_DELETED_FUNCTION;
   DWARFContext &operator=(DWARFContext &) LLVM_DELETED_FUNCTION;
@@ -148,6 +149,9 @@ public:
   /// Get a pointer to the parsed dwo abbreviations object.
   const DWARFDebugAbbrev *getDebugAbbrevDWO();
 
+  /// Get a pointer to the parsed DebugLoc object.
+  const DWARFDebugLocDWO *getDebugLocDWO();
+
   /// Get a pointer to the parsed DebugAranges object.
   const DWARFDebugAranges *getDebugAranges();
 
@@ -173,6 +177,7 @@ public:
   virtual const TypeSectionMap &getTypesSections() = 0;
   virtual StringRef getAbbrevSection() = 0;
   virtual const Section &getLocSection() = 0;
+  virtual const Section &getLocDWOSection() = 0;
   virtual StringRef getARangeSection() = 0;
   virtual StringRef getDebugFrameSection() = 0;
   virtual const Section &getLineSection() = 0;
@@ -216,6 +221,7 @@ class DWARFContextInMemory : public DWARFContext {
   TypeSectionMap TypesSections;
   StringRef AbbrevSection;
   Section LocSection;
+  Section LocDWOSection;
   StringRef ARangeSection;
   StringRef DebugFrameSection;
   Section LineSection;
@@ -246,6 +252,7 @@ public:
   const TypeSectionMap &getTypesSections() override { return TypesSections; }
   StringRef getAbbrevSection() override { return AbbrevSection; }
   const Section &getLocSection() override { return LocSection; }
+  const Section &getLocDWOSection() override { return LocDWOSection; }
   StringRef getARangeSection() override { return ARangeSection; }
   StringRef getDebugFrameSection() override { return DebugFrameSection; }
   const Section &getLineSection() override { return LineSection; }

llvm/lib/DebugInfo/DWARFDebugLoc.cpp

@@ -11,6 +11,7 @@
 #include "llvm/Support/Compiler.h"
 #include "llvm/Support/Format.h"
 #include "llvm/Support/raw_ostream.h"
+#include "llvm/Support/Dwarf.h"
 
 using namespace llvm;
 
@@ -74,3 +75,54 @@ void DWARFDebugLoc::parse(DataExtractor data, unsigned AddressSize) {
   if (data.isValidOffset(Offset))
     llvm::errs() << "error: failed to consume entire .debug_loc section\n";
 }
+
+void DWARFDebugLocDWO::parse(DataExtractor data) {
+  uint32_t Offset = 0;
+  while (data.isValidOffset(Offset)) {
+    Locations.resize(Locations.size() + 1);
+    LocationList &Loc = Locations.back();
+    Loc.Offset = Offset;
+    dwarf::LocationListEntry Kind;
+    while ((Kind = static_cast<dwarf::LocationListEntry>(
+                data.getU8(&Offset))) != dwarf::DW_LLE_end_of_list_entry) {
+
+      if (Kind != dwarf::DW_LLE_start_length_entry) {
+        llvm::errs() << "error: dumping support for LLE of kind " << (int)Kind
+                     << " not implemented\n";
+        return;
+      }
+
+      Entry E;
+
+      E.Start = data.getULEB128(&Offset);
+      E.Length = data.getU32(&Offset);
+
+      unsigned Bytes = data.getU16(&Offset);
+      // A single location description describing the location of the object...
+      StringRef str = data.getData().substr(Offset, Bytes);
+      Offset += Bytes;
+      E.Loc.resize(str.size());
+      std::copy(str.begin(), str.end(), E.Loc.begin());
+
+      Loc.Entries.push_back(std::move(E));
+    }
+  }
+}
+
+void DWARFDebugLocDWO::dump(raw_ostream &OS) const {
+  for (const LocationList &L : Locations) {
+    OS << format("0x%8.8x: ", L.Offset);
+    const unsigned Indent = 12;
+    for (const Entry &E : L.Entries) {
+      if (&E != L.Entries.begin())
+        OS.indent(Indent);
+      OS << "Beginning address index: " << E.Start << '\n';
+      OS.indent(Indent) << "                 Length: " << E.Length << '\n';
+      OS.indent(Indent) << "   Location description: ";
+      for (unsigned char Loc : E.Loc)
+        OS << format("%2.2x ", Loc);
+      OS << "\n\n";
+    }
+  }
+}
+

llvm/lib/DebugInfo/DWARFDebugLoc.h

@@ -55,6 +55,27 @@ public:
   /// specified address size to interpret the address ranges.
   void parse(DataExtractor data, unsigned AddressSize);
 };
+
+class DWARFDebugLocDWO {
+  struct Entry {
+    uint64_t Start;
+    uint32_t Length;
+    SmallVector<unsigned char, 4> Loc;
+  };
+
+  struct LocationList {
+    unsigned Offset;
+    SmallVector<Entry, 2> Entries;
+  };
+
+  typedef SmallVector<LocationList, 4> LocationLists;
+
+  LocationLists Locations;
+
+public:
+  void parse(DataExtractor data);
+  void dump(raw_ostream &OS) const;
+};
 }
 
 #endif

llvm/test/DebugInfo/X86/fission-ranges.ll

@@ -1,14 +1,39 @@
 ; RUN: llc -split-dwarf=Enable -O0 %s -mtriple=x86_64-unknown-linux-gnu -filetype=obj -o %t
-; RUN: llvm-dwarfdump -debug-dump=info.dwo %t | FileCheck %s
-; RUN: llvm-objdump -r %t | FileCheck --check-prefix=RELA %s
+; RUN: llvm-dwarfdump %t | FileCheck %s
+; RUN: llvm-objdump -h %t | FileCheck --check-prefix=HDR %s
 
+; CHECK: .debug_info.dwo contents:
+; CHECK: DW_AT_location [DW_FORM_sec_offset]   ([[A:0x[0-9a-z]*]])
+; CHECK: DW_AT_location [DW_FORM_sec_offset]   ([[E:0x[0-9a-z]*]])
+; CHECK: DW_AT_location [DW_FORM_sec_offset]   ([[B:0x[0-9a-z]*]])
+; CHECK: DW_AT_location [DW_FORM_sec_offset]   ([[D:0x[0-9a-z]*]])
 ; CHECK: DW_AT_ranges [DW_FORM_sec_offset]   (0x000000a0)
+; CHECK: .debug_loc contents:
+; CHECK-NOT: Beginning address offset
+; CHECK: .debug_loc.dwo contents:
+
+; Don't assume these locations are entirely correct - feel free to update them
+; if they've changed due to a bugfix, change in register allocation, etc.
+
+; CHECK: [[A]]: Beginning address index: 2
+; CHECK-NEXT:                    Length: 199
+; CHECK-NEXT:      Location description: 10 00
+; CHECK-NEXT: {{^$}}
+; CHECK-NEXT:   Beginning address index: 3
+; CHECK-NEXT:                    Length: 23
+; CHECK-NEXT:      Location description: 50 93 04
+; CHECK: [[E]]: Beginning address index: 4
+; CHECK-NEXT:                    Length: 21
+; CHECK-NEXT:      Location description: 50 93 04
+; CHECK: [[B]]: Beginning address index: 5
+; CHECK-NEXT:                    Length: 19
+; CHECK-NEXT:      Location description: 50 93 04
+; CHECK: [[D]]: Beginning address index: 6
+; CHECK-NEXT:                    Length: 23
+; CHECK-NEXT:      Location description: 50 93 04
 
 ; Make sure we don't produce any relocations in any .dwo section (though in particular, debug_info.dwo)
-; FIXME: There should be no relocations in .dwo sections at all, but for now there are debug_loc relocs here.
-; RELA: RELOCATION RECORDS FOR [.rela.debug_info.dwo]
-; RELA-NOT: R_X86_64_32 .debug_ranges
-; RELA: RELOCATION RECORDS FOR
+; HDR-NOT: .rela.{{.*}}.dwo
 
 ; From the code:
 

llvm/tools/llvm-dwarfdump/llvm-dwarfdump.cpp

@@ -66,6 +66,7 @@ DumpType("debug-dump", cl::init(DIDT_All),
         clEnumValN(DIDT_Line, "line", ".debug_line"),
         clEnumValN(DIDT_LineDwo, "line.dwo", ".debug_line.dwo"),
         clEnumValN(DIDT_Loc, "loc", ".debug_loc"),
+        clEnumValN(DIDT_LocDwo, "loc.dwo", ".debug_loc.dwo"),
         clEnumValN(DIDT_Frames, "frames", ".debug_frame"),
         clEnumValN(DIDT_Ranges, "ranges", ".debug_ranges"),
         clEnumValN(DIDT_Pubnames, "pubnames", ".debug_pubnames"),