[GOLD] PowerPC TLS fixes

[H.J. Lu]

On Mon, Sep 10, 2012 at 10:04 AM, Ian Lance Taylor <iant@google.com> wrote:
> On Sun, Sep 9, 2012 at 7:42 AM, Alan Modra <amodra@gmail.com> wrote:
>>
>>         * output.h (Output_data_got::add_global_tls, add_local_tls,
>>         add_local_tls_pair): New functions.
>>         (Output_data_got::add_local_pair_with_rel): Remove second
>>         reloc param.  Expand comment.
>>         (Output_data_got::Got_entry::write): Add got_index param.
>>         * output.cc (Output_data_got::add_global_tls, add_local_tls,
>>         add_local_tls_pair): New functions.
>>         (Output_data_got::Got_entry::write): Handle tls symbols
>>         with use_plt_offset_ set specially.
>>         (Output_data_got::add_local_pair_with_rel): Only one reloc.
>>         (Output_data_got::do_write): Replace iterator with index, pass
>>         index to entry write function.
>>         * target.h (Target::tls_offset_for_local, tls_offset_for_global,
>>         do_tls_offset_for_local, do_tls_offset_for_global): New functions.
>>         * arm.cc (Target_arm::Scan::local): Update add_local_pair_with_rel
>>         call.
>>         * i386.cc (Target_i386::Scan::local): Likewise.
>>         * sparc.cc (Target_sparc::Scan::local): Likewise.
>>         * x86_64.cc (Target_x86_64::Scan::local): Likewise.
>
> Unimportant, but I've always been of the opinion that it's OK to just
> write "Update all calls" in the change to the function definition.
>
>>         * powerpc.cc (Target_powerpc::do_tls_offset_for_local,
>>         do_tls_offset_for_global): New functions.
>>         (Target_powerpc::Scan::local): Correct TLS relocations and got
>>         entry values.
>>         (Target_powerpc::Scan::global): Don't emit unnecessary
>>         dynamic relocations on TLS GOT entries.
>
>>    // Add a pair of entries for a local symbol to the GOT, and add
>> -  // dynamic relocations of type R_TYPE_1 and R_TYPE_2, respectively.
>> +  // a dynamic relocations of type R_TYPE using the section symbol of
>> +  // the output section to which input section SHNDX maps, on the first.
>> +  // The first got entry will have a value of zero, the second the
>> +  // value of the local symbol.
>
> s/dynamic relocations/dynamic relocation/
>
>
>>      // Whether to use the PLT offset of the symbol if it has one.
>> +    // For TLS symbols, whether to offset the symbol value.
>>      bool use_plt_offset_ : 1;
>
> If we're going to change the meaning of the field, let's rename it.
> Otherwise the uses can be very confusing.  I suggest simply
> use_plt_or_tls_offset_.
>
>
>>  // Add a pair of entries for a local symbol to the GOT, and add
>> -// dynamic relocations of type R_TYPE_1 and R_TYPE_2, respectively.
>> -// If R_TYPE_2 == 0, add the second entry with no relocation.
>> +// a dynamic relocations of type R_TYPE using the section symbol of
>> +// the output section to which input section SHNDX maps, on the first.
>> +// The first got entry will have a value of zero, the second the
>> +// value of the local symbol.
>
> s/dynamic relocations/dynamic relocation/
>
>> +  // Return the offset to use for the GOT_INDX'th got entry which is
>> +  // for a local tls symbol specified by OBJECT, SYMNDX
>
> Missing period at end of comment.
>
>> +  // Return the offset to use for the GOT_INDX'th got entry which is
>> +  // for global tls symbol GSYM
>
> Missing period here too.
>
> This is OK with those changes.
>

This breaks native x32 gold:

gcc -mx32  -W -Wall -Wstrict-prototypes -Wmissing-prototypes -Wshadow
-Werror -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -fmerge-constants
-g -O2   -o ifuncmain7pic -Bgcctestdir/ ifuncmain7pic.o
gcctestdir/ld: internal error in do_tls_offset_for_local, at
/export/gnu/import/git/binutils/gold/target.h:565
collect2: error: ld returned 1 exit status
make[3]: *** [ifuncmain7pic] Error 1
make[3]: *** Waiting for unfinished jobs....



-- 
H.J.