Re: 2.6.21-rc2-mm2

[Jens Axboe <jens.axboe@oracle.com>]

On Tue, Mar 06 2007, J.A. Magallón wrote:
> On Tue, 6 Mar 2007 00:44:08 -0800, Andrew Morton <akpm@linux-foundation.org> wrote:
> 
> > 
> > Temporarily at
> > 
> >   http://userweb.kernel.org/~akpm/2.6.21-rc2-mm2/
> > 
> 
> Does this include a fix for the NFS problem ?
> 
> BTW, I have in my kernel a patch like this below, isn't it needed ?
> Original thread:
> 
> http://marc.theaimsgroup.com/?l=linux-kernel&m=117189051227292&w=2
> 
> --- a/block/ll_rw_blk.c
> +++ b/block/ll_rw_blk.c
> @@ -2919,14 +2919,14 @@ static int __make_request(request_queue_
>  	 */
>  	blk_queue_bounce(q, &bio);
>  
> +	spin_lock_irq(q->queue_lock);
>  	/*
>  	 * Check if we can merge with the plugged list before grabbing
>  	 * any locks.
>  	 */
>  	if (!check_plug_merge(q, ioc, bio))
> -		goto out;
> +		goto out_unlock;
>  
> -	spin_lock_irq(q->queue_lock);
>  	el_ret = elv_merge(q, &req, bio);
>  	if (el_ret == ELEVATOR_BACK_MERGE) {
>  		if (bio_attempt_back_merge(q, req, bio)) {
> @@ -2984,7 +2984,6 @@ out_unlock:
>  		list_add_tail(&req->queuelist, &ioc->plugged_list);
>  	}
>  
> -out:
>  	return 0;
>  
>  end_io_eopnotsupp:

No it's not, plus andrew didn't include git-block in this release so
you'd have a hard time even applying it. The above patch would also
eliminate 50% of the win of per-process plugging, if you go and grab the
queue lock anyway.

-- 
Jens Axboe