[lldb] Support .debug_rnglists.dwo sections in dwp file

[kimanh]

Summary

This patch considers the CU index entry
when reading the .debug_rnglists.dwo section.

[kimanh]

Include all commits.

[kimanh]

Simplifying test and add comments.

[jankratochvil]

lldb/source/Plugins/SymbolFile/DWARF/DWARFUnit.cpp

524

Here could be an error message.

546

I have found this getContribution adjustment duplication, are there some reasons it is not unified? https://www.jankratochvil.net/t/D107456-unify.patch
The issue is the code then handles two different .debug_rnglists DWARFDataExtractors with different offsets.

553

One such reason can be missing DWP absolute offset for the error report. That could be returned from GetRnglistData().

[kimanh]

• Incorporating Jan's patch
• Adding error message

[kimanh]

Thanks a lot for the review! Very sorry for my late update, I was out on vacation and then had to catch up with mails. Updated the revision now.

lldb/source/Plugins/SymbolFile/DWARF/DWARFUnit.cpp

546

Thanks a lot for the patch, I unified it in the latest update. IIRC (some time ago, so I don't recall completely) there was a mismatch in offsets maybe because I missed to update GetRnglistOffset to use getRnglistData too.

553

If I understand your comment correctly, you are suggesting to incorporate the contribution offset into the error message of GetRnglistData, is that correct? However, GetRnglistData will only return an error message if the contribution offset cannot be read, so the contribution offset cannot be specified in the error message (and thus the absolute offset cannot be inferred) . Or maybe I'm misunderstanding your comment here?

[kimanh]

Friendly ping!

[jankratochvil]

lldb/source/Plugins/SymbolFile/DWARF/DWARFUnit.cpp

553

Not so correct. The callers of GetRnglistData need to know the contribution offset to properly report it during parsing errors in the callers. GetRnglistTable can report error if extractHeaderAndOffsets finds invalid header data. In such case LLDB will report (with https://www.jankratochvil.net/t/debug_rnglists-dwp.s.patch):

error: debug_rnglists-dwp.s.tmp Failed to extract range list table at offset 0xc: parsing .debug_rnglists table at offset 0x0: unsupported reserved unit length of value 0xfffffff0

And one has no idea which DWO in that DWP file was invalid, that 0xc is relative to start of contribution, not to start of the .debug_rnglists.dwo section. GetRnglistData knows the contribution offset but its caller DWARFDebugRnglistTable does not.
I am not sure how perfect the error messages should be so giving an approval (if I am authorized to give one) but it would be nice to improve the error reporting.

[kimanh]

• Updating error messages

[kimanh]

lldb/source/Plugins/SymbolFile/DWARF/DWARFUnit.cpp

553

Ahh, okay, thanks a lot for the clarification! I updated a patch to address the error message. Please take a look if the updated patch is what you'd like to see.

[kimanh]

• splitting up new code (for error handling) into a different CL
• keep code that was already accepted

[kimanh]

I've split up the code that addresses the .debug_rnglists from the code that takes care of better error handling (https://reviews.llvm.org/D109231). I'll commit the reviewed part (this).

lldb/source/Plugins/SymbolFile/DWARF/DWARFUnit.h

const DWARFDebugInfoEntry *DIEPtr() {
     ExtractDIEsIfNeeded();
     if (m_die_array.empty())
       return NULL;
     return &m_die_array[0];
   }
 
   const llvm::Optional<llvm::DWARFDebugRnglistTable> &GetRnglistTable();
 
+  lldb_private::DWARFDataExtractor GetRnglistData() const;
+
   SymbolFileDWARF &m_dwarf;
   std::shared_ptr<DWARFUnit> m_dwo;
   DWARFUnitHeader m_header;
   const DWARFAbbreviationDeclarationSet *m_abbrevs = nullptr;
   void *m_user_data = nullptr;
   // The compile unit debug information entry item
   DWARFDebugInfoEntry::collection m_die_array;
   mutable llvm::sys::RWMutex m_die_array_mutex;

lldb/source/Plugins/SymbolFile/DWARF/DWARFUnit.cpp

if (const auto *contribution = entry->getContribution(
             GetVersion() >= 5 ? llvm::DW_SECT_LOCLISTS : llvm::DW_SECT_EXT_LOC))
       return DWARFDataExtractor(data, contribution->Offset,
                                 contribution->Length);
     return DWARFDataExtractor();
   }
   return data;
 }
 
+DWARFDataExtractor DWARFUnit::GetRnglistData() const {
+  DWARFContext &Ctx = GetSymbolFileDWARF().GetDWARFContext();
+  const DWARFDataExtractor &data = Ctx.getOrLoadRngListsData();
+  if (const llvm::DWARFUnitIndex::Entry *entry = m_header.GetIndexEntry()) {
+    if (const auto *contribution =
+            entry->getContribution(llvm::DW_SECT_RNGLISTS))
+      return DWARFDataExtractor(data, contribution->Offset,
+                                contribution->Length);
+    GetSymbolFileDWARF().GetObjectFile()->GetModule()->ReportError(
+        "Failed to find range list contribution for CU with signature "
+        "0x%" PRIx64,
+        entry->getSignature());
+
+    return DWARFDataExtractor();
+  }
+  return data;
+}
+
 void DWARFUnit::SetRangesBase(dw_addr_t ranges_base) {
   lldbassert(!m_rnglist_table_done);
 
   m_ranges_base = ranges_base;
 }
 
 const llvm::Optional<llvm::DWARFDebugRnglistTable> &
 DWARFUnit::GetRnglistTable() {
   if (GetVersion() >= 5 && !m_rnglist_table_done) {
     m_rnglist_table_done = true;
     if (auto table_or_error =
             ParseListTableHeader<llvm::DWARFDebugRnglistTable>(
-                m_dwarf.GetDWARFContext().getOrLoadRngListsData().GetAsLLVM(),
-                m_ranges_base, DWARF32))
+                GetRnglistData().GetAsLLVM(), m_ranges_base, DWARF32))
       m_rnglist_table = std::move(table_or_error.get());
     else
       GetSymbolFileDWARF().GetObjectFile()->GetModule()->ReportError(
           "Failed to extract range list table at offset 0x%" PRIx64 ": %s",
           m_ranges_base, toString(table_or_error.takeError()).c_str());
   }
   return m_rnglist_table;
 }
 
 // This function is called only for DW_FORM_rnglistx.
 llvm::Expected<uint64_t> DWARFUnit::GetRnglistOffset(uint32_t Index) {
   if (!GetRnglistTable())
     return llvm::createStringError(errc::invalid_argument,
                                    "missing or invalid range list table");
   if (!m_ranges_base)
     return llvm::createStringError(errc::invalid_argument,
                                    "DW_FORM_rnglistx cannot be used without "
                                    "DW_AT_rnglists_base for CU at 0x%8.8x",
                                    GetOffset());
   if (llvm::Optional<uint64_t> off = GetRnglistTable()->getOffsetEntry(
-          m_dwarf.GetDWARFContext().getOrLoadRngListsData().GetAsLLVM(), Index))
+          GetRnglistData().GetAsLLVM(), Index))
     return *off + m_ranges_base;
   return llvm::createStringError(
       errc::invalid_argument,
       "invalid range list table index %u; OffsetEntryCount is %u, "
       "DW_AT_rnglists_base is %" PRIu64,
       Index, GetRnglistTable()->getOffsetEntryCount(), m_ranges_base);
 }
 
if (GetVersion() <= 4) {
     debug_ranges->FindRanges(this, offset, ranges);
     return ranges;
   }
 
   if (!GetRnglistTable())
     return llvm::createStringError(errc::invalid_argument,
                                    "missing or invalid range list table");
 
-  llvm::DWARFDataExtractor data =
-      m_dwarf.GetDWARFContext().getOrLoadRngListsData().GetAsLLVM();
+  llvm::DWARFDataExtractor data = GetRnglistData().GetAsLLVM();
 
   // As DW_AT_rnglists_base may be missing we need to call setAddressSize.
   data.setAddressSize(m_header.GetAddressByteSize());
   auto range_list_or_error = GetRnglistTable()->findList(data, offset);
   if (!range_list_or_error)
     return range_list_or_error.takeError();
 
   llvm::Expected<llvm::DWARFAddressRangesVector> llvm_ranges =

lldb/test/Shell/SymbolFile/DWARF/x86/debug_rnglists-dwp.s

+## This tests if .debug_rnglists.dwo are correctly read if they are part
+## of a dwp file.
+
+# RUN: llvm-mc -triple=x86_64-pc-linux -filetype=obj %s --defsym MAIN=0 > %t
+# RUN: llvm-mc -triple=x86_64-pc-linux -filetype=obj %s > %t.dwp
+# RUN: %lldb %t -o "image lookup -v -s lookup_rnglists" -o exit | FileCheck %s
+
+# CHECK-LABEL: image lookup -v -s lookup_rnglists
+# CHECK:  Function: id = {{.*}}, name = "rnglists", range = [0x0000000000000000-0x0000000000000003)
+# CHECK:    Blocks: id = {{.*}}, range = [0x00000000-0x00000003)
+# CHECK-NEXT:       id = {{.*}}, range = [0x00000001-0x00000002)
+
+        .text
+rnglists:
+        nop
+.Lblock1_begin:
+lookup_rnglists:
+        nop
+.Lblock1_end:
+        nop
+.Lrnglists_end:
+
+## The main file.
+.ifdef MAIN
+        .section        .debug_abbrev,"",@progbits
+        .byte   1                       # Abbreviation Code
+        .byte   17                      # DW_TAG_compile_unit
+        .byte   0                       # DW_CHILDREN_no
+        .byte   0x76                    # DW_AT_dwo_name
+        .byte   8                       # DW_FORM_string
+        .byte   115                     # DW_AT_addr_base
+        .byte   23                      # DW_FORM_sec_offset
+        .byte   85                      # DW_AT_ranges
+        .byte   35                      # DW_FORM_rnglistx
+        .byte   116                     # DW_AT_rnglists_base
+        .byte   23                      # DW_FORM_sec_offset
+        .byte   0                       # EOM(1)
+        .byte   0                       # EOM(2)
+        .byte   0                       # EOM(3)
+
+        .section        .debug_info,"",@progbits
+.Lcu_begin0:
+        .long   .Ldebug_info_end0-.Ldebug_info_start0 # Length of Unit
+.Ldebug_info_start0:
+        .short  5                       # DWARF version number
+        .byte   4                       # DWARF Unit Type
+        .byte   8                       # Address Size (in bytes)
+        .long   .debug_abbrev           # Offset Into Abbrev. Section
+        .quad   1026699901672188186     # DWO id
+        .byte   1                       # Abbrev [1] DW_TAG_compile_unit
+        .asciz  "debug_rnglists-dwp.s.tmp.dwo"  # DW_AT_dwo_name
+        .long   .Laddr_table_base0      # DW_AT_addr_base
+        .byte   0                       # DW_AT_ranges
+        .long   .Lskel_rnglists_table_base # DW_AT_rnglists_base
+.Ldebug_info_end0:
+
+        .section        .debug_addr,"",@progbits
+        .long   .Ldebug_addr_end0-.Ldebug_addr_start0 # Length of contribution
+.Ldebug_addr_start0:
+        .short  5                       # DWARF version number
+        .byte   8                       # Address size
+        .byte   0                       # Segment selector size
+.Laddr_table_base0:
+        .quad   rnglists
+        .quad   .Lblock1_begin
+.Ldebug_addr_end0:
+
+        .section        .debug_rnglists,"",@progbits
+        .long   .Lskel_rnglist_table_end-.Lskel_rnglist_table_start # Length
+.Lskel_rnglist_table_start:
+        .short  5                       # Version
+        .byte   8                       # Address size
+        .byte   0                       # Segment selector size
+        .long   1                       # Offset entry count
+.Lskel_rnglists_table_base:
+        .long   .Lskel_ranges0-.Lskel_rnglists_table_base
+.Lskel_ranges0:
+        .byte   7                       # DW_RLE_start_length
+        .quad   rnglists
+        .uleb128   .Lrnglists_end-rnglists
+        .byte   0                       # DW_RLE_end_of_list
+.Lskel_rnglist_table_end:
+ .else
+ ## DWP file starts here.
+        .section        .debug_abbrev.dwo,"e",@progbits
+.LAbbrevBegin:
+        .byte   1                       # Abbreviation Code
+        .byte   17                      # DW_TAG_compile_unit
+        .byte   1                       # DW_CHILDREN_yes
+        .byte   37                      # DW_AT_producer
+        .byte   8                       # DW_FORM_string
+        .byte   0                       # EOM(1)
+        .byte   0                       # EOM(2)
+        .byte   2                       # Abbreviation Code
+        .byte   46                      # DW_TAG_subprogram
+        .byte   1                       # DW_CHILDREN_yes
+        .byte   17                      # DW_AT_low_pc
+        .byte   27                      # DW_FORM_addrx
+        .byte   18                      # DW_AT_high_pc
+        .byte   6                       # DW_FORM_data4
+        .byte   3                       # DW_AT_name
+        .byte   8                       # DW_FORM_string
+        .byte   0                       # EOM(1)
+        .byte   0                       # EOM(2)
+        .byte   5                       # Abbreviation Code
+        .byte   11                      # DW_TAG_lexical_block
+        .byte   0                       # DW_CHILDREN_no
+        .byte   85                      # DW_AT_ranges
+        .byte   35                      # DW_FORM_rnglistx
+        .byte   0                       # EOM(1)
+        .byte   0                       # EOM(2)
+        .byte   0                       # EOM(3)
+.LAbbrevEnd:
+        .section        .debug_info.dwo,"e",@progbits
+.LCUBegin:
+        .long   .Ldebug_info_end1-.Ldebug_info_start1 # Length of Unit
+.Ldebug_info_start1:
+        .short  5                       # DWARF version number
+        .byte   5                       # DWARF Unit Type
+        .byte   8                       # Address Size (in bytes)
+        .long   0                       # Offset Into Abbrev. Section
+        .quad   1026699901672188186      # DWO id
+        .byte   1                       # Abbrev [1] DW_TAG_compile_unit
+        .asciz  "Hand-written DWARF"    # DW_AT_producer
+        .byte   2                       # Abbrev [2] DW_TAG_subprogram
+        .byte   0                       # DW_AT_low_pc
+        .long   .Lrnglists_end-rnglists # DW_AT_high_pc
+        .asciz  "rnglists"              # DW_AT_name
+        .byte   5                       # Abbrev [5] DW_TAG_lexical_block
+        .byte   0                       # DW_AT_ranges
+        .byte   0                       # End Of Children Mark
+        .byte   0                       # End Of Children Mark
+.Ldebug_info_end1:
+.LCUEnd:
+        .section        .debug_rnglists.dwo,"e",@progbits
+## Fake rnglists to check if the cu index is taken into account
+        .long   .Lfake_rnglist_end-.Lfake_rnglist_start # Length
+.Lfake_rnglist_start:
+        .short  5                       # Version
+        .byte   8                       # Address size
+        .byte   0                       # Segment selector size
+        .long   0                       # Offset entry count
+        .byte   0                       # DW_RLE_end_of_list
+.Lfake_rnglist_end:
+.LRLBegin:
+        .long   .Ldwo_rnglist_table_end-.Ldwo_rnglist_table_start # Length
+.Ldwo_rnglist_table_start:
+        .short  5                       # Version
+        .byte   8                       # Address size
+        .byte   0                       # Segment selector size
+        .long   1                       # Offset entry count
+.Ldwo_rnglists_table_base:
+        .long   .Ldwo_ranges-.Ldwo_rnglists_table_base
+.Ldwo_ranges:
+        .byte   3                       # DW_RLE_startx_length
+        .uleb128 1
+        .uleb128 .Lblock1_end-.Lblock1_begin
+        .byte   0                       # DW_RLE_end_of_list
+.Ldwo_rnglist_table_end:
+.LRLEnd:
+        .section .debug_cu_index, "", @progbits
+## Header:
+        .short 5                        # Version
+        .short 0                        # Padding
+        .long 3                         # Section count
+        .long 1                         # Unit count
+        .long 2                         # Slot count
+## Hash Table of Signatures:
+        .quad 1026699901672188186
+        .quad 0
+## Parallel Table of Indexes:
+        .long 1
+        .long 0
+## Table of Section Offsets:
+## Row 0:
+        .long 1                         # DW_SECT_INFO
+        .long 3                         # DW_SECT_ABBREV
+        .long 8                         # DW_SECT_RNGLISTS
+## Row 1:
+        .long 0                         # Offset in .debug_info.dwo
+        .long 0                         # Offset in .debug_abbrev.dwo
+        .long .LRLBegin-.debug_rnglists.dwo # Offset in .debug_rnglists.dwo
+## Table of Section Sizes:
+        .long .LCUEnd-.LCUBegin         # Size in .debug_info.dwo
+        .long .LAbbrevEnd-.LAbbrevBegin # Size in .debug_abbrev.dwo
+        .long .LRLEnd-.LRLBegin         # Size in .debug_rnglists.dwo
+.endif