[PATCH] D79311: [lld-macho] Support X86_64_RELOC_SIGNED_{1,2,4}

[Fangrui Song via Phabricator via llvm-commits]

MaskRay marked an inline comment as done.
MaskRay added inline comments.


================
Comment at: lld/MachO/Arch/X86_64.cpp:53
+  case X86_64_RELOC_SIGNED_2:
+  case X86_64_RELOC_SIGNED_4:
   case X86_64_RELOC_GOT_LOAD:
----------------
MaskRay wrote:
> thakis wrote:
> > Is this correct? Looking at `lld/lib/ReaderWriter/MachO/ArchHandler_x86_64.cpp`, it has this handy table:
> > 
> > ```
> >     ripRel32,              /// ex: movq _foo(%rip), %rax
> >     ripRel32Minus1,        /// ex: movb $0x12, _foo(%rip)
> >     ripRel32Minus2,        /// ex: movw $0x1234, _foo(%rip)
> >     ripRel32Minus4,        /// ex: movl $0x12345678, _foo(%rip)
> >     ripRel32Anon,          /// ex: movq L1(%rip), %rax
> >     ripRel32Minus1Anon,    /// ex: movb $0x12, L1(%rip)
> >     ripRel32Minus2Anon,    /// ex: movw $0x1234, L1(%rip)
> >     ripRel32Minus4Anon,    /// ex: movw $0x12345678, L1(%rip)
> > 
> > ```
> > 
> > and it maps SIGNED(_1/2/4) to these and then it does relocations with offsets for the offset relocation types.
> This matches my observation of  ld64 and X86MachObjectWriter.cpp
> 
> Reading the implicit addend back may seem weird but I believe `X86_64_RELOC_SIGNED_1` requires the implicit addend to be 1.
> 
Sorry, I mean the implicit addend is -1 (not 1).

```
       f: c6 05 ff ff ff ff 45          movb    $69, -1(%rip)  # 15 <__text+0x15>
                0000000000000011:  X86_64_RELOC_SIGNED_1        _s-1
```


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D79311/new/

https://reviews.llvm.org/D79311