[Xen-devel] [PATCH RFC 2/3] Virtio draft II: example block driver

[Rusty Russell <rusty@xxxxxxxxxxxxxxx>]

The block driver uses outbufs with sg[0] being the request information
(struct virtio_blk_outhdr) with the type, sector and inbuf id.  For a
write, the rest of the sg will contain the data to be written.

The first segment of the inbuf is a result code (struct
virtio_blk_inhdr).  For a read, the rest of the sg points to the input
buffer.

TODO:
        1) Ordered tag support.

Signed-off-by: Rusty Russell <rusty@xxxxxxxxxxxxxxx>
---
 drivers/block/Makefile     |    1 
 drivers/block/virtio_blk.c |  299 ++++++++++++++++++++++++++++++++++++++++++++
 include/linux/virtio_blk.h |   28 ++++
 3 files changed, 328 insertions(+)

===================================================================
--- a/drivers/block/Makefile
+++ b/drivers/block/Makefile
@@ -20,6 +20,7 @@ obj-$(CONFIG_BLK_CPQ_CISS_DA)  += cciss.
 obj-$(CONFIG_BLK_CPQ_CISS_DA)  += cciss.o
 obj-$(CONFIG_BLK_DEV_DAC960)   += DAC960.o
 obj-$(CONFIG_CDROM_PKTCDVD)    += pktcdvd.o
+obj-y                          += virtio_blk.o
 
 obj-$(CONFIG_BLK_DEV_UMEM)     += umem.o
 obj-$(CONFIG_BLK_DEV_NBD)      += nbd.o
===================================================================
--- /dev/null
+++ b/drivers/block/virtio_blk.c
@@ -0,0 +1,299 @@
+//#define DEBUG
+#include <linux/spinlock.h>
+#include <linux/blkdev.h>
+#include <linux/hdreg.h>
+#include <linux/virtio.h>
+#include <linux/virtio_blk.h>
+
+static unsigned char virtblk_index = 'a';
+struct virtio_blk
+{
+       struct virtio_device *vdev;
+
+       /* The disk structure for the kernel. */
+       struct gendisk *disk;
+
+       /* Request tracking. */
+       struct list_head reqs;
+
+       mempool_t *pool;
+
+       /* Scatterlist: can be too big for stack. */
+       struct scatterlist sg[1+MAX_PHYS_SEGMENTS];
+};
+
+struct virtblk_req
+{
+       struct list_head list;
+       struct request *req;
+       unsigned long out_id;
+       bool out_done, in_done;
+       bool failed;
+       struct virtio_blk_outhdr out_hdr;
+       struct virtio_blk_inhdr in_hdr;
+};
+
+/* Jens gave me this nice helper to end all chunks of a request. */
+static void end_dequeued_request(struct request *req, int uptodate)
+{
+       if (end_that_request_first(req, uptodate, req->hard_nr_sectors))
+               BUG();
+       add_disk_randomness(req->rq_disk);
+       end_that_request_last(req, uptodate);
+}
+
+static void finish(struct virtio_blk *vblk, struct virtblk_req *vbr)
+{
+       end_dequeued_request(vbr->req, !vbr->failed);
+       list_del(&vbr->list);
+       mempool_free(vbr, vblk->pool);
+       /* In case queue is stopped waiting for more buffers. */
+       blk_start_queue(vblk->disk->queue);
+}
+
+/* We make sure they finished both the input and output buffers: otherwise
+ * they might still have read access after we free them. */
+static void blk_out_done(struct virtio_device *vdev, void *_vbr, unsigned len)
+{
+       struct virtblk_req *vbr = _vbr;
+       struct virtio_blk *vblk = vdev->priv;
+
+       assert_spin_locked(&vblk->vdev->lock);
+
+       BUG_ON(vbr->out_done);
+       vbr->out_done = true;
+       if (vbr->in_done)
+               finish(vblk, vbr);
+}
+
+static void blk_in_done(struct virtio_device *vdev, void *_vbr, unsigned len)
+{
+       struct virtblk_req *vbr = _vbr;
+       struct virtio_blk *vblk = vdev->priv;
+       unsigned long expected_len;
+
+       assert_spin_locked(&vblk->vdev->lock);
+
+       expected_len = sizeof(vbr->in_hdr);
+       if (vbr->out_hdr.type == READ)
+               expected_len += vbr->req->hard_nr_sectors*512;
+
+       if (unlikely(len != expected_len)) {
+               dev_err(vblk->vdev->dev, "short reply %u not %lu",
+                       len, expected_len);
+               vbr->failed = true;
+       } else if (unlikely(vbr->in_hdr.status != 1)) {
+               vbr->failed = true;
+       }
+
+       BUG_ON(vbr->in_done);
+       vbr->in_done = true;
+       if (vbr->out_done)
+               finish(vblk, vbr);
+}
+
+static bool do_write(request_queue_t *q, struct virtio_blk *vblk,
+                    struct virtblk_req *vbr)
+{
+       unsigned long num;
+
+       /* Set up for reply. */
+       vblk->sg[0].page = virt_to_page(&vbr->in_hdr);
+       vblk->sg[0].offset = offset_in_page(&vbr->in_hdr);
+       vblk->sg[0].length = sizeof(vbr->in_hdr);
+       vbr->out_hdr.id = vblk->vdev->ops->add_inbuf(vblk->vdev, vblk->sg, 1,
+                                                    blk_in_done, vbr);
+       if (IS_ERR_VALUE(vbr->out_hdr.id))
+               goto full;
+
+       /* First sg element points to output header. */
+       vblk->sg[0].page = virt_to_page(&vbr->out_hdr);
+       vblk->sg[0].offset = offset_in_page(&vbr->out_hdr);
+       vblk->sg[0].length = sizeof(vbr->out_hdr);
+
+       num = blk_rq_map_sg(q, vbr->req, vblk->sg+1);
+       vbr->out_done = vbr->in_done = false;
+       vbr->failed = false;
+       vbr->out_id = vblk->vdev->ops->add_outbuf(vblk->vdev, vblk->sg, 1+num,
+                                                 blk_out_done, vbr);
+       if (IS_ERR_VALUE(vbr->out_id))
+               goto detach_inbuf_full;
+
+       pr_debug("Write: %p in=%lu out=%lu\n", vbr,
+                vbr->out_hdr.id, vbr->out_id);
+       list_add_tail(&vbr->list, &vblk->reqs);
+       return true;
+
+detach_inbuf_full:
+       vblk->vdev->ops->detach_inbuf(vblk->vdev, vbr->out_hdr.id);
+full:
+       return false;
+}
+
+static bool do_read(request_queue_t *q, struct virtio_blk *vblk,
+                   struct virtblk_req *vbr)
+{
+       unsigned long num;
+
+       /* Set up for reply. */
+       vblk->sg[0].page = virt_to_page(&vbr->in_hdr);
+       vblk->sg[0].offset = offset_in_page(&vbr->in_hdr);
+       vblk->sg[0].length = sizeof(vbr->in_hdr);
+       num = blk_rq_map_sg(q, vbr->req, vblk->sg+1);
+       vbr->out_hdr.id = vblk->vdev->ops->add_inbuf(vblk->vdev, vblk->sg,
+                                                    1+num, blk_in_done, vbr);
+       if (IS_ERR_VALUE(vbr->out_hdr.id))
+               goto full;
+
+       vblk->sg[0].page = virt_to_page(&vbr->out_hdr);
+       vblk->sg[0].offset = offset_in_page(&vbr->out_hdr);
+       vblk->sg[0].length = sizeof(vbr->out_hdr);
+
+       vbr->out_done = vbr->in_done = false;
+       vbr->failed = false;
+       vbr->out_id = vblk->vdev->ops->add_outbuf(vblk->vdev, vblk->sg, 1,
+                                                 blk_out_done, vbr);
+       if (IS_ERR_VALUE(vbr->out_id))
+               goto detach_inbuf_full;
+
+       pr_debug("Read: %p in=%lu out=%lu\n", vbr,
+                vbr->out_hdr.id, vbr->out_id);
+       list_add_tail(&vbr->list, &vblk->reqs);
+       return true;
+
+detach_inbuf_full:
+       vblk->vdev->ops->detach_inbuf(vblk->vdev, vbr->out_hdr.id);
+full:
+       return false;
+}
+
+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;
+
+               /* FIXME: handle these iff capable. */
+               if (!blk_fs_request(req)) {
+                       pr_debug("Got non-command 0x%08x\n", req->cmd_type);
+                       req->errors++;
+                       blkdev_dequeue_request(req);
+                       end_dequeued_request(req, 0);
+                       continue;
+               }
+
+               vbr = mempool_alloc(vblk->pool, GFP_ATOMIC);
+               if (!vbr)
+                       goto stop;
+
+               BUG_ON(req->nr_phys_segments > ARRAY_SIZE(vblk->sg));
+               vbr->req = req;
+               vbr->out_hdr.type = rq_data_dir(req);
+               vbr->out_hdr.sector = req->sector;
+
+               if (rq_data_dir(req) == WRITE) {
+                       if (!do_write(q, vblk, vbr))
+                               goto stop;
+               } else {
+                       if (!do_read(q, vblk, vbr))
+                               goto stop;
+               }
+               blkdev_dequeue_request(req);
+       }
+
+sync:
+       if (vblk)
+               vblk->vdev->ops->sync(vblk->vdev);
+       return;
+
+stop:
+       /* Queue full?  Wait. */
+       blk_stop_queue(q);
+       mempool_free(vbr, vblk->pool);
+       goto sync;
+}
+
+static struct block_device_operations virtblk_fops = {
+       .owner = THIS_MODULE,
+};
+
+struct gendisk *virtblk_probe(struct virtio_device *vdev)
+{
+       struct virtio_blk *vblk;
+       int err, major;
+
+       vblk = kmalloc(sizeof(*vblk), GFP_KERNEL);
+       if (!vblk) {
+               err = -ENOMEM;
+               goto out;
+       }
+
+       INIT_LIST_HEAD(&vblk->reqs);
+       vblk->vdev = vdev;
+       vdev->priv = vblk;
+
+       vblk->pool = mempool_create_kmalloc_pool(1,sizeof(struct virtblk_req));
+       if (!vblk->pool) {
+               err = -ENOMEM;
+               goto out_free_vblk;
+       }
+
+       major = register_blkdev(0, "virtblk");
+       if (major < 0) {
+               err = major;
+               goto out_mempool;
+       }
+
+       /* FIXME: How many partitions?  How long is a piece of string? */
+       vblk->disk = alloc_disk(1 << 3);
+       if (!vblk->disk) {
+               err = -ENOMEM;
+               goto out_unregister_blkdev;
+       }
+
+       vblk->disk->queue = blk_init_queue(do_virtblk_request,
+                                          &vblk->vdev->lock);
+       if (!vblk->disk->queue) {
+               err = -ENOMEM;
+               goto out_put_disk;
+       }
+
+       sprintf(vblk->disk->disk_name, "vb%c", virtblk_index++);
+       vblk->disk->major = major;
+       vblk->disk->first_minor = 0;
+       vblk->disk->private_data = vblk;
+       vblk->disk->fops = &virtblk_fops;
+
+       /* Caller can do blk_queue_max_hw_segments(), set_capacity()
+        * etc then add_disk(). */
+       return vblk->disk;
+
+out_put_disk:
+       put_disk(vblk->disk);
+out_unregister_blkdev:
+       unregister_blkdev(major, "virtblk");
+out_mempool:
+       mempool_destroy(vblk->pool);
+out_free_vblk:
+       kfree(vblk);
+out:
+       return ERR_PTR(err);
+}
+EXPORT_SYMBOL_GPL(virtblk_probe);
+
+void virtblk_remove(struct gendisk *disk)
+{
+       struct virtio_blk *vblk = disk->private_data;
+       int major = vblk->disk->major;
+
+       BUG_ON(!list_empty(&vblk->reqs));
+       blk_cleanup_queue(vblk->disk->queue);
+       put_disk(vblk->disk);
+       unregister_blkdev(major, "virtblk");
+       mempool_destroy(vblk->pool);
+       kfree(vblk);
+}
+EXPORT_SYMBOL_GPL(virtblk_remove);
===================================================================
--- /dev/null
+++ b/include/linux/virtio_blk.h
@@ -0,0 +1,28 @@
+#ifndef _LINUX_VIRTIO_BLK_H
+#define _LINUX_VIRTIO_BLK_H
+#include <linux/types.h>
+struct gendisk;
+struct virtio_device;
+struct hd_geometry;
+
+/* This is the first element of the scatter-gather list. */
+struct virtio_blk_outhdr
+{
+       /* 0 == read, 1 == write */
+       u32 type;
+       /* Sector (ie. 512 byte offset) */
+       unsigned long sector;
+       /* Where to put reply. */
+       unsigned long id;
+};
+
+struct virtio_blk_inhdr
+{
+       /* 1 = OK, 0 = not ok. */
+       unsigned long status;
+};
+
+struct gendisk *virtblk_probe(struct virtio_device *vdev);
+void virtblk_remove(struct gendisk *disk);
+
+#endif /* _LINUX_VIRTIO_BLK_H */



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel