Re: [PATCH] gold: Don't adjust local symbol values in relocatable link twice

["H.J. Lu" <hjl dot tools at gmail dot com>]

On Wed, Nov 8, 2017 at 3:16 PM, Cary Coutant <ccoutant@gmail.com> wrote:
>> The fix committed for PR gold/19291 ended up breaking other cases. The
>> commit added adjustment code to write_local_symbols, but in many cases
>> compute_final_local_value_internal had already subtracted the output
>> section's address. To fix this, all other adjustments are now removed, so
>> only the one in write_local_symbols is left.
>>
>> gold/
>>         PR gold/22266
>>         * object.cc (Sized_relobj_file::compute_final_local_value_internal):
>>         Drop relocatable parameter and stop adjusting output value based on
>>         it.
>>         (Sized_relobj_file::compute_final_local_value): Stop passing
>>         relocatable to compute_final_local_value_internal.
>>         (Sized_relobj_file::do_finalize_local_symbols): Ditto.
>>         * object.h (Sized_relobj_file::compute_final_local_value_internal):
>>         Drop relocatable parameter.
>> ---
>>
>> A simple test is available at http://paste.debian.net/990796/, which is
>> much simpler than the real-world example in PR gold/22266. You can see
>> that int_from_a_1 ends up having a negative (well, large positive since
>> it's unsigned) symbol value when gold is used before this patch, so the
>> value printed by the program is not the expected 11223344 but whatever
>> happens to be before it in memory.
>
> Thanks! I've added the test case from the link you gave, and committed
> this on your behalf.
>

This patch has broken -r for all REL targets, including i386.


-- 
H.J.