Re: Changes to Linux/SCSI target mode infrastructure for v2.6.28

[Tejun Heo <tj@kernel.org>]

Hello, Nicholas.

Nicholas A. Bellinger wrote:
> Ok, I am up and running using the following patch against v2.6.28-rc6
> (along with Tejun's patch).   Comments please..?
> 
> diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c
> index f5d3b96..77f1fe0 100644
> --- a/drivers/scsi/scsi_lib.c
> +++ b/drivers/scsi/scsi_lib.c
> @@ -1505,7 +1507,7 @@ static void scsi_kill_request(struct request *req, struct request_queue *q)
>         struct scsi_target *starget = scsi_target(sdev);
>         struct Scsi_Host *shost = sdev->host;
>  
> -       blkdev_dequeue_request(req);
> +       elv_dequeue_request(req->q, req);

I don't think this really matters.  Either one wouldn't really matter.

>         if (unlikely(cmd == NULL)) {
>                 printk(KERN_CRIT "impossible request in %s.\n",
> @@ -1634,7 +1636,7 @@ static void scsi_request_fn(struct request_queue *q)
>                  * Remove the request from the request list.
>                  */
>                 if (!(blk_queue_tagged(q) && !blk_queue_start_tag(q, req)))
> -                       blkdev_dequeue_request(req);
> +                       elv_dequeue_request(req->q, req);
>                 sdev->device_busy++;
>  
>                 spin_unlock(q->queue_lock);

And this change is incorrect.  Timer should start here.  I don't think
you're seeing the same problem.  I'll reply to the original message.

Thanks.

-- 
tejun