netdev - RE: [PATCH] net: check the return value of the copy_from

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: <0858b2bde3f54c9da0f655a09bab7dab@AcuMS.aculab.com>
Date: Wed, 18 Sep 2024 13:11:34 +0000
From: David Laight <David.Laight@...LAB.COM>
To: 'Qianqiang Liu' <qianqiang.liu@....com>, "xiyou.wangcong@...il.com"
	<xiyou.wangcong@...il.com>, "davem@...emloft.net" <davem@...emloft.net>,
	"edumazet@...gle.com" <edumazet@...gle.com>, "kuba@...nel.org"
	<kuba@...nel.org>, "pabeni@...hat.com" <pabeni@...hat.com>
CC: "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: RE: [PATCH] net: check the return value of the copy_from_sockptr

From: Qianqiang Liu
> Sent: 11 September 2024 06:05
> 
> We must check the return value of the copy_from_sockptr. Otherwise, it
> may cause some weird issues.

Actually there is no point doing the copy for optlen in absolutely every function.
'optlen' can be passed as a kernel address and any user copy done by the caller.
Someone should have spotted that before sockptr_t was used for optlen.

The wrapper code can then also do a correct check for (optlen >= 0) which has
been pretty much broken in most protocols for ~ever.
(I wonder if any (important) userspace relies on a negative optlen
being treated as 4.)

It might mean that the final 'copy_to_user' are done in the opposite
order so that an kernel side side effects aren't reversed if the length
can't be copied out - but I suspect that doesn't matter and the paths
are likely to be untested and buggy.

I have toyed with making the getsockopt() functions return a +ve length
or -ve error - but there are a few strange places that need to update
the length and return an error.

	David

> 
> Signed-off-by: Qianqiang Liu <qianqiang.liu@....com>
> ---
>  net/socket.c | 7 +++++--
>  1 file changed, 5 insertions(+), 2 deletions(-)
> 
> diff --git a/net/socket.c b/net/socket.c
> index 0a2bd22ec105..6b9a414d01d5 100644
> --- a/net/socket.c
> +++ b/net/socket.c
> @@ -2370,8 +2370,11 @@ int do_sock_getsockopt(struct socket *sock, bool compat, int level,
>  	if (err)
>  		return err;
> 
> -	if (!compat)
> -		copy_from_sockptr(&max_optlen, optlen, sizeof(int));
> +	if (!compat) {
> +		err = copy_from_sockptr(&max_optlen, optlen, sizeof(int));
> +		if (err)
> +			return -EFAULT;
> +	}
> 
>  	ops = READ_ONCE(sock->ops);
>  	if (level == SOL_SOCKET) {
> --
> 2.39.2
> 

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)