[PATCH] Implement x86 SIZE32/SIZE64 relocations

[David Miller]

If anything, for relocations meant to be found in shared objects, it
is the dynamic linker which is absolutely the first place support
should show up.  It should not be dependent upon binutils support at
all.

GLIBC dynamic linker support is the one that needs the largest amount
of lead time, and the one for which it is the most difficult for a
developer or use to integrate into their environment on their own.

This even applies to the person who writes the binutils support for
emitting the relocations themselves.

So I definitely think HJ's patch should go in as soon as possible.