[PowerPC, Mach-O] PPCMachObjectWriter.cpp, at long last

David Fang fang at csl.cornell.edu
Thu Aug 8 11:33:46 PDT 2013


Rafael,
 	Thanks for getting around to another review.  As requested, here's 
the otool output for the object file produced by the hello-reloc.s test 
case:

% otool -rvV hello-reloc.o
hello-reloc.o:
External relocation information 0 entries
address  pcrel length extern type    scattered symbolnum/value
Local relocation information 0 entries
address  pcrel length extern type    scattered symbolnum/value
Relocation information (__TEXT,__text) 5 entries
address  pcrel length extern type    scattered symbolnum/value
00000034 True  long   False  BR24    False     3 (__TEXT,__picsymbolstub1)
00000030 False long   n/a    LO16DIF True      0x00000074
          False long   n/a    PAIR    True      0x00000014 other_half = 
0x0000
0000002c False long   n/a    HA16DIF True      0x00000074
          False long   n/a    PAIR    True      0x00000014 other_half = 
0x0060
Relocation information (__TEXT,__picsymbolstub1) 4 entries
address  pcrel length extern type    scattered symbolnum/value
00000014 False long   n/a    LO16DIF True      0x00000070
          False long   n/a    PAIR    True      0x00000058 other_half = 
0x0000
0000000c False long   n/a    HA16DIF True      0x00000070
          False long   n/a    PAIR    True      0x00000058 other_half = 
0x0018
Relocation information (__DATA,__la_symbol_ptr) 1 entries
address  pcrel length extern type    scattered symbolnum/value
00000000 False long   True   VANILLA False     dyld_stub_binding_helper

It looks like it matches the FileCheck DARWIN-G4-DUMP text, as far as hex 
values of addresses goes.
Should I get any more reviewers?

Fang

> On 2 August 2013 18:25, David Fang <fang at csl.cornell.edu> wrote:
>> Hi,
>>         Patch updated again.  For the test I just used llvm-readobj
>> -relocations.  Let me know how this looks.
>
> LGTM!
>
> One lest request would be checking with otool if llvm-readobject is
> printing the correct information (the ppc macho support in lib/Object
> is not very well tested).
>
>
>> Fang
>
> Cheers,
> Rafael
>

-- 
David Fang
http://www.csl.cornell.edu/~fang/




More information about the llvm-commits mailing list