Re: [PATCH] flow_dissector: fix uninit-value in __skb_flow_dissect() for ETH_ADDRS

[Jiayuan Chen <jiayuan.chen@linux.dev>]

On 6/3/26 4:33 PM, Eric Dumazet wrote:
> On Wed, Jun 3, 2026 at 1:16 AM Zhou, Yun <yun.zhou@windriver.com> wrote:
>>
>>
>> On 6/3/26 13:44, Eric Dumazet wrote:
>>> CAUTION: This email comes from a non Wind River email account!
>>> Do not click links or open attachments unless you recognize the sender and know the content is safe.
>>>
>>> On Tue, Jun 2, 2026 at 8:08 PM Yun Zhou <yun.zhou@windriver.com> wrote:
>>> Please add a Closes: tag
>> I will add a Closes link in v2.
>> Closes: https://syzkaller.appspot.com/bug?extid=fa2f5b1fb06147be5e16
>>> I found some not relevant syzbot report :
>>>
>>> https://lore.kernel.org/netdev/6a196faf.c16d89a8.217f2c.0002.GAE@google.com/
>> This should be the same issue. And it can be reproduced by
>> https://syzkaller.appspot.com/text?tag=ReproC&x=12924152580000
> Please investigate which device allowed to send an Ethernet packet
> smaller than the ethernet header.
>
> We do not want to add tests all over the places, we should fix the origin.
>
> Look for dev->min_header_len
>
> Thanks.


It's TUN.

I think we should reject loading "tc filter ... flower eth_dst 
aa:bb:cc:dd:ee:ff ... " if dev->hard_header_len < 
sizeof(flow_dissector_key_eth_addrs)