Date:	Fri,  1 Apr 2016 15:23:05 +0100
From:	Stefan Hajnoczi <stefanha@...hat.com>
To:	kvm@...r.kernel.org
Cc:	netdev@...r.kernel.org, "Michael S. Tsirkin" <mst@...hat.com>,
	Matt Benjamin <mbenjamin@...hat.com>,
	Christoffer Dall <christoffer.dall@...aro.org>,
	Alex Bennée <alex.bennee@...aro.org>,
	marius vlad <marius.vlad0@...il.com>, areis@...hat.com,
	Claudio Imbrenda <imbrenda@...ux.vnet.ibm.com>,
	Greg Kurz <gkurz@...ux.vnet.ibm.com>,
	Ian Campbell <ian.campbell@...ker.com>,
	virtualization@...ts.linux-foundation.org,
	Asias He <asias@...hat.com>,
	Stefan Hajnoczi <stefanha@...hat.com>
Subject: [RFC v5 3/5] VSOCK: Introduce virtio_transport.ko

From: Asias He <asias@...hat.com>

VM sockets virtio transport implementation.  This driver runs in the
guest.

Signed-off-by: Asias He <asias@...hat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@...hat.com>
---
v5:
 * Add transport reset event handling
 * Drop ctrl virtqueue
v4:
 * Add MAINTAINERS file entry
 * Drop short/long rx packets
 * checkpatch.pl cleanups
 * Clarify locking in struct virtio_vsock
 * Narrow local variable scopes as suggested by Alex Bennee
 * Call wake_up() after decrementing total_tx_buf to avoid deadlock
v2:
 * Fix total_tx_buf accounting
 * Add virtio_transport global mutex to prevent races
---
 MAINTAINERS                      |   1 +
 net/vmw_vsock/virtio_transport.c | 584 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 585 insertions(+)
 create mode 100644 net/vmw_vsock/virtio_transport.c

diff --git a/MAINTAINERS b/MAINTAINERS
index fb9c47a..1ed7364 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -11611,6 +11611,7 @@ S:	Maintained
 F:	include/linux/virtio_vsock.h
 F:	include/uapi/linux/virtio_vsock.h
 F:	net/vmw_vsock/virtio_transport_common.c
+F:	net/vmw_vsock/virtio_transport.c
 
 VIRTUAL SERIO DEVICE DRIVER
 M:	Stephen Chandler Paul <thatslyude@...il.com>
diff --git a/net/vmw_vsock/virtio_transport.c b/net/vmw_vsock/virtio_transport.c
new file mode 100644
index 0000000..45472e0
--- /dev/null
+++ b/net/vmw_vsock/virtio_transport.c
@@ -0,0 +1,584 @@
+/*
+ * virtio transport for vsock
+ *
+ * Copyright (C) 2013-2015 Red Hat, Inc.
+ * Author: Asias He <asias@...hat.com>
+ *         Stefan Hajnoczi <stefanha@...hat.com>
+ *
+ * Some of the code is take from Gerd Hoffmann <kraxel@...hat.com>'s
+ * early virtio-vsock proof-of-concept bits.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2.
+ */
+#include <linux/spinlock.h>
+#include <linux/module.h>
+#include <linux/list.h>
+#include <linux/virtio.h>
+#include <linux/virtio_ids.h>
+#include <linux/virtio_config.h>
+#include <linux/virtio_vsock.h>
+#include <net/sock.h>
+#include <linux/mutex.h>
+#include <net/af_vsock.h>
+
+static struct workqueue_struct *virtio_vsock_workqueue;
+static struct virtio_vsock *the_virtio_vsock;
+static DEFINE_MUTEX(the_virtio_vsock_mutex); /* protects the_virtio_vsock */
+static void virtio_vsock_rx_fill(struct virtio_vsock *vsock);
+
+struct virtio_vsock {
+	struct virtio_device *vdev;
+	struct virtqueue *vqs[VSOCK_VQ_MAX];
+
+	/* Virtqueue processing is deferred to a workqueue */
+	struct work_struct tx_work;
+	struct work_struct rx_work;
+	struct work_struct event_work;
+
+	wait_queue_head_t tx_wait;	/* for waiting for tx resources */
+
+	/* The following fields are protected by tx_lock.  vqs[VSOCK_VQ_TX]
+	 * must be accessed with tx_lock held.
+	 */
+	struct mutex tx_lock;
+	u32 total_tx_buf;
+
+	/* The following fields are protected by rx_lock.  vqs[VSOCK_VQ_RX]
+	 * must be accessed with rx_lock held.
+	 */
+	struct mutex rx_lock;
+	int rx_buf_nr;
+	int rx_buf_max_nr;
+
+	/* The following fields are protected by event_lock.
+	 * vqs[VSOCK_VQ_EVENT] must be accessed with event_lock held.
+	 */
+	struct mutex event_lock;
+	struct virtio_vsock_event event_list[8];
+
+	u32 guest_cid;
+};
+
+static struct virtio_vsock *virtio_vsock_get(void)
+{
+	return the_virtio_vsock;
+}
+
+static u32 virtio_transport_get_local_cid(void)
+{
+	struct virtio_vsock *vsock = virtio_vsock_get();
+
+	return vsock->guest_cid;
+}
+
+static int
+virtio_transport_send_one_pkt(struct virtio_vsock *vsock,
+			      struct virtio_vsock_pkt *pkt)
+{
+	struct scatterlist hdr, buf, *sgs[2];
+	int ret, in_sg = 0, out_sg = 0;
+	struct virtqueue *vq;
+	DEFINE_WAIT(wait);
+
+	vq = vsock->vqs[VSOCK_VQ_TX];
+
+	/* Put pkt in the virtqueue */
+	sg_init_one(&hdr, &pkt->hdr, sizeof(pkt->hdr));
+	sgs[out_sg++] = &hdr;
+	if (pkt->buf) {
+		sg_init_one(&buf, pkt->buf, pkt->len);
+		sgs[out_sg++] = &buf;
+	}
+
+	mutex_lock(&vsock->tx_lock);
+	while ((ret = virtqueue_add_sgs(vq, sgs, out_sg, in_sg, pkt,
+					GFP_KERNEL)) < 0) {
+		prepare_to_wait_exclusive(&vsock->tx_wait, &wait,
+					  TASK_UNINTERRUPTIBLE);
+		mutex_unlock(&vsock->tx_lock);
+		schedule();
+		mutex_lock(&vsock->tx_lock);
+		finish_wait(&vsock->tx_wait, &wait);
+	}
+	virtqueue_kick(vq);
+	mutex_unlock(&vsock->tx_lock);
+
+	return pkt->len;
+}
+
+static int
+virtio_transport_send_pkt_no_sock(struct virtio_vsock_pkt *pkt)
+{
+	struct virtio_vsock *vsock;
+
+	vsock = virtio_vsock_get();
+	if (!vsock) {
+		virtio_transport_free_pkt(pkt);
+		return -ENODEV;
+	}
+
+	return virtio_transport_send_one_pkt(vsock, pkt);
+}
+
+static int
+virtio_transport_send_pkt(struct vsock_sock *vsk,
+			  struct virtio_vsock_pkt_info *info)
+{
+	u32 src_cid, src_port, dst_cid, dst_port;
+	struct virtio_vsock_sock *vvs;
+	struct virtio_vsock_pkt *pkt;
+	struct virtio_vsock *vsock;
+	u32 pkt_len = info->pkt_len;
+	DEFINE_WAIT(wait);
+
+	vsock = virtio_vsock_get();
+	if (!vsock)
+		return -ENODEV;
+
+	src_cid	= virtio_transport_get_local_cid();
+	src_port = vsk->local_addr.svm_port;
+	if (!info->remote_cid) {
+		dst_cid	= vsk->remote_addr.svm_cid;
+		dst_port = vsk->remote_addr.svm_port;
+	} else {
+		dst_cid = info->remote_cid;
+		dst_port = info->remote_port;
+	}
+
+	vvs = vsk->trans;
+
+	if (pkt_len > VIRTIO_VSOCK_DEFAULT_RX_BUF_SIZE)
+		pkt_len = VIRTIO_VSOCK_DEFAULT_RX_BUF_SIZE;
+	pkt_len = virtio_transport_get_credit(vvs, pkt_len);
+	/* Do not send zero length OP_RW pkt*/
+	if (pkt_len == 0 && info->op == VIRTIO_VSOCK_OP_RW)
+		return pkt_len;
+
+	/* Respect global tx buf limitation */
+	mutex_lock(&vsock->tx_lock);
+	while (pkt_len + vsock->total_tx_buf > VIRTIO_VSOCK_MAX_TX_BUF_SIZE) {
+		prepare_to_wait_exclusive(&vsock->tx_wait, &wait,
+					  TASK_UNINTERRUPTIBLE);
+		mutex_unlock(&vsock->tx_lock);
+		schedule();
+		mutex_lock(&vsock->tx_lock);
+		finish_wait(&vsock->tx_wait, &wait);
+	}
+	vsock->total_tx_buf += pkt_len;
+	mutex_unlock(&vsock->tx_lock);
+
+	pkt = virtio_transport_alloc_pkt(info, pkt_len,
+					 src_cid, src_port,
+					 dst_cid, dst_port);
+	if (!pkt) {
+		mutex_lock(&vsock->tx_lock);
+		vsock->total_tx_buf -= pkt_len;
+		mutex_unlock(&vsock->tx_lock);
+		virtio_transport_put_credit(vvs, pkt_len);
+		wake_up(&vsock->tx_wait);
+		return -ENOMEM;
+	}
+
+	virtio_transport_inc_tx_pkt(vvs, pkt);
+
+	return virtio_transport_send_one_pkt(vsock, pkt);
+}
+
+static void virtio_vsock_rx_fill(struct virtio_vsock *vsock)
+{
+	int buf_len = VIRTIO_VSOCK_DEFAULT_RX_BUF_SIZE;
+	struct virtio_vsock_pkt *pkt;
+	struct scatterlist hdr, buf, *sgs[2];
+	struct virtqueue *vq;
+	int ret;
+
+	vq = vsock->vqs[VSOCK_VQ_RX];
+
+	do {
+		pkt = kzalloc(sizeof(*pkt), GFP_KERNEL);
+		if (!pkt)
+			break;
+
+		pkt->buf = kmalloc(buf_len, GFP_KERNEL);
+		if (!pkt->buf) {
+			virtio_transport_free_pkt(pkt);
+			break;
+		}
+
+		pkt->len = buf_len;
+
+		sg_init_one(&hdr, &pkt->hdr, sizeof(pkt->hdr));
+		sgs[0] = &hdr;
+
+		sg_init_one(&buf, pkt->buf, buf_len);
+		sgs[1] = &buf;
+		ret = virtqueue_add_sgs(vq, sgs, 0, 2, pkt, GFP_KERNEL);
+		if (ret) {
+			virtio_transport_free_pkt(pkt);
+			break;
+		}
+		vsock->rx_buf_nr++;
+	} while (vq->num_free);
+	if (vsock->rx_buf_nr > vsock->rx_buf_max_nr)
+		vsock->rx_buf_max_nr = vsock->rx_buf_nr;
+	virtqueue_kick(vq);
+}
+
+static void virtio_transport_send_pkt_work(struct work_struct *work)
+{
+	struct virtio_vsock *vsock =
+		container_of(work, struct virtio_vsock, tx_work);
+	struct virtqueue *vq;
+	bool added = false;
+
+	vq = vsock->vqs[VSOCK_VQ_TX];
+	mutex_lock(&vsock->tx_lock);
+	do {
+		struct virtio_vsock_pkt *pkt;
+		unsigned int len;
+
+		virtqueue_disable_cb(vq);
+		while ((pkt = virtqueue_get_buf(vq, &len)) != NULL) {
+			vsock->total_tx_buf -= pkt->len;
+			virtio_transport_free_pkt(pkt);
+			added = true;
+		}
+	} while (!virtqueue_enable_cb(vq));
+	mutex_unlock(&vsock->tx_lock);
+
+	if (added)
+		wake_up(&vsock->tx_wait);
+}
+
+static void virtio_transport_recv_pkt_work(struct work_struct *work)
+{
+	struct virtio_vsock *vsock =
+		container_of(work, struct virtio_vsock, rx_work);
+	struct virtqueue *vq;
+
+	vq = vsock->vqs[VSOCK_VQ_RX];
+	mutex_lock(&vsock->rx_lock);
+	do {
+		struct virtio_vsock_pkt *pkt;
+		unsigned int len;
+
+		virtqueue_disable_cb(vq);
+		while ((pkt = virtqueue_get_buf(vq, &len)) != NULL) {
+			vsock->rx_buf_nr--;
+
+			/* Drop short/long packets */
+			if (unlikely(len < sizeof(pkt->hdr) ||
+				     len > sizeof(pkt->hdr) + pkt->len)) {
+				virtio_transport_free_pkt(pkt);
+				continue;
+			}
+
+			pkt->len = len - sizeof(pkt->hdr);
+			virtio_transport_recv_pkt(pkt);
+		}
+	} while (!virtqueue_enable_cb(vq));
+
+	if (vsock->rx_buf_nr < vsock->rx_buf_max_nr / 2)
+		virtio_vsock_rx_fill(vsock);
+	mutex_unlock(&vsock->rx_lock);
+}
+
+/* event_lock must be held */
+static int virtio_vsock_event_fill_one(struct virtio_vsock *vsock,
+				       struct virtio_vsock_event *event)
+{
+	struct scatterlist sg;
+	struct virtqueue *vq;
+
+	vq = vsock->vqs[VSOCK_VQ_EVENT];
+
+	sg_init_one(&sg, event, sizeof(*event));
+
+	return virtqueue_add_inbuf(vq, &sg, 1, event, GFP_KERNEL);
+}
+
+/* event_lock must be held */
+static void virtio_vsock_event_fill(struct virtio_vsock *vsock)
+{
+	size_t i;
+
+	for (i = 0; i < ARRAY_SIZE(vsock->event_list); i++) {
+		struct virtio_vsock_event *event = &vsock->event_list[i];
+
+		virtio_vsock_event_fill_one(vsock, event);
+	}
+
+	virtqueue_kick(vsock->vqs[VSOCK_VQ_EVENT]);
+}
+
+static void virtio_vsock_reset_sock(struct sock *sk)
+{
+	lock_sock(sk);
+	sk->sk_state = SS_UNCONNECTED;
+	sk->sk_err = ECONNRESET;
+	sk->sk_error_report(sk);
+	release_sock(sk);
+}
+
+static void virtio_vsock_update_guest_cid(struct virtio_vsock *vsock)
+{
+	struct virtio_device *vdev = vsock->vdev;
+	u32 guest_cid;
+
+	vdev->config->get(vdev, offsetof(struct virtio_vsock_config, guest_cid),
+			  &guest_cid, sizeof(guest_cid));
+	vsock->guest_cid = le32_to_cpu(guest_cid);
+}
+
+/* event_lock must be held */
+static void virtio_vsock_event_handle(struct virtio_vsock *vsock,
+				      struct virtio_vsock_event *event)
+{
+	switch (le32_to_cpu(event->id)) {
+	case VIRTIO_VSOCK_EVENT_TRANSPORT_RESET:
+		virtio_vsock_update_guest_cid(vsock);
+		vsock_for_each_connected_socket(virtio_vsock_reset_sock);
+		break;
+	}
+}
+
+static void virtio_transport_event_work(struct work_struct *work)
+{
+	struct virtio_vsock *vsock =
+		container_of(work, struct virtio_vsock, event_work);
+	struct virtqueue *vq;
+
+	vq = vsock->vqs[VSOCK_VQ_EVENT];
+
+	mutex_lock(&vsock->event_lock);
+
+	do {
+		struct virtio_vsock_event *event;
+		unsigned int len;
+
+		virtqueue_disable_cb(vq);
+		while ((event = virtqueue_get_buf(vq, &len)) != NULL) {
+			if (len == sizeof(*event))
+				virtio_vsock_event_handle(vsock, event);
+
+			virtio_vsock_event_fill_one(vsock, event);
+		}
+	} while (!virtqueue_enable_cb(vq));
+
+	virtqueue_kick(vsock->vqs[VSOCK_VQ_EVENT]);
+
+	mutex_unlock(&vsock->event_lock);
+}
+
+static void virtio_vsock_event_done(struct virtqueue *vq)
+{
+	struct virtio_vsock *vsock = vq->vdev->priv;
+
+	if (!vsock)
+		return;
+	queue_work(virtio_vsock_workqueue, &vsock->event_work);
+}
+
+static void virtio_vsock_tx_done(struct virtqueue *vq)
+{
+	struct virtio_vsock *vsock = vq->vdev->priv;
+
+	if (!vsock)
+		return;
+	queue_work(virtio_vsock_workqueue, &vsock->tx_work);
+}
+
+static void virtio_vsock_rx_done(struct virtqueue *vq)
+{
+	struct virtio_vsock *vsock = vq->vdev->priv;
+
+	if (!vsock)
+		return;
+	queue_work(virtio_vsock_workqueue, &vsock->rx_work);
+}
+
+static struct virtio_transport virtio_transport = {
+	.transport = {
+		.get_local_cid            = virtio_transport_get_local_cid,
+
+		.init                     = virtio_transport_do_socket_init,
+		.destruct                 = virtio_transport_destruct,
+		.release                  = virtio_transport_release,
+		.connect                  = virtio_transport_connect,
+		.shutdown                 = virtio_transport_shutdown,
+
+		.dgram_bind               = virtio_transport_dgram_bind,
+		.dgram_dequeue            = virtio_transport_dgram_dequeue,
+		.dgram_enqueue            = virtio_transport_dgram_enqueue,
+		.dgram_allow              = virtio_transport_dgram_allow,
+
+		.stream_dequeue           = virtio_transport_stream_dequeue,
+		.stream_enqueue           = virtio_transport_stream_enqueue,
+		.stream_has_data          = virtio_transport_stream_has_data,
+		.stream_has_space         = virtio_transport_stream_has_space,
+		.stream_rcvhiwat          = virtio_transport_stream_rcvhiwat,
+		.stream_is_active         = virtio_transport_stream_is_active,
+		.stream_allow             = virtio_transport_stream_allow,
+
+		.notify_poll_in           = virtio_transport_notify_poll_in,
+		.notify_poll_out          = virtio_transport_notify_poll_out,
+		.notify_recv_init         = virtio_transport_notify_recv_init,
+		.notify_recv_pre_block    = virtio_transport_notify_recv_pre_block,
+		.notify_recv_pre_dequeue  = virtio_transport_notify_recv_pre_dequeue,
+		.notify_recv_post_dequeue = virtio_transport_notify_recv_post_dequeue,
+		.notify_send_init         = virtio_transport_notify_send_init,
+		.notify_send_pre_block    = virtio_transport_notify_send_pre_block,
+		.notify_send_pre_enqueue  = virtio_transport_notify_send_pre_enqueue,
+		.notify_send_post_enqueue = virtio_transport_notify_send_post_enqueue,
+
+		.set_buffer_size          = virtio_transport_set_buffer_size,
+		.set_min_buffer_size      = virtio_transport_set_min_buffer_size,
+		.set_max_buffer_size      = virtio_transport_set_max_buffer_size,
+		.get_buffer_size          = virtio_transport_get_buffer_size,
+		.get_min_buffer_size      = virtio_transport_get_min_buffer_size,
+		.get_max_buffer_size      = virtio_transport_get_max_buffer_size,
+	},
+
+	.send_pkt		= virtio_transport_send_pkt,
+	.send_pkt_no_sock	= virtio_transport_send_pkt_no_sock,
+};
+
+static int virtio_vsock_probe(struct virtio_device *vdev)
+{
+	vq_callback_t *callbacks[] = {
+		virtio_vsock_rx_done,
+		virtio_vsock_tx_done,
+		virtio_vsock_event_done,
+	};
+	static const char * const names[] = {
+		"rx",
+		"tx",
+		"event",
+	};
+	struct virtio_vsock *vsock = NULL;
+	int ret;
+
+	ret = mutex_lock_interruptible(&the_virtio_vsock_mutex);
+	if (ret)
+		return ret;
+
+	/* Only one virtio-vsock device per guest is supported */
+	if (the_virtio_vsock) {
+		ret = -EBUSY;
+		goto out;
+	}
+
+	vsock = kzalloc(sizeof(*vsock), GFP_KERNEL);
+	if (!vsock) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	vsock->vdev = vdev;
+
+	ret = vsock->vdev->config->find_vqs(vsock->vdev, VSOCK_VQ_MAX,
+					    vsock->vqs, callbacks, names);
+	if (ret < 0)
+		goto out;
+
+	virtio_vsock_update_guest_cid(vsock);
+
+	ret = vsock_core_init(&virtio_transport.transport);
+	if (ret < 0)
+		goto out_vqs;
+
+	vsock->rx_buf_nr = 0;
+	vsock->rx_buf_max_nr = 0;
+
+	vdev->priv = vsock;
+	the_virtio_vsock = vsock;
+	init_waitqueue_head(&vsock->tx_wait);
+	mutex_init(&vsock->tx_lock);
+	mutex_init(&vsock->rx_lock);
+	mutex_init(&vsock->event_lock);
+	INIT_WORK(&vsock->rx_work, virtio_transport_recv_pkt_work);
+	INIT_WORK(&vsock->tx_work, virtio_transport_send_pkt_work);
+	INIT_WORK(&vsock->event_work, virtio_transport_event_work);
+
+	mutex_lock(&vsock->rx_lock);
+	virtio_vsock_rx_fill(vsock);
+	mutex_unlock(&vsock->rx_lock);
+
+	mutex_lock(&vsock->event_lock);
+	virtio_vsock_event_fill(vsock);
+	mutex_unlock(&vsock->event_lock);
+
+	mutex_unlock(&the_virtio_vsock_mutex);
+	return 0;
+
+out_vqs:
+	vsock->vdev->config->del_vqs(vsock->vdev);
+out:
+	kfree(vsock);
+	mutex_unlock(&the_virtio_vsock_mutex);
+	return ret;
+}
+
+static void virtio_vsock_remove(struct virtio_device *vdev)
+{
+	struct virtio_vsock *vsock = vdev->priv;
+
+	flush_work(&vsock->rx_work);
+	flush_work(&vsock->tx_work);
+	flush_work(&vsock->event_work);
+
+	vdev->config->reset(vdev);
+
+	mutex_lock(&the_virtio_vsock_mutex);
+	the_virtio_vsock = NULL;
+	vsock_core_exit();
+	mutex_unlock(&the_virtio_vsock_mutex);
+
+	vdev->config->del_vqs(vdev);
+
+	kfree(vsock);
+}
+
+static struct virtio_device_id id_table[] = {
+	{ VIRTIO_ID_VSOCK, VIRTIO_DEV_ANY_ID },
+	{ 0 },
+};
+
+static unsigned int features[] = {
+};
+
+static struct virtio_driver virtio_vsock_driver = {
+	.feature_table = features,
+	.feature_table_size = ARRAY_SIZE(features),
+	.driver.name = KBUILD_MODNAME,
+	.driver.owner = THIS_MODULE,
+	.id_table = id_table,
+	.probe = virtio_vsock_probe,
+	.remove = virtio_vsock_remove,
+};
+
+static int __init virtio_vsock_init(void)
+{
+	int ret;
+
+	virtio_vsock_workqueue = alloc_workqueue("virtio_vsock", 0, 0);
+	if (!virtio_vsock_workqueue)
+		return -ENOMEM;
+	ret = register_virtio_driver(&virtio_vsock_driver);
+	if (ret)
+		destroy_workqueue(virtio_vsock_workqueue);
+	return ret;
+}
+
+static void __exit virtio_vsock_exit(void)
+{
+	unregister_virtio_driver(&virtio_vsock_driver);
+	destroy_workqueue(virtio_vsock_workqueue);
+}
+
+module_init(virtio_vsock_init);
+module_exit(virtio_vsock_exit);
+MODULE_LICENSE("GPL v2");
+MODULE_AUTHOR("Asias He");
+MODULE_DESCRIPTION("virtio transport for vsock");
+MODULE_DEVICE_TABLE(virtio, id_table);
-- 
2.5.5

Powered by blists - more mailing lists