Re: [PATCH] superbh, fractured blocks, and grouped io

From: Jens Axboe (axboe@suse.de)
Date: Tue Oct 15 2002 - 02:44:23 EST


On Mon, Oct 14 2002, Joel Becker wrote:
> On Mon, Oct 14, 2002 at 03:51:00PM +0200, Jens Axboe wrote:
>
> Just a couple niggles, really. Looking good.
>
> > /*
> > - * First step, 'identity mapping' - RAID or LVM might
> > - * further remap this.
> > + * detach each bh and resubmit, or completely and if its a grouped bh
> > */
>
> The last line of the comment means "completely fail if its
> grouped", right?

Oops, the 'and' should be an 'end'. So your wording is correct.

> > +#define MAX_SUPERBH 65535 /* must fit info ->b_size right now */
>
> Why not sizeof(b_size) in case we ever care?

We don't want to make them too large anyway, and I think that 64k-1 is
more than enough. Maybe it would even be better to simply use an even
32kb. Consider someone writing in chunks of 64kb. It would simply suck
to get one 65024b request followed by a 512b one.

> > +extern int submit_bh_linked(int, struct buffer_head *);
> > +extern int submit_bh_grouped(int, struct buffer_head *);
>
> Why aren't these EXPORT_SYMBOL(), given that a third party
> driver may wish to use them (eg, a filesystem doing its own O_DIRECT
> work)?

Sure yes they will be, posted code was a first draft :-)

-- 
Jens Axboe

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



This archive was generated by hypermail 2b29 : Tue Oct 15 2002 - 22:00:54 EST