linux-kernel - Re: [PATCH] mm: vmscan: protect shrinker idr replace with CONFIG

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:   Thu, 5 Dec 2019 13:00:31 +0300
From:   Kirill Tkhai <ktkhai@...tuozzo.com>
To:     Michal Hocko <mhocko@...nel.org>
Cc:     Yang Shi <yang.shi@...ux.alibaba.com>, hannes@...xchg.org,
        shakeelb@...gle.com, guro@...com, akpm@...ux-foundation.org,
        linux-mm@...ck.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] mm: vmscan: protect shrinker idr replace with
 CONFIG_MEMCG

On 05.12.2019 12:43, Michal Hocko wrote:
> On Thu 05-12-19 11:23:28, Kirill Tkhai wrote:
>> On 04.12.2019 22:16, Yang Shi wrote:
>>> Since commit 0a432dcbeb32edcd211a5d8f7847d0da7642a8b4 ("mm: shrinker:
>>> make shrinker not depend on memcg kmem"), shrinkers' idr is protected by
>>> CONFIG_MEMCG instead of CONFIG_MEMCG_KMEM, so it makes no sense to
>>> protect shrinker idr replace with CONFIG_MEMCG_KMEM.
>>>
>>> Cc: Kirill Tkhai <ktkhai@...tuozzo.com>
>>> Cc: Johannes Weiner <hannes@...xchg.org>
>>> Cc: Michal Hocko <mhocko@...e.com>
>>> Cc: Shakeel Butt <shakeelb@...gle.com>
>>> Cc: Roman Gushchin <guro@...com>
>>> Signed-off-by: Yang Shi <yang.shi@...ux.alibaba.com>
>>
>> It looks like that in CONFIG_SLOB case we do not even call some shrinkers
>> for subordinate mem cgroups (i.e., we don't call deferred_split_shrinker),
>> since they never become completely registered.
>>
>> Fixes: 0a432dcbeb32edcd211a5d8f7847d0da7642a8b4 ("mm: shrinker: make shrinker not depend on memcg kmem")
> 
> I am confused. Why the Fixes tag? Nothing should be really broken with
> KMEM config guard right?

idr_replace() is disabled in CONFIG_MEMCG && CONFIG_SLOB case, and this is
wrong.

0a432dcbeb32edcd211a5d8f7847d0da7642a8b4 goes in the series, which enables
shrinker_idr infrastructure for huge_memory.c's deferred_split_shrinker
in CONFIG_MEMCG case. Previously, all SHRINKER_MEMCG_AWARE shrinkers were
based on LRUs, and they remain to base of CONFIG_MEMCG_KMEM.
But deferred_split_shrinker is an exception.

In CONFIG_MEMCG && CONFIG_SLOB case, shrinker_idr contains only shrinker,
and it is deferred_split_shrinker. But it is never actually called, since
idr_replace() is never compiled. deferred_split_shrinker all the time is
staying in half-registered state, and it's never called for subordinate
mem cgroups.

So, this is a BUG, and this should go to stable.

> This is a mere clean up AFAICS.
> 
>> Reviewed-by: Kirill Tkhai <ktkhai@...tuozzo.com>
>>
>>> ---
>>>  mm/vmscan.c | 2 +-
>>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/mm/vmscan.c b/mm/vmscan.c
>>> index ee4eecc..e7f10c4 100644
>>> --- a/mm/vmscan.c
>>> +++ b/mm/vmscan.c
>>> @@ -422,7 +422,7 @@ void register_shrinker_prepared(struct shrinker *shrinker)
>>>  {
>>>  	down_write(&shrinker_rwsem);
>>>  	list_add_tail(&shrinker->list, &shrinker_list);
>>> -#ifdef CONFIG_MEMCG_KMEM
>>> +#ifdef CONFIG_MEMCG
>>>  	if (shrinker->flags & SHRINKER_MEMCG_AWARE)
>>>  		idr_replace(&shrinker_idr, shrinker, shrinker->id);
>>>  #endif
>>>
>