linux-kernel - Re: Re: [PATCH 1/6] res_counter: handle limit change

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

Date:	Mon, 16 Jun 2008 16:39:37 +0900 (JST)
From:	kamezawa.hiroyu@...fujitsu.com
To:	Pavel Emelyanov <xemul@...nvz.org>
Cc:	KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>,
	linux-mm@...ck.org, LKML <linux-kernel@...r.kernel.org>,
	menage@...gle.com, balbir@...ux.vnet.ibm.com,
	yamamoto@...inux.co.jp, nishimura@....nes.nec.co.jp,
	lizf@...fujitsu.com
Subject: Re: Re: [PATCH 1/6] res_counter:  handle limit change

----- Original Message -----
>> +	 * registered callbacks etc...for res_counter.
>> +	 */
>> +	struct res_counter_ops ops;
>> +	/*
>
Now, write to limit is done in following path.
sys_write() -> write_func of subsys -> write in res_counter -> 
strategy callback -> set limit -> return

Because stragety callback is called in res_counter, we can only do 
something after set-limit without callback. So res_counter should call
another callback before set-limit if it can fail.

>Why would we need such? All res_counter.limit update comes via the appropiate
>cgroup's files, so it can do whatever it needs w/o any callbacks?
>

First reason is that this allows us to implement generic algorithm to
handle limit change. Second is that generic algorithm can be a stack of
functions. I don't like to pass function pointers through several stack
of functions. (And this design allow the code to be much easier to read.
My first version used an argument of function pointer but it was verrry ugly.)

I think when I did all in memcg, someone will comment that "why do that
all in memcg ? please implement generic one to avoid code duplication"

>And (if we definitely need one) isn't it better to make it a
>	struct res_counter_ops *ops;
>pointer?
>
My first version did that. When I added hierarchy_model to ops(see later patch
), I made use of copy of ops. But maybe you're right. Keeping 
res_counter small is important. I'll use pointer in v5.  

Thanks,
-Kame-

--
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/