[DebugInfo] Parse StrOffsets section if needed

[fdeazeve]

Summary

Querying the debug_str_offsets section requires parsing the top level DIE of the
CU (as well as the section itself); the current getter, however, assumes this is
done elsewhere. This patch changes the getter behavior to match what is done in
other getter methods (e.g. getCompilationDir or getVariableForAddress), in
other words, extractDIEsIfNeeded is now called prior to returning the
debug_str_offsets contributions for the Unit.

One way in which this bug manifested is when dwarfdump --debug-str-offsets is
invoked: because the DIEs are never parsed, we incorrectly print an empty
section (with no warnings or errors).

[fdeazeve]

Fix incorrect comment in commit message.

[probinson]

Please use GitHub for new patches.

[aprantl]

llvm/include/llvm/DebugInfo/DWARF/DWARFUnit.h

404

CUDIEOnly or CUDieOnly

[fdeazeve]

In D159484#4641955, @probinson wrote:

Please use GitHub for new patches.

Apologies, I had in my mind that September was the in-between month where both were allowed. Just re-read the discourse thread and I am indeed wrong

llvm/include/llvm/DebugInfo/DWARF/DWARFUnit.h

public:
   /// .debug_ranges section. If the extraction is unsuccessful, an error
   /// is returned. Successful extraction requires that the compile unit
   /// has already been extracted.
   Error extractRangeList(uint64_t RangeListOffset,
                          DWARFDebugRangeList &RangeList) const;
   void clear();
 
   const std::optional<StrOffsetsContributionDescriptor> &
-  getStringOffsetsTableContribution() const {
+  getStringOffsetsTableContribution() {
+    extractDIEsIfNeeded(true /*CUDIeOnly*/);
     return StringOffsetsTableContribution;
   }
 
   uint8_t getDwarfStringOffsetsByteSize() const {
     assert(StringOffsetsTableContribution);
     return StringOffsetsTableContribution->getDwarfOffsetByteSize();
   }
 

llvm/test/DebugInfo/X86/dwarfdump-str-offsets.s

 # RUN: llvm-mc -triple x86_64-unknown-linux %s -filetype=obj -o %t.o
-# RUN: llvm-dwarfdump -v %t.o 2> %t.err | FileCheck --check-prefix=COMMON --check-prefix=SPLIT %s
+# RUN: llvm-dwarfdump -v %t.o 2> %t.err | FileCheck --check-prefixes=COMMON,SPLIT,OFFSETS %s
 # RUN: llvm-dwarfdump -verify %t.o | FileCheck --check-prefix=VERIFY %s
+# RUN: llvm-dwarfdump -debug-str-offsets %t.o | FileCheck --check-prefix=OFFSETS %s
 # 
 # Check that we don't report an error on a non-existent range list table.
 # RUN: FileCheck -allow-empty --check-prefix ERR %s < %t.err
 
 # Test object to verify dwarfdump handles v5 string offset tables.
 # We have 2 v5 CUs, a v5 TU, and a split v5 CU and TU.
 #
 

 # The split type unit
 # SPLIT:       .debug_types.dwo contents:
 # SPLIT:       DW_TAG_type_unit
 # SPLIT-NEXT:  DW_AT_name [DW_FORM_strx] (indexed (00000003) string = "V5_split_type_unit")
 # SPLIT:       DW_TAG_structure_type
 # SPLIT-NEXT:  DW_AT_name [DW_FORM_strx] (indexed (00000004) string = "V5_split_Mystruct")
 # 
 # The .debug_str_offsets section
-# COMMON:      .debug_str_offsets contents:
-# COMMON-NEXT: 0x00000000: Contribution size = 32, Format = DWARF32, Version = 5
-# COMMON-NEXT: 0x00000008: 00000000 "Handmade DWARF producer"
-# COMMON-NEXT: 0x0000000c: 00000018 "Compile_Unit_1"
-# COMMON-NEXT: 0x00000010: 00000027 "/home/test/CU1"
-# COMMON-NEXT: 0x00000014: 00000067 "MyFunc"
-# COMMON-NEXT: 0x00000018: 0000006e "MyVar1"
-# COMMON-NEXT: 0x0000001c: 00000075 "MyVar2"
-# COMMON-NEXT: 0x00000020: 0000007c "MyVar3"
-# COMMON-NEXT: Gap, length = 4
-# COMMON-NEXT: 0x00000028: Contribution size = 28, Format = DWARF64, Version = 5
-# COMMON-NEXT: 0x00000038: 0000000000000000 "Handmade DWARF producer"
-# COMMON-NEXT: 0x00000040: 0000000000000036 "Compile_Unit_2"
-# COMMON-NEXT: 0x00000048: 0000000000000045 "/home/test/CU2"
-# COMMON-NEXT: 0x00000050: Contribution size = 12, Format = DWARF32, Version = 5
-# COMMON-NEXT: 0x00000058: 00000054 "Type_Unit"
-# COMMON-NEXT: 0x0000005c: 0000005e "MyStruct"
+# OFFSETS:      .debug_str_offsets contents:
+# OFFSETS-NEXT: 0x00000000: Contribution size = 32, Format = DWARF32, Version = 5
+# OFFSETS-NEXT: 0x00000008: 00000000 "Handmade DWARF producer"
+# OFFSETS-NEXT: 0x0000000c: 00000018 "Compile_Unit_1"
+# OFFSETS-NEXT: 0x00000010: 00000027 "/home/test/CU1"
+# OFFSETS-NEXT: 0x00000014: 00000067 "MyFunc"
+# OFFSETS-NEXT: 0x00000018: 0000006e "MyVar1"
+# OFFSETS-NEXT: 0x0000001c: 00000075 "MyVar2"
+# OFFSETS-NEXT: 0x00000020: 0000007c "MyVar3"
+# OFFSETS-NEXT: Gap, length = 4
+# OFFSETS-NEXT: 0x00000028: Contribution size = 28, Format = DWARF64, Version = 5
+# OFFSETS-NEXT: 0x00000038: 0000000000000000 "Handmade DWARF producer"
+# OFFSETS-NEXT: 0x00000040: 0000000000000036 "Compile_Unit_2"
+# OFFSETS-NEXT: 0x00000048: 0000000000000045 "/home/test/CU2"
+# OFFSETS-NEXT: 0x00000050: Contribution size = 12, Format = DWARF32, Version = 5
+# OFFSETS-NEXT: 0x00000058: 00000054 "Type_Unit"
+# OFFSETS-NEXT: 0x0000005c: 0000005e "MyStruct"
 # 
-# SPLIT:       .debug_str_offsets.dwo contents:
-# SPLIT-NEXT:  0x00000000: Contribution size = 24, Format = DWARF32, Version = 5
-# SPLIT-NEXT:  0x00000008: 00000000 "Handmade split DWARF producer"
-# SPLIT-NEXT:  0x0000000c: 0000001e "V5_split_compile_unit"
-# SPLIT-NEXT:  0x00000010: 00000034 "/home/test/splitCU"
-# SPLIT-NEXT:  0x00000014: 00000047 "V5_split_type_unit"
-# SPLIT-NEXT:  0x00000018: 0000005a "V5_split_Mystruct"
+# OFFSETS:       .debug_str_offsets.dwo contents:
+# OFFSETS-NEXT:  0x00000000: Contribution size = 24, Format = DWARF32, Version = 5
+# OFFSETS-NEXT:  0x00000008: 00000000 "Handmade split DWARF producer"
+# OFFSETS-NEXT:  0x0000000c: 0000001e "V5_split_compile_unit"
+# OFFSETS-NEXT:  0x00000010: 00000034 "/home/test/splitCU"
+# OFFSETS-NEXT:  0x00000014: 00000047 "V5_split_type_unit"
+# OFFSETS-NEXT:  0x00000018: 0000005a "V5_split_Mystruct"
 
 # VERIFY: No errors.
 
 # ERR-NOT: parsing a range list table: