On Mon, 2007-07-09 at 16:17 +0200, Martin Peschke wrote:
>  > +static void do_virtblk_request(request_queue_t *q)
>  > +{
>  > +  struct virtio_blk *vblk = NULL;
>  > +  struct request *req;
>  > +  struct virtblk_req *vbr;
>  > +
>  > +  while ((req = elv_next_request(q)) != NULL) {
>  > +          vblk = req->rq_disk->private_data;
>  > +
>  > +          vbr = mempool_alloc(vblk->pool, GFP_ATOMIC);
>  > +          if (!vbr)
>  > +                  goto stop;
>  > +
>  > +          BUG_ON(req->nr_phys_segments > ARRAY_SIZE(vblk->sg));
>  > +          vbr->req = req;
>  > +          if (!do_req(q, vblk, vbr))
>  > +                  goto stop;
>  > +          blkdev_dequeue_request(req);
>  > +  }
>  > +
>  > +sync:
>  > +  if (vblk)
> 
> 
> this check looks bogus, as vblk->pool has been accessed unconditionally above


Sorry, I have been seduced by the code to misread it.
The check is correct and needed.

Looks like we can make it more readable, though.

- get rid of convoluted goto's
- no need for calling mempool_free() with NULL-pointer
- read private data from q->queuedata once instead of poking inside
  every request
- use a local variable to track issued requests and the need for
  a sync-operation
- pass req to do_req() as it is used there throughout the function
  (nice symmetry in parameter list, isn't it?)

It's just an untested and incomplete sketch done while I am acquainting
myself with the virtio code without having set up lguest yet.


static void do_virtblk_request(request_queue_t *q)
{
        struct virtio_blk *vblk = q->queuedata;
        struct request *req;
        struct virtblk_req *vbr;
        int issued = 0;

        while ((req = elv_next_request(q)) != NULL) {
                vbr = mempool_alloc(vblk->pool, GFP_ATOMIC);
                if (!vbr) {
                        blk_stop_queue(q);
                        break;
                }

                BUG_ON(req->nr_phys_segments > ARRAY_SIZE(vblk->sg));
                vbr->req = req;
                if (!do_req(q, vblk, req, vbr)) {
                        /* Queue full?  Wait. */
                        blk_stop_queue(q);
                        mempool_free(vbr, vblk->pool);
                        break;
                }
                blkdev_dequeue_request(req);
                issued++;
        }

        if (issued)
                vblk->vq->ops->sync(vblk->vq);
}



_______________________________________________
Virtualization mailing list
[email protected]
https://lists.linux-foundation.org/mailman/listinfo/virtualization

Reply via email to