Problem with x32 pointer_guard

[Roland McGrath]

> roland/x32-setjmp-mangle branch looks OK, except for that we
> also need to make similar change in:
> 
> ./sysdeps/unix/sysv/linux/x86_64/____longjmp_chk.S

Ah.  Done.

> If we want to be clever, we can reorder sysdeps/x86_64/setjmp.S to
> use misaligned store for upper 32 bits of RBP instead of
> shift and 32-bit store.  But I don't think it is really needed.

It's not a misaligned store, it's just a partially-redundant store, right?
The does seem more clever, and it's fewer instructions.  I'll take your
word on whether the redundant store is better than two more register-only
instructions for performance.

Please verify the new version on the branch.


Thanks,
Roland