Re: [PATCH] NVMe: Fix compilation on architecturs without readq/writeq

[Ingo Molnar <mingo@elte.hu>]

* Alan Cox <alan@lxorguk.ukuu.org.uk> wrote:

> On Tue, 31 Jan 2012 13:09:22 +0100
> Ingo Molnar <mingo@elte.hu> wrote:
> 
> > 
> > * Alan Cox <alan@lxorguk.ukuu.org.uk> wrote:
> > 
> > > >   u64 val;
> > > >   val = readl(addr);
> > > >   val |= readl(addr+4) << 32;
> > > > 
> > > > is well-defined and must read the low word first - both at the C level
> > > > *and* at the CPU level. Anything else would be a bug in the
> > > > architecture "readl()" implementation or the hardware.
> > > 
> > > That doesn't make the access atomic to hardware however as a true 64bit
> > > readq/writeq would be ?
> > > 
> > > It seems to me the two are not quite the same semantically
> > 
> > Correct, and that's what the:
> > 
> > 	#include <asm/io-inatomic.h>
> > 
> > line in the driver would express.
> 
> Why would "inatomic" indicate that - I'm confused. It would 
> imply to me they were extra specially atomic ?

Yeah, s/inatomic/non-atomic.

inatomic would be doubly confusing for the reason that it's 
already used as an 'in atomic section' sense in the kernel.

> (atomos if from the Greek so in- as a prefix isn't the same 
> in- as in many other words, welcome to English hell - who 
> needs perl)
> 
> non-atomic.h might be better, or 'un-atomic' or 'multi-read' 
> or something ?

non-atomic sounds good to me too.

Thanks,

	Ingo