Olof Johansson <[EMAIL PROTECTED]> :
> On Mon, Jan 29, 2007 at 11:35:06PM +0100, Francois Romieu wrote:
[...]
> > - The driver does not contain a single SMP locking instruction but
> > http://www.pasemi.com claims the platform to be multicore.
> > Is the driver really designed to be lockless ?
>
> Unless I misunderstood something, NAPI drivers that don't set NETIF_F_LLTX
> will have all locking taken care of by higher layers, no?
It is not necessarily _that_ simple (it would be cool though :o) ).
For instance, what does prevent pasemi_mac_clean_tx() to be issued
from IRQ context (pasemi_mac_tx_intr) and from the xmit handler
(pasemi_mac_start_tx) at the same time ?
[...]
> > unsigned int is supposed to save some cycles on ppc.
>
> Who told you that? That's not true.
Jon D Mason <[EMAIL PROTECTED]> on 25/08/2004 about ppc64 (not ppc, sorry).
[...]
> > > +#define DESCR(ring, i) ((ring)->desc[i % ((ring)->count)])
> > > +#define BUFF(ring, i) ((ring)->buffers[i % ((ring)->count)])
> > > +#define INFO(ring, i) ((ring)->desc_info[i % ((ring)->count)])
> >
> > A bit ugly/obfuscating/name clash prone imvho.
> >
> > Use local variables ?
>
> I'm open for suggestions here, not sure how local variables will help though?
struct pas_dma_xct_descr *desc = ring->desc[i % ring->count];
--
Ueimor
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
