> On Wed, 24 Mar 1999, Stephen C. Tweedie wrote:
>
> > Hi,
> >
> > On Tue, 23 Mar 1999 21:41:56 +0100 (MET), Gerard Roudier
> > <groudier@club-internet.fr> said:
> >
> > > A) Linux is doing IOs per block or per page from the kernel and we need
> > > to coalesce and sort this small virtual IOs somewhere.
> >
> > > B) The Linux design is, at the moment, to perform this work from
> > > ll_rw_blk.
> >
> > > The get_queue() only function looks like some band-aid for IDE to me.
> > > Since too partial design is bad design I am not going to use this method.
> >
> > You really have to justify that statement!
>
> Indeed. It seems fairly clean to me - I was actually thinking
It is currently as fairly clean as the following, but this will probably
become worse later, IMO, if it not _really_ cleaned as it deserves:
./drivers/block/ide.c: if (bdev->current_request != &bdev->plug)
./drivers/block/ide.c: if (bdev->current_request != &bdev->plug) /* FIXME: this will do for now */
./drivers/block/ide.c: if (bdev->current_request == &bdev->plug) /* FIXME: paranoia */
./drivers/block/ide.c: printk("%s: Huh? nuking plugged queue\n", drive->name);
./drivers/block/raid5.c:static void unplug_devices(struct stripe_head *sh)
./drivers/block/raid5.c: unplug_device(blk_dev + MAJOR(raid_conf->disks[i].dev));
./drivers/block/raid5.c: int i, handled = 0, unplug = 0;
./drivers/block/raid5.c: unplug = 1;
./drivers/block/raid5.c: unplug = 1;
./drivers/block/raid5.c: if (unplug) {
./drivers/block/raid5.c: PRINTK(("unplugging devices, sector == %lu, count == %d\n", sh->sector, raid_conf->sector_count));
./drivers/block/raid5.c: unplug_devices(sh);
./drivers/block/raid5.c: unplug = 0;
Note that in the raid code, this fairly clean thing is just #ifed ZERO.
> that the idea could be significantly enhanced such that get_queue instead
> returns a pointer to a structure:
>
> struct request_queue
> {
> struct request * current_request;
> request_fn_proc * request_fn;
> make_request_fn * merge_fn;
> spinlock_t io_request_lock;
> void * queuedata;
> };
>
> such that each queue has it's own queue head (obviously - this is
> effectively all this structure contained before), function pointers for
> requeust queueing, mergeing of new commands, a lock (to be used in place
> of io_request_lock), and finally an abstract pointer that request_fn()
> could use (for whatever purpose the actual queue handler wanted).
>
> -Eric
Gérard.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/