linux-kernel - Re: [PATCH 05/18] xen/pvcalls: connect to a frontend

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <alpine.DEB.2.10.1705161312470.4103@sstabellini-ThinkPad-X260>
Date:   Tue, 16 May 2017 13:23:22 -0700 (PDT)
From:   Stefano Stabellini <sstabellini@...nel.org>
To:     Boris Ostrovsky <boris.ostrovsky@...cle.com>
cc:     Stefano Stabellini <sstabellini@...nel.org>,
        xen-devel@...ts.xen.org, linux-kernel@...r.kernel.org,
        jgross@...e.com, Stefano Stabellini <stefano@...reto.com>
Subject: Re: [PATCH 05/18] xen/pvcalls: connect to a frontend

On Mon, 15 May 2017, Boris Ostrovsky wrote:
> On 05/15/2017 04:35 PM, Stefano Stabellini wrote:
> > Introduce a per-frontend data structure named pvcalls_back_priv. It
> > contains pointers to the command ring, its event channel, a list of
> > active sockets and a tree of passive sockets (passing sockets need to be
> > looked up from the id on listen, accept and poll commands, while active
> > sockets only on release).
> 
> It would be useful to put this into a comment in pvcalls_back_priv definition.

I'll do that.


> > It also has an unbound workqueue to schedule the work of parsing and
> > executing commands on the command ring. pvcallss_lock protects the two
> > lists. In pvcalls_back_global, keep a list of connected frontends.
> > 
> > Signed-off-by: Stefano Stabellini <stefano@...reto.com>
> > CC: boris.ostrovsky@...cle.com
> > CC: jgross@...e.com
> > ---
> >  drivers/xen/pvcalls-back.c | 87
> > ++++++++++++++++++++++++++++++++++++++++++++++
> >  1 file changed, 87 insertions(+)
> > 
> > diff --git a/drivers/xen/pvcalls-back.c b/drivers/xen/pvcalls-back.c
> > index 86eca19..876e577 100644
> > --- a/drivers/xen/pvcalls-back.c
> > +++ b/drivers/xen/pvcalls-back.c
> > @@ -44,13 +44,100 @@ struct pvcalls_back_global {
> >  	struct rw_semaphore privs_lock;
> >  } pvcalls_back_global;
> > 
> > +struct pvcalls_back_priv {
> > +	struct list_head list;
> > +	struct xenbus_device *dev;
> > +	struct xen_pvcalls_sring *sring;
> > +	struct xen_pvcalls_back_ring ring;
> > +	int irq;
> > +	struct list_head socket_mappings;
> > +	struct radix_tree_root socketpass_mappings;
> > +	struct rw_semaphore pvcallss_lock;
> 
> Same question as before regarding using rw semaphore --- I only see
> down/up_writes.

And again, you are right. I'll switch it to a regular semaphore.


> And what does the name (pvcallss) stand for?
> 
> 
> > +	atomic_t work;
> > +	struct workqueue_struct *wq;
> > +	struct work_struct register_work;
> > +};
> > +
> >  static void pvcalls_back_ioworker(struct work_struct *work)
> >  {
> >  }
> > 
> > +static void pvcalls_back_work(struct work_struct *work)
> > +{
> > +}
> > +
> > +static irqreturn_t pvcalls_back_event(int irq, void *dev_id)
> > +{
> > +	return IRQ_HANDLED;
> > +}
> > +
> >  static int backend_connect(struct xenbus_device *dev)
> >  {
> > +	int err, evtchn;
> > +	grant_ref_t ring_ref;
> > +	void *addr = NULL;
> > +	struct pvcalls_back_priv *priv = NULL;
> > +
> > +	priv = kzalloc(sizeof(struct pvcalls_back_priv), GFP_KERNEL);
> > +	if (!priv)
> > +		return -ENOMEM;
> > +
> > +	err = xenbus_scanf(XBT_NIL, dev->otherend, "port", "%u",
> > +			   &evtchn);
> > +	if (err != 1) {
> > +		err = -EINVAL;
> > +		xenbus_dev_fatal(dev, err, "reading %s/event-channel",
> > +				 dev->otherend);
> > +		goto error;
> > +	}
> > +
> > +	err = xenbus_scanf(XBT_NIL, dev->otherend, "ring-ref", "%u",
> > &ring_ref);
> > +	if (err != 1) {
> > +		err = -EINVAL;
> > +		xenbus_dev_fatal(dev, err, "reading %s/ring-ref",
> > +				 dev->otherend);
> > +		goto error;
> > +	}
> > +
> > +	err = xenbus_map_ring_valloc(dev, &ring_ref, 1, &addr);
> > +	if (err < 0)
> > +		goto error;
> 
> 
> I'd move this closer to first use, below.

Sure


> > +
> > +	err = bind_interdomain_evtchn_to_irqhandler(dev->otherend_id, evtchn,
> > +						    pvcalls_back_event, 0,
> > +						    "pvcalls-backend", dev);
> > +	if (err < 0)
> > +		goto error;
> > +
> > +	priv->wq = alloc_workqueue("pvcalls_back_wq", WQ_UNBOUND, 1);
> > +	if (!priv->wq) {
> > +		err = -ENOMEM;
> > +		goto error;
> > +	}
> > +	INIT_WORK(&priv->register_work, pvcalls_back_work);
> > +	priv->dev = dev;
> > +	priv->sring = addr;
> > +	BACK_RING_INIT(&priv->ring, priv->sring, XEN_PAGE_SIZE * 1);
> > +	priv->irq = err;
> > +	INIT_LIST_HEAD(&priv->socket_mappings);
> > +	INIT_RADIX_TREE(&priv->socketpass_mappings, GFP_KERNEL);
> > +	init_rwsem(&priv->pvcallss_lock);
> > +	dev_set_drvdata(&dev->dev, priv);
> > +	down_write(&pvcalls_back_global.privs_lock);
> > +	list_add_tail(&priv->list, &pvcalls_back_global.privs);
> > +	up_write(&pvcalls_back_global.privs_lock);
> > +	queue_work(priv->wq, &priv->register_work);
> > +
> >  	return 0;
> > +
> > + error:
> > +	if (addr != NULL)
> > +		xenbus_unmap_ring_vfree(dev, addr);
> > +	if (priv->wq)
> > +		destroy_workqueue(priv->wq);
> > +	unbind_from_irqhandler(priv->irq, dev);
> > +	kfree(priv);
> > +	return err;
> >  }
> > 
> >  static int backend_disconnect(struct xenbus_device *dev)
> > 
>