[PATCH/RFC] mac80211: wake queues if ieee80211_suspend failed

[PATCH/RFC] mac80211: wake queues if ieee80211_suspend failed

[Vitaly Wool]

Hi,

Sometimes I'm seeing that after a rejected suspend which was
rejected by the driver's suspend routine, the device can not
associate to any AP any more. Investigation shows that it's
due to the fact that the queues are left stopped with the
reason IEEE80211_QUEUE_STOP_REASON_SUSPEND.

With that known, it looks like we need to wake queues on an
unsuccessful suspend attempt explicitly.

---
  net/mac80211/pm.c           |    2 ++
  1 file changed, 2 insertions(+), 0 deletions(-)

diff --git a/net/mac80211/pm.c b/net/mac80211/pm.c
index 844cefb..fd57913 100644
--- a/net/mac80211/pm.c
+++ b/net/mac80211/pm.c
@@ -96,6 +96,8 @@ int __ieee80211_suspend(struct ieee80211_hw *hw, struct cfg80211_wowlan *wowlan)
  		int err = drv_suspend(local, wowlan);
  		if (err) {
  			local->quiescing = false;
+			ieee80211_wake_queues_by_reason(hw,
+					IEEE80211_QUEUE_STOP_REASON_SUSPEND);
  			return err;
  		}
  		list_for_each_entry(sdata,&local->interfaces, list) {
-- 
1.7.5.4

-- 
Vitaly Wool<vitalywool@gmail.com>

Re: [PATCH/RFC] mac80211: wake queues if ieee80211_suspend failed

[Eliad Peller]

hi Vitaly,

On Mon, Apr 23, 2012 at 5:57 PM, Vitaly Wool <vitalywool@gmail.com> wrote:
> Sometimes I'm seeing that after a rejected suspend which was
> rejected by the driver's suspend routine, the device can not
> associate to any AP any more. Investigation shows that it's
> due to the fact that the queues are left stopped with the
> reason IEEE80211_QUEUE_STOP_REASON_SUSPEND.
>
> With that known, it looks like we need to wake queues on an
> unsuccessful suspend attempt explicitly.
>
> ---
>  net/mac80211/pm.c           |    2 ++
>  1 file changed, 2 insertions(+), 0 deletions(-)
>
> diff --git a/net/mac80211/pm.c b/net/mac80211/pm.c
> index 844cefb..fd57913 100644
> --- a/net/mac80211/pm.c
> +++ b/net/mac80211/pm.c
> @@ -96,6 +96,8 @@ int __ieee80211_suspend(struct ieee80211_hw *hw, struct
> cfg80211_wowlan *wowlan)
>                int err = drv_suspend(local, wowlan);
>                if (err) {
>                        local->quiescing = false;
> +                       ieee80211_wake_queues_by_reason(hw,
> +
> IEEE80211_QUEUE_STOP_REASON_SUSPEND);
>                        return err;
>                }

it seems right.
looks like there's a similar issue with WLAN_STA_BLOCK_BA being set
for all the stations and not cleared on error.

Eliad.

Re: [PATCH/RFC] mac80211: wake queues if ieee80211_suspend failed

[Vitaly Wool]

Hi Eliad,

On Mon, Apr 23, 2012 at 5:30 PM, Eliad Peller <eliad@wizery.com> wrote:
> hi Vitaly,
>
> On Mon, Apr 23, 2012 at 5:57 PM, Vitaly Wool <vitalywool@gmail.com> wrote:
>> Sometimes I'm seeing that after a rejected suspend which was
>> rejected by the driver's suspend routine, the device can not
>> associate to any AP any more. Investigation shows that it's
>> due to the fact that the queues are left stopped with the
>> reason IEEE80211_QUEUE_STOP_REASON_SUSPEND.
>>
>> With that known, it looks like we need to wake queues on an
>> unsuccessful suspend attempt explicitly.
>>
>> ---
>>  net/mac80211/pm.c           |    2 ++
>>  1 file changed, 2 insertions(+), 0 deletions(-)
>>
>> diff --git a/net/mac80211/pm.c b/net/mac80211/pm.c
>> index 844cefb..fd57913 100644
>> --- a/net/mac80211/pm.c
>> +++ b/net/mac80211/pm.c
>> @@ -96,6 +96,8 @@ int __ieee80211_suspend(struct ieee80211_hw *hw, struct
>> cfg80211_wowlan *wowlan)
>>                int err = drv_suspend(local, wowlan);
>>                if (err) {
>>                        local->quiescing = false;
>> +                       ieee80211_wake_queues_by_reason(hw,
>> +
>> IEEE80211_QUEUE_STOP_REASON_SUSPEND);
>>                        return err;
>>                }
>
> it seems right.
> looks like there's a similar issue with WLAN_STA_BLOCK_BA being set
> for all the stations and not cleared on error..

thanks for the feedback. Johannes, any objections/additions? Should I
resubmit the patch?

Thanks,
   Vitaly

Re: [PATCH/RFC] mac80211: wake queues if ieee80211_suspend failed

[Johannes Berg]

On Wed, 2012-05-02 at 21:29 +0200, Vitaly Wool wrote:
> Hi Eliad,
> 
> On Mon, Apr 23, 2012 at 5:30 PM, Eliad Peller <eliad@wizery.com> wrote:
> > hi Vitaly,
> >
> > On Mon, Apr 23, 2012 at 5:57 PM, Vitaly Wool <vitalywool@gmail.com> wrote:
> >> Sometimes I'm seeing that after a rejected suspend which was
> >> rejected by the driver's suspend routine, the device can not
> >> associate to any AP any more. Investigation shows that it's
> >> due to the fact that the queues are left stopped with the
> >> reason IEEE80211_QUEUE_STOP_REASON_SUSPEND.
> >>
> >> With that known, it looks like we need to wake queues on an
> >> unsuccessful suspend attempt explicitly.
> >>
> >> ---
> >>  net/mac80211/pm.c           |    2 ++
> >>  1 file changed, 2 insertions(+), 0 deletions(-)
> >>
> >> diff --git a/net/mac80211/pm.c b/net/mac80211/pm.c
> >> index 844cefb..fd57913 100644
> >> --- a/net/mac80211/pm.c
> >> +++ b/net/mac80211/pm.c
> >> @@ -96,6 +96,8 @@ int __ieee80211_suspend(struct ieee80211_hw *hw, struct
> >> cfg80211_wowlan *wowlan)
> >>                int err = drv_suspend(local, wowlan);
> >>                if (err) {
> >>                        local->quiescing = false;
> >> +                       ieee80211_wake_queues_by_reason(hw,
> >> +
> >> IEEE80211_QUEUE_STOP_REASON_SUSPEND);
> >>                        return err;
> >>                }
> >
> > it seems right.
> > looks like there's a similar issue with WLAN_STA_BLOCK_BA being set
> > for all the stations and not cleared on error..
> 
> thanks for the feedback. Johannes, any objections/additions? Should I
> resubmit the patch?

No, no objections -- as for resubmitting I don't know, I've been on
vacation (and really still am until Monday)

johannes