Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Subject: Re: [PATCH] mm/mlock: fix potential imbalanced rlimit ucounts
 adjustment
To:     Hugh Dickins <hughd@google.com>
CC:     <akpm@linux-foundation.org>, <Herbert.van.den.Bergh@oracle.com>,
        Alexey Gladkov <legion@kernel.org>,
        "Eric W. Biederman" <ebiederm@xmission.com>, <linux-mm@kvack.org>,
        <linux-kernel@vger.kernel.org>
References: <20220310132417.41189-1-linmiaohe@huawei.com>
 <268b3146-2963-15b6-6d6-95a96853314@google.com>
 <bcfc2048-93ce-b357-8671-7070614db36a@huawei.com>
 <1e758c8f-c87b-50ef-52db-942c8a86a32a@google.com>
From:   Miaohe Lin <linmiaohe@huawei.com>
Message-ID: <291c20b2-0559-5226-0718-2d7bb1571079@huawei.com>
Date:   Mon, 14 Mar 2022 11:49:41 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:78.0) Gecko/20100101
 Thunderbird/78.6.0
MIME-Version: 1.0
In-Reply-To: <1e758c8f-c87b-50ef-52db-942c8a86a32a@google.com>
Content-Type: text/plain; charset="utf-8"
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 2022/3/14 11:11, Hugh Dickins wrote:
> On Mon, 14 Mar 2022, Miaohe Lin wrote:
>> On 2022/3/14 10:40, Hugh Dickins wrote:
>>> On Thu, 10 Mar 2022, Miaohe Lin wrote:
>>>
>>>> user_shm_lock forgets to set allowed to 0 when get_ucounts fails. So
>>>> the later user_shm_unlock might do the extra dec_rlimit_ucounts. Fix
>>>> this by resetting allowed to 0.
>>>>
>>>> Fixes: 5ed44a401ddf ("do not limit locked memory when RLIMIT_MEMLOCK is RLIM_INFINITY")
>>>> Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
>>>
>>> NAK.  user_shm_lock() remembers to declare "int allowed = 0" on entry.
>>>
>>
>> If lock_limit is RLIM_INFINITY, "allowed" will be set to 1. And if get_ucounts fails
>> in some corner cases, "allowed" will remain to be 1 while the user_shm_lock ops indeed
>> fails. Or am I miss something?
> 
> You are right, I am wrong: sorry.
> Thanks for pointing now to that RLIM_INFINITY case.
> 
> But then the Fixes tag is wrong: it should be
> Fixes: d7c9e99aee48 ("Reimplement RLIMIT_MEMLOCK on top of ucounts")
>        which introduced the possibility of error down there.
> 

You're right. commit 5ed44a401ddf ("do not limit locked memory when RLIMIT_MEMLOCK is RLIM_INFINITY")
introduced RLIM_INFINITY and set "allowed" to 1 but there is no possibility of error down there.
Will change this in V2.

> With that,
> Acked-by: Hugh Dickins <hughd@google.com>

Many thanks!

> 
>>
>> Many thanks for comment.
>>
>>>> ---
>>>>  mm/mlock.c | 1 +
>>>>  1 file changed, 1 insertion(+)
>>>>
>>>> diff --git a/mm/mlock.c b/mm/mlock.c
>>>> index 29372c0eebe5..efd2dd2943de 100644
>>>> --- a/mm/mlock.c
>>>> +++ b/mm/mlock.c
>>>> @@ -733,6 +733,7 @@ int user_shm_lock(size_t size, struct ucounts *ucounts)
>>>>  	}
>>>>  	if (!get_ucounts(ucounts)) {
>>>>  		dec_rlimit_ucounts(ucounts, UCOUNT_RLIMIT_MEMLOCK, locked);
>>>> +		allowed = 0;
>>>>  		goto out;
>>>>  	}
>>>>  	allowed = 1;
>>>> -- 
>>>> 2.23.0
> .
>