[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20071018171429.GH10674@kernel.dk>
Date: Thu, 18 Oct 2007 19:14:29 +0200
From: Jens Axboe <jens.axboe@...cle.com>
To: Arjan van de Ven <arjan@...radead.org>
Cc: Linus Torvalds <torvalds@...ux-foundation.org>,
David Miller <davem@...emloft.net>,
fujita.tomonori@....ntt.co.jp, mingo@...e.hu,
linux-kernel@...r.kernel.org, jgarzik@...ox.com,
alan@...rguk.ukuu.org.uk, tomof@....org
Subject: Re: [bug] ata subsystem related crash with latest -git
On Thu, Oct 18 2007, Arjan van de Ven wrote:
> On Thu, 18 Oct 2007 09:55:04 -0700 (PDT)
> Linus Torvalds <torvalds@...ux-foundation.org> wrote:
>
> >
> >
> > On Thu, 18 Oct 2007, Jens Axboe wrote:
> > > - unsigned long addr = page_to_phys(s->page) +
> > > s->offset;
> > > + unsigned long addr = page_to_phys(sg_page(s)) +
> > > s->offset;
> >
> > Umm. May I suggest (I haven't read the whole thread yet, maybe
> > somebody else already did) that
> >
> > static inline unsigned long sg_phys(struct scatterlist *sg)
> > {
> > return page_to_phys(sg_page(sg)) + sg->offset;
> > }
> >
> > would be a good thing to have?
> >
> > Very few drivers should care so much about the *page* itself (or the
> > offset). That's something that the generic allocation code etc cares
> > about, but the driver is almost bound to care mostly about the actual
> > DMA address
>
> .... but will that work for systems with IOMMU ? or is it fundamentally
> the wrong interface
They use foo_to_bus() on the address. sg_phys() should of course only be
used where the user previously did page_to_phys() on the sg page.
--
Jens Axboe
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists