netpoll: more efficient locking

Callers of netpoll_poll_lock() own NAPI_STATE_SCHED

Callers of netpoll_poll_unlock() have BH blocked between
the NAPI_STATE_SCHED being cleared and poll_lock is released.

We can avoid the spinlock which has no contention, and use cmpxchg()
on poll_owner which we need to set anyway.

This removes a possible lockdep violation after the cited commit,
since sk_busy_loop() re-enables BH before calling busy_poll_stop()

Fixes: 217f69743681 ("net: busy-poll: allow preemption in sk_busy_loop()")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 3 insertions, 3 deletions

diff --git a/net/core/netpoll.c b/net/core/netpoll.c
index 53599bd0c82d..9424673009c1 100644
--- a/net/core/netpoll.c
+++ b/net/core/netpoll.c
@@ -171,12 +171,12 @@ static void poll_one_napi(struct napi_struct *napi)
 static void poll_napi(struct net_device *dev)
 {
         struct napi_struct *napi;
+        int cpu = smp_processor_id();
 
         list_for_each_entry(napi, &dev->napi_list, dev_list) {
-                if (napi->poll_owner != smp_processor_id() &&
-                    spin_trylock(&napi->poll_lock)) {
+                if (cmpxchg(&napi->poll_owner, -1, cpu) == -1) {
                         poll_one_napi(napi);
-                        spin_unlock(&napi->poll_lock);
+                        smp_store_release(&napi->poll_owner, -1);
                 }
         }
 }