vhost: replace vhost_workqueue with per-vhost kthread

Replace vhost_workqueue with per-vhost kthread.  Other than callback
argument change from struct work_struct * to struct vhost_work *,
there's no visible change to vhost_poll_*() interface.

This conversion is to make each vhost use a dedicated kthread so that
resource control via cgroup can be applied.

Partially based on Sridhar Samudrala's patch.

* Updated to use sub structure vhost_work instead of directly using
  vhost_poll at Michael's suggestion.

* Added flusher wake_up() optimization at Michael's suggestion.

Changes by MST:
* Converted atomics/barrier use to a spinlock.
* Create thread on SET_OWNER
* Fix flushing

Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Sridhar Samudrala <samudrala.sridhar@gmail.com>

3 files changed, 164 insertions, 73 deletions

diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c
index f11e6bb..d395b592 100644
--- a/drivers/vhost/net.c
+++ b/drivers/vhost/net.c
@@ -302,54 +302,58 @@ static void handle_rx(struct vhost_net *net)
 	unuse_mm(net->dev.mm);
 }
 
-static void handle_tx_kick(struct work_struct *work)
+static void handle_tx_kick(struct vhost_work *work)
 {
-	struct vhost_virtqueue *vq;
-	struct vhost_net *net;
-	vq = container_of(work, struct vhost_virtqueue, poll.work);
-	net = container_of(vq->dev, struct vhost_net, dev);
+	struct vhost_virtqueue *vq = container_of(work, struct vhost_virtqueue,
+						  poll.work);
+	struct vhost_net *net = container_of(vq->dev, struct vhost_net, dev);
+
 	handle_tx(net);
 }
 
-static void handle_rx_kick(struct work_struct *work)
+static void handle_rx_kick(struct vhost_work *work)
 {
-	struct vhost_virtqueue *vq;
-	struct vhost_net *net;
-	vq = container_of(work, struct vhost_virtqueue, poll.work);
-	net = container_of(vq->dev, struct vhost_net, dev);
+	struct vhost_virtqueue *vq = container_of(work, struct vhost_virtqueue,
+						  poll.work);
+	struct vhost_net *net = container_of(vq->dev, struct vhost_net, dev);
+
 	handle_rx(net);
 }
 
-static void handle_tx_net(struct work_struct *work)
+static void handle_tx_net(struct vhost_work *work)
 {
-	struct vhost_net *net;
-	net = container_of(work, struct vhost_net, poll[VHOST_NET_VQ_TX].work);
+	struct vhost_net *net = container_of(work, struct vhost_net,
+					     poll[VHOST_NET_VQ_TX].work);
 	handle_tx(net);
 }
 
-static void handle_rx_net(struct work_struct *work)
+static void handle_rx_net(struct vhost_work *work)
 {
-	struct vhost_net *net;
-	net = container_of(work, struct vhost_net, poll[VHOST_NET_VQ_RX].work);
+	struct vhost_net *net = container_of(work, struct vhost_net,
+					     poll[VHOST_NET_VQ_RX].work);
 	handle_rx(net);
 }
 
 static int vhost_net_open(struct inode *inode, struct file *f)
 {
 	struct vhost_net *n = kmalloc(sizeof *n, GFP_KERNEL);
+	struct vhost_dev *dev;
 	int r;
+
 	if (!n)
 		return -ENOMEM;
+
+	dev = &n->dev;
 	n->vqs[VHOST_NET_VQ_TX].handle_kick = handle_tx_kick;
 	n->vqs[VHOST_NET_VQ_RX].handle_kick = handle_rx_kick;
-	r = vhost_dev_init(&n->dev, n->vqs, VHOST_NET_VQ_MAX);
+	r = vhost_dev_init(dev, n->vqs, VHOST_NET_VQ_MAX);
 	if (r < 0) {
 		kfree(n);
 		return r;
 	}
 
-	vhost_poll_init(n->poll + VHOST_NET_VQ_TX, handle_tx_net, POLLOUT);
-	vhost_poll_init(n->poll + VHOST_NET_VQ_RX, handle_rx_net, POLLIN);
+	vhost_poll_init(n->poll + VHOST_NET_VQ_TX, handle_tx_net, POLLOUT, dev);
+	vhost_poll_init(n->poll + VHOST_NET_VQ_RX, handle_rx_net, POLLIN, dev);
 	n->tx_poll_state = VHOST_NET_POLL_DISABLED;
 
 	f->private_data = n;
@@ -656,25 +660,13 @@ static struct miscdevice vhost_net_misc = {
 
 static int vhost_net_init(void)
 {
-	int r = vhost_init();
-	if (r)
-		goto err_init;
-	r = misc_register(&vhost_net_misc);
-	if (r)
-		goto err_reg;
-	return 0;
-err_reg:
-	vhost_cleanup();
-err_init:
-	return r;
-
+	return misc_register(&vhost_net_misc);
 }
 module_init(vhost_net_init);
 
 static void vhost_net_exit(void)
 {
 	misc_deregister(&vhost_net_misc);
-	vhost_cleanup();
 }
 module_exit(vhost_net_exit);
 
diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index 248ed2d..30d93c2 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -17,12 +17,12 @@
 #include <linux/mm.h>
 #include <linux/miscdevice.h>
 #include <linux/mutex.h>
-#include <linux/workqueue.h>
 #include <linux/rcupdate.h>
 #include <linux/poll.h>
 #include <linux/file.h>
 #include <linux/highmem.h>
 #include <linux/slab.h>
+#include <linux/kthread.h>
 
 #include <linux/net.h>
 #include <linux/if_packet.h>
@@ -37,8 +37,6 @@ enum {
 	VHOST_MEMORY_F_LOG = 0x1,
 };
 
-static struct workqueue_struct *vhost_workqueue;
-
 static void vhost_poll_func(struct file *file, wait_queue_head_t *wqh,
 			    poll_table *pt)
 {
@@ -52,23 +50,31 @@ static void vhost_poll_func(struct file *file, wait_queue_head_t *wqh,
 static int vhost_poll_wakeup(wait_queue_t *wait, unsigned mode, int sync,
 			     void *key)
 {
-	struct vhost_poll *poll;
-	poll = container_of(wait, struct vhost_poll, wait);
+	struct vhost_poll *poll = container_of(wait, struct vhost_poll, wait);
+
 	if (!((unsigned long)key & poll->mask))
 		return 0;
 
-	queue_work(vhost_workqueue, &poll->work);
+	vhost_poll_queue(poll);
 	return 0;
 }
 
 /* Init poll structure */
-void vhost_poll_init(struct vhost_poll *poll, work_func_t func,
-		     unsigned long mask)
+void vhost_poll_init(struct vhost_poll *poll, vhost_work_fn_t fn,
+		     unsigned long mask, struct vhost_dev *dev)
 {
-	INIT_WORK(&poll->work, func);
+	struct vhost_work *work = &poll->work;
+
 	init_waitqueue_func_entry(&poll->wait, vhost_poll_wakeup);
 	init_poll_funcptr(&poll->table, vhost_poll_func);
 	poll->mask = mask;
+	poll->dev = dev;
+
+	INIT_LIST_HEAD(&work->node);
+	work->fn = fn;
+	init_waitqueue_head(&work->done);
+	work->flushing = 0;
+	work->queue_seq = work->done_seq = 0;
 }
 
 /* Start polling a file. We add ourselves to file's wait queue. The caller must
@@ -92,12 +98,40 @@ void vhost_poll_stop(struct vhost_poll *poll)
  * locks that are also used by the callback. */
 void vhost_poll_flush(struct vhost_poll *poll)
 {
-	flush_work(&poll->work);
+	struct vhost_work *work = &poll->work;
+	unsigned seq;
+	int left;
+	int flushing;
+
+	spin_lock_irq(&poll->dev->work_lock);
+	seq = work->queue_seq;
+	work->flushing++;
+	spin_unlock_irq(&poll->dev->work_lock);
+	wait_event(work->done, ({
+		   spin_lock_irq(&poll->dev->work_lock);
+		   left = seq - work->done_seq <= 0;
+		   spin_unlock_irq(&poll->dev->work_lock);
+		   left;
+	}));
+	spin_lock_irq(&poll->dev->work_lock);
+	flushing = --work->flushing;
+	spin_unlock_irq(&poll->dev->work_lock);
+	BUG_ON(flushing < 0);
 }
 
 void vhost_poll_queue(struct vhost_poll *poll)
 {
-	queue_work(vhost_workqueue, &poll->work);
+	struct vhost_dev *dev = poll->dev;
+	struct vhost_work *work = &poll->work;
+	unsigned long flags;
+
+	spin_lock_irqsave(&dev->work_lock, flags);
+	if (list_empty(&work->node)) {
+		list_add_tail(&work->node, &dev->work_list);
+		work->queue_seq++;
+		wake_up_process(dev->worker);
+	}
+	spin_unlock_irqrestore(&dev->work_lock, flags);
 }
 
 static void vhost_vq_reset(struct vhost_dev *dev,
@@ -125,10 +159,51 @@ static void vhost_vq_reset(struct vhost_dev *dev,
 	vq->log_ctx = NULL;
 }
 
+static int vhost_worker(void *data)
+{
+	struct vhost_dev *dev = data;
+	struct vhost_work *work = NULL;
+	unsigned uninitialized_var(seq);
+
+	for (;;) {
+		/* mb paired w/ kthread_stop */
+		set_current_state(TASK_INTERRUPTIBLE);
+
+		spin_lock_irq(&dev->work_lock);
+		if (work) {
+			work->done_seq = seq;
+			if (work->flushing)
+				wake_up_all(&work->done);
+		}
+
+		if (kthread_should_stop()) {
+			spin_unlock_irq(&dev->work_lock);
+			__set_current_state(TASK_RUNNING);
+			return 0;
+		}
+		if (!list_empty(&dev->work_list)) {
+			work = list_first_entry(&dev->work_list,
+						struct vhost_work, node);
+			list_del_init(&work->node);
+			seq = work->queue_seq;
+		} else
+			work = NULL;
+		spin_unlock_irq(&dev->work_lock);
+
+		if (work) {
+			__set_current_state(TASK_RUNNING);
+			work->fn(work);
+		} else
+			schedule();
+
+	}
+}
+
 long vhost_dev_init(struct vhost_dev *dev,
 		    struct vhost_virtqueue *vqs, int nvqs)
 {
 	int i;
+
 	dev->vqs = vqs;
 	dev->nvqs = nvqs;
 	mutex_init(&dev->mutex);
@@ -136,6 +211,9 @@ long vhost_dev_init(struct vhost_dev *dev,
 	dev->log_file = NULL;
 	dev->memory = NULL;
 	dev->mm = NULL;
+	spin_lock_init(&dev->work_lock);
+	INIT_LIST_HEAD(&dev->work_list);
+	dev->worker = NULL;
 
 	for (i = 0; i < dev->nvqs; ++i) {
 		dev->vqs[i].dev = dev;
@@ -143,9 +221,9 @@ long vhost_dev_init(struct vhost_dev *dev,
 		vhost_vq_reset(dev, dev->vqs + i);
 		if (dev->vqs[i].handle_kick)
 			vhost_poll_init(&dev->vqs[i].poll,
-					dev->vqs[i].handle_kick,
-					POLLIN);
+					dev->vqs[i].handle_kick, POLLIN, dev);
 	}
+
 	return 0;
 }
 
@@ -159,12 +237,31 @@ long vhost_dev_check_owner(struct vhost_dev *dev)
 /* Caller should have device mutex */
 static long vhost_dev_set_owner(struct vhost_dev *dev)
 {
+	struct task_struct *worker;
+	int err;
 	/* Is there an owner already? */
-	if (dev->mm)
-		return -EBUSY;
+	if (dev->mm) {
+		err = -EBUSY;
+		goto err_mm;
+	}
 	/* No owner, become one */
 	dev->mm = get_task_mm(current);
+	worker = kthread_create(vhost_worker, dev, "vhost-%d", current->pid);
+	if (IS_ERR(worker)) {
+		err = PTR_ERR(worker);
+		goto err_worker;
+	}
+
+	dev->worker = worker;
+	wake_up_process(worker);	/* avoid contributing to loadavg */
+
 	return 0;
+err_worker:
+	if (dev->mm)
+		mmput(dev->mm);
+	dev->mm = NULL;
+err_mm:
+	return err;
 }
 
 /* Caller should have device mutex */
@@ -217,6 +314,9 @@ void vhost_dev_cleanup(struct vhost_dev *dev)
 	if (dev->mm)
 		mmput(dev->mm);
 	dev->mm = NULL;
+
+	WARN_ON(!list_empty(&dev->work_list));
+	kthread_stop(dev->worker);
 }
 
 static int log_access_ok(void __user *log_base, u64 addr, unsigned long sz)
@@ -1115,16 +1215,3 @@ void vhost_disable_notify(struct vhost_virtqueue *vq)
 		vq_err(vq, "Failed to enable notification at %p: %d\n",
 		       &vq->used->flags, r);
 }
-
-int vhost_init(void)
-{
-	vhost_workqueue = create_singlethread_workqueue("vhost");
-	if (!vhost_workqueue)
-		return -ENOMEM;
-	return 0;
-}
-
-void vhost_cleanup(void)
-{
-	destroy_workqueue(vhost_workqueue);
-}
diff --git a/drivers/vhost/vhost.h b/drivers/vhost/vhost.h
index 11ee13d..3693327 100644
--- a/drivers/vhost/vhost.h
+++ b/drivers/vhost/vhost.h
@@ -5,13 +5,13 @@
 #include <linux/vhost.h>
 #include <linux/mm.h>
 #include <linux/mutex.h>
-#include <linux/workqueue.h>
 #include <linux/poll.h>
 #include <linux/file.h>
 #include <linux/skbuff.h>
 #include <linux/uio.h>
 #include <linux/virtio_config.h>
 #include <linux/virtio_ring.h>
+#include <asm/atomic.h>
 
 struct vhost_device;
 
@@ -20,19 +20,31 @@ enum {
 	VHOST_NET_MAX_SG = MAX_SKB_FRAGS + 2,
 };
 
+struct vhost_work;
+typedef void (*vhost_work_fn_t)(struct vhost_work *work);
+
+struct vhost_work {
+	struct list_head	  node;
+	vhost_work_fn_t		  fn;
+	wait_queue_head_t	  done;
+	int			  flushing;
+	unsigned		  queue_seq;
+	unsigned		  done_seq;
+};
+
 /* Poll a file (eventfd or socket) */
 /* Note: there's nothing vhost specific about this structure. */
 struct vhost_poll {
 	poll_table                table;
 	wait_queue_head_t        *wqh;
 	wait_queue_t              wait;
-	/* struct which will handle all actual work. */
-	struct work_struct        work;
+	struct vhost_work	  work;
 	unsigned long		  mask;
+	struct vhost_dev	 *dev;
 };
 
-void vhost_poll_init(struct vhost_poll *poll, work_func_t func,
-		     unsigned long mask);
+void vhost_poll_init(struct vhost_poll *poll, vhost_work_fn_t fn,
+		     unsigned long mask, struct vhost_dev *dev);
 void vhost_poll_start(struct vhost_poll *poll, struct file *file);
 void vhost_poll_stop(struct vhost_poll *poll);
 void vhost_poll_flush(struct vhost_poll *poll);
@@ -63,7 +75,7 @@ struct vhost_virtqueue {
 	struct vhost_poll poll;
 
 	/* The routine to call when the Guest pings us, or timeout. */
-	work_func_t handle_kick;
+	vhost_work_fn_t handle_kick;
 
 	/* Last available index we saw. */
 	u16 last_avail_idx;
@@ -86,11 +98,11 @@ struct vhost_virtqueue {
 	struct iovec hdr[VHOST_NET_MAX_SG];
 	size_t hdr_size;
 	/* We use a kind of RCU to access private pointer.
-	 * All readers access it from workqueue, which makes it possible to
-	 * flush the workqueue instead of synchronize_rcu. Therefore readers do
+	 * All readers access it from worker, which makes it possible to
+	 * flush the vhost_work instead of synchronize_rcu. Therefore readers do
 	 * not need to call rcu_read_lock/rcu_read_unlock: the beginning of
-	 * work item execution acts instead of rcu_read_lock() and the end of
-	 * work item execution acts instead of rcu_read_lock().
+	 * vhost_work execution acts instead of rcu_read_lock() and the end of
+	 * vhost_work execution acts instead of rcu_read_lock().
 	 * Writers use virtqueue mutex. */
 	void *private_data;
 	/* Log write descriptors */
@@ -110,6 +122,9 @@ struct vhost_dev {
 	int nvqs;
 	struct file *log_file;
 	struct eventfd_ctx *log_ctx;
+	spinlock_t work_lock;
+	struct list_head work_list;
+	struct task_struct *worker;
 };
 
 long vhost_dev_init(struct vhost_dev *, struct vhost_virtqueue *vqs, int nvqs);
@@ -136,9 +151,6 @@ bool vhost_enable_notify(struct vhost_virtqueue *);
 int vhost_log_write(struct vhost_virtqueue *vq, struct vhost_log *log,
 		    unsigned int log_num, u64 len);
 
-int vhost_init(void);
-void vhost_cleanup(void);
-
 #define vq_err(vq, fmt, ...) do {                                  \
 		pr_debug(pr_fmt(fmt), ##__VA_ARGS__);       \
 		if ((vq)->error_ctx)                               \