From: Vladimir Oltean <vladimir.oltean@xxxxxxx>
In the effort of making .ndo_get_stats64 be able to sleep, we need to
ensure the callers of dev_get_stats do not use atomic context.
I need to preface this by saying that I have no idea why netstat_show
takes the dev_base_lock rwlock. Two things can be observed:
(a) it does not appear to be due to dev_isalive requiring it for some
reason, because broadcast_show() also calls dev_isalive() and has
had no problem existing since the beginning of git.
(b) the dev_get_stats function definitely does not need dev_base_lock
protection either. In fact, holding the dev_base_lock is the entire
problem here, because we want to make dev_get_stats sleepable, and
holding a rwlock gives us atomic context.
So since no protection seems to be necessary, just run unlocked while
retrieving the /sys/class/net/eth0/statistics/* values.
Cc: Christian Brauner <christian.brauner@xxxxxxxxxx>
Cc: Eric Dumazet <edumazet@xxxxxxxxxx>
Signed-off-by: Vladimir Oltean <vladimir.oltean@xxxxxxx>
---
net/core/net-sysfs.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/net/core/net-sysfs.c b/net/core/net-sysfs.c
index 999b70c59761..0782a476b424 100644
--- a/net/core/net-sysfs.c
+++ b/net/core/net-sysfs.c
@@ -585,14 +585,13 @@ static ssize_t netstat_show(const struct device *d,
WARN_ON(offset > sizeof(struct rtnl_link_stats64) ||
offset % sizeof(u64) != 0);
- read_lock(&dev_base_lock);
if (dev_isalive(dev)) {
struct rtnl_link_stats64 temp;
const struct rtnl_link_stats64 *stats = dev_get_stats(dev, &temp);
ret = sprintf(buf, fmt_u64, *(u64 *)(((u8 *)stats) + offset));
}
- read_unlock(&dev_base_lock);
+
return ret;
}
--
2.25.1
