Re: [PATCH 15/17] x86, mpx: do 32-bit-only cmpxchg for 32-bit apps

[Borislav Petkov <bp@alien8.de>]

On Mon, Mar 30, 2015 at 09:57:14AM -0700, Dave Hansen wrote:
> > +	if (is_64bit_mm(mm))
> > +		return user_atomic_cmpxchg_inatomic(actual_old_val_ptr,
> > +						   bd_entry_addr,
> > +						   expected_old_val,
> > +						   new_bd_entry);
> > +	else
> > +		return futex_atomic_cmpxchg_inatomic((u32 *)actual_old_val_ptr,
> > +						    (u32 __user *)bd_entry_addr,
> > +						    expected_old_val,
> > +						    new_bd_entry);
> >  }
> 
> That does look tempting, and I appreciate the analysis.
> 
> But, I'd really rather not hide this behind another layer of abstraction
> in order to save a few variable declarations.  It's definitely _smaller_
> code, but it's a little less obvious what is going on.

If you rename futex_atomic_cmpxchg_inatomic to
atomic_cmpxchg_inatomic_32 or so, it is perfectly clear what's going on.

	if (is_64bit_mm())
		return user_atomic_cmpxchg_inatomic()
	else /* 32-bit */
		return user_atomic_cmpxchg_inatomic_32()

It can't get any more obvious than that.

:-D

-- 
Regards/Gruss,
    Boris.

ECO tip #101: Trim your mails when you reply.
--