linux-kernel - Re: [PATCH 3/4] cpuset: Fix unchecked calls to NODEMASK

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]

Message-ID: <4D5DDDD7.509@cn.fujitsu.com>
Date:	Fri, 18 Feb 2011 10:47:51 +0800
From:	Li Zefan <lizf@...fujitsu.com>
To:	Paul Menage <menage@...gle.com>
CC:	Andrew Morton <akpm@...ux-foundation.org>,
	LKML <linux-kernel@...r.kernel.org>,
	David Rientjes <rientjes@...gle.com>,
	缪 勰 <miaox@...fujitsu.com>,
	linux-mm@...ck.org
Subject: Re: [PATCH 3/4] cpuset: Fix unchecked calls to NODEMASK_ALLOC()

Paul Menage wrote:
> On Thu, Feb 17, 2011 at 2:46 PM, Andrew Morton
> <akpm@...ux-foundation.org> wrote:
>> On Thu, 17 Feb 2011 09:50:09 +0800
>> Li Zefan <lizf@...fujitsu.com> wrote:
>>
>>> +/*
>>> + * In functions that can't propogate errno to users, to avoid declaring a
>>> + * nodemask_t variable, and avoid using NODEMASK_ALLOC that can return
>>> + * -ENOMEM, we use this global cpuset_mems.
>>> + *
>>> + * It should be used with cgroup_lock held.
>>
>> I'll do s/should/must/ - that would be a nasty bug.
>>
>> I'd be more comfortable about the maintainability of this optimisation
>> if we had
>>
>>        WARN_ON(!cgroup_is_locked());
>>
>> at each site.
>>
> 
> Agreed - that was my first thought on reading the patch. How about:
> 
> static nodemask_t *cpuset_static_nodemask() {

Then this should be 'noinline', otherwise we'll have one copy for each
function that calls it.

>   static nodemask_t nodemask;
>   WARN_ON(!cgroup_is_locked());
>   return &nodemask;
> }
> 
> and then just call cpuset_static_nodemask() in the various locations
> being patched?
> 

I think a defect of this is people might call it twice in one function
but don't know it returns the same variable?

For example in cpuset_attach():

void cpuset_attach(...)
{
	nodemask_t *from = cpuset_static_nodemask();
	nodemask_t *to = cpuset_static_nodemask();
	...
}
--
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/