On 03/19 02:55, Kohei Enju wrote:
> On 03/18 17:35, Alexander Lobakin wrote:
> > Queue management ops unconditionally enable netdev locking. The same
> > lock is taken by default by several NAPI configuration functions,
> > such as napi_enable() and netif_napi_set_irq().
> > Request ops locking in advance and make sure we use the _locked
> > counterparts of those functions to avoid deadlocks, taking the lock
> > manually where needed (suspend/resume, queue rebuild and resets).
>
> Hi Alexander,
> After applying this patch (3/5) along with the preceding ones on top of
> net-next, I got some WARNING splats when changing the admin state
> (up/down) using the ip link command. [1, 2]
>
> Since I haven't looked into this series in detail, I'm reporting the
> splats anyway.
> I'm wondering why I haven't seen anyone report this type of issue up to
> v3. Maybe there is something wrong with my setup or devices?
>
> Device: Intel Corporation Ethernet Controller E810-XXV for SFP (rev 02)
Ah, I think I figured out the reason. My adapter accidentally fell into
safe mode. When the adapter is in the safe mode, netdev->queue_mgmt_ops
== NULL and netdev->request_ops_lock == false, so
netdev_assert_locked_or_invisible() complains about not holding the
netdev lock.
Setting netdev->request_ops_lock = true in the safe mode path also
worked fine for me.
---
diff --git a/drivers/net/ethernet/intel/ice/ice_main.c
b/drivers/net/ethernet/intel/ice/ice_main.c
index 9ef258d5ab48..3477c53316ba 100644
--- a/drivers/net/ethernet/intel/ice/ice_main.c
+++ b/drivers/net/ethernet/intel/ice/ice_main.c
@@ -3519,6 +3519,7 @@ static void ice_set_ops(struct ice_vsi *vsi)
if (ice_is_safe_mode(pf)) {
netdev->netdev_ops = &ice_netdev_safe_mode_ops;
+ netdev->request_ops_lock = true;
ice_set_ethtool_safe_mode_ops(netdev);
return;
}
