linux-kernel - Re: [PATCH] cgroup: Fixes the un-paired cgroup lock problem

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: Windows password security audit tool. GUI, reports in PDF.

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

Message-ID: <4AF43E0F.2090000@tmr.com>
Date:	Fri, 06 Nov 2009 10:17:35 -0500
From:	Bill Davidsen <davidsen@....com>
To:	Matt Helsley <matthltc@...ibm.com>
CC:	Li Zefan <lizf@...fujitsu.com>, containers@...ts.osdl.org,
	Paul Menage <menage@...gle.com>,
	Liu Aleaxander <aleaxander@...il.com>,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH] cgroup: Fixes the un-paired cgroup lock problem

Matt Helsley wrote:
> On Thu, Nov 05, 2009 at 07:14:11PM -0500, Bill Davidsen wrote:
>   
>> Li Zefan wrote:
>>     
>>> Bill Davidsen wrote:
>>>   
>>>       
>>>> Li Zefan wrote:
>>>>     
>>>>         
>>>>> Liu Aleaxander wrote:
>>>>>       
>>>>>           
>>>>>> From: Liu Aleaxander <Aleaxander@...il.com>
>>>>>> Date: Wed, 4 Nov 2009 09:27:06 +0800
>>>>>> Subject: [PATCH] Fixes the un-paired cgroup lock problem
>>>>>>
>>>>>> In cgroup_lock_live_group, it locks the cgroup by mutex_lock, while
>>>>>> in the
>>>>>> cgroup_tasks_write, it unlock it by cgroup_unlock. Even though they are
>>>>>> equal, but I do think we should make it pair.
>>>>>>
>>>>>> BTW, should we replace others with cgroup_lock and cgroup_unlock?
>>>>>> Since we already have a wrapper one and it's meaningful.
>>>>>>
>>>>>>         
>>>>>>             
>>>>> Before I read the email body, I thought there is a bug where
>>>>> there is a lock without unlock or vise versa.
>>>>>
>>>>> I agree the case here can be called "unpaired", but I'm not
>>>>> convinced this patch is needed. The code is not buggy or
>>>>> confusing. So the patch neither fixes a bug nor make the code
>>>>> more readable.
>>>>>
>>>>>       
>>>>>           
>>>> I would say it fixes a bug, the one that would be introduced when the
>>>> two methods are no longer compatible and essentially two names for the
>>>> same thing. And while you may know the code so well that you knew
>>>> without looking that this was (currently) okay, there will be lots of
>>>> eyes on this code over the years, I think most people would find use of
>>>> cgroup_lock to lock the cgroup a LOT more readable.
>>>>
>>>> While you can't go back in time to murder your grandfather, it creates
>>>> no paradox to fix a bug before someone writes it.
>>>>
>>>>     
>>>>         
>>> cgroup_lock() is not necessarily more readable than mutex_lock(&cgroup_mutex),
>>> at least the former doesn't tell you the lock is a spin_lock or a mutex.
>>>
>>>   
>>>       
>> That's the point, cgroup_lock() is an abstraction, you want to lock the 
>> cgroup, you call the macro, the macro handles the details, and if 
>> thinking (or the most common cache configurations) change, the code 
>> still works.
>>     
>
> Except it doesn't really "lock the cgroup" as you've been saying -- else
> it would take the cgroup to lock as a parameter. Instead it locks
> "all cgroups". Clearly there's room for misunderstanding even with
> cgroup_lock().
>   

Now that seems to be a good argument for better naming of the locks, 
something like lock_all_cgroups or whatever.

It does seem to support my point of using a macro named after "waht 
operation you are doing" rather than "how you are doing it today."

Thanks for the clarification.

-- 
Bill Davidsen <davidsen@....com>
  Unintended results are the well-earned reward for incompetence.

--
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/