linux-kernel - Re: [PATCH 05/12] workqueue: don't wake up other workers in rescuer

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 28 Sep 2012 18:18:27 +0800
From:	Lai Jiangshan <laijs@...fujitsu.com>
To:	Tejun Heo <tj@...nel.org>
CC:	linux-kernel@...r.kernel.org, Ray Jui <rjui@...adcom.com>
Subject: Re: [PATCH 05/12] workqueue: don't wake up other workers in rescuer

On 09/27/2012 02:34 AM, Tejun Heo wrote:
> (cc'ing Ray Jui)
> 
> On Thu, Sep 27, 2012 at 01:20:36AM +0800, Lai Jiangshan wrote:
>> rescuer is NOT_RUNNING, so there is no sense when it wakes up other workers,
>> if there are available normal workers, they are already woken up when needed.
>>
>> Signed-off-by: Lai Jiangshan <laijs@...fujitsu.com>
>> ---
>>  kernel/workqueue.c |    8 --------
>>  1 files changed, 0 insertions(+), 8 deletions(-)
>>
>> diff --git a/kernel/workqueue.c b/kernel/workqueue.c
>> index c718b94..6c339bf 100644
>> --- a/kernel/workqueue.c
>> +++ b/kernel/workqueue.c
>> @@ -2438,14 +2438,6 @@ repeat:
>>  
>>  		process_scheduled_works(rescuer);
>>  
>> -		/*
>> -		 * Leave this gcwq.  If keep_working() is %true, notify a
>> -		 * regular worker; otherwise, we end up with 0 concurrency
>> -		 * and stalling the execution.
>> -		 */
>> -		if (keep_working(pool))
>> -			wake_up_worker(pool);
>> -
> 
> This was added by 7576958a9d5a4a6 ("workqueue: wake up a worker when a
> rescuer is leaving a gcwq") to fix a bug reported by Ray Jui.
> 
>   http://thread.gmane.org/gmane.linux.kernel/1098131
> 
> I'm fairly sure it was a valid bug report.  I don't think the
> depletion comes from concurrency management.  It's just the lack of
> chaining which could lead to stall.  What am I missing here?
> 

"http://thread.gmane.org/gmane.linux.kernel/1098131" does not find out the root cause.
I find out the root cause with hard searching from the code, I will describe it later.

Thanks,
Lai

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/