linux-kernel - Re: [PATCH] landlock: Initialize kernel stack variables properly

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

Date:   Fri, 5 Nov 2021 09:58:47 -0700
From:   J Freyensee <why2jjj.linux@...il.com>
To:     Austin Kim <austindh.kim@...il.com>,
        Mickaël Salaün <mic@...ikod.net>
Cc:     "Serge E. Hallyn" <serge@...lyn.com>,
        linux-security-module <linux-security-module@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] landlock: Initialize kernel stack variables properly



On 11/4/21 4:41 AM, Austin Kim wrote:
> 2021년 11월 3일 (수) 오후 9:14, Mickaël Salaün <mic@...ikod.net>님이 작성:
>> Hi Austin,
>>
>> On 03/11/2021 08:14, Austin Kim wrote:
>>> In case kernel stack variables are not initialized properly, there might
>>> be a little chance of kernel information disclosure. So it is better for
>>> kernel stack variables to be initialized with null characters.
>>>
>>> Signed-off-by: Austin Kim <austindh.kim@...il.com>
>>> ---
>>>   security/landlock/syscalls.c | 2 ++
>>>   1 file changed, 2 insertions(+)
>>>
>>> diff --git a/security/landlock/syscalls.c b/security/landlock/syscalls.c
>>> index 32396962f04d..50a6f7091428 100644
>>> --- a/security/landlock/syscalls.c
>>> +++ b/security/landlock/syscalls.c
>>> @@ -320,6 +320,8 @@ SYSCALL_DEFINE4(landlock_add_rule,
>>>        if (rule_type != LANDLOCK_RULE_PATH_BENEATH)
>>>                return -EINVAL;
>>>
>>> +     memset(&path_beneath_attr, 0, sizeof(path_beneath_attr));
>>> +
>> This memset is already done with the copy_from_user() call just below.
>>
> It seems that memset() is done inside copy_from_user().
> Thanks for feedback.

If you are really sensitive of what information may be disclosed in this 
case you can consider memzero_explicit() to use instead:

https://www.kernel.org/doc/htmldocs/kernel-api/API-memzero-explicit.html

Jay

>
> BR,
> Austin Kim
>
>>>        /* Copies raw user space buffer, only one type for now. */
>>>        res = copy_from_user(&path_beneath_attr, rule_attr,
>>>                        sizeof(path_beneath_attr));
>>>