linux-kernel - Re: [PATCH] mm/shrinker_debug: Fix possible memory leak in shrinker_debugfs

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>] [day] [month] [year] [list]

Message-ID: <da4ae4e2-4cda-4a86-b71d-f6b46e6f9772@bytedance.com>
Date: Wed, 5 Mar 2025 14:28:38 +0800
From: Qi Zheng <zhengqi.arch@...edance.com>
To: liuye <liuye@...inos.cn>
Cc: akpm@...ux-foundation.org, david@...morbit.com, roman.gushchin@...ux.dev,
 linux-mm@...ck.org, linux-kernel@...r.kernel.org,
 Muchun Song <muchun.song@...ux.dev>
Subject: Re: [PATCH] mm/shrinker_debug: Fix possible memory leak in
 shrinker_debugfs_rename function.



On 3/5/25 2:10 PM, liuye wrote:
> 
> 在 2025/3/5 11:26, Qi Zheng 写道:
>>
>>
>> On 3/5/25 11:17 AM, Muchun Song wrote:
>>>
>>>
>>>> On Mar 5, 2025, at 10:01, Liu Ye <liuye@...inos.cn> wrote:
>>>>
>>>> After calling debugfs_change_name function, the return value should be
>>>> checked and the old name restored. If debugfs_change_name fails, the 
>>>> new
>>>> name memory should be freed.
>>>
>>> Seems it is not a big problem, no memory leak at least. The effect is 
>>> that
>>> the shrinker->name is not consistent with the name displayed in debugfs.
>>> Right? But the improvement LGTM. So:
>>
>> Right, so the subject needs to be changed.
>>
>> Maybe:
>>
>> mm: shrinker: fix name consistency issue in shrinker_debugfs_rename()
>>
>> ?
> 
>   I will send a new patch using this subject later.
> And add  Reviewed-by：Qi Zheng <zhengqi.arch@...edance.com> ？

Sure, with the changes to the subject:

Reviewed-by：Qi Zheng <zhengqi.arch@...edance.com>

Thanks.

> 
> 
>>
>> BTW, it seems that the callers of shrinker_debugfs_rename() did not
>> process the return value of the function?
> 
> Yes,  At the same time, I also found that many positions using
> debugfs_change_name did not determine the return value.
> 
>>
>>>
>>> Reviewed-by: Muchun Song <muchun.song@...ux.dev>
>>>
>>> Thanks.
> Thanks.
>>>
>>>>
>>>> Signed-off-by: Liu Ye <liuye@...inos.cn>
>>>> ---
>>>> mm/shrinker_debug.c | 8 ++++++--
>>>> 1 file changed, 6 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/mm/shrinker_debug.c b/mm/shrinker_debug.c
>>>> index 794bd433cce0..20eaee3e97f7 100644
>>>> --- a/mm/shrinker_debug.c
>>>> +++ b/mm/shrinker_debug.c
>>>> @@ -214,10 +214,14 @@ int shrinker_debugfs_rename(struct shrinker 
>>>> *shrinker, const char *fmt, ...)
>>>> ret = debugfs_change_name(shrinker->debugfs_entry, "%s-%d",
>>>> shrinker->name, shrinker->debugfs_id);
>>>>
>>>> +     if (ret) {
>>>> +         shrinker->name = old;
>>>> +         kfree_const(new);
>>>> +     } else {
>>>> +         kfree_const(old);
>>>> +     }
>>>>     mutex_unlock(&shrinker_mutex);
>>>>
>>>> -     kfree_const(old);
>>>> -
>>>>     return ret;
>>>> }
>>>> EXPORT_SYMBOL(shrinker_debugfs_rename);
>>>> -- 
>>>> 2.25.1
>>>>
>>>
>>