mips64 n32 and n64 support in dl-machine.h

[Alexandre Oliva]

On Mar 14, 2003, Andreas Jaeger <aj@suse.de> wrote:

> Getting rid of the mips64-specific file is definitly fine.  I don't
> think it ever worked.

Should I put it in in advance, or wait to check both files in at once?

>> [...]
>> Index: sysdeps/mips/dl-machine.h
>> [...]
>> -/* The MIPS never uses Elfxx_Rela relocations.  */
>> +/* No mips dynamic relocations are RELA.  We do have code in this file
>> +   is conditional to ! ELF_MACHINE_NO_RELA.  It was never

> Shouldn't this be: "We do have code in this file that is..." ?

Yes, thanks.  I'm taking it out, along with the code referenced in the
new comment.

>> +	__builtin_memcpy (&reloc_value, reloc_addr, sizeof (reloc_value));

> Why is this line needed?  I don't understand the need for it.

To support mis-aligned relocations.  According to Ralf Baechle, we
must support them.  I had this in the ChangeLog, but somehow got rid
of it while cleaning up some duplication.

> And why a memcpy and not just an assignment?  Please add a comment to
> explain this line - and the other memcpy line below.

Done.

> I suggest to leave this out completly since it's not used anywhere.

Ok.  I'm adding comments for any differences I had found out between
them.


Here's the revised patch.  Ok?

-------------- next part --------------
A non-text attachment was scrubbed...
Name: mips64-dl-machine.patch
Type: text/x-patch
Size: 33157 bytes
Desc: not available
URL: <http://sourceware.org/pipermail/libc-alpha/attachments/20030314/8d8f6261/attachment.bin>
-------------- next part --------------

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                 aoliva@{redhat.com, gcc.gnu.org}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist                Professional serial bug killer