netdev - Re: [PATCH] net: tls: fix possible race condition between do_tls_getsockopt_conf() and do_tls_setsockopt

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: <20230224120606.GI26596@breakpoint.cc>
Date:   Fri, 24 Feb 2023 13:06:06 +0100
From:   Florian Westphal <fw@...len.de>
To:     Hangyu Hua <hbh25y@...il.com>
Cc:     borisp@...dia.com, john.fastabend@...il.com, kuba@...nel.org,
        davem@...emloft.net, edumazet@...gle.com, pabeni@...hat.com,
        davejwatson@...com, aviadye@...lanox.com, ilyal@...lanox.com,
        sd@...asysnail.net, netdev@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] net: tls: fix possible race condition between
 do_tls_getsockopt_conf() and do_tls_setsockopt_conf()

Hangyu Hua <hbh25y@...il.com> wrote:
> ctx->crypto_send.info is not protected by lock_sock in
> do_tls_getsockopt_conf(). A race condition between do_tls_getsockopt_conf()
> and do_tls_setsockopt_conf() can cause a NULL point dereference or
> use-after-free read when memcpy.

Its good practice to quote the relevant parts of the splat here.

> Fixes: 3c4d7559159b ("tls: kernel TLS support")
> Signed-off-by: Hangyu Hua <hbh25y@...il.com>
> ---
>  net/tls/tls_main.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/net/tls/tls_main.c b/net/tls/tls_main.c
> index 3735cb00905d..4956f5149b8e 100644
> --- a/net/tls/tls_main.c
> +++ b/net/tls/tls_main.c
> @@ -374,6 +374,7 @@ static int do_tls_getsockopt_conf(struct sock *sk, char __user *optval,
>  	}
>  
>  	/* get user crypto info */
> +	lock_sock(sk);
>  	if (tx) {
>  		crypto_info = &ctx->crypto_send.info;
>  		cctx = &ctx->tx;
> @@ -381,6 +382,7 @@ static int do_tls_getsockopt_conf(struct sock *sk, char __user *optval,
>  		crypto_info = &ctx->crypto_recv.info;
>  		cctx = &ctx->rx;
>  	}
> +	release_sock(sk);

Could you elaborate how this fixes a UAF bug?

AFAIU do_tls_setsockopt_conf uses ctx-> as scratch space, but this means
that getsockopt can see partial states.

If so, can the setsockopt part be changed so that reads only see
consistent states?