Re: [PATCH] atomic_inc_return() for i386[1/5] (Re:atomic_inc_return)

[Andrew Morton]

Andi Kleen <ak@xxxxxx> wrote:
>
> On Sat, Sep 11, 2004 at 04:05:32PM -0700, Andrew Morton wrote:
> > kaigai@xxxxxxxxxxxxx (Kaigai Kohei) wrote:
> > >
> > > 
> > > [1/5] atomic_inc_return-linux-2.6.9-rc1.i386.patch
> > >   This patch implements atomic_inc_return() and so on for i386,
> > >   and includes runtime check whether CPU is legacy 386.
> > >   It is same as I posted to LKML and Andi Kleen at '04/09/01.
> > > 
> > 
> > Can we not use the `alternative instruction' stuff to eliminate the runtime
> > test?
> 
> Yes, we could. I suggested this to Kaigai-san earlier, but
> he decided that it was too complicated because he would have needed
> to add an additional alternative() macro with enough parameters.
> 
> Given that atomic instructions are quite costly anyways and the jump
> should be very predictable he's probably right that it wouldn't 
> be worth the effort. 
> 

Hm.  Well if these things only have a few callsites then OK.  But if we go
and do something like implementing atomic_inc() or up_read() or whatever
with atomic_add_return() then we'd need to do something from a codesize
POV.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/