RE: [PATCH v2] zd1211rw: replace ZD_ASSERT with lockdep_assert_held()

[Sharma, Sanjeev]

-----Original Message-----
From: Julian Calaby [mailto:julian.calaby@xxxxxxxxx] 
Sent: Monday, September 15, 2014 11:32 AM
To: Sharma, Sanjeev
Cc: Johannes Berg; dsd@xxxxxxxxxx; kune@xxxxxxxxxxxxxx; linux-wireless@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; netdev@xxxxxxxxxxxxxxx
Subject: Re: [PATCH v2] zd1211rw: replace ZD_ASSERT with lockdep_assert_held()

Hi Sanjeev,

On Thu, Sep 11, 2014 at 8:36 PM, Sharma, Sanjeev <Sanjeev_Sharma@xxxxxxxxxx> wrote:
> -----Original Message-----
> From: Johannes Berg [mailto:johannes@xxxxxxxxxxxxxxxx]
> Sent: Thursday, September 11, 2014 3:42 PM
> To: Sharma, Sanjeev
> Cc: dsd@xxxxxxxxxx; kune@xxxxxxxxxxxxxx; 
> linux-wireless@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; 
> netdev@xxxxxxxxxxxxxxx
> Subject: Re: [PATCH v2] zd1211rw: replace ZD_ASSERT with 
> lockdep_assert_held()
>
> On Thu, 2014-09-11 at 15:39 +0530, Sanjeev Sharma wrote:
>> on some architecture spin_is_locked() always return false in 
>> uniprocessor configuration and therefore it would be advise to 
>> replace with lockdep_assert_held().
>>
>> Signed-off-by: Sanjeev Sharma <Sanjeev_Sharma@xxxxxxxxxx>
>> ---
>> Changes in v2:
>>   - corrected the typo
>
>> Now it compiles, but you got the logic wrong.
>
>> +++ b/drivers/net/wireless/zd1211rw/zd_mac.c
>> @@ -235,7 +235,7 @@ void zd_mac_clear(struct zd_mac *mac)  {
>>       flush_workqueue(zd_workqueue);
>>       zd_chip_clear(&mac->chip);
>> -     ZD_ASSERT(!spin_is_locked(&mac->lock));
>> +     lockdep_assert_held(&mac->lock);
>>       ZD_MEMCLEAR(mac, sizeof(struct zd_mac));  }
>
>>Look closely at this again.
>
> I didn't understand where I  put wrong logic  ?

I find it helps to spell out what code is doing in words.

E.g. the line you're removing is:
ZD_ASSERT(!spin_is_locked(&mac->lock));

So, we'll assert when spin_is_locked(&mac->lock) is false, i.e. when
mac->lock is not spin locked.

This isn't the same as what you're replacing it with.

 I  feel logic is absolutely correct and if you expand it will looks like ..

+#define lockdep_assert_held(l)	do {	\ 
+	WARN_ON(debug_locks && !lockdep_is_held(l));	\ 
+	} while (0)

Please refer http://lkml.iu.edu/hypermail/linux/kernel/1203.2/00369.html and also see include/linux/lockdep.h for more detail.

Thanks
Sanjeev Sharma