From 172687f53a09dbb27e2951db6747ace5cf7dec08 Mon Sep 17 00:00:00 2001
From: Michael Kolupaev <michael.kolupaev@clickhouse.com>
Date: Sat, 23 Dec 2023 21:43:32 +0000
Subject: [PATCH] Fix DWARFBlockInputFormat failing on DWARF 5 unit address
 ranges

---
 .../Formats/Impl/DWARFBlockInputFormat.cpp    | 41 +++++++++++++++----
 .../Formats/Impl/DWARFBlockInputFormat.h      |  1 +
 2 files changed, 35 insertions(+), 7 deletions(-)

diff --git a/src/Processors/Formats/Impl/DWARFBlockInputFormat.cpp b/src/Processors/Formats/Impl/DWARFBlockInputFormat.cpp
index 22ccbe03f0f..b9f65dd49b0 100644
--- a/src/Processors/Formats/Impl/DWARFBlockInputFormat.cpp
+++ b/src/Processors/Formats/Impl/DWARFBlockInputFormat.cpp
@@ -390,6 +390,40 @@ Chunk DWARFBlockInputFormat::parseEntries(UnitState & unit)
             if (need[COL_TAG])
                 col_tag->insertValue(tag);
 
+            if (tag == llvm::dwarf::DW_TAG_compile_unit)
+            {
+                /// Pre-parse DW_AT_addr_base and DW_AT_rnglists_base because other attributes may
+                /// rely on them. (Why couldn't DWARF just promise that these attributes must appear
+                /// before any attributes that depend on them?)
+                uint64_t offset = unit.offset;
+                for (auto attr : abbrev->attributes())
+                {
+                    if (attr.Attr == llvm::dwarf::DW_AT_addr_base ||
+                        attr.Attr == llvm::dwarf::DW_AT_rnglists_base)
+                    {
+                        auto val = llvm::DWARFFormValue::createFromSValue(
+                            attr.Form, attr.isImplicitConst() ? attr.getImplicitConstValue() : 0);
+                        if (!val.extractValue(*extractor, &offset, form_params, unit.dwarf_unit))
+                            throw Exception(ErrorCodes::CANNOT_PARSE_DWARF,
+                                "Failed to parse attribute {} of form {} at offset {}",
+                                llvm::dwarf::AttributeString(attr.Attr), attr.Form, unit.offset);
+                        uint64_t v = val.getRawUValue();
+                        if (attr.Attr == llvm::dwarf::DW_AT_addr_base)
+                            unit.addr_base = v;
+                        else
+                            unit.rnglists_base = v;
+                    }
+                    else
+                    {
+                        if (!llvm::DWARFFormValue::skipValue(
+                                attr.Form, *extractor, &offset, form_params))
+                            throw Exception(ErrorCodes::CANNOT_PARSE_DWARF,
+                                "Failed to skip attribute {} of form {} at offset {}",
+                                llvm::dwarf::AttributeString(attr.Attr), attr.Form, offset);
+                    }
+                }
+            }
+
             bool need_name = need[COL_NAME];
             bool need_linkage_name = need[COL_LINKAGE_NAME];
             bool need_decl_file = need[COL_DECL_FILE];
@@ -452,13 +486,6 @@ Chunk DWARFBlockInputFormat::parseEntries(UnitState & unit)
                         if (attr.Attr == llvm::dwarf::DW_AT_decl_line && std::exchange(need_decl_line, false))
                             col_decl_line->insertValue(static_cast<UInt32>(val.getRawUValue()));
 
-                        /// Starting offset of this unit's data in .debug_addr section.
-                        if (attr.Attr == llvm::dwarf::DW_AT_addr_base)
-                            unit.addr_base = val.getRawUValue();
-                        /// Same for .debug_rnglists section.
-                        if (attr.Attr == llvm::dwarf::DW_AT_rnglists_base)
-                            unit.rnglists_base = val.getRawUValue();
-
                         if (attr.Attr == llvm::dwarf::DW_AT_high_pc)
                         {
                             high_pc = val.getRawUValue();
diff --git a/src/Processors/Formats/Impl/DWARFBlockInputFormat.h b/src/Processors/Formats/Impl/DWARFBlockInputFormat.h
index 07c00656e4d..366671210bb 100644
--- a/src/Processors/Formats/Impl/DWARFBlockInputFormat.h
+++ b/src/Processors/Formats/Impl/DWARFBlockInputFormat.h
@@ -53,6 +53,7 @@ private:
         std::string unit_name;
         ColumnPtr filename_table; // from .debug_line
         size_t filename_table_size = 0;
+        /// Starting offset of this unit's data in .debug_addr and .debug_rnglists sections.
         uint64_t addr_base = UINT64_MAX;
         uint64_t rnglists_base = UINT64_MAX;