Issue 243558
Summary [lld][MachO] ld64.lld applies a relocation whose r_address is past the end of its section: out-of-section write or crash
Labels lld:MachO, crash
Assignees
Reporter AnantaSrikar
    `ObjFile::parseRelocations` copies a relocation's `r_address` into `Relocation::offset` without checking it against the section size. When no subsection contains the offset, `findContainingSubsection` returns the last subsection that starts before it, again without an upper-bound check. `ConcatInputSection::writeTo` then applies the relocation at `buf + r.offset`. With a large `r_address`, `ld64.lld` segfaults. With an `r_address` just past the end of the section, it links with status 0 and writes the relocated bytes after the section in the output.

**Version**: main at c7014922f582 (2026-11-16), built from the Git repository with `-DCMAKE_BUILD_TYPE=Release -DLLVM_ENABLE_ASSERTIONS=ON -DLLVM_ENABLE_PROJECTS=lld`, on x86_64 Linux.

### Reproducer

```sh
cat > raddr.yaml <<'YAML'
--- !mach-o
IsLittleEndian:  true
FileHeader:
 magic:           0xFEEDFACF
  cputype:         0x100000C
  cpusubtype: 0x0
  filetype:        0x1
  ncmds:           2
  sizeofcmds:      176
 flags:           0x0
  reserved:        0x0
LoadCommands:
  - cmd: LC_SEGMENT_64
    cmdsize:         152
    segname:         ''
 vmaddr:          0
    vmsize:          9
    fileoff:         208
 filesize:        9
    maxprot:         7
    initprot:        7
 nsects:          1
    flags:           0
    Sections:
      - sectname: __text
        segname:         __TEXT
        addr:            0x0
 size:            9
        offset:          0xD0
        align: 2
        reloff:          0xE0
        nreloc:          1
 flags:           0x80000400
        reserved1:       0x0
        reserved2: 0x0
        reserved3:       0x0
        content: 00000094C0035FD600
        relocations:
          - address: 0x40000000
            symbolnum:       1
            pcrel: true
            length:          2
            extern:          true
 type:            2
            scattered:       false
 value:           0
  - cmd:             LC_SYMTAB
    cmdsize:         24
 symoff:          232
    nsyms:           2
    stroff:          264
 strsize:         34
LinkEditData:
  NameList:
    - n_strx:          1
 n_type:          0xF
      n_sect:          1
      n_desc: 0
      n_value:         0
    - n_strx:          17
      n_type: 0x1
      n_sect:          0
      n_desc:          0
      n_value: 0
  StringTable:
    - ''
    - _entry_00000000
    - _target_00000000
...
YAML
yaml2obj raddr.yaml -o raddr.o
ld64.lld -arch arm64 -platform_version macos 11.0 11.0 -dylib -undefined dynamic_lookup -o raddr.dylib raddr.o
```

It segfaults (status 139). With `address: 0x0`, the same object links (status 0).

With `address: 0x8` (the 4-byte `ARM64_RELOC_BRANCH26` fixup then starts at byte 8 of the 9-byte `__text`), and the second symbol made absolute so the branch value isn't zero (`n_type: 0x3`, `n_value: 1048576`), the link succeeds and the two bytes after `__text` in the output are `ff 03`. With `address: 0x0` they are `00 00`.

### Observed

```
Segmentation fault (core dumped)
```

With ASan, the fault is a 4-byte read through `llvm::support::endian::read` while applying the relocation.

### Expected

An error that the relocation's offset is outside its section.

### Analysis

https://github.com/llvm/llvm-project/blob/c7014922f5829426952a38d079220c21b0a8ebe8/lld/MachO/InputFiles.cpp#L619

https://github.com/llvm/llvm-project/blob/c7014922f5829426952a38d079220c21b0a8ebe8/lld/MachO/InputFiles.cpp#L660-L666

https://github.com/llvm/llvm-project/blob/c7014922f5829426952a38d079220c21b0a8ebe8/lld/MachO/InputSection.cpp#L232-L234

Checking in `validateRelocationInfo` (or before `findContainingSubsection`) that `r_address + (1 << r_length) <= sec.size`, and reporting an error otherwise, would fix both cases. #46325 is the same missing check in the ELF port.

This report was drafted with help from an AI assistant (Claude); I reviewed it and reproduced it myself.
_______________________________________________
llvm-bugs mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs

Reply via email to