Received-SPF: pass (google.com: domain of linux-wireless-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Subject: Re: [PATCH] mac80211: Ensure vif queues are operational after start
To:     Alexander Wetzel <alexander@wetzel-home.de>,
        Felix Fietkau <nbd@nbd.name>, linux-wireless@vger.kernel.org
Cc:     Johannes Berg <johannes@sipsolutions.net>
References: <20220915130946.302803-1-alexander@wetzel-home.de>
 <26e9ae91-8e13-df45-815c-cb45c1911032@nbd.name>
 <9e36f219-14a6-8960-a5fd-cb9f708237ed@wetzel-home.de>
 <0b342479-d04a-a45e-d63b-73eec5a1fb40@wetzel-home.de>
From:   Ben Greear <greearb@candelatech.com>
Organization: Candela Technologies
Message-ID: <4f77c10b-a6ea-2b79-cada-deefb2a3a07b@candelatech.com>
Date:   Thu, 15 Sep 2022 13:39:55 -0700
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.2.2
MIME-Version: 1.0
In-Reply-To: <0b342479-d04a-a45e-d63b-73eec5a1fb40@wetzel-home.de>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 9/15/22 1:06 PM, Alexander Wetzel wrote:
>>
>> I've got some doubts that my fix is correct...
>> While it fixes the problem in my tests it looks like we'll need another queue restart to get the queues working again.
>>
>> After all IEEE80211_TXQ_STOP_NETIF_TX will not be cleared when it has been set by __ieee80211_stop_queue().
>>
>> I'll update the patch and skip setting vif.txqs_stopped when SDATA_STATE_RUNNING is not set. Not having IEEE80211_TXQ_STOP_NETIF_TX set looks harmless, having 
>> it set when it should less problematic...
> 
> Scratch that: The patch should be ok as it is: IEEE80211_TXQ_STOP_NETIF_TX is not set on stop, the patch should be ok as it is.
> 
> Sorry for the noise.
> 
> Alexander
> 

To add to the noise...

 From reading the original patch description, it was to stop an NPE when AP was stopped.  I have been testing
this patch below and it fixes the problems I saw with multiple vdevs.  I was worried that
the code in the 'list_for_each_entry_rcu(sta, &local->sta_list, list) {' might still need to run
to keep everything in sync (and my patch allows that to happen), but I do not know if that is true or not.

diff --git a/net/mac80211/util.c b/net/mac80211/util.c
index c768e583aad4..2b5dafe9f4cc 100644
--- a/net/mac80211/util.c
+++ b/net/mac80211/util.c
@@ -370,12 +370,6 @@ static void __ieee80211_wake_txqs(struct ieee80211_sub_if_data *sdata, int ac)
         local_bh_disable();
         spin_lock(&fq->lock);

-       if (!test_bit(SDATA_STATE_RUNNING, &sdata->state))
-               goto out;
-
-       if (sdata->vif.type == NL80211_IFTYPE_AP)
-               ps = &sdata->bss->ps;
-
         sdata->vif.txqs_stopped[ac] = false;

         list_for_each_entry_rcu(sta, &local->sta_list, list) {
@@ -408,6 +402,10 @@ static void __ieee80211_wake_txqs(struct ieee80211_sub_if_data *sdata, int ac)

         txqi = to_txq_info(vif->txq);

+       if (test_bit(SDATA_STATE_RUNNING, &sdata->state))
+               if (sdata->vif.type == NL80211_IFTYPE_AP)
+                       ps = &sdata->bss->ps;
+
         if (!test_and_clear_bit(IEEE80211_TXQ_STOP_NETIF_TX, &txqi->flags) ||
             (ps && atomic_read(&ps->num_sta_ps)) || ac != vif->txq->ac)
                 goto out;


Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com