From: Vladimir Oltean <olteanv@gmail.com>
To: "David S . Miller" <davem@davemloft.net>,
Jakub Kicinski <kuba@kernel.org>
Cc: netdev@vger.kernel.org, Andrew Lunn <andrew@lunn.ch>,
Florian Fainelli <f.fainelli@gmail.com>,
Cong Wang <xiyou.wangcong@gmail.com>,
Stephen Hemminger <stephen@networkplumber.org>,
Eric Dumazet <edumazet@google.com>,
George McCollister <george.mccollister@gmail.com>,
Oleksij Rempel <o.rempel@pengutronix.de>,
Jay Vosburgh <j.vosburgh@gmail.com>,
Veaceslav Falico <vfalico@gmail.com>,
Andy Gospodarek <andy@greyhouse.net>,
Arnd Bergmann <arnd@arndb.de>, Taehee Yoo <ap420073@gmail.com>,
Jiri Pirko <jiri@resnulli.us>, Florian Westphal <fw@strlen.de>,
Nikolay Aleksandrov <nikolay@nvidia.com>,
Pravin B Shelar <pshelar@ovn.org>,
Sridhar Samudrala <sridhar.samudrala@intel.com>,
Saeed Mahameed <saeedm@nvidia.com>
Subject: [PATCH v6 net-next 13/15] net: net_failover: ensure .ndo_get_stats64 can sleep
Date: Sat, 9 Jan 2021 19:26:22 +0200 [thread overview]
Message-ID: <20210109172624.2028156-14-olteanv@gmail.com> (raw)
In-Reply-To: <20210109172624.2028156-1-olteanv@gmail.com>
From: Vladimir Oltean <vladimir.oltean@nxp.com>
The failover framework sets up a virtio_net interface [ when it has the
VIRTIO_NET_F_STANDBY feature ] and a VF interface, having the same MAC
address, in a standby/active relationship. When the active VF is
unplugged, the standby virtio_net temporarily kicks in.
The failover framework registers a common upper for the active and the
standby interface, which is what the application layer uses. This is
similar to bonding/team. The statistics of the upper interface are the
sum of the statistics of the active and of the standby interface.
There is an effort to convert .ndo_get_stats64 to sleepable context, and
for that to work, we need to prevent callers of dev_get_stats from using
atomic locking. The failover driver needs protection via an RCU
read-side critical section to access the standby and the active
interface. This is ok because it is reentrant, and generally speaking,
dev_get_stats is recursive. But it is also not ok, because holding
rcu_read_lock() while calling dev_get_stats will incur atomic context
upon it, and that needs to change.
The existing logic can be rehashed just a little bit such that the
recursive dev_get_stats call will not be under any lock. We can achieve
that by "cheating" a little bit and use dev_hold() to take a reference
on the active and backup interfaces, and netdev_wait_allrefs() will just
have to wait until dev_get_stats() finishes.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
Changes in v6:
Slightly touch up the commit message.
Changes in v5:
Use rcu_read_lock() and do not change the locking architecture of the
driver.
Changes in v4:
Now there is code to propagate errors.
Changes in v3:
None.
Changes in v2:
Switched to the new scheme of holding just a refcnt to the slave
interfaces while recursing with dev_get_stats.
drivers/net/net_failover.c | 64 ++++++++++++++++++++++++++++----------
1 file changed, 47 insertions(+), 17 deletions(-)
diff --git a/drivers/net/net_failover.c b/drivers/net/net_failover.c
index 2815228a34d5..8b7a6d1eab30 100644
--- a/drivers/net/net_failover.c
+++ b/drivers/net/net_failover.c
@@ -183,38 +183,64 @@ static int net_failover_get_stats(struct net_device *dev,
struct rtnl_link_stats64 *stats)
{
struct net_failover_info *nfo_info = netdev_priv(dev);
- struct rtnl_link_stats64 temp;
- struct net_device *slave_dev;
+ struct rtnl_link_stats64 primary_stats;
+ struct rtnl_link_stats64 standby_stats;
+ struct net_device *primary_dev;
+ struct net_device *standby_dev;
int err = 0;
- spin_lock(&nfo_info->stats_lock);
- memcpy(stats, &nfo_info->failover_stats, sizeof(*stats));
-
rcu_read_lock();
- slave_dev = rcu_dereference(nfo_info->primary_dev);
- if (slave_dev) {
- err = dev_get_stats(slave_dev, &temp);
+ primary_dev = rcu_dereference(nfo_info->primary_dev);
+ if (primary_dev)
+ dev_hold(primary_dev);
+
+ standby_dev = rcu_dereference(nfo_info->standby_dev);
+ if (standby_dev)
+ dev_hold(standby_dev);
+
+ rcu_read_unlock();
+
+ /* Don't hold rcu_read_lock while calling dev_get_stats, just a
+ * reference to ensure they won't get unregistered.
+ */
+ if (primary_dev) {
+ err = dev_get_stats(primary_dev, &primary_stats);
if (err)
goto out;
- net_failover_fold_stats(stats, &temp, &nfo_info->primary_stats);
- memcpy(&nfo_info->primary_stats, &temp, sizeof(temp));
}
- slave_dev = rcu_dereference(nfo_info->standby_dev);
- if (slave_dev) {
- err = dev_get_stats(slave_dev, &temp);
+ if (standby_dev) {
+ err = dev_get_stats(standby_dev, &standby_stats);
if (err)
goto out;
- net_failover_fold_stats(stats, &temp, &nfo_info->standby_stats);
- memcpy(&nfo_info->standby_stats, &temp, sizeof(temp));
}
-out:
- rcu_read_unlock();
+ spin_lock(&nfo_info->stats_lock);
+
+ memcpy(stats, &nfo_info->failover_stats, sizeof(*stats));
+
+ if (primary_dev) {
+ net_failover_fold_stats(stats, &primary_stats,
+ &nfo_info->primary_stats);
+ memcpy(&nfo_info->primary_stats, &primary_stats,
+ sizeof(primary_stats));
+ }
+ if (standby_dev) {
+ net_failover_fold_stats(stats, &standby_stats,
+ &nfo_info->standby_stats);
+ memcpy(&nfo_info->standby_stats, &standby_stats,
+ sizeof(standby_stats));
+ }
memcpy(&nfo_info->failover_stats, stats, sizeof(*stats));
+
spin_unlock(&nfo_info->stats_lock);
+out:
+ if (primary_dev)
+ dev_put(primary_dev);
+ if (standby_dev)
+ dev_put(standby_dev);
return err;
}
@@ -728,6 +754,7 @@ static struct failover_ops net_failover_ops = {
struct failover *net_failover_create(struct net_device *standby_dev)
{
struct device *dev = standby_dev->dev.parent;
+ struct net_failover_info *nfo_info;
struct net_device *failover_dev;
struct failover *failover;
int err;
@@ -772,6 +799,9 @@ struct failover *net_failover_create(struct net_device *standby_dev)
failover_dev->min_mtu = standby_dev->min_mtu;
failover_dev->max_mtu = standby_dev->max_mtu;
+ nfo_info = netdev_priv(failover_dev);
+ spin_lock_init(&nfo_info->stats_lock);
+
err = register_netdev(failover_dev);
if (err) {
dev_err(dev, "Unable to register failover_dev!\n");
--
2.25.1
next prev parent reply other threads:[~2021-01-09 17:28 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-01-09 17:26 [PATCH v6 net-next 00/15] Make .ndo_get_stats64 sleepable Vladimir Oltean
2021-01-09 17:26 ` [PATCH v6 net-next 01/15] net: mark dev_base_lock for deprecation Vladimir Oltean
2021-01-09 17:26 ` [PATCH v6 net-next 02/15] net: introduce a mutex for the netns interface lists Vladimir Oltean
2021-01-09 17:26 ` [PATCH v6 net-next 03/15] net: procfs: hold netif_lists_lock when retrieving device statistics Vladimir Oltean
2021-01-11 23:46 ` Saeed Mahameed
2021-01-12 13:44 ` Vladimir Oltean
2021-01-12 20:06 ` Saeed Mahameed
2021-01-09 17:26 ` [PATCH v6 net-next 04/15] net: sysfs: don't hold dev_base_lock while " Vladimir Oltean
2021-01-09 17:26 ` [PATCH v6 net-next 05/15] s390/appldata_net_sum: hold the netdev lists lock when " Vladimir Oltean
2021-01-09 17:26 ` [PATCH v6 net-next 06/15] parisc/led: " Vladimir Oltean
2021-01-09 17:26 ` [PATCH v6 net-next 07/15] net: remove return value from dev_get_stats Vladimir Oltean
2021-01-11 22:08 ` Saeed Mahameed
2021-01-09 17:26 ` [PATCH v6 net-next 08/15] net: allow ndo_get_stats64 to return an int error code Vladimir Oltean
2021-01-11 22:24 ` Saeed Mahameed
2021-01-11 22:43 ` Saeed Mahameed
2021-01-09 17:26 ` [PATCH v6 net-next 09/15] scsi: fcoe: propagate errors from dev_get_stats Vladimir Oltean
2021-01-09 17:26 ` [PATCH v6 net-next 10/15] net: openvswitch: " Vladimir Oltean
2021-01-09 17:26 ` [PATCH v6 net-next 11/15] net: catch " Vladimir Oltean
2021-01-11 22:54 ` Saeed Mahameed
2021-01-11 23:15 ` Vladimir Oltean
2021-01-11 23:51 ` Saeed Mahameed
2021-01-18 10:31 ` Vladimir Oltean
2021-01-09 17:26 ` [PATCH v6 net-next 12/15] net: openvswitch: ensure dev_get_stats can sleep Vladimir Oltean
2021-01-09 17:26 ` Vladimir Oltean [this message]
2021-01-09 17:26 ` [PATCH v6 net-next 14/15] net: bonding: ensure .ndo_get_stats64 " Vladimir Oltean
2021-01-11 23:38 ` Saeed Mahameed
2021-01-12 14:37 ` Vladimir Oltean
2021-01-12 20:10 ` Saeed Mahameed
2021-01-13 2:00 ` Jay Vosburgh
2021-01-09 17:26 ` [PATCH v6 net-next 15/15] net: mark ndo_get_stats64 as being able to sleep Vladimir Oltean
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=20210109172624.2028156-14-olteanv@gmail.com \
--to=olteanv@gmail.com \
--cc=andrew@lunn.ch \
--cc=andy@greyhouse.net \
--cc=ap420073@gmail.com \
--cc=arnd@arndb.de \
--cc=davem@davemloft.net \
--cc=edumazet@google.com \
--cc=f.fainelli@gmail.com \
--cc=fw@strlen.de \
--cc=george.mccollister@gmail.com \
--cc=j.vosburgh@gmail.com \
--cc=jiri@resnulli.us \
--cc=kuba@kernel.org \
--cc=netdev@vger.kernel.org \
--cc=nikolay@nvidia.com \
--cc=o.rempel@pengutronix.de \
--cc=pshelar@ovn.org \
--cc=saeedm@nvidia.com \
--cc=sridhar.samudrala@intel.com \
--cc=stephen@networkplumber.org \
--cc=vfalico@gmail.com \
--cc=xiyou.wangcong@gmail.com \
/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;
as well as URLs for NNTP newsgroup(s).