netdev - Re: [PATCH net] macvlan: Return error on register_netdevice

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]

Message-Id: <20240725.230542.2159475514876691347.syoshida@redhat.com>
Date: Thu, 25 Jul 2024 23:05:42 +0900 (JST)
From: Shigeru Yoshida <syoshida@...hat.com>
To: pabeni@...hat.com, edumazet@...gle.com
Cc: davem@...emloft.net, kuba@...nel.org, netdev@...r.kernel.org,
 linux-kernel@...r.kernel.org
Subject: Re: [PATCH net] macvlan: Return error on
 register_netdevice_notifier() failure

Hi Eric, Paolo,

On Thu, 25 Jul 2024 13:53:27 +0200, Eric Dumazet wrote:
> On Thu, Jul 25, 2024 at 12:13 PM Paolo Abeni <pabeni@...hat.com> wrote:
>>
>>
>>
>> On 7/25/24 11:44, Eric Dumazet wrote:
>> > On Wed, Jul 24, 2024 at 3:56 PM Shigeru Yoshida <syoshida@...hat.com> wrote:
>> >>
>> >> register_netdevice_notifier() may fail, but macvlan_init_module() does
>> >> not handle the failure.  Handle the failure by returning an error.
>> >
>> > How could this fail exactly ? Please provide details, because I do not
>> > think it can.
>>
>> Yup, it looks like the registration can't fail for macvlan.
>>
>> It's better to avoid adding unneeded checks, to reduce noise on the
>> tree, keep stable backport easy, etc.

Thank you for your comments, and it's my bad, sorry.

I happened to look at macvlan's code and found this.  I compared this
with others like macvtap and I assumed that registering for notifier
needs error handling without checking carefully.

> Shigeru, you could send a debug patch when net-next reopens next week,
> so that we do not get another attempt
> on fixing a non-existent bug.

I think we don't need this DEBUG_NET_WARN_ON_ONCE() because it makes
the source code a little messy.

Next time I find something, I'll be more careful.

Thanks,
Shigeru

> diff --git a/drivers/net/macvlan.c b/drivers/net/macvlan.c
> index 24298a33e0e94851ebf9c704c723f25ac7bf5eec..0803fcf8df4c56ede10597c862288c7aa887160e
> 100644
> --- a/drivers/net/macvlan.c
> +++ b/drivers/net/macvlan.c
> @@ -1849,7 +1849,8 @@ static int __init macvlan_init_module(void)
>  {
>         int err;
> 
> -       register_netdevice_notifier(&macvlan_notifier_block);
> +       err = register_netdevice_notifier(&macvlan_notifier_block);
> +       DEBUG_NET_WARN_ON_ONCE(err < 0);
> 
>         err = macvlan_link_register(&macvlan_link_ops);
>         if (err < 0)
>