Merge tag 'net-6.17-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/netHEADtorvalds/mastertorvalds/HEADmaster

Pull networking fixes from Jakub Kicinski:
  Previous releases - regressions:

   - netlink: avoid infinite retry looping in netlink_unicast()

  Previous releases - always broken:

   - packet: fix a race in packet_set_ring() and packet_notifier()

   - ipv6: reject malicious packets in ipv6_gso_segment()

   - sched: mqprio: fix stack out-of-bounds write in tc entry parsing

   - net: drop UFO packets (injected via virtio) in udp_rcv_segment()

   - eth: mlx5: correctly set gso_segs when LRO is used, avoid false
     positive checksum validation errors

   - netpoll: prevent hanging NAPI when netcons gets enabled

   - phy: mscc: fix parsing of unicast frames for PTP timestamping

   - a number of device tree / OF reference leak fixes"

* tag 'net-6.17-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (44 commits)
  pptp: fix pptp_xmit() error path
  net: ti: icssg-prueth: Fix skb handling for XDP_PASS
  net: Update threaded state in napi config in netif_set_threaded
  selftests: netdevsim: Xfail nexthop test on slow machines
  eth: fbnic: Lock the tx_dropped update
  eth: fbnic: Fix tx_dropped reporting
  eth: fbnic: remove the debugging trick of super high page bias
  net: ftgmac100: fix potential NULL pointer access in ftgmac100_phy_disconnect
  dt-bindings: net: Replace bouncing Alexandru Tachici emails
  dpll: zl3073x: ZL3073X_I2C and ZL3073X_SPI should depend on NET
  net/sched: mqprio: fix stack out-of-bounds write in tc entry parsing
  Revert "net: mdio_bus: Use devm for getting reset GPIO"
  selftests: net: packetdrill: xfail all problems on slow machines
  net/packet: fix a race in packet_set_ring() and packet_notifier()
  benet: fix BUG when creating VFs
  net: airoha: npu: Add missing MODULE_FIRMWARE macros
  net: devmem: fix DMA direction on unmapping
  ipa: fix compile-testing with qcom-mdt=m
  eth: fbnic: unlink NAPIs from queues on error to open
  net: Add locking to protect skb->dev access in ip_output
  ...

1 files changed, 9 insertions, 17 deletions

diff --git a/net/core/dev.c b/net/core/dev.c
index b28ce68830b2..68dc47d7e700 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -6978,6 +6978,12 @@ int napi_set_threaded(struct napi_struct *napi,
 	if (napi->config)
 		napi->config->threaded = threaded;
 
+	/* Setting/unsetting threaded mode on a napi might not immediately
+	 * take effect, if the current napi instance is actively being
+	 * polled. In this case, the switch between threaded mode and
+	 * softirq mode will happen in the next round of napi_schedule().
+	 * This should not cause hiccups/stalls to the live traffic.
+	 */
 	if (!threaded && napi->thread) {
 		napi_stop_kthread(napi);
 	} else {
@@ -7011,23 +7017,9 @@ int netif_set_threaded(struct net_device *dev,
 
 	WRITE_ONCE(dev->threaded, threaded);
 
-	/* Make sure kthread is created before THREADED bit
-	 * is set.
-	 */
-	smp_mb__before_atomic();
-
-	/* Setting/unsetting threaded mode on a napi might not immediately
-	 * take effect, if the current napi instance is actively being
-	 * polled. In this case, the switch between threaded mode and
-	 * softirq mode will happen in the next round of napi_schedule().
-	 * This should not cause hiccups/stalls to the live traffic.
-	 */
-	list_for_each_entry(napi, &dev->napi_list, dev_list) {
-		if (!threaded && napi->thread)
-			napi_stop_kthread(napi);
-		else
-			assign_bit(NAPI_STATE_THREADED, &napi->state, threaded);
-	}
+	/* The error should not occur as the kthreads are already created. */
+	list_for_each_entry(napi, &dev->napi_list, dev_list)
+		WARN_ON_ONCE(napi_set_threaded(napi, threaded));
 
 	return err;
 }