From mboxrd@z Thu Jan  1 00:00:00 1970
From: bjorn@mork.no (=?utf-8?Q?Bj=C3=B8rn_Mork?=)
Date: Wed, 11 Mar 2015 16:09:25 +0100
Subject: confusing code....whats the point of this construct ?
In-Reply-To: <55475.1426084817@turing-police.cc.vt.edu> (Valdis Kletnieks's
	message of "Wed, 11 Mar 2015 10:40:17 -0400")
References: <20150311141744.GC23845@opentech.at>
	<55475.1426084817@turing-police.cc.vt.edu>
Message-ID: <87twxr4llm.fsf@nemi.mork.no>
To: kernelnewbies@lists.kernelnewbies.org
List-Id: kernelnewbies.lists.kernelnewbies.org

Valdis.Kletnieks at vt.edu writes:

> On Wed, 11 Mar 2015 15:17:44 +0100, Nicholas Mc Guire said:
>
>> So the wait_event_timeout condition here ends up being (empty || skip)
>> but what is the point of puting this code into the parameter list of
>> wait_event_timeout() ?
>>
>> Would it not be equivalent to:
>>
>> 	bool empty;
>> 	...
>>
>> 	spin_lock_bh(&ar->htt.tx_lock);
>> 	empty = (ar->htt.num_pending_tx == 0);
>> 	spin_unlock_bh(&ar->htt.tx_lock);
>>
>> 	skip = (ar->state == ATH10K_STATE_WEDGED) ||
>> 		test_bit(ATH10K_FLAG_CRASH_FLUSH,
>> 		&ar->dev_flags);
>>
>> 	ret = wait_event_timeout(ar->htt.empty_tx_wq, (empty || skip),
>> 				 ATH10K_FLUSH_TIMEOUT_HZ);
>>
>> What am I missing here ?
>
> Umm... a Signed-off-by: and formatting it as an actual patch? :)
>
> Seriously - you're right, it's ugly code that needs fixing...

Huh?

The condition needs to be re-evaluated every time the process wakes up.
Evaluating it once and then reusing that result is not the same.
Something elseis supposed to modify ar->htt.num_pending_tx, ar->state or
ar->dev_flags while we are waiting.



Bj?rn