Allow copy relocations with pie links

[H.J. Lu]

On Thu, May 8, 2014 at 11:48 AM, Sriraman Tallam <tmsriram@google.com> wrote:
> On Wed, May 7, 2014 at 4:46 PM, Cary Coutant <ccoutant@google.com> wrote:
>>> * symtab.h (may_need_copy_reloc): Extra default parameter to
>>> indicate if this is a PC relative relocation.
>>
>> In the PC-rel case, you need to check for output_is_executable. With
>> your patch, we will now allow copy relocs for pc-relative relocations
>> even in shared libraries.
>>
>> I'd prefer to move the check for output format to the points of call;
>> there are two calls in each target, one for pc-relative relocs, and
>> one for absolute relocs. For the absolute reloc, use:
>>
>>   if (!parameters->options().output_is_position_independent()
>>       && gsym->may_need_copy_reloc())
>>
>> For the pc-relative reloc, use:
>>
>>   if (parameters->options().output_is_executable()
>>       && gsym->may_need_copy_reloc())
>>
>>> * x86_64.cc (Target_x86_64<size>::Scan::global): Call
>>> may_need_copy_reloc with parameter set to true.
>>
>> Please update the other targets, too.
>
> All changes done and new patch attached.  Tested *only* for x86_64.
> Appreciate help in testing  other targets.
>
>
> * symtab.h (may_need_copy_reloc): Remove check for position independent
> code.
> * x86_64.cc (Target_x86_64<size>::Scan::global): Add check for no
> position independence before pc absolute may_need_copy_reloc call.
> Add check for executable output befor pc relative may_need_copy_reloc
> call.
> * i386.cc: Ditto.
> * arm.cc: Ditto.
> * sparc.cc: Ditto.
> * powerpc.cc: Ditto.
> * tilegx.cc: Ditto.
> * testsuite/pie_copyrelocs_test.cc: New file.
> * testsuite/pie_copyrelocs_shared_test.cc: New file.
> * Makefile.am (pie_copyrelocs_test): New test.
> * Makefile.in: Regenerate.
>

x86 doesn't have PC-relative data access.  x86 needs
to use GOT for all data access in PIE unless run-time
text relocation is allowed.  Normally x86 shouldn't have
copy relocation in PIE.

-- 
H.J.