netdev - Re: [PATCH] net: Finish up ->msg_control{,

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]

Date:   Wed, 12 Apr 2023 17:25:48 +0200
From:   Christoph Hellwig <hch@....de>
To:     Kevin Brodsky <kevin.brodsky@....com>
Cc:     netdev@...r.kernel.org, linux-kernel@...r.kernel.org,
        Christoph Hellwig <hch@....de>,
        Eric Dumazet <edumazet@...gle.com>,
        "David S. Miller" <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>
Subject: Re: [PATCH] net: Finish up ->msg_control{,_user} split

On Tue, Apr 11, 2023 at 01:26:25PM +0100, Kevin Brodsky wrote:
> This patch is attempting to complete the split. Most issues are about
> msg_control being used when in fact a user pointer is stored in the
> union; msg_control_user is now used instead. An exception is made
> for null checks, as it should be safe to use msg_control
> unconditionally for that purpose.

So all of the fixes looks good to me.

> Additionally, a special situation in
> cmsghdr_from_user_compat_to_kern() is addressed. There the input
> struct msghdr holds a user pointer (msg_control_user), but a kernel
> pointer is stored in msg_control when returning. msg_control_is_user
> is now updated accordingly.

But this is a small isolated real bugfix.  So I'd suggest to split
this into a simple and easily backportable patch, and do the rest
in another.

> diff --git a/net/ipv6/ipv6_sockglue.c b/net/ipv6/ipv6_sockglue.c
> index 2917dd8d198c..ae818ff46224 100644
> --- a/net/ipv6/ipv6_sockglue.c
> +++ b/net/ipv6/ipv6_sockglue.c
> @@ -716,6 +716,7 @@ int do_ipv6_setsockopt(struct sock *sk, int level, int optname,
>  			goto done;
>  
>  		msg.msg_controllen = optlen;
> +		msg.msg_control_is_user = false;

And this is another one that has a real effect.