Re: [PATCH, RFC]: per queue request tables, 2nd cut
Jens Axboe (axboe@suse.de)
Fri, 14 Apr 2000 19:11:18 +0200
On Fri, Apr 14 2000, Steve Lord wrote:
>
> So surely the next step to take with this is to break run_task_queue(&tq_disk)
> apart. There are really two uses for this call at the moment -
>
> 1. There is a shortage of request queue entries or memory
>
> 2. A specific page or buffer head is needed, push the task queue to
> ensure the I/O gets done.
>
> Once a system has multiple devices the second use is really a sledge-hammer,
> all plugged request queues get unplugged and fire at once. Just prodding
> the specific device into life would be much more preferable.
I agree completely, running tq_disk is over kill in most situations now.
I'm reworking parts of it right now, along with splitting the free list
in a read/write to kill xxx_get_queue from end_that_request_last.
--
* Jens Axboe <axboe@suse.de>
* Linux CD/DVD-ROM, SuSE Labs
* http://kernel.dk
-
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/