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.
We accept the normal SCSI ioctls: they get handed through to the other
side which can then handle it or reply that it's unsupported.
Although we try to reply -ENOTTY on unsupported commands, the block
layer in its infinite wisdom suppressed the error so ioctl(fd,
CDROMEJECT) returns success to userspace.
Signed-off-by: Rusty Russell <rusty@xxxxxxxxxxxxxxx>
---
drivers/block/Makefile | 1
drivers/block/virtio_blk.c | 385 ++++++++++++++++++++++++++++++++++++++++++++
include/linux/Kbuild | 1
include/linux/virtio_blk.h | 39 ++++
4 files changed, 426 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 +1,383 @@
+#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
+{
+ spinlock_t lock;
+
+ 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[2+MAX_PHYS_SEGMENTS];
+};
+
+struct virtblk_req
+{
+ struct list_head list;
+ struct request *req;
+ unsigned long out_id;
+ bool out_done, in_done;
+ int uptodate;
+ struct virtio_blk_outhdr out_hdr;
+ struct virtio_blk_inhdr in_hdr;
+};
+
+static void end_dequeued_request(struct request *req,
+ request_queue_t *q, int uptodate)
+{
+ /* And so the insanity of the block layer infects us here. */
+ int nsectors = req->hard_nr_sectors;
+
+ if (blk_pc_request(req)) {
+ nsectors = (req->data_len + 511) >> 9;
+ if (!nsectors)
+ nsectors = 1;
+ }
+ if (end_that_request_first(req, uptodate, nsectors))
+ BUG();
+ add_disk_randomness(req->rq_disk);
+ end_that_request_last(req, uptodate);
+}
+
+static bool finish(struct virtio_blk *vblk, struct virtblk_req *vbr)
+{
+ if (!vbr->in_done || !vbr->out_done)
+ return false;
+ end_dequeued_request(vbr->req, vblk->disk->queue, vbr->uptodate);
+ list_del(&vbr->list);
+ mempool_free(vbr, vblk->pool);
+ return true;
+}
+
+/* We make sure they finished both the input and output buffers: otherwise
+ * they might still have read access after we free them. */
+static bool blk_out_done(struct virtio_device *vdev)
+{
+ struct virtio_blk *vblk = vdev->priv;
+ struct virtblk_req *vbr;
+ unsigned int len, finished = 0;
+ unsigned long flags;
+
+ spin_lock_irqsave(&vblk->lock, flags);
+ while ((vbr = vdev->ops->get_outbuf(vdev, &len)) != NULL) {
+ BUG_ON(vbr->out_done);
+ vbr->out_done = true;
+ finished += finish(vblk, vbr);
+ }
+ /* In case queue is stopped waiting for more buffers. */
+ if (finished)
+ blk_start_queue(vblk->disk->queue);
+ spin_unlock_irqrestore(&vblk->lock, flags);
+ return true;
+}
+
+static bool blk_in_done(struct virtio_device *vdev)
+{
+ struct virtio_blk *vblk = vdev->priv;
+ struct virtblk_req *vbr;
+ unsigned int len, finished = 0;
+ unsigned long flags;
+
+ spin_lock_irqsave(&vblk->lock, flags);
+
+ while ((vbr = vdev->ops->get_inbuf(vdev, &len)) != NULL) {
+ BUG_ON(vbr->in_done);
+
+ switch (vbr->in_hdr.status) {
+ case VIRTIO_BLK_S_OK:
+ vbr->uptodate = 1;
+ break;
+ case VIRTIO_BLK_S_UNSUPP:
+ vbr->uptodate = -ENOTTY;
+ break;
+ default:
+ vbr->uptodate = 0;
+ break;
+ }
+ vbr->in_done = true;
+ finished += finish(vblk, vbr);
+ }
+
+ /* In case queue is stopped waiting for more buffers. */
+ if (finished)
+ blk_start_queue(vblk->disk->queue);
+ spin_unlock_irqrestore(&vblk->lock, flags);
+ return true;
+}
+
+static bool do_write(request_queue_t *q, struct virtio_blk *vblk,
+ struct virtblk_req *vbr)
+{
+ unsigned long num;
+
+ vbr->out_hdr.type |= VIRTIO_BLK_T_WRITE;
+
+ /* 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,
+ 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_id = vblk->vdev->ops->add_outbuf(vblk->vdev, vblk->sg, 1+num,
+ vbr);
+ if (IS_ERR_VALUE(vbr->out_id))
+ goto detach_inbuf_full;
+
+ pr_debug("Write: %p in=%lu out=%lu\n", vbr,
+ (long)vbr->out_hdr.id, (long)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;
+
+ vbr->out_hdr.type |= VIRTIO_BLK_T_READ;
+
+ /* 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, 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_id = vblk->vdev->ops->add_outbuf(vblk->vdev, vblk->sg, 1,
+ vbr);
+ if (IS_ERR_VALUE(vbr->out_id))
+ goto detach_inbuf_full;
+
+ pr_debug("Read: %p in=%lu out=%lu\n", vbr,
+ (long)vbr->out_hdr.id, (long)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_scsi(request_queue_t *q, struct virtio_blk *vblk,
+ struct virtblk_req *vbr)
+{
+ unsigned long num;
+
+ vbr->out_hdr.type |= VIRTIO_BLK_T_SCSI_CMD;
+
+ /* 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,
+ 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);
+ vblk->sg[1].page = virt_to_page(vbr->req->cmd);
+ vblk->sg[1].offset = offset_in_page(vbr->req->cmd);
+ vblk->sg[1].length = vbr->req->cmd_len;
+
+ num = blk_rq_map_sg(q, vbr->req, vblk->sg+1);
+ vbr->out_id = vblk->vdev->ops->add_outbuf(vblk->vdev, vblk->sg, 2+num,
+ vbr);
+ if (IS_ERR_VALUE(vbr->out_id))
+ goto detach_inbuf_full;
+
+ pr_debug("Scsi: %p in=%lu out=%lu\n", vbr,
+ (long)vbr->out_hdr.id, (long)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;
+
+ vbr = mempool_alloc(vblk->pool, GFP_ATOMIC);
+ if (!vbr)
+ goto stop;
+
+ BUG_ON(req->nr_phys_segments > ARRAY_SIZE(vblk->sg));
+ vbr->req = req;
+ /* Actual type gets or'ed in do_scsi/do_write/do_read */
+ vbr->out_hdr.type = blk_barrier_rq(req)?VIRTIO_BLK_T_BARRIER:0;
+ vbr->out_hdr.sector = req->sector;
+ vbr->out_hdr.ioprio = req->ioprio;
+ vbr->out_done = vbr->in_done = false;
+
+ if (blk_pc_request(req)) {
+ if (!do_scsi(q, vblk, vbr))
+ goto stop;
+ } else if (blk_fs_request(req)) {
+ if (rq_data_dir(req) == WRITE) {
+ if (!do_write(q, vblk, vbr))
+ goto stop;
+ } else {
+ if (!do_read(q, vblk, vbr))
+ goto stop;
+ }
+ } else
+ /* We don't put anything else in the queue. */
+ BUG();
+ blkdev_dequeue_request(req);
+ }
+
+sync:
+ if (vblk)
+ vblk->vdev->ops->sync(vblk->vdev, VIRTIO_IN|VIRTIO_OUT);
+ return;
+
+stop:
+ /* Queue full? Wait. */
+ blk_stop_queue(q);
+ mempool_free(vbr, vblk->pool);
+ goto sync;
+}
+
+static int virtblk_ioctl(struct inode *inode, struct file *filp,
+ unsigned cmd, unsigned long data)
+{
+ return scsi_cmd_ioctl(filp, inode->i_bdev->bd_disk, cmd,
+ (void __user *)data);
+}
+
+static struct virtio_driver_ops virtblk_ops = {
+ .in = blk_in_done,
+ .out = blk_out_done,
+};
+
+
+static struct block_device_operations virtblk_fops = {
+ .ioctl = virtblk_ioctl,
+ .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);
+ spin_lock_init(&vblk->lock);
+ vblk->vdev = vdev;
+ vdev->priv = vblk;
+ vdev->driver_ops = &virtblk_ops;
+
+ 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->lock);
+ if (!vblk->disk->queue) {
+ err = -ENOMEM;
+ goto out_put_disk;
+ }
+
+ sprintf(vblk->disk->disk_name, "vd%c", virtblk_index++);
+ vblk->disk->major = major;
+ vblk->disk->first_minor = 0;
+ vblk->disk->private_data = vblk;
+ vblk->disk->fops = &virtblk_fops;
+
+ blk_queue_ordered(vblk->disk->queue, QUEUE_ORDERED_TAG, NULL);
+
+ /* 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);
===================================================================
--- a/include/linux/Kbuild
+++ b/include/linux/Kbuild
@@ -341,6 +341,7 @@ unifdef-y += utsname.h
unifdef-y += utsname.h
unifdef-y += videodev2.h
unifdef-y += videodev.h
+unifdef-y += virtio_blk.h
unifdef-y += wait.h
unifdef-y += wanrouter.h
unifdef-y += watchdog.h
===================================================================
--- /dev/null
+++ b/include/linux/virtio_blk.h
@@ -0,0 +1,39 @@
+#ifndef _LINUX_VIRTIO_BLK_H
+#define _LINUX_VIRTIO_BLK_H
+#include <linux/types.h>
+
+#define VIRTIO_BLK_T_READ 0
+#define VIRTIO_BLK_T_WRITE 1
+#define VIRTIO_BLK_T_SCSI_CMD 3
+#define VIRTIO_BLK_T_BARRIER 0x80000000 /* Barrier before this op. */
+
+/* This is the first element of the scatter-gather list. */
+struct virtio_blk_outhdr
+{
+ /* VIRTIO_BLK_T* */
+ __u32 type;
+ /* io priority. */
+ __u32 ioprio;
+ /* Sector (ie. 512 byte offset) */
+ __u64 sector;
+ /* Where to put reply. */
+ __u64 id;
+};
+
+#define VIRTIO_BLK_S_OK 0
+#define VIRTIO_BLK_S_IOERR 1
+#define VIRTIO_BLK_S_UNSUPP 2
+
+struct virtio_blk_inhdr
+{
+ unsigned char status;
+};
+
+#ifdef __KERNEL__
+struct gendisk;
+struct virtio_device;
+
+struct gendisk *virtblk_probe(struct virtio_device *vdev);
+void virtblk_remove(struct gendisk *disk);
+#endif /* __KERNEL__ */
+#endif /* _LINUX_VIRTIO_BLK_H */
_______________________________________________
Virtualization mailing list
Virtualization@xxxxxxxxxxxxxxxxxxxxxxxxxx
https://lists.linux-foundation.org/mailman/listinfo/virtualization
