On Thu, Sep 15, 2011 at 12:49 PM, Felix Fietkau <nbd@xxxxxxxxxxx> wrote:
> On 2011-09-15 9:14 AM, Mohammed Shafi wrote:
>>
>> Hi Felix,
>>
>> On Thu, Sep 15, 2011 at 12:53 AM, Felix Fietkau<nbd@xxxxxxxxxxx> wrote:
>>>
>>> The interrupt handler increases the interrupt disable refcount, so the
>>> tasklet needs to always call ath9k_hw_enable_interrupts.
>>>
>>> Signed-off-by: Felix Fietkau<nbd@xxxxxxxxxxx>
>>> ---
>>> drivers/net/wireless/ath/ath9k/main.c | 9 +++++----
>>> 1 files changed, 5 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/drivers/net/wireless/ath/ath9k/main.c
>>> b/drivers/net/wireless/ath/ath9k/main.c
>>> index 8a17cff5..a3c3316 100644
>>> --- a/drivers/net/wireless/ath/ath9k/main.c
>>> +++ b/drivers/net/wireless/ath/ath9k/main.c
>>> @@ -669,15 +669,15 @@ void ath9k_tasklet(unsigned long data)
>>> u32 status = sc->intrstatus;
>>> u32 rxmask;
>>>
>>> + ath9k_ps_wakeup(sc);
>>
>> this is done in ath_reset, is there any corner cases that I had missed ?
>>
>>> + spin_lock(&sc->sc_pcu_lock);
>>
>> spin_lock_bh(&sc->sc_pcu_lock) in ath_reset_internal, won't be a
>> problem know, or should we need to move this lock
>>
>>> +
>>> if ((status& ATH9K_INT_FATAL) ||
>>> (status& ATH9K_INT_BB_WATCHDOG)) {
>>> ieee80211_queue_work(sc->hw,&sc->hw_reset_work);
>>> - return;
>>> + goto out;
>>> }
>>>
>>> - ath9k_ps_wakeup(sc);
>>> - spin_lock(&sc->sc_pcu_lock);
>>> -
>>> /*
>>> * Only run the baseband hang check if beacons stop working in AP
>>> or
>>> * IBSS mode, because it has a high false positive rate. For
>>> station
>>> @@ -725,6 +725,7 @@ void ath9k_tasklet(unsigned long data)
>>> if (status& ATH9K_INT_GENTIMER)
>>> ath_gen_timer_isr(sc->sc_ah);
>>>
>>> +out:
>>> /* re-enable hardware interrupt */
>>> ath9k_hw_enable_interrupts(ah);
>>
>>
>> this is done in ath_complete_reset, should we enable interrupts if
>> some times ath9k_hw_reset fails?
>> please correct me if I had understood wrongly. thanks!
>
> ath9k_hw_enable_interrupts was changed to use refcounting to keep interrupts
> blocked. The interrupt handler increses the refcount, so the tasklet needs
> to decrease it. When a reset is issued, it starts by disabling interrupts,
> then re-enables them after completing the reset.
> Disabling interrupts in the ISR, but not enabling them in the tasklet causes
> an imbalance which leaves interrupts disabled after the reset is done.
Felix, thanks for your explanation. i understood it somewhat :). i
will take a look at this more carefully, especially intr_ref_cnt.
>
> - Felix
>
--
shafi
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
