Re: KMSAN: uninit-value in tomoyo_check_inet_address

From: Tetsuo Handa
Date: Thu May 16 2019 - 09:39:42 EST


On 2019/05/16 22:19, Alexander Potapenko wrote:
>> commit e6193f78bb689f3f424559bb45f4a091c8b314df
>> Author: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx>
>> Date: Fri Apr 12 19:59:36 2019 +0900
>>
>> tomoyo: Check address length before reading address family
>>
>> KMSAN will complain if valid address length passed to bind()/connect()/
>> sendmsg() is shorter than sizeof("struct sockaddr"->sa_family) bytes.
>>
>> Signed-off-by: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx>
>> Signed-off-by: James Morris <jamorris@xxxxxxxxxxxxxxxxxxx>
> Apparently the fix didn't make it to 5.1, I'll cherry-pick it to KMSAN tree.

Together with three more patches listed bottom, please.

> OOC, how did you know about this bug?

I did a tree-wide check when writing a patch for

KMSAN: uninit-value in rds_bind
KMSAN: uninit-value in rds_connect

reports. Patches for LSM part just did not make it to 5.1.



commit b9ef5513c99bf9c8bfd9c9e8051b67f52b2dee1e
Author: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx>
Date: Fri Apr 12 19:59:35 2019 +0900

smack: Check address length before reading address family

commit 619ae03e922b65a1a5d4269ceae1e9e13a058d6b
Author: Casey Schaufler <casey@xxxxxxxxxxxxxxxx>
Date: Tue Apr 30 14:13:32 2019 -0700

Smack: Fix kbuild reported build error

commit c750e6929d3c76d13d1d0ba475989d6dd74785d5
Author: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx>
Date: Fri Apr 12 19:59:34 2019 +0900

selinux: Check address length before reading address family