Re: [RFC] change of ->bd_op->open() semantics

Martin Dalecki (dalecki@evision-ventures.com)
Fri, 24 May 2002 23:15:56 +0200


Uz.ytkownik Linus Torvalds napisa?:
>
> On Fri, 24 May 2002, Alexander Viro wrote:
>
>> It has an additional benefit of killing the array of default
>>queues on the same pass - a thing we will need to do sooner or later
>>anyway.
>
>
> I'd like to see this, because we want to make the "find the right queue" a
> much more expensive operation (no longer some fairly simple mapping from
> major number - a more dynamic and general "register this queue for minors
> xxxx-yyyy of major zzz").
>
> Doing it just once at open() time allows for that to happen without any
> performance downside.
>
> Linus

Current ATA code says about this:
/*
* Returns the queue which corresponds to a given device.
*
* FIXME: this should take struct block_device * as argument in future.
*/
static request_queue_t *ata_get_queue(kdev_t dev)
{
struct ata_channel *ch = (struct ata_channel *)blk_dev[major(dev)].data;

/* FIXME: ALLERT: This discriminates between master and slave! */
return &ch->drives[DEVICE_NR(dev) & 1].queue;
}

Guess who did spill the FIXME: entiers there?
So of course plese plese go ahead as soon as possible!.
I would love to don't have to look at the above any longer.

-
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/