Fix an integer trunctation issues for the DW_AT_frame_base DWARF loca…

[clayborg]

…tion expression.

This patch allows offsets to the DW_AT_frame_base to exceed 4GB. Prior to this, for any .debug_info.dwo offset that exceeded 4GB, we would truncate the offset to the DW_AT_frame_base expression bytes to be 32 bit only. Changing the offset to 64 bits restores correct functionality.

No test for this as we don't want to create a .dwp file that has a .debug_info.dwo size that is over 4GB.

[llvmbot]

@llvm/pr-subscribers-lldb

Author: Greg Clayton (clayborg)

Changes

…tion expression.

This patch allows offsets to the DW_AT_frame_base to exceed 4GB. Prior to this, for any .debug_info.dwo offset that exceeded 4GB, we would truncate the offset to the DW_AT_frame_base expression bytes to be 32 bit only. Changing the offset to 64 bits restores correct functionality.

No test for this as we don't want to create a .dwp file that has a .debug_info.dwo size that is over 4GB.

---

Full diff: https://github.com/llvm/llvm-project/pull/110388.diff

1 Files Affected:

• (modified) lldb/source/Plugins/SymbolFile/DWARF/DWARFDebugInfoEntry.cpp (+2-2)

diff --git a/lldb/source/Plugins/SymbolFile/DWARF/DWARFDebugInfoEntry.cpp b/lldb/source/Plugins/SymbolFile/DWARF/DWARFDebugInfoEntry.cpp
index 77ef59d605c9c4..66d0bc4b90cb52 100644
--- a/lldb/source/Plugins/SymbolFile/DWARF/DWARFDebugInfoEntry.cpp
+++ b/lldb/source/Plugins/SymbolFile/DWARF/DWARFDebugInfoEntry.cpp
@@ -229,9 +229,9 @@ bool DWARFDebugInfoEntry::GetDIENamesAndRanges(
         case DW_AT_frame_base:
           if (frame_base) {
             if (form_value.BlockData()) {
-              uint32_t block_offset =
+              uint64_t block_offset =
                   form_value.BlockData() - data.GetDataStart();
-              uint32_t block_length = form_value.Unsigned();
+              uint64_t block_length = form_value.Unsigned();
               *frame_base =
                   DWARFExpressionList(module,
                                       DWARFExpression(DataExtractor(

[clayborg]

Since .dwp files can get large, it is ok for the .debug_info.dwo section to exceed 4GB as we can parse the linked list of dwarf units in the .debug_info.dwo and ignore the CU and TU indexes which are limited to 4GB. We have work arounds already checked into LLVM's DWARF code that detects this issues and reconstructs the CU and TU index to be 64 bit safe. So this fix allows variables to show correctly when they have a DW_AT_frame_base attribute on the DW_TAG_subprogram were the .debug_info.dwo offset of the DW_TAG_subprogram is over 4GB.