netdev - Re: [PATCH v2 net 0/3] Introduce refcount_t for reference counting of rose

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]

Message-ID: <58ba5453-52a2-4d26-9a5d-647967c8ede1@free.fr>
Date: Tue, 2 Sep 2025 19:18:52 +0200
From: F6BVP <f6bvp@...e.fr>
To: Takamitsu Iwai <takamitz@...zon.co.jp>
Cc: linux-hams@...r.kernel.org, netdev@...r.kernel.org, davem@...emloft.net,
 edumazet@...gle.com, kuba@...nel.org, pabeni@...hat.com, horms@...nel.org,
 enjuk@...zon.com, mingo@...nel.org, tglx@...utronix.de, hawk@...nel.org,
 n.zhandarovich@...tech.ru, kuniyu@...gle.com
Subject: Re: [PATCH v2 net 0/3] Introduce refcount_t for reference counting of
 rose_neigh

Hi,

I am facing an issue while trying to apply refcount rose patchs to 
latest stable release 6.16.4

In rose_in.c the call to sk_filter_trim_cap function is using an extra 
argument that is not declared in 6.16.4  ~/include/linux/filter.h but 
appears in 6.17.0-rc.

As a result I had to apply the following patch in order to be able to 
build kernel 6.16.4 with refcount patches.

Otherwise ROSE module refcount patchs would prevent building rose module 
in stable kernel

Is there any other solution ?

Regards,

Bernard Pidoux,
F6BVP


Le 27/08/2025 à 16:50, patchwork-bot+netdevbpf@...nel.org a écrit :
> Hello:
> 
> This series was applied to netdev/net.git (main)
> by Jakub Kicinski <kuba@...nel.org>:
> 
> On Sat, 23 Aug 2025 17:58:54 +0900 you wrote:
>> The current implementation of rose_neigh uses 'use' and 'count' field of
>> type unsigned short as a reference count. This approach lacks atomicity,
>> leading to potential race conditions. As a result, syzbot has reported
>> slab-use-after-free errors due to unintended removals.
>>
>> This series introduces refcount_t for reference counting to ensure
>> atomicity and prevent race conditions. The patches are structured as
>> follows:
>>
>> [...]
> 
> Here is the summary with links:
>    - [v2,net,1/3] net: rose: split remove and free operations in rose_remove_neigh()
>      https://git.kernel.org/netdev/net/c/dcb34659028f
>    - [v2,net,2/3] net: rose: convert 'use' field to refcount_t
>      https://git.kernel.org/netdev/net/c/d860d1faa6b2
>    - [v2,net,3/3] net: rose: include node references in rose_neigh refcount
>      https://git.kernel.org/netdev/net/c/da9c9c877597
> 
> You are awesome, thank you!

View attachment "rose_in_reason_dr_ignored.patch" of type "text/plain" (925 bytes)