mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Patrick Mullaney <pmullaney@novell.com>
To: alacrityvm-devel@lists.sourceforge.net
Cc: linux-kernel@vger.kernel.org, netdev@vger.kernel.org
Subject: [PATCH 2/2] venet: Fixes for shutdown and re-registration of venet devices.
Date: Wed, 02 Dec 2009 16:10:33 -0500	[thread overview]
Message-ID: <20091202211033.5116.63285.stgit@mimic.site> (raw)
In-Reply-To: <20091202204722.5116.33183.stgit@mimic.site>

Fixed a problem with simple re-registration of the venet netdevice. The
netdev core infrastructure appears to have a problem with re-registering
a previously registered netdevice(reg_state and kobj.state_initialized
need to be reset to their default values). I now do this prior to calling
register_netdev but this probably should be moved to net/core. Fixed an
issue with the removal of the lower level device of a venet-macvlan. I'm
planning a cleaner solution in the future. Fixed a lock issue with
venet-macvlan stop method that was causing a hang.

Signed-off-by: Patrick Mullaney <pmullaney@novell.com>
---

 kernel/vbus/devices/venet/device.c      |   11 +++++-
 kernel/vbus/devices/venet/macvlan.c     |   55 +++++++++++++++++++++++++------
 kernel/vbus/devices/venet/venetdevice.h |    7 +++-
 3 files changed, 58 insertions(+), 15 deletions(-)

diff --git a/kernel/vbus/devices/venet/device.c b/kernel/vbus/devices/venet/device.c
index 01d6f1c..b4cf397 100644
--- a/kernel/vbus/devices/venet/device.c
+++ b/kernel/vbus/devices/venet/device.c
@@ -2300,7 +2300,7 @@ client_mac_show(struct vbus_device *dev, struct vbus_device_attribute *attr,
 struct vbus_device_attribute attr_cmac =
 	__ATTR(client_mac, S_IRUGO | S_IWUSR, client_mac_show, cmac_store);
 
-static ssize_t
+ssize_t
 enabled_show(struct vbus_device *dev, struct vbus_device_attribute *attr,
 	 char *buf)
 {
@@ -2309,7 +2309,7 @@ enabled_show(struct vbus_device *dev, struct vbus_device_attribute *attr,
 	return snprintf(buf, PAGE_SIZE, "%d\n", priv->netif.enabled);
 }
 
-static ssize_t
+ssize_t
 enabled_store(struct vbus_device *dev, struct vbus_device_attribute *attr,
 	      const char *buf, size_t count)
 {
@@ -2323,8 +2323,13 @@ enabled_store(struct vbus_device *dev, struct vbus_device_attribute *attr,
 	if (enabled != 0 && enabled != 1)
 		return -EINVAL;
 
-	if (enabled && !priv->netif.enabled)
+	if (enabled && !priv->netif.enabled) {
+		/* need to un-initialize certain fields of the
+		   net_device so that it may be re-registered */
+		priv->netif.dev->reg_state = NETREG_UNINITIALIZED;
+		priv->netif.dev->dev.kobj.state_initialized = 0;
 		ret = register_netdev(priv->netif.dev);
+	}
 
 	if (!enabled && priv->netif.enabled)
 		venetdev_netdev_unregister(priv);
diff --git a/kernel/vbus/devices/venet/macvlan.c b/kernel/vbus/devices/venet/macvlan.c
index 432ff5d..60479a6 100644
--- a/kernel/vbus/devices/venet/macvlan.c
+++ b/kernel/vbus/devices/venet/macvlan.c
@@ -137,7 +137,7 @@ venetmacv_vlink_up(struct venetdev *vdev)
 		ret = macv->macvlan_netdev_ops->ndo_open(vdev->netif.dev);
 		rtnl_unlock();
 		if (ret)
-			printk(KERN_ERR "macvlanopen failed %d!\n", ret);
+			printk(KERN_ERR "macvlan_open failed %d!\n", ret);
 	}
 }
 
@@ -320,8 +320,10 @@ venetmacv_device_release(struct vbus_device *dev)
 {
 	struct venetmacv *macv = vbusdev_to_macv(dev);
 
-	if (macv->mdev.lowerdev)
+	if (macv->mdev.lowerdev) {
 		dev_put(macv->mdev.lowerdev);
+		macv->mdev.lowerdev = NULL;
+	}
 
 	venetdev_netdev_unregister(&macv->dev);
 	free_netdev(macv->mdev.dev);
@@ -397,12 +399,31 @@ clientmac_store(struct vbus_device *dev, struct vbus_device_attribute *attr,
 	return ret;
 }
 
-struct vbus_device_attribute attr_clientmac =
+static struct vbus_device_attribute attr_clientmac =
 	__ATTR(client_mac, S_IRUGO | S_IWUSR, client_mac_show, clientmac_store);
 
+
+static ssize_t
+macv_enabled_store(struct vbus_device *dev, struct vbus_device_attribute *attr,
+		   const char *buf, size_t count)
+{
+	struct venetmacv *macv = vbusdev_to_macv(dev);
+	int ret;
+
+	if (!macv->mdev.lowerdev)
+		return -ENXIO;
+
+	ret = attr_enabled.store(dev, attr, buf, count);
+
+	return ret;
+}
+
+static struct vbus_device_attribute attr_macv_enabled =
+	__ATTR(enabled, S_IRUGO | S_IWUSR, enabled_show, macv_enabled_store);
+
 static struct attribute *attrs[] = {
 	&attr_clientmac.attr,
-	&attr_enabled.attr,
+	&attr_macv_enabled.attr,
 	&attr_burstthresh.attr,
 	&attr_txmitigation.attr,
 	&attr_ifname.attr,
@@ -423,9 +444,7 @@ venetmacv_netdev_open(struct net_device *dev)
 	venetdev_open(&priv->dev);
 
 	if (priv->dev.vbus.link) {
-		rtnl_lock();
 		ret = priv->macvlan_netdev_ops->ndo_open(priv->mdev.dev);
-		rtnl_unlock();
 	}
 
 	return ret;
@@ -443,15 +462,26 @@ venetmacv_netdev_stop(struct net_device *dev)
 
 	venetdev_stop(&priv->dev);
 
-	if (priv->dev.vbus.link && needs_stop) {
-		rtnl_lock();
+	if (priv->dev.vbus.link && needs_stop)
 		ret = priv->macvlan_netdev_ops->ndo_stop(dev);
-		rtnl_unlock();
-	}
 
 	return ret;
 }
 
+static void
+venetmacv_netdev_uninit(struct net_device *dev)
+{
+	struct venetmacv *macv = netdev_priv(dev);
+
+	if (macv->mdev.lowerdev) {
+		dev_put(macv->mdev.lowerdev);
+		macv->mdev.lowerdev = NULL;
+		memset(macv->ll_ifname, '\0', IFNAMSIZ);
+	}
+
+	macv->dev.netif.enabled = 0;
+}
+
 /*
  * out routine for macvlan
  */
@@ -527,9 +557,9 @@ static struct net_device_ops venetmacv_netdev_ops = {
 	.ndo_start_xmit         = venetmacv_netdev_tx,
 	.ndo_do_ioctl           = venetdev_netdev_ioctl,
 	.ndo_get_stats          = venetmacv_netdev_stats,
+	.ndo_uninit             = venetmacv_netdev_uninit,
 };
 
-
 /*
  * This is called whenever the admin instantiates our devclass via
  * "mkdir /config/vbus/devices/$(inst)/venet-tap"
@@ -545,6 +575,9 @@ venetmacv_device_create(struct vbus_devclass *dc,
 	dev = alloc_netdev(sizeof(struct venetmacv), "macvenet%d",
 					   macvlan_setup);
 
+
+	dev->destructor = NULL;
+
 	if (!dev)
 		return -ENOMEM;
 
diff --git a/kernel/vbus/devices/venet/venetdevice.h b/kernel/vbus/devices/venet/venetdevice.h
index 1a74723..edee570 100644
--- a/kernel/vbus/devices/venet/venetdevice.h
+++ b/kernel/vbus/devices/venet/venetdevice.h
@@ -150,7 +150,7 @@ struct net_device_stats *venetdev_get_stats(struct venetdev *dev);
 static inline void venetdev_netdev_unregister(struct venetdev *priv)
 {
 	if (priv->netif.enabled) {
-		venetdev_netdev_stop(priv->netif.dev);
+		venetdev_stop(priv);
 		unregister_netdev(priv->netif.dev);
 	}
 }
@@ -173,6 +173,11 @@ extern struct vbus_device_attribute attr_ifname;
 extern struct vbus_device_attribute attr_txmitigation;
 extern struct vbus_device_attribute attr_zcthresh;
 
+ssize_t enabled_store(struct vbus_device *dev,
+		struct vbus_device_attribute *attr,
+		 const char *buf, size_t count);
+ssize_t enabled_show(struct vbus_device *dev,
+			struct vbus_device_attribute *attr, char *buf);
 ssize_t cmac_store(struct vbus_device *dev,
 		struct vbus_device_attribute *attr,
 		 const char *buf, size_t count);


      parent reply	other threads:[~2009-12-02 21:10 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-12-02 21:10 [PATCH 0/2] registration fixes for vbus venet macvlan driver Patrick Mullaney
2009-12-02 21:10 ` [PATCH 1/2] venet-macvlan: correctly remove newline from lower device name Patrick Mullaney
2009-12-02 21:10 ` Patrick Mullaney [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20091202211033.5116.63285.stgit@mimic.site \
    --to=pmullaney@novell.com \
    --cc=alacrityvm-devel@lists.sourceforge.net \
    --cc=linux-kernel@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®