[WebAssembly] Fix offsets for data relocations

[sbc100]

Summary

The code relocations getSectionOffset() returns the offset into the code
section for the data relocations it returns the offset into the data payload
which doesn't take into account the header of the data section.

As an alternative to this approach we could fixup all the FixupSection
fields of the data relocations once we know the header size.

The added test fails without this change (generates an invalid wasm
file).

[sbc100]

• Merge remote-tracking branch 'origin/master' into fix_data_relocations
• Add test

[sbc100]

• update comment

[dschuff]

Don't have a strong opinion but this seems better than fixing up the relocations separately. Maybe the comment on the SectionOffset field of MCSectionWasm should be updated (or some other place) to say that the section offsets don't include headers. Also it says "code section" when it's really more general than that now.

[sbc100]

• simplify

[sbc100]

• Merge remote-tracking branch 'origin/master' into fix_data_relocations
• add offsets
• Merge remote-tracking branch 'origin/master' into fix_data_relocations
• Merge remote-tracking branch 'origin/master' into fix_data_relocations
• use signed addend

[sbc100]

• formatting + comment fix

[sbc100]

In D32513#738751, @dschuff wrote:

Don't have a strong opinion but this seems better than fixing up the relocations separately. Maybe the comment on the SectionOffset field of MCSectionWasm should be updated (or some other place) to say that the section offsets don't include headers. Also it says "code section" when it's really more general than that now.

Done.

[dschuff]

LGTM, this is even better.

lib/MC/WasmObjectWriter.cpp

474

should these be on separate lines to match the rest?

[sbc100]

• Merge remote-tracking branch 'origin/master' into fix_data_relocations
• formatting

[sbc100]

lib/MC/WasmObjectWriter.cpp

474

Ran clang format to fix correctly.

lib/MC/WasmObjectWriter.cpp

 }
 
 // Apply the portions of the relocation records that we can handle ourselves
 // directly.
 static void ApplyRelocations(
     ArrayRef<WasmRelocationEntry> Relocations,
     raw_pwrite_stream &Stream,
     DenseMap<const MCSymbolWasm *, uint32_t> &SymbolIndices,
-    uint64_t ContentsOffset)
+    uint64_t ContentsOffset, uint64_t HeaderSize)
 {
   for (const WasmRelocationEntry &RelEntry : Relocations) {
     uint64_t Offset = ContentsOffset +
                       RelEntry.FixupSection->getSectionOffset() +
-                      RelEntry.Offset;
+                      RelEntry.Offset + HeaderSize;
     switch (RelEntry.Type) {
     case wasm::R_WEBASSEMBLY_FUNCTION_INDEX_LEB: {
       uint32_t Index = SymbolIndices[RelEntry.Symbol];
       assert(RelEntry.Addend == 0);
 
       WritePatchableLEB(Stream, Index, Offset);
       break;
     }
static void ApplyRelocations(
   }
 }
 
 // Write out the portions of the relocation records that the linker will
 // need to handle.
 static void WriteRelocations(
     ArrayRef<WasmRelocationEntry> Relocations,
     raw_pwrite_stream &Stream,
-    DenseMap<const MCSymbolWasm *, uint32_t> &SymbolIndices)
+    DenseMap<const MCSymbolWasm *, uint32_t> &SymbolIndices, uint64_t HeaderSize)
 {
   for (const WasmRelocationEntry RelEntry : Relocations) {
     encodeULEB128(RelEntry.Type, Stream);
 
     uint64_t Offset = RelEntry.Offset +
-                      RelEntry.FixupSection->getSectionOffset();
+                      RelEntry.FixupSection->getSectionOffset() + HeaderSize;
     uint32_t Index = SymbolIndices[RelEntry.Symbol];
     int64_t Addend = RelEntry.Addend;
 
     switch (RelEntry.Type) {
     case wasm::R_WEBASSEMBLY_FUNCTION_INDEX_LEB:
     case wasm::R_WEBASSEMBLY_TABLE_INDEX_SLEB:
     case wasm::R_WEBASSEMBLY_TABLE_INDEX_I32:
       encodeULEB128(Offset, Stream);
for (size_t i = 0, e = TypeIndexFixups.size(); i < e; ++i) {
       unsigned SizeLen = encodeULEB128(Type, Buffer, Padding);
       assert(SizeLen == 5);
       getStream().pwrite((char *)Buffer, SizeLen,
                          Section.ContentsOffset + Offset);
     }
 
     // Apply fixups.
     ApplyRelocations(CodeRelocations, getStream(), SymbolIndices,
-                     Section.ContentsOffset);
+                     Section.ContentsOffset, 0);
 
     endSection(Section);
   }
 
   // === Data Section ==========================================================
+  uint32_t DataSectionHeaderSize = 0;
   if (!DataBytes.empty()) {
     startSection(Section, wasm::WASM_SEC_DATA);
 
     encodeULEB128(1, getStream()); // count
     encodeULEB128(0, getStream()); // memory index
     write8(wasm::WASM_OPCODE_I32_CONST);
     encodeSLEB128(0, getStream()); // offset
     write8(wasm::WASM_OPCODE_END);
     encodeULEB128(DataBytes.size(), getStream()); // size
+    DataSectionHeaderSize = getStream().tell() - Section.ContentsOffset;
     writeBytes(DataBytes); // data
 
     // Apply fixups.
     ApplyRelocations(DataRelocations, getStream(), SymbolIndices,
-                     Section.ContentsOffset);
+                     Section.ContentsOffset, DataSectionHeaderSize);
 
     endSection(Section);
   }
 
   // === Name Section ==========================================================
   uint32_t TotalFunctions = NumFuncImports + Functions.size();
   if (TotalFunctions != 0) {
     startSection(Section, wasm::WASM_SEC_CUSTOM, "name");
#endif
   // === Code Reloc Section ====================================================
   if (!CodeRelocations.empty()) {
     startSection(Section, wasm::WASM_SEC_CUSTOM, "reloc.CODE");
 
     encodeULEB128(wasm::WASM_SEC_CODE, getStream());
 
     encodeULEB128(CodeRelocations.size() + TypeIndexFixups.size(), getStream());
 
-    WriteRelocations(CodeRelocations, getStream(), SymbolIndices);
+    WriteRelocations(CodeRelocations, getStream(), SymbolIndices, 0);
     WriteTypeRelocations(TypeIndexFixups, TypeIndexFixupTypes, getStream());
 
     endSection(Section);
   }
 
   // === Data Reloc Section ====================================================
   if (!DataRelocations.empty()) {
     startSection(Section, wasm::WASM_SEC_CUSTOM, "reloc.DATA");
 
     encodeULEB128(wasm::WASM_SEC_DATA, getStream());
 
     encodeULEB128(DataRelocations.size(), getStream());
 
-    WriteRelocations(DataRelocations, getStream(), SymbolIndices);
+    WriteRelocations(DataRelocations, getStream(), SymbolIndices, DataSectionHeaderSize);
 
     endSection(Section);
   }
 
   // === Linking Metadata Section ==============================================
   if (HasStackPointer) {
     startSection(Section, wasm::WASM_SEC_CUSTOM, "linking");
 

test/MC/WebAssembly/reloc-data.ll

+; RUN: llc -O0 -mtriple wasm32-unknown-unknown-wasm -filetype=obj %s -o - | llvm-readobj -r -expand-relocs | FileCheck %s
+
+; bar is a pointer initialised to the value of foo
+@foo = external global i32, align 4
+@bar = external global i64, align 4
+@a = global i32* @foo, align 8
+@b = global i64* @bar, align 8
+
+; CHECK: Format: WASM
+; CHECK: Relocations [
+; CHECK:   Section (6) DATA {
+; CHECK:     Relocation {
+; CHECK:       Type: R_WEBASSEMBLY_GLOBAL_ADDR_I32 (5)
+; CHECK:       Offset: 0x6
+; CHECK:       Index: 0x0
+; CHECK:       Addend: 0x0
+; CHECK:     }
+; CHECK:     Relocation {
+; CHECK:       Type: R_WEBASSEMBLY_GLOBAL_ADDR_I32 (5)
+; CHECK:       Offset: 0xE
+; CHECK:       Index: 0x1
+; CHECK:       Addend: 0x0
+; CHECK:     }
+; CHECK:   }
+; CHECK: ]