Skip to content

Conversation

JDevlieghere
Copy link
Member

This is a continuation of #153494. In a WebAssembly file, the "name" section contains names for the segments in the data section (WASM_NAMES_DATA_SEGMENT). We already parse these as symbols, and with this PR, we now also create sub-sections for each of the segments.

This is a continuation of llvm#153494. In a WebAssembly file, the "name"
section contains names for the segments in the data section
(WASM_NAMES_DATA_SEGMENT). We already parse these as sections, as with
this PR, we also create sub-sections for the data segments.
@llvmbot
Copy link
Member

llvmbot commented Aug 14, 2025

@llvm/pr-subscribers-lldb

Author: Jonas Devlieghere (JDevlieghere)

Changes

This is a continuation of #153494. In a WebAssembly file, the "name" section contains names for the segments in the data section (WASM_NAMES_DATA_SEGMENT). We already parse these as symbols, and with this PR, we now also create sub-sections for each of the segments.


Full diff: https://github.com/llvm/llvm-project/pull/153634.diff

2 Files Affected:

  • (modified) lldb/source/Plugins/ObjectFile/wasm/ObjectFileWasm.cpp (+51-19)
  • (modified) lldb/test/Shell/Symtab/symtab-wasm.test (+13-7)
diff --git a/lldb/source/Plugins/ObjectFile/wasm/ObjectFileWasm.cpp b/lldb/source/Plugins/ObjectFile/wasm/ObjectFileWasm.cpp
index b3144f28f4913..dc0b0241d1f24 100644
--- a/lldb/source/Plugins/ObjectFile/wasm/ObjectFileWasm.cpp
+++ b/lldb/source/Plugins/ObjectFile/wasm/ObjectFileWasm.cpp
@@ -281,7 +281,16 @@ ParseFunctions(SectionSP code_section_sp) {
   return functions;
 }
 
-static llvm::Expected<std::vector<AddressRange>>
+struct WasmSegment {
+  WasmSegment(SectionSP section_sp, lldb::offset_t offset, uint32_t size,
+              uint32_t flags)
+      : address_range(section_sp, offset, size), flags(flags) {};
+  std::string name;
+  AddressRange address_range;
+  uint32_t flags = 0;
+};
+
+static llvm::Expected<std::vector<WasmSegment>>
 ParseData(SectionSP data_section_sp) {
   DataExtractor data;
   data_section_sp->GetSectionData(data);
@@ -292,7 +301,7 @@ ParseData(SectionSP data_section_sp) {
   if (segment_count > std::numeric_limits<uint32_t>::max())
     return llvm::createStringError("segment count overflows uint32_t");
 
-  std::vector<AddressRange> segments;
+  std::vector<WasmSegment> segments;
   segments.reserve(segment_count);
 
   for (uint32_t i = 0; i < segment_count; ++i) {
@@ -304,7 +313,7 @@ ParseData(SectionSP data_section_sp) {
     if (flags > std::numeric_limits<uint32_t>::max())
       return llvm::createStringError("segment size overflows uint32_t");
 
-    segments.emplace_back(data_section_sp, offset, segment_size);
+    segments.emplace_back(data_section_sp, offset, segment_size, flags);
 
     std::optional<lldb::offset_t> next_offset =
         llvm::checkedAddUnsigned(offset, segment_size);
@@ -319,7 +328,7 @@ ParseData(SectionSP data_section_sp) {
 static llvm::Expected<std::vector<Symbol>>
 ParseNames(SectionSP name_section_sp,
            const std::vector<AddressRange> &function_ranges,
-           const std::vector<AddressRange> &segment_ranges) {
+           std::vector<WasmSegment> &segments) {
   DataExtractor name_section_data;
   name_section_sp->GetSectionData(name_section_data);
 
@@ -358,12 +367,14 @@ ParseNames(SectionSP name_section_sp,
       for (uint64_t i = 0; c && i < count; ++i) {
         const uint64_t idx = data.getULEB128(c);
         const std::optional<std::string> name = GetWasmString(data, c);
-        if (!name || idx >= segment_ranges.size())
+        if (!name || idx >= segments.size())
           continue;
+        // Update the segment name.
+        segments[i].name = *name;
         symbols.emplace_back(
             symbols.size(), Mangled(*name), lldb::eSymbolTypeData,
             /*external=*/false, /*is_debug=*/false, /*is_trampoline=*/false,
-            /*is_artificial=*/false, segment_ranges[idx],
+            /*is_artificial=*/false, segments[i].address_range,
             /*size_is_valid=*/true, /*contains_linker_annotations=*/false,
             /*flags=*/0);
       }
@@ -391,33 +402,34 @@ void ObjectFileWasm::ParseSymtab(Symtab &symtab) {
 
   // The name section contains names and indexes. First parse the data from the
   // relevant sections so we can access it by its index.
-  std::vector<AddressRange> function_ranges;
-  std::vector<AddressRange> segment_ranges;
+  std::vector<AddressRange> functions;
+  std::vector<WasmSegment> segments;
 
   // Parse the code section.
   if (SectionSP code_section_sp =
           m_sections_up->FindSectionByType(lldb::eSectionTypeCode, false)) {
-    llvm::Expected<std::vector<AddressRange>> functions =
+    llvm::Expected<std::vector<AddressRange>> maybe_functions =
         ParseFunctions(code_section_sp);
-    if (!functions) {
-      LLDB_LOG_ERROR(log, functions.takeError(),
+    if (!maybe_functions) {
+      LLDB_LOG_ERROR(log, maybe_functions.takeError(),
                      "Failed to parse Wasm code section: {0}");
       return;
     }
-    function_ranges = *functions;
+    functions = *maybe_functions;
   }
 
   // Parse the data section.
-  if (SectionSP data_section_sp =
-          m_sections_up->FindSectionByType(lldb::eSectionTypeData, false)) {
-    llvm::Expected<std::vector<AddressRange>> segments =
+  SectionSP data_section_sp =
+      m_sections_up->FindSectionByType(lldb::eSectionTypeData, false);
+  if (data_section_sp) {
+    llvm::Expected<std::vector<WasmSegment>> maybe_segments =
         ParseData(data_section_sp);
-    if (!segments) {
-      LLDB_LOG_ERROR(log, segments.takeError(),
+    if (!maybe_segments) {
+      LLDB_LOG_ERROR(log, maybe_segments.takeError(),
                      "Failed to parse Wasm data section: {0}");
       return;
     }
-    segment_ranges = *segments;
+    segments = *maybe_segments;
   }
 
   // Parse the name section.
@@ -429,7 +441,7 @@ void ObjectFileWasm::ParseSymtab(Symtab &symtab) {
   }
 
   llvm::Expected<std::vector<Symbol>> symbols =
-      ParseNames(name_section_sp, function_ranges, segment_ranges);
+      ParseNames(name_section_sp, functions, segments);
   if (!symbols) {
     LLDB_LOG_ERROR(log, symbols.takeError(), "Failed to parse Wasm names: {0}");
     return;
@@ -438,6 +450,26 @@ void ObjectFileWasm::ParseSymtab(Symtab &symtab) {
   for (const Symbol &symbol : *symbols)
     symtab.AddSymbol(symbol);
 
+  lldb::user_id_t segment_id = 0;
+  for (const WasmSegment &segment : segments) {
+    const lldb::addr_t segment_addr =
+        segment.address_range.GetBaseAddress().GetFileAddress();
+    const size_t segment_size = segment.address_range.GetByteSize();
+    SectionSP segment_sp = std::make_shared<Section>(
+        /*parent_section_sp=*/data_section_sp, GetModule(),
+        /*obj_file=*/this,
+        ++segment_id << 8, // 1-based segment index, shifted by 8 bits to avoid
+                           // collision with section IDs.
+        ConstString(segment.name), eSectionTypeData,
+        /*file_vm_addr=*/segment_addr,
+        /*vm_size=*/segment_size,
+        /*file_offset=*/segment_addr,
+        /*file_size=*/segment_size,
+        /*log2align=*/0, segment.flags);
+    m_sections_up->AddSection(segment_sp);
+    GetModule()->GetSectionList()->AddSection(segment_sp);
+  }
+
   symtab.Finalize();
 }
 
diff --git a/lldb/test/Shell/Symtab/symtab-wasm.test b/lldb/test/Shell/Symtab/symtab-wasm.test
index 5374b0c2f2892..5e7c7cabc5280 100644
--- a/lldb/test/Shell/Symtab/symtab-wasm.test
+++ b/lldb/test/Shell/Symtab/symtab-wasm.test
@@ -1,9 +1,15 @@
 # RUN: yaml2obj %S/Inputs/simple.wasm.yaml -o %t.wasm
-# RUN: %lldb %t.wasm -o 'image dump symtab'
+# RUN: %lldb %t.wasm -o 'image dump symtab' -o 'image dump sections' | FileCheck %s
 
-# CHECK: Code 0x0000000000000002 0x0000000000000002 {{.*}} __wasm_call_ctors
-# CHECK: Code 0x0000000000000005 0x0000000000000029 {{.*}} add
-# CHECK: Code 0x000000000000002f 0x000000000000004c {{.*}} __original_main
-# CHECK: Code 0x000000000000007c 0x0000000000000009 {{.*}} main
-# CHECK: Data 0x000000000000022f 0x0000000000000041 {{.*}} .rodata
-# CHECK: Data 0x0000000000000270 0x0000000000000000 {{.*}} .data
+CHECK: Code 0x0000000000000002 0x0000000000000002 0x00000000 __wasm_call_ctors
+CHECK: Code 0x0000000000000005 0x0000000000000029 0x00000000 add
+CHECK: Code 0x000000000000002f 0x000000000000004c 0x00000000 __original_main
+CHECK: Code 0x000000000000007c 0x0000000000000009 0x00000000 main
+CHECK: Data 0x000000000000022f 0x0000000000000041 0x00000000 .rodata
+CHECK: Data 0x0000000000000270 0x0000000000000000 0x00000000 .data
+
+CHECK: 0x0000000000000001 code       {{.*}} 0x000001a1 0x00000085 0x00000000 symtab-wasm.test.tmp.wasm.code
+CHECK: 0x0000000000000003 data       {{.*}} 0x0000022c 0x0000001a 0x00000000 symtab-wasm.test.tmp.wasm.data
+CHECK: 0x0000000000000040 wasm-name  {{.*}} 0x00000251 0x00000059 0x00000000 symtab-wasm.test.tmp.wasm.name
+CHECK: 0x0000000000000100 data       {{.*}} 0x0000022f 0x00000041 0x00000000 symtab-wasm.test.tmp.wasm.data..rodata
+CHECK: 0x0000000000000200 data       {{.*}} 0x00000270 0x00000000 0x00000000 symtab-wasm.test.tmp.wasm.data..data

@JDevlieghere JDevlieghere changed the title [lldb] Create sections fro Wasm segments [lldb] Create sections for Wasm segments Aug 14, 2025
Copy link
Collaborator

@DavidSpickett DavidSpickett left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Don't have much expertise here, defer to Adrian to approve.

@@ -304,7 +313,7 @@ ParseData(SectionSP data_section_sp) {
if (flags > std::numeric_limits<uint32_t>::max())
return llvm::createStringError("segment size overflows uint32_t");
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This error should say flags, right?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Actually, no, the error message is correct, but the check above is not (it should check segment_size).

// Skip over the constant expression.
for (uint8_t b = 0; b != llvm::wasm::WASM_OPCODE_END;)
b = data.GetU8(&offset);
}
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This looks correct.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

FWIW, there might be a place for a wrapper function (or macro?) that does effectively a GetULEB32OrError()

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah, let me clean that up in a follow-up for the whole file.

@JDevlieghere JDevlieghere merged commit 8bd9897 into llvm:main Aug 19, 2025
9 checks passed
@JDevlieghere JDevlieghere deleted the wasm-data-segments branch August 19, 2025 18:08
JDevlieghere added a commit to swiftlang/llvm-project that referenced this pull request Aug 27, 2025
This is a continuation of llvm#153494. In a WebAssembly file, the "name"
section contains names for the segments in the data section
(WASM_NAMES_DATA_SEGMENT). We already parse these as symbols, and with
this PR, we now also create sub-sections for each of the segments.

(cherry picked from commit 8bd9897)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants