Re: [PATCH v5.1-rc] iwlwifi: make locking in iwl_mvm_tx_mpdu() BH-safe
From: Johannes Berg
Date: Tue Oct 01 2019 - 05:46:40 EST
Hi Jiri,
Sorry for the long delay.
> CPU: 1 PID: 28401 Comm: kworker/u8:2 Tainted: G W 5.3.0-rc8 #3
> Hardware name: LENOVO 20K5S22R00/20K5S22R00, BIOS R0IET38W (1.16 ) 05/31/2017
> Workqueue: phy0 ieee80211_beacon_connection_loss_work [mac80211]
> Call Trace:
> dump_stack+0x78/0xb3
> mark_lock+0x28a/0x2a0
> __lock_acquire+0x568/0x1020
> ? iwl_mvm_set_tx_cmd+0x1c5/0x400 [iwlmvm]
> lock_acquire+0xbd/0x1e0
> ? iwl_mvm_tx_mpdu+0xae/0x600 [iwlmvm]
> _raw_spin_lock+0x35/0x50
> ? iwl_mvm_tx_mpdu+0xae/0x600 [iwlmvm]
> iwl_mvm_tx_mpdu+0xae/0x600 [iwlmvm]
> ? ieee80211_tx_h_select_key+0xf1/0x4a0 [mac80211]
> iwl_mvm_tx_skb+0x1f8/0x460 [iwlmvm]
> iwl_mvm_mac_itxq_xmit+0xcc/0x200 [iwlmvm]
> ? iwl_mvm_mac_itxq_xmit+0x55/0x200 [iwlmvm]
> _ieee80211_wake_txqs+0x2cf/0x660 [mac80211]
> ? _ieee80211_wake_txqs+0x5/0x660 [mac80211]
> ? __ieee80211_wake_queue+0x219/0x340 [mac80211]
> ieee80211_wake_queues_by_reason+0x64/0xa0 [mac80211]
>
I'm a bit confused by this.
ieee80211_wake_queues_by_reason() does
spin_lock_irqsave()/spin_unlock_irqrestore() - why is that "{SOFTIRQ-ON-
W} usage"?
Or what did you snip?
johannes