[LLVM][DWARF] Add support for .debug_names with split dwarf #73872
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
@llvm/pr-subscribers-debuginfo Author: Alexander Yermolovich (ayermolo) ChangesEnables Type Units with DWARF5 accelerator tables for split dwarf. It is still Full diff: https://github.com/llvm/llvm-project/pull/73872.diff 4 Files Affected:
diff --git a/llvm/include/llvm/CodeGen/AccelTable.h b/llvm/include/llvm/CodeGen/AccelTable.h
index 6a69a01a8c786c4..0f35fd3514fae74 100644
--- a/llvm/include/llvm/CodeGen/AccelTable.h
+++ b/llvm/include/llvm/CodeGen/AccelTable.h
@@ -297,15 +297,15 @@ class DWARF5AccelTableData : public AccelTableData {
};
struct TypeUnitMetaInfo {
- // Symbol for start of the TU section.
- MCSymbol *Label;
+ // Symbol for start of the TU section or signature if this is SplitDwarf.
+ std::variant<MCSymbol *, uint64_t> LabelOrSignature;
// Unique ID of Type Unit.
unsigned UniqueID;
};
using TUVectorTy = SmallVector<TypeUnitMetaInfo, 1>;
class DWARF5AccelTable : public AccelTable<DWARF5AccelTableData> {
// Symbols to start of all the TU sections that were generated.
- TUVectorTy TUSymbols;
+ TUVectorTy TUSymbolsOrHashes;
public:
struct UnitIndexAndEncoding {
@@ -313,9 +313,11 @@ class DWARF5AccelTable : public AccelTable<DWARF5AccelTableData> {
DWARF5AccelTableData::AttributeEncoding Endoding;
};
/// Returns type units that were constructed.
- const TUVectorTy &getTypeUnitsSymbols() { return TUSymbols; }
+ const TUVectorTy &getTypeUnitsSymbols() { return TUSymbolsOrHashes; }
/// Add a type unit start symbol.
void addTypeUnitSymbol(DwarfTypeUnit &U);
+ /// Add a type unit Signature.
+ void addTypeUnitSignature(DwarfTypeUnit &U);
/// Convert DIE entries to explicit offset.
/// Needs to be called after DIE offsets are computed.
void convertDieToOffset() {
diff --git a/llvm/lib/CodeGen/AsmPrinter/AccelTable.cpp b/llvm/lib/CodeGen/AsmPrinter/AccelTable.cpp
index 60c54687c96662d..d6f487c18b0301e 100644
--- a/llvm/lib/CodeGen/AsmPrinter/AccelTable.cpp
+++ b/llvm/lib/CodeGen/AsmPrinter/AccelTable.cpp
@@ -201,9 +201,11 @@ class Dwarf5AccelTableWriter : public AccelTableWriter {
char AugmentationString[8] = {'L', 'L', 'V', 'M', '0', '7', '0', '0'};
Header(uint32_t CompUnitCount, uint32_t LocalTypeUnitCount,
- uint32_t BucketCount, uint32_t NameCount)
+ uint32_t ForeignTypeUnitCount, uint32_t BucketCount,
+ uint32_t NameCount)
: CompUnitCount(CompUnitCount), LocalTypeUnitCount(LocalTypeUnitCount),
- BucketCount(BucketCount), NameCount(NameCount) {}
+ ForeignTypeUnitCount(ForeignTypeUnitCount), BucketCount(BucketCount),
+ NameCount(NameCount) {}
void emit(Dwarf5AccelTableWriter &Ctx);
};
@@ -220,6 +222,8 @@ class Dwarf5AccelTableWriter : public AccelTableWriter {
MCSymbol *AbbrevStart = Asm->createTempSymbol("names_abbrev_start");
MCSymbol *AbbrevEnd = Asm->createTempSymbol("names_abbrev_end");
MCSymbol *EntryPool = Asm->createTempSymbol("names_entries");
+ // Indicates if this module is built with Split Dwarf enabled.
+ bool IsSplitDwarf = false;
void populateAbbrevsMap();
@@ -238,7 +242,8 @@ class Dwarf5AccelTableWriter : public AccelTableWriter {
ArrayRef<std::variant<MCSymbol *, uint64_t>> TypeUnits,
llvm::function_ref<
std::optional<DWARF5AccelTable::UnitIndexAndEncoding>(const DataT &)>
- getIndexForEntry);
+ getIndexForEntry,
+ bool IsSplitDwarf);
void emit();
};
@@ -450,6 +455,8 @@ void Dwarf5AccelTableWriter<DataT>::emitTUList() const {
Asm->OutStreamer->AddComment("Type unit " + Twine(TU.index()));
if (std::holds_alternative<MCSymbol *>(TU.value()))
Asm->emitDwarfSymbolReference(std::get<MCSymbol *>(TU.value()));
+ else if (IsSplitDwarf)
+ Asm->emitInt64(std::get<uint64_t>(TU.value()));
else
Asm->emitDwarfLengthOrOffset(std::get<uint64_t>(TU.value()));
}
@@ -551,12 +558,15 @@ Dwarf5AccelTableWriter<DataT>::Dwarf5AccelTableWriter(
ArrayRef<std::variant<MCSymbol *, uint64_t>> TypeUnits,
llvm::function_ref<
std::optional<DWARF5AccelTable::UnitIndexAndEncoding>(const DataT &)>
- getIndexForEntry)
+ getIndexForEntry,
+ bool IsSplitDwarf)
: AccelTableWriter(Asm, Contents, false),
- Header(CompUnits.size(), TypeUnits.size(), Contents.getBucketCount(),
+ Header(CompUnits.size(), IsSplitDwarf ? 0 : TypeUnits.size(),
+ IsSplitDwarf ? TypeUnits.size() : 0, Contents.getBucketCount(),
Contents.getUniqueNameCount()),
CompUnits(CompUnits), TypeUnits(TypeUnits),
- getIndexForEntry(std::move(getIndexForEntry)) {
+ getIndexForEntry(std::move(getIndexForEntry)),
+ IsSplitDwarf(IsSplitDwarf) {
populateAbbrevsMap();
}
@@ -608,7 +618,10 @@ void llvm::emitDWARF5AccelTable(
for (const auto &TU : TUSymbols) {
TUIndex[TU.UniqueID] = TUCount++;
- TypeUnits.push_back(TU.Label);
+ if (DD.useSplitDwarf())
+ TypeUnits.push_back(std::get<uint64_t>(TU.LabelOrSignature));
+ else
+ TypeUnits.push_back(std::get<MCSymbol *>(TU.LabelOrSignature));
}
if (CompUnits.empty())
@@ -633,12 +646,17 @@ void llvm::emitDWARF5AccelTable(
return {{CUIndex[Entry.getUnitID()],
{dwarf::DW_IDX_compile_unit, CUIndexForm}}};
return std::nullopt;
- })
+ },
+ DD.useSplitDwarf())
.emit();
}
void DWARF5AccelTable::addTypeUnitSymbol(DwarfTypeUnit &U) {
- TUSymbols.push_back({U.getLabelBegin(), U.getUniqueID()});
+ TUSymbolsOrHashes.push_back({U.getLabelBegin(), U.getUniqueID()});
+}
+
+void DWARF5AccelTable::addTypeUnitSignature(DwarfTypeUnit &U) {
+ TUSymbolsOrHashes.push_back({U.getTypeSignature(), U.getUniqueID()});
}
void llvm::emitDWARF5AccelTable(
@@ -650,7 +668,7 @@ void llvm::emitDWARF5AccelTable(
std::vector<std::variant<MCSymbol *, uint64_t>> TypeUnits;
Contents.finalize(Asm, "names");
Dwarf5AccelTableWriter<DWARF5AccelTableData>(Asm, Contents, CUs, TypeUnits,
- getIndexForEntry)
+ getIndexForEntry, false)
.emit();
}
diff --git a/llvm/lib/CodeGen/AsmPrinter/DwarfDebug.cpp b/llvm/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
index f311931a41aa5df..ab29020bf1d7bb0 100644
--- a/llvm/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
+++ b/llvm/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
@@ -305,7 +305,6 @@ void Loc::MMI::addFrameIndexExpr(const DIExpression *Expr, int FI) {
static AccelTableKind computeAccelTableKind(unsigned DwarfVersion,
bool GenerateTypeUnits,
- bool HasSplitDwarf,
DebuggerKind Tuning,
const Triple &TT) {
// Honor an explicit request.
@@ -314,8 +313,7 @@ static AccelTableKind computeAccelTableKind(unsigned DwarfVersion,
// Generating DWARF5 acceleration table.
// Currently Split dwarf and non ELF format is not supported.
- if (GenerateTypeUnits &&
- (DwarfVersion < 5 || HasSplitDwarf || !TT.isOSBinFormatELF()))
+ if (GenerateTypeUnits && (DwarfVersion < 5 || !TT.isOSBinFormatELF()))
return AccelTableKind::None;
// Accelerator tables get emitted if targetting DWARF v5 or LLDB. DWARF v5
@@ -403,9 +401,8 @@ DwarfDebug::DwarfDebug(AsmPrinter *A)
A->TM.getTargetTriple().isOSBinFormatWasm()) &&
GenerateDwarfTypeUnits;
- TheAccelTableKind =
- computeAccelTableKind(DwarfVersion, GenerateTypeUnits, HasSplitDwarf,
- DebuggerTuning, A->TM.getTargetTriple());
+ TheAccelTableKind = computeAccelTableKind(
+ DwarfVersion, GenerateTypeUnits, DebuggerTuning, A->TM.getTargetTriple());
// Work around a GDB bug. GDB doesn't support the standard opcode;
// SCE doesn't support GNU's; LLDB prefers the standard opcode, which
@@ -3532,8 +3529,12 @@ void DwarfDebug::addDwarfTypeUnitType(DwarfCompileUnit &CU,
InfoHolder.computeSizeAndOffsetsForUnit(TU.first.get());
InfoHolder.emitUnit(TU.first.get(), useSplitDwarf());
if (getDwarfVersion() >= 5 &&
- getAccelTableKind() == AccelTableKind::Dwarf)
- AccelDebugNames.addTypeUnitSymbol(*TU.first);
+ getAccelTableKind() == AccelTableKind::Dwarf) {
+ if (useSplitDwarf())
+ AccelDebugNames.addTypeUnitSignature(*TU.first);
+ else
+ AccelDebugNames.addTypeUnitSymbol(*TU.first);
+ }
}
AccelTypeUnitsDebugNames.convertDieToOffset();
AccelDebugNames.addTypeEntries(AccelTypeUnitsDebugNames);
diff --git a/llvm/test/DebugInfo/X86/debug-names-types.ll b/llvm/test/DebugInfo/X86/debug-names-types.ll
index fc23e604b386365..d32691305d1c101 100644
--- a/llvm/test/DebugInfo/X86/debug-names-types.ll
+++ b/llvm/test/DebugInfo/X86/debug-names-types.ll
@@ -1,15 +1,10 @@
; UNSUPPORTED: system-windows
-; This checks that .debug_names can be generated with monolithic -fdebug-type-sections, and does not generate when split-dwarf is enabled.
+; This checks that .debug_names can be generated with monolithic, and split-dwarf, when -fdebug-type-sections is enabled.
; Generated with: clang++ main.cpp -g2 -gdwarf-5 -gpubnames -fdebug-types-section
; RUN: llc -mtriple=x86_64 -generate-type-units -dwarf-version=5 -filetype=obj %s -o %t
; RUN: llvm-dwarfdump -debug-info -debug-names %t | FileCheck %s
-; RUN: llc -mtriple=x86_64 -generate-type-units -dwarf-version=5 -filetype=obj -split-dwarf-file=%t.mainTypes.dwo --split-dwarf-output=%t.mainTypes.dwo %s -o %t
-; RUN: llvm-readelf --sections %t | FileCheck %s --check-prefixes=CHECK-SPLIT
-
-; CHECK-SPLIT-NOT: .debug_names
-
; CHECK: .debug_info contents:
; CHECK: DW_TAG_type_unit
; CHECK-NEXT: DW_AT_language (DW_LANG_C_plus_plus_14)
@@ -117,6 +112,99 @@
; CHECK-NEXT: ]
; CHECK-NEXT: }
+; RUN: llc -mtriple=x86_64 -generate-type-units -dwarf-version=5 -filetype=obj -split-dwarf-file=%t.mainTypes.dwo --split-dwarf-output=%t.mainTypes.dwo %s -o %t
+; RUN: llvm-dwarfdump -debug-names %t | FileCheck %s --check-prefixes=CHECK-SPLIT
+
+; CHECK-SPLIT: .debug_names contents
+; CHECK-SPLIT: Foreign TU count: 1
+; CHECK-SPLIT-NEXT: Bucket count: 4
+; CHECK-SPLIT-NEXT: Name count: 4
+; CHECK-SPLIT-NEXT: Abbreviations table size: 0x28
+; CHECK-SPLIT-NEXT: Augmentation: 'LLVM0700'
+; CHECK-SPLIT-NEXT: }
+; CHECK-SPLIT-NEXT: Compilation Unit offsets [
+; CHECK-SPLIT-NEXT: CU[0]: 0x00000000
+; CHECK-SPLIT-NEXT: ]
+; CHECK-SPLIT-NEXT: Foreign Type Unit signatures [
+; CHECK-SPLIT-NEXT: ForeignTU[0]: 0x675d23e4f33235f2
+; CHECK-SPLIT-NEXT: ]
+; CHECK-SPLIT-NEXT: Abbreviations [
+; CHECK-SPLIT-NEXT: Abbreviation [[ABBREV:0x[0-9a-f]*]] {
+; CHECK-SPLIT-NEXT: Tag: DW_TAG_structure_type
+; CHECK-SPLIT-NEXT: DW_IDX_die_offset: DW_FORM_ref4
+; CHECK-SPLIT-NEXT: }
+; CHECK-SPLIT-NEXT: Abbreviation [[ABBREV1:0x[0-9a-f]*]] {
+; CHECK-SPLIT-NEXT: Tag: DW_TAG_structure_type
+; CHECK-SPLIT-NEXT: DW_IDX_type_unit: DW_FORM_data1
+; CHECK-SPLIT-NEXT: DW_IDX_die_offset: DW_FORM_ref4
+; CHECK-SPLIT-NEXT: }
+; CHECK-SPLIT-NEXT: Abbreviation [[ABBREV2:0x[0-9a-f]*]] {
+; CHECK-SPLIT-NEXT: Tag: DW_TAG_base_type
+; CHECK-SPLIT-NEXT: DW_IDX_die_offset: DW_FORM_ref4
+; CHECK-SPLIT-NEXT: }
+; CHECK-SPLIT-NEXT: Abbreviation [[ABBREV3:0x[0-9a-f]*]] {
+; CHECK-SPLIT-NEXT: Tag: DW_TAG_subprogram
+; CHECK-SPLIT-NEXT: DW_IDX_die_offset: DW_FORM_ref4
+; CHECK-SPLIT-NEXT: }
+; CHECK-SPLIT-NEXT: Abbreviation [[ABBREV4:0x[0-9a-f]*]] {
+; CHECK-SPLIT-NEXT: Tag: DW_TAG_base_type
+; CHECK-SPLIT-NEXT: DW_IDX_type_unit: DW_FORM_data1
+; CHECK-SPLIT-NEXT: DW_IDX_die_offset: DW_FORM_ref4
+; CHECK-SPLIT-NEXT: }
+; CHECK-SPLIT-NEXT: ]
+; CHECK-SPLIT-NEXT: Bucket 0 [
+; CHECK-SPLIT-NEXT: Name 1 {
+; CHECK-SPLIT-NEXT: Hash: 0xB888030
+; CHECK-SPLIT-NEXT: String: {{.+}} "int"
+; CHECK-SPLIT-NEXT: Entry @ {{.+}} {
+; CHECK-SPLIT-NEXT: Abbrev: [[ABBREV2]]
+; CHECK-SPLIT-NEXT: Tag: DW_TAG_base_type
+; CHECK-SPLIT-NEXT: DW_IDX_die_offset: 0x00000035
+; CHECK-SPLIT-NEXT: }
+; CHECK-SPLIT-NEXT: }
+; CHECK-SPLIT-NEXT: ]
+; CHECK-SPLIT-NEXT: Bucket 1 [
+; CHECK-SPLIT-NEXT: Name 2 {
+; CHECK-SPLIT-NEXT: Hash: 0xB887389
+; CHECK-SPLIT-NEXT: String: {{.+}} "Foo"
+; CHECK-SPLIT-NEXT: Entry @ {{.+}} {
+; CHECK-SPLIT-NEXT: Abbrev: [[ABBREV1]]
+; CHECK-SPLIT-NEXT: Tag: DW_TAG_structure_type
+; CHECK-SPLIT-NEXT: DW_IDX_type_unit: 0x00
+; CHECK-SPLIT-NEXT: DW_IDX_die_offset: 0x0000001f
+; CHECK-SPLIT-NEXT: }
+; CHECK-SPLIT-NEXT: Entry @ 0xae {
+; CHECK-SPLIT-NEXT: Abbrev: [[ABBREV]]
+; CHECK-SPLIT-NEXT: Tag: DW_TAG_structure_type
+; CHECK-SPLIT-NEXT: DW_IDX_die_offset: 0x00000039
+; CHECK-SPLIT-NEXT: }
+; CHECK-SPLIT-NEXT: }
+; CHECK-SPLIT-NEXT: ]
+; CHECK-SPLIT-NEXT: Bucket 2 [
+; CHECK-SPLIT-NEXT: Name 3 {
+; CHECK-SPLIT-NEXT: Hash: 0x7C9A7F6A
+; CHECK-SPLIT-NEXT: String: {{.+}} "main"
+; CHECK-SPLIT-NEXT: Entry @ {{.+}} {
+; CHECK-SPLIT-NEXT: Abbrev: [[ABBREV3]]
+; CHECK-SPLIT-NEXT: Tag: DW_TAG_subprogram
+; CHECK-SPLIT-NEXT: DW_IDX_die_offset: 0x0000001a
+; CHECK-SPLIT-NEXT: }
+; CHECK-SPLIT-NEXT: }
+; CHECK-SPLIT-NEXT: ]
+; CHECK-SPLIT-NEXT: Bucket 3 [
+; CHECK-SPLIT-NEXT: Name 4 {
+; CHECK-SPLIT-NEXT: Hash: 0x7C952063
+; CHECK-SPLIT-NEXT: String: {{.+}} "char"
+; CHECK-SPLIT-NEXT: Entry @ {{.+}} {
+; CHECK-SPLIT-NEXT: Abbrev: [[ABBREV4]]
+; CHECK-SPLIT-NEXT: Tag: DW_TAG_base_type
+; CHECK-SPLIT-NEXT: DW_IDX_type_unit: 0x00
+; CHECK-SPLIT-NEXT: DW_IDX_die_offset: 0x00000034
+; CHECK-SPLIT-NEXT: }
+; CHECK-SPLIT-NEXT: }
+; CHECK-SPLIT-NEXT: ]
+; CHECK-SPLIT-NEXT: }
+
; ModuleID = 'main.cpp'
source_filename = "main.cpp"
|
ayermolo
force-pushed
the
debugNamesSplitDwarf
branch
2 times, most recently
from
366f614
to
d26811e
Compare
Enables Type Units with DWARF5 accelerator tables for split dwarf. It is still under discussion what is the best way to implement support for de-duplication in DWP. This will be in follow up PR.
ayermolo
force-pushed
the
debugNamesSplitDwarf
branch
from
d26811e
to
d5170f5
Compare
|
@dwblaikie ping :) |
dwblaikie
approved these changes
Dec 4, 2023
There was a problem hiding this comment.
Looks good to me.
The DWARF spec I think is a smidge vague on how the IDX_type_unit works with the local and foreign type units. (like if you had 5 local and 5 foreign type units (won't happen for us with object files, but would happen with the linked index) - presumably an IDX_type_unit value of 5 means the first foreign type unit? and 4 means the last local one?))
(also, total aside: be interesting to measure the size of .debug_names relative to .debug_gnu_pubnames and consider including a flag for producing .debug_names without the hash table in object files (if it's going to be merged into a unified index by the linker anyway, we don't need the hash table in the object file, probably))
Yes. I am currently working on support for it in BOLT which have similar output as with "smart linking" since at that point binary is fully linked. |
Ah, great! Thanks for the chapter-and-verse!
Because the hash table is there for fast lookup - but if you're merging tables, you don't need fast lookup/you can walk the entries linearly when building a merged table. So far as I understand/can think of. @cmtice - looking at |
felipepiovezan
pushed a commit
to felipepiovezan/llvm-project
that referenced
this pull request
Feb 2, 2024
Enables Type Units with DWARF5 accelerator tables for split dwarf. It is still under discussion what is the best way to implement support for de-duplication in DWP. This will be in follow up PR. (cherry picked from commit e8f3ccd)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Enables Type Units with DWARF5 accelerator tables for split dwarf. It is still
under discussion what is the best way to implement support for de-duplication in
DWP. This will be in follow up PR.