netdev - Re: [PATCH net 1/1] sctp: Start shutdown on association restart if in SHUTDOWN-SENT state and socket is closed

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: <20200520194629.GS2491@localhost.localdomain>
Date:   Wed, 20 May 2020 16:46:29 -0300
From:   Marcelo Ricardo Leitner <marcelo.leitner@...il.com>
To:     Jere Leppänen <jere.leppanen@...ia.com>
Cc:     linux-sctp@...r.kernel.org, netdev@...r.kernel.org,
        Vlad Yasevich <vyasevich@...il.com>,
        Neil Horman <nhorman@...driver.com>,
        "David S . Miller" <davem@...emloft.net>,
        David Laight <David.Laight@...lab.com>
Subject: Re: [PATCH net 1/1] sctp: Start shutdown on association restart if
 in SHUTDOWN-SENT state and socket is closed

On Wed, May 20, 2020 at 06:15:31PM +0300, Jere Leppänen wrote:
> Commit bdf6fa52f01b ("sctp: handle association restarts when the
> socket is closed.") starts shutdown when an association is restarted,
> if in SHUTDOWN-PENDING state and the socket is closed. However, the
> rationale stated in that commit applies also when in SHUTDOWN-SENT
> state - we don't want to move an association to ESTABLISHED state when
> the socket has been closed, because that results in an association
> that is unreachable from user space.
> 
> The problem scenario:
> 
> 1.  Client crashes and/or restarts.
> 
> 2.  Server (using one-to-one socket) calls close(). SHUTDOWN is lost.
> 
> 3.  Client reconnects using the same addresses and ports.
> 
> 4.  Server's association is restarted. The association and the socket
>     move to ESTABLISHED state, even though the server process has
>     closed its descriptor.
> 
> Also, after step 4 when the server process exits, some resources are
> leaked in an attempt to release the underlying inet sock structure in
> ESTABLISHED state:
> 
>     IPv4: Attempt to release TCP socket in state 1 00000000377288c7
> 
> Fix by acting the same way as in SHUTDOWN-PENDING state. That is, if
> an association is restarted in SHUTDOWN-SENT state and the socket is
> closed, then start shutdown and don't move the association or the
> socket to ESTABLISHED state.
> 
> Fixes: bdf6fa52f01b ("sctp: handle association restarts when the socket is closed.")
> Signed-off-by: Jere Leppänen <jere.leppanen@...ia.com>
> ---
>  net/sctp/sm_statefuns.c | 9 +++++----
>  1 file changed, 5 insertions(+), 4 deletions(-)
> 
> diff --git a/net/sctp/sm_statefuns.c b/net/sctp/sm_statefuns.c
> index 26788f4a3b9e..e86620fbd90f 100644
> --- a/net/sctp/sm_statefuns.c
> +++ b/net/sctp/sm_statefuns.c
> @@ -1856,12 +1856,13 @@ static enum sctp_disposition sctp_sf_do_dupcook_a(
>  	/* Update the content of current association. */
>  	sctp_add_cmd_sf(commands, SCTP_CMD_UPDATE_ASSOC, SCTP_ASOC(new_asoc));
>  	sctp_add_cmd_sf(commands, SCTP_CMD_EVENT_ULP, SCTP_ULPEVENT(ev));
> -	if (sctp_state(asoc, SHUTDOWN_PENDING) &&
> +	if ((sctp_state(asoc, SHUTDOWN_PENDING) ||
> +	     sctp_state(asoc, SHUTDOWN_SENT)) &&
>  	    (sctp_sstate(asoc->base.sk, CLOSING) ||
>  	     sock_flag(asoc->base.sk, SOCK_DEAD))) {
> -		/* if were currently in SHUTDOWN_PENDING, but the socket
> -		 * has been closed by user, don't transition to ESTABLISHED.
> -		 * Instead trigger SHUTDOWN bundled with COOKIE_ACK.
> +		/* If the socket has been closed by user, don't
> +		 * transition to ESTABLISHED. Instead trigger SHUTDOWN
> +		 * bundled with COOKIE_ACK.
>  		 */
>  		sctp_add_cmd_sf(commands, SCTP_CMD_REPLY, SCTP_CHUNK(repl));
>  		return sctp_sf_do_9_2_start_shutdown(net, ep, asoc,
> 
> base-commit: 20a785aa52c82246055a089e55df9dac47d67da1

This last line is not standard, but git didn't complain about it here.

Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@...il.com>