Re: [PATCH v7 3/3] tcm_vhost: Add hotplug/hotunplug support

["Michael S. Tsirkin" <mst@redhat.com>]

On Thu, Apr 18, 2013 at 04:59:08PM +0800, Asias He wrote:
> On Thu, Apr 18, 2013 at 10:34:29AM +0300, Michael S. Tsirkin wrote:
> > On Thu, Apr 18, 2013 at 09:05:54AM +0800, Asias He wrote:
> > > In commit 365a7150094 ([SCSI] virtio-scsi: hotplug support for
> > > virtio-scsi), hotplug support is added to virtio-scsi.
> > > 
> > > This patch adds hotplug and hotunplug support to tcm_vhost.
> > > 
> > > You can create or delete a LUN in targetcli to hotplug or hotunplug a
> > > LUN in guest.
> > > 
> > > Changes in v7:
> > > - Add vhost_work_flush for vs->vs_event_work to this series
> > > 
> > > Changes in v6:
> > > - Pass tcm_vhost_evt to tcm_vhost_do_evt_work
> > > 
> > > Changes in v5:
> > > - Switch to int from u64 to vs_events_nr
> > > - Set s->vs_events_dropped flag in tcm_vhost_allocate_evt
> > > - Do not nest dev mutex within vq mutex
> > > - Use vs_events_lock to protect vs_events_dropped and vs_events_nr
> > > - Rebase to target/master
> > > 
> > > Changes in v4:
> > > - Drop tcm_vhost_check_endpoint in tcm_vhost_send_evt
> > > - Add tcm_vhost_check_endpoint in vhost_scsi_evt_handle_kick
> > > 
> > > Changes in v3:
> > > - Separate the bug fix to another thread
> > > 
> > > Changes in v2:
> > > - Remove code duplication in tcm_vhost_{hotplug,hotunplug}
> > > - Fix racing of vs_events_nr
> > > - Add flush fix patch to this series
> > > 
> > > Signed-off-by: Asias He <asias@redhat.com>
> > > Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
> > > ---
> > >  drivers/vhost/tcm_vhost.c | 208 +++++++++++++++++++++++++++++++++++++++++++++-
> > >  drivers/vhost/tcm_vhost.h |  10 +++
> > >  2 files changed, 214 insertions(+), 4 deletions(-)
> > > 
> > > diff --git a/drivers/vhost/tcm_vhost.c b/drivers/vhost/tcm_vhost.c
> > > index 8f05528..da2021b 100644
> > > --- a/drivers/vhost/tcm_vhost.c
> > > +++ b/drivers/vhost/tcm_vhost.c
> > > @@ -66,11 +66,13 @@ enum {
> > >   * TODO: debug and remove the workaround.
> > >   */
> > >  enum {
> > > -	VHOST_SCSI_FEATURES = VHOST_FEATURES & (~VIRTIO_RING_F_EVENT_IDX)
> > > +	VHOST_SCSI_FEATURES = (VHOST_FEATURES & (~VIRTIO_RING_F_EVENT_IDX)) |
> > > +			      (1ULL << VIRTIO_SCSI_F_HOTPLUG)
> > >  };
> > >  
> > >  #define VHOST_SCSI_MAX_TARGET	256
> > >  #define VHOST_SCSI_MAX_VQ	128
> > > +#define VHOST_SCSI_MAX_EVENT	128
> > >  
> > >  struct vhost_scsi {
> > >  	/* Protected by vhost_scsi->dev.mutex */
> > > @@ -82,6 +84,13 @@ struct vhost_scsi {
> > >  
> > >  	struct vhost_work vs_completion_work; /* cmd completion work item */
> > >  	struct llist_head vs_completion_list; /* cmd completion queue */
> > > +
> > > +	struct vhost_work vs_event_work; /* evt injection work item */
> > > +	struct llist_head vs_event_list; /* evt injection queue */
> > > +
> > > +	struct mutex vs_events_lock; /* protect vs_events_dropped,events_nr */
> > 
> > Looking at this code, there are just so many locks now.
> > This does not make me feel safe :)
> > At least, please document lock nesting.
> > 
> 
> Do you see a real problem?

Complexity is a real problem. My head already spins. No I don't see a
bug, but we need to simplify locking.

And I think I see a nice way to do this:
1. move away from a global work to per-event work - so no list
2. remove dynamic allocation of events - so no events_nr
3. set/test overrun flag under the appropriate vq mutex

I think that's ideal.  But we can move there in small steps.  As a first
step - why can't we always take the vq mutex lock and drop
vs_events_lock?

> > > +	bool vs_events_dropped; /* any missed events */
> > > +	int vs_events_nr; /* num of pending events */
> > >  };
> > >  
> > >  /* Local pointer to allocated TCM configfs fabric module */
> > > @@ -129,6 +138,17 @@ static bool tcm_vhost_check_endpoint(struct vhost_virtqueue *vq)
> > >  	return ret;
> > >  }
> > >  
> > > +static bool tcm_vhost_check_events_dropped(struct vhost_scsi *vs)
> > > +{
> > > +	bool ret;
> > > +
> > > +	mutex_lock(&vs->vs_events_lock);
> > > +	ret = vs->vs_events_dropped;
> > > +	mutex_unlock(&vs->vs_events_lock);
> > > +
> > > +	return ret;
> > > +}
> > > +
> > >  static int tcm_vhost_check_true(struct se_portal_group *se_tpg)
> > >  {
> > >  	return 1;
> > > @@ -379,6 +399,37 @@ static int tcm_vhost_queue_tm_rsp(struct se_cmd *se_cmd)
> > >  	return 0;
> > >  }
> > >  
> > > +static void tcm_vhost_free_evt(struct vhost_scsi *vs, struct tcm_vhost_evt *evt)
> > > +{
> > > +	mutex_lock(&vs->vs_events_lock);
> > > +	vs->vs_events_nr--;
> > > +	kfree(evt);
> > > +	mutex_unlock(&vs->vs_events_lock);
> > > +}
> > > +
> > > +static struct tcm_vhost_evt *tcm_vhost_allocate_evt(struct vhost_scsi *vs,
> > > +	u32 event, u32 reason)
> > > +{
> > > +	struct tcm_vhost_evt *evt;
> > > +
> > > +	mutex_lock(&vs->vs_events_lock);
> > > +	if (vs->vs_events_nr > VHOST_SCSI_MAX_EVENT) {
> > > +		vs->vs_events_dropped = true;
> > > +		mutex_unlock(&vs->vs_events_lock);
> > > +		return NULL;
> > > +	}
> > > +
> > > +	evt = kzalloc(sizeof(*evt), GFP_KERNEL);
> > 
> > I think kzalloc not needed here, you init all fields.
> 
> Not really! evt->event.lun[4-7] is not initialized. It needs to be 0.

So that is 4 bytes just init them when you set rest of lun.

> > Also, this basically means if backend does plug/unplug very quickly,
> > we start consuming kernel memory without a limit. Not good.
> > How about we allocate the event as part of target?
> > There shouldn't ever be more than one hotplug
> > event in flight per target, right?
> 
> It is limited by VHOST_SCSI_MAX_EVENT.

OK I missed that. But where does VHOST_SCSI_MAX_EVENT come from by the
way?  I don't see it in spec or did I miss it?  It seems it's best not
to lost events as long as there are descriptors in the event vq.

> > 
> > > +	if (evt) {
> > 
> > Let's do clean error handling here and above:
> > if (!evt)
> > 	goto err;
> 
> We can you do in err? You simply unlock and return. Why bother? How
> cleaner it will be.

There's another error above and we'll share code.
It's not a big issue, just a bit nicer IMHO.

> > > +		evt->event.event = event;
> > > +		evt->event.reason = reason;
> > > +		vs->vs_events_nr++;
> > > +	}
> > > +	mutex_unlock(&vs->vs_events_lock);
> > > +
> > > +	return evt;
> > > +}
> > > +
> > >  static void vhost_scsi_free_cmd(struct tcm_vhost_cmd *tv_cmd)
> > >  {
> > >  	struct se_cmd *se_cmd = &tv_cmd->tvc_se_cmd;
> > > @@ -397,6 +448,75 @@ static void vhost_scsi_free_cmd(struct tcm_vhost_cmd *tv_cmd)
> > >  	kfree(tv_cmd);
> > >  }
> > >  
> > > +static void tcm_vhost_do_evt_work(struct vhost_scsi *vs,
> > > +	struct tcm_vhost_evt *evt)
> > > +{
> > > +	struct vhost_virtqueue *vq = &vs->vqs[VHOST_SCSI_VQ_EVT];
> > > +	struct virtio_scsi_event *event = &evt->event;
> > > +	struct virtio_scsi_event __user *eventp;
> > > +	unsigned out, in;
> > > +	int head, ret;
> > > +
> > > +	if (!tcm_vhost_check_endpoint(vq))
> > > +		return;
> > > +
> > > +	mutex_lock(&vs->vs_events_lock);
> > > +	mutex_lock(&vq->mutex);
> > > +again:
> > > +	vhost_disable_notify(&vs->dev, vq);
> > > +	head = vhost_get_vq_desc(&vs->dev, vq, vq->iov,
> > > +			ARRAY_SIZE(vq->iov), &out, &in,
> > > +			NULL, NULL);
> > > +	if (head < 0) {
> > > +		vs->vs_events_dropped = true;
> > > +		goto out;
> > > +	}
> > > +	if (head == vq->num) {
> > > +		if (vhost_enable_notify(&vs->dev, vq))
> > > +			goto again;
> > 
> > Please write loops with while() or for().
> > Not with goto. goto is for error handling.
> 
> This makes extra indention which is more ugly.

I don't care. No loops with goto and that's a hard rule.

> > > +		vs->vs_events_dropped = true;
> > > +		goto out;
> > > +	}
> > > +
> > > +	if ((vq->iov[out].iov_len != sizeof(struct virtio_scsi_event))) {
> > > +		vq_err(vq, "Expecting virtio_scsi_event, got %zu bytes\n",
> > > +				vq->iov[out].iov_len);
> > > +		vs->vs_events_dropped = true;
> > > +		goto out;
> > > +	}
> > > +
> > > +	if (vs->vs_events_dropped) {
> > > +		event->event |= VIRTIO_SCSI_T_EVENTS_MISSED;
> > > +		vs->vs_events_dropped = false;
> > > +	}
> > > +
> > > +	eventp = vq->iov[out].iov_base;
> > > +	ret = __copy_to_user(eventp, event, sizeof(*event));
> > > +	if (!ret)
> > > +		vhost_add_used_and_signal(&vs->dev, vq, head, 0);
> > > +	else
> > > +		vq_err(vq, "Faulted on tcm_vhost_send_event\n");
> > > +out:
> > > +	mutex_unlock(&vq->mutex);
> > > +	mutex_unlock(&vs->vs_events_lock);
> > > +}
> > > +
> > > +static void tcm_vhost_evt_work(struct vhost_work *work)
> > > +{
> > > +	struct vhost_scsi *vs = container_of(work, struct vhost_scsi,
> > > +					vs_event_work);
> > > +	struct tcm_vhost_evt *evt;
> > > +	struct llist_node *llnode;
> > > +
> > > +	llnode = llist_del_all(&vs->vs_event_list);
> > 
> > The assumption is that this is slow path thing, no need to worry about
> > speed, yes? so let's do simple list_add, list_del etc.
> 
> Why it is simpler?

simple list_ with a lock is easier to use correctly.

> We are using llist for cmd. Why it is better using
> one for evt and the different for cmd? Similar code makes people easier
> to read.

OK fair enough. But my idea is above to use a work
structure instead of a list of events.
This way we don't need extra list at all,
no new locks, nothing.

> > > +	while (llnode) {
> > > +		evt = llist_entry(llnode, struct tcm_vhost_evt, list);
> > > +		llnode = llist_next(llnode);
> > > +		tcm_vhost_do_evt_work(vs, evt);
> > > +		tcm_vhost_free_evt(vs, evt);
> > > +	}
> > > +}
> > > +
> > >  /* Fill in status and signal that we are done processing this command
> > >   *
> > >   * This is scheduled in the vhost work queue so we are called with the owner
> > > @@ -807,9 +927,42 @@ static void vhost_scsi_ctl_handle_kick(struct vhost_work *work)
> > >  	pr_debug("%s: The handling func for control queue.\n", __func__);
> > >  }
> > >  
> > > +static int tcm_vhost_send_evt(struct vhost_scsi *vs, struct tcm_vhost_tpg *tpg,
> > > +	struct se_lun *lun, u32 event, u32 reason)
> > > +{
> > > +	struct tcm_vhost_evt *evt;
> > > +
> > > +	evt = tcm_vhost_allocate_evt(vs, event, reason);
> > > +	if (!evt)
> > > +		return -ENOMEM;
> > > +
> > > +	if (tpg && lun) {
> > > +		/* TODO: share lun setup code with virtio-scsi.ko */
> > > +		evt->event.lun[0] = 0x01;
> > > +		evt->event.lun[1] = tpg->tport_tpgt & 0xFF;
> > > +		if (lun->unpacked_lun >= 256)
> > > +			evt->event.lun[2] = lun->unpacked_lun >> 8 | 0x40 ;
> > > +		evt->event.lun[3] = lun->unpacked_lun & 0xFF;
> > > +	}
> > > +
> > > +	llist_add(&evt->list, &vs->vs_event_list);
> > > +	vhost_work_queue(&vs->dev, &vs->vs_event_work);
> > > +
> > > +	return 0;
> > > +}
> > > +
> > >  static void vhost_scsi_evt_handle_kick(struct vhost_work *work)
> > >  {
> > > -	pr_debug("%s: The handling func for event queue.\n", __func__);
> > > +	struct vhost_virtqueue *vq = container_of(work, struct vhost_virtqueue,
> > > +						poll.work);
> > > +	struct vhost_scsi *vs = container_of(vq->dev, struct vhost_scsi, dev);
> > > +
> > > +	if (!tcm_vhost_check_endpoint(vq))
> > > +		return;
> > 
> > Again just drop this check.
> 
> Why?

Same as above.
Because it's not safe to do this outside the vq mutex.

> > > +
> > > +	if (tcm_vhost_check_events_dropped(vs))
> > > +		tcm_vhost_send_evt(vs, NULL, NULL, VIRTIO_SCSI_T_NO_EVENT, 0);
> > > +
> > >  }
> > >  
> > >  static void vhost_scsi_handle_kick(struct vhost_work *work)
> > > @@ -833,6 +986,7 @@ static void vhost_scsi_flush(struct vhost_scsi *vs)
> > >  	for (i = 0; i < VHOST_SCSI_MAX_VQ; i++)
> > >  		vhost_scsi_flush_vq(vs, i);
> > >  	vhost_work_flush(&vs->dev, &vs->vs_completion_work);
> > > +	vhost_work_flush(&vs->dev, &vs->vs_event_work);
> > >  }
> > >  
> > >  /*
> > > @@ -891,6 +1045,7 @@ static int vhost_scsi_set_endpoint(
> > >  				return -EEXIST;
> > >  			}
> > >  			tv_tpg->tv_tpg_vhost_count++;
> > > +			tv_tpg->vhost_scsi = vs;
> > >  			vs_tpg[tv_tpg->tport_tpgt] = tv_tpg;
> > >  			smp_mb__after_atomic_inc();
> > >  			match = true;
> > > @@ -974,6 +1129,7 @@ static int vhost_scsi_clear_endpoint(
> > >  			goto err_tpg;
> > >  		}
> > >  		tv_tpg->tv_tpg_vhost_count--;
> > > +		tv_tpg->vhost_scsi = NULL;
> > >  		vs->vs_tpg[target] = NULL;
> > >  		match = true;
> > >  		mutex_unlock(&tv_tpg->tv_tpg_mutex);
> > > @@ -1033,6 +1189,11 @@ static int vhost_scsi_open(struct inode *inode, struct file *f)
> > >  		return -ENOMEM;
> > >  
> > >  	vhost_work_init(&s->vs_completion_work, vhost_scsi_complete_cmd_work);
> > > +	vhost_work_init(&s->vs_event_work, tcm_vhost_evt_work);
> > > +
> > > +	s->vs_events_nr = 0;
> > > +	s->vs_events_dropped = false;
> > > +	mutex_init(&s->vs_events_lock);
> > >  
> > >  	s->vqs[VHOST_SCSI_VQ_CTL].handle_kick = vhost_scsi_ctl_handle_kick;
> > >  	s->vqs[VHOST_SCSI_VQ_EVT].handle_kick = vhost_scsi_evt_handle_kick;
> > > @@ -1163,6 +1324,42 @@ static char *tcm_vhost_dump_proto_id(struct tcm_vhost_tport *tport)
> > >  	return "Unknown";
> > >  }
> > >  
> > > +static int tcm_vhost_do_plug(struct tcm_vhost_tpg *tpg,
> > > +	struct se_lun *lun, bool plug)
> > > +{
> > > +
> > > +	struct vhost_scsi *vs = tpg->vhost_scsi;
> > > +	u32 reason;
> > > +
> > > +	mutex_lock(&tpg->tv_tpg_mutex);
> > > +	vs = tpg->vhost_scsi;
> > > +	mutex_unlock(&tpg->tv_tpg_mutex);
> > > +	if (!vs)
> > > +		return -EOPNOTSUPP;
> > 
> > Why EOPNOTSUPP? When does this happen?
> 
> When tpg->vhost_scsi has not been setup. E.g. no one starts a vhost-scsi
> guest.

So ENODEV or something.

> > > +
> > > +	if (!tcm_vhost_check_feature(vs, VIRTIO_SCSI_F_HOTPLUG))
> > > +		return -EOPNOTSUPP;
> > > +
> > > +	if (plug)
> > > +		reason = VIRTIO_SCSI_EVT_RESET_RESCAN;
> > > +	else
> > > +		reason = VIRTIO_SCSI_EVT_RESET_REMOVED;
> > > +
> > 
> > You have dropped the tpg lock so tpg->vhost_scsi can become
> > NULL now. Why is this safe?
> >
> > > +	return tcm_vhost_send_evt(vs, tpg, lun,
> > > +			VIRTIO_SCSI_T_TRANSPORT_RESET,
> > > +			reason);
> > > +}
> > > +
> > > +static int tcm_vhost_hotplug(struct tcm_vhost_tpg *tpg, struct se_lun *lun)
> > > +{
> > > +	return tcm_vhost_do_plug(tpg, lun, true);
> > > +}
> > > +
> > > +static int tcm_vhost_hotunplug(struct tcm_vhost_tpg *tpg, struct se_lun *lun)
> > > +{
> > > +	return tcm_vhost_do_plug(tpg, lun, false);
> > > +}
> > > +
> > 
> > What are these wrappers for? Seem useless ...
> 
> 
> It make people easier to understand what's the true and false is about
> in tcm_vhost_do_plug.

So just pass in the reason and not a bool.
When readers see
VIRTIO_SCSI_EVT_RESET_RESCAN/VIRTIO_SCSI_EVT_RESET_REMOVED
they will know what is going on.

> > >  static int tcm_vhost_port_link(struct se_portal_group *se_tpg,
> > >  	struct se_lun *lun)
> > >  {
> > > @@ -1173,18 +1370,21 @@ static int tcm_vhost_port_link(struct se_portal_group *se_tpg,
> > >  	tv_tpg->tv_tpg_port_count++;
> > >  	mutex_unlock(&tv_tpg->tv_tpg_mutex);
> > >  
> > > +	tcm_vhost_hotplug(tv_tpg, lun);
> > > +
> > >  	return 0;
> > >  }
> > >  
> > >  static void tcm_vhost_port_unlink(struct se_portal_group *se_tpg,
> > > -	struct se_lun *se_lun)
> > > +	struct se_lun *lun)
> > >  {
> > >  	struct tcm_vhost_tpg *tv_tpg = container_of(se_tpg,
> > >  				struct tcm_vhost_tpg, se_tpg);
> > > -
> > >  	mutex_lock(&tv_tpg->tv_tpg_mutex);
> > >  	tv_tpg->tv_tpg_port_count--;
> > >  	mutex_unlock(&tv_tpg->tv_tpg_mutex);
> > > +
> > > +	tcm_vhost_hotunplug(tv_tpg, lun);
> > >  }
> > >  
> > >  static struct se_node_acl *tcm_vhost_make_nodeacl(
> > > diff --git a/drivers/vhost/tcm_vhost.h b/drivers/vhost/tcm_vhost.h
> > > index 1d2ae7a..94e9ee53 100644
> > > --- a/drivers/vhost/tcm_vhost.h
> > > +++ b/drivers/vhost/tcm_vhost.h
> > > @@ -53,6 +53,7 @@ struct tcm_vhost_nacl {
> > >  	struct se_node_acl se_node_acl;
> > >  };
> > >  
> > > +struct vhost_scsi;
> > >  struct tcm_vhost_tpg {
> > >  	/* Vhost port target portal group tag for TCM */
> > >  	u16 tport_tpgt;
> > > @@ -70,6 +71,8 @@ struct tcm_vhost_tpg {
> > >  	struct tcm_vhost_tport *tport;
> > >  	/* Returned by tcm_vhost_make_tpg() */
> > >  	struct se_portal_group se_tpg;
> > > +	/* Pointer back to struct vhost_scsi, protected by tv_tpg_mutex */
> > > +	struct vhost_scsi *vhost_scsi;
> > >  };
> > >  
> > >  struct tcm_vhost_tport {
> > > @@ -83,6 +86,13 @@ struct tcm_vhost_tport {
> > >  	struct se_wwn tport_wwn;
> > >  };
> > >  
> > > +struct tcm_vhost_evt {
> > > +	/* virtio_scsi event */
> > > +	struct virtio_scsi_event event;
> > > +	/* virtio_scsi event list, serviced from vhost worker thread */
> > > +	struct llist_node list;
> > > +};
> > > +
> > >  /*
> > >   * As per request from MST, keep TCM_VHOST related ioctl defines out of
> > >   * linux/vhost.h (user-space) for now..
> > > -- 
> > > 1.8.1.4
> 
> -- 
> Asias