virtio_blk: don't bounce highmem requests
Christoph Hellwig [Sat, 18 Jul 2009 03:47:45 +0000 (21:47 -0600)]
By default a block driver bounces highmem requests, but virtio-blk is
perfectly fine with any request that fit into it's 64 bit addressing scheme,
mapped in the kernel virtual space or not.

Besides improving performance on highmem systems this also makes the
reproducible oops in __bounce_end_io go away (but hiding the real cause).

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

drivers/block/virtio_blk.c

index 43db3ea..4c47859 100644 (file)
@@ -360,6 +360,9 @@ static int __devinit virtblk_probe(struct virtio_device *vdev)
        blk_queue_max_phys_segments(vblk->disk->queue, vblk->sg_elems-2);
        blk_queue_max_hw_segments(vblk->disk->queue, vblk->sg_elems-2);
 
+       /* No need to bounce any requests */
+       blk_queue_bounce_limit(vblk->disk->queue, BLK_BOUNCE_ANY);
+
        /* No real sector limit. */
        blk_queue_max_sectors(vblk->disk->queue, -1U);