[PATCH v6 net-next 15/15] net: mark ndo_get_stats64 as being able to sleep

[Vladimir Oltean <olteanv@gmail.com>]

From: Vladimir Oltean <vladimir.oltean@nxp.com>

Now that all callers have been converted to not use atomic context when
calling dev_get_stats, it is time to update the documentation and put a
notice in the function that it expects process context.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
Changes in v6:
Rebase on top of Jakub's patch 9f9d41f03bb0 ("docs: net: fix
documentation on .ndo_get_stats").

Changes in v5:
None.

Changes in v4:
None.

Changes in v3:
None.

Changes in v2:
Updated the documentation.

 Documentation/networking/netdevices.rst | 8 ++++++--
 Documentation/networking/statistics.rst | 9 ++++-----
 net/core/dev.c                          | 2 ++
 3 files changed, 12 insertions(+), 7 deletions(-)

diff --git a/Documentation/networking/netdevices.rst b/Documentation/networking/netdevices.rst
index e65665c5ab50..944599722c76 100644
--- a/Documentation/networking/netdevices.rst
+++ b/Documentation/networking/netdevices.rst
@@ -64,8 +64,12 @@ ndo_do_ioctl:
 	Context: process
 
 ndo_get_stats:
-	Synchronization: rtnl_lock() semaphore, dev_base_lock rwlock, or RCU.
-	Context: atomic (can't sleep under rwlock or RCU)
+	Synchronization:
+		none. netif_lists_lock(net) might be held, but not guaranteed.
+		It is illegal to hold rtnl_lock() in this method, since it will
+		cause a lock inversion with netif_lists_lock and a deadlock.
+	Context:
+		process
 
 ndo_start_xmit:
 	Synchronization: __netif_tx_lock spinlock.
diff --git a/Documentation/networking/statistics.rst b/Documentation/networking/statistics.rst
index 234abedc29b2..ad3e353df0dd 100644
--- a/Documentation/networking/statistics.rst
+++ b/Documentation/networking/statistics.rst
@@ -155,11 +155,10 @@ Drivers must ensure best possible compliance with
 Please note for example that detailed error statistics must be
 added into the general `rx_error` / `tx_error` counters.
 
-The `.ndo_get_stats64` callback can not sleep because of accesses
-via `/proc/net/dev`. If driver may sleep when retrieving the statistics
-from the device it should do so periodically asynchronously and only return
-a recent copy from `.ndo_get_stats64`. Ethtool interrupt coalescing interface
-allows setting the frequency of refreshing statistics, if needed.
+Drivers may sleep when retrieving the statistics from the device, or they might
+read the counters periodically and only return in `.ndo_get_stats64` a recent
+copy collected asynchronously. In the latter case, the ethtool interrupt
+coalescing interface allows setting the frequency of refreshing statistics.
 
 Retrieving ethtool statistics is a multi-syscall process, drivers are advised
 to keep the number of statistics constant to avoid race conditions with
diff --git a/net/core/dev.c b/net/core/dev.c
index 30facac95d5e..afd0e226efd4 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -10409,6 +10409,8 @@ int __must_check dev_get_stats(struct net_device *dev,
 	const struct net_device_ops *ops = dev->netdev_ops;
 	int err = 0;
 
+	might_sleep();
+
 	if (ops->ndo_get_stats64) {
 		memset(storage, 0, sizeof(*storage));
 		err = ops->ndo_get_stats64(dev, storage);
-- 
2.25.1