netdev - Re: [PATCH net] sctp: fix an issue that plpmtu can never go to complete state

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: <5a49aee53de52cf3c24246ccf18391aabc0c5e50.camel@redhat.com>
Date: Thu, 18 May 2023 12:18:18 +0200
From: Paolo Abeni <pabeni@...hat.com>
To: Xin Long <lucien.xin@...il.com>, network dev <netdev@...r.kernel.org>, 
	linux-sctp@...r.kernel.org
Cc: davem@...emloft.net, kuba@...nel.org, Eric Dumazet
 <edumazet@...gle.com>,  Marcelo Ricardo Leitner <marcelo.leitner@...il.com>
Subject: Re: [PATCH net] sctp: fix an issue that plpmtu can never go to
 complete state

On Tue, 2023-05-16 at 21:22 -0400, Xin Long wrote:
> When doing plpmtu probe, the probe size is growing every time when it
> receives the ACK during the Search state until the probe fails. When
> the failure occurs, pl.probe_high is set and it goes to the Complete
> state.
> 
> However, if the link pmtu is huge, like 65535 in loopback_dev, the probe
> eventually keeps using SCTP_MAX_PLPMTU as the probe size and never fails.
> Because of that, pl.probe_high can not be set, and the plpmtu probe can
> never go to the Complete state.
> 
> Fix it by setting pl.probe_high to SCTP_MAX_PLPMTU when the probe size
> grows to SCTP_MAX_PLPMTU in sctp_transport_pl_recv(). Also, increase
> the probe size only when the next is less than SCTP_MAX_PLPMTU.
> 
> Fixes: b87641aff9e7 ("sctp: do state transition when a probe succeeds on HB ACK recv path")
> Signed-off-by: Xin Long <lucien.xin@...il.com>
> ---
>  net/sctp/transport.c | 11 +++++++----
>  1 file changed, 7 insertions(+), 4 deletions(-)
> 
> diff --git a/net/sctp/transport.c b/net/sctp/transport.c
> index 2f66a2006517..b0ccfaa4c1d1 100644
> --- a/net/sctp/transport.c
> +++ b/net/sctp/transport.c
> @@ -324,9 +324,11 @@ bool sctp_transport_pl_recv(struct sctp_transport *t)
>  		t->pl.probe_size += SCTP_PL_BIG_STEP;
>  	} else if (t->pl.state == SCTP_PL_SEARCH) {
>  		if (!t->pl.probe_high) {
> -			t->pl.probe_size = min(t->pl.probe_size + SCTP_PL_BIG_STEP,
> -					       SCTP_MAX_PLPMTU);
> -			return false;
> +			if (t->pl.probe_size + SCTP_PL_BIG_STEP < SCTP_MAX_PLPMTU) {
> +				t->pl.probe_size += SCTP_PL_BIG_STEP;
> +				return false;
> +			}
> +			t->pl.probe_high = SCTP_MAX_PLPMTU;

It looks like this way the probed mtu can't reach SCTP_MAX_PLPMTU
anymore, while it was possible before.

What about something alike:

		if (!t->pl.probe_high) {
			if (t->pl.probe_size < SCTP_MAX_PLPMTU) {
				t->pl.probe_size = min(t->pl.probe_size + SCTP_PL_BIG_STEP,
						       SCTP_MAX_PLPMTU);
				return false;
			}
			t->pl.probe_high = SCTP_MAX_PLPMTU;
>  		}
>  		t->pl.probe_size += SCTP_PL_MIN_STEP;
>  		if (t->pl.probe_size >= t->pl.probe_high) {
> @@ -341,7 +343,8 @@ bool sctp_transport_pl_recv(struct sctp_transport *t)
>  	} else if (t->pl.state == SCTP_PL_COMPLETE) {
>  		/* Raise probe_size again after 30 * interval in Search Complete */
>  		t->pl.state = SCTP_PL_SEARCH; /* Search Complete -> Search */
> -		t->pl.probe_size += SCTP_PL_MIN_STEP;
> +		if (t->pl.probe_size + SCTP_PL_MIN_STEP < SCTP_MAX_PLPMTU)
> +			t->pl.probe_size += SCTP_PL_MIN_STEP;

In a similar way, should the above check be:

		if (t->pl.probe_size + SCTP_PL_MIN_STEP <= SCTP_MAX_PLPMTU)
			t->pl.probe_size += SCTP_PL_MIN_STEP;

or simply:
		t->pl.probe_size = min(t->pl.probe_size + SCTP_PL_MIN_STEP, SCTP_MAX_PLPMTU)
> 
Cheers,

Paolo