linux-kernel - [PATCH 4.19 38/88] tcp: fix potential xmit stalls caused by TCP_NOTSENT

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: <20220510130734.852784026@linuxfoundation.org>
Date:   Tue, 10 May 2022 15:07:23 +0200
From:   Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To:     linux-kernel@...r.kernel.org
Cc:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        stable@...r.kernel.org, Eric Dumazet <edumazet@...gle.com>,
        Doug Porter <dsp@...com>,
        Soheil Hassas Yeganeh <soheil@...gle.com>,
        Neal Cardwell <ncardwell@...gle.com>,
        "David S. Miller" <davem@...emloft.net>,
        Sasha Levin <sashal@...nel.org>
Subject: [PATCH 4.19 38/88] tcp: fix potential xmit stalls caused by TCP_NOTSENT_LOWAT

From: Eric Dumazet <edumazet@...gle.com>

[ Upstream commit 4bfe744ff1644fbc0a991a2677dc874475dd6776 ]

I had this bug sitting for too long in my pile, it is time to fix it.

Thanks to Doug Porter for reminding me of it!

We had various attempts in the past, including commit
0cbe6a8f089e ("tcp: remove SOCK_QUEUE_SHRUNK"),
but the issue is that TCP stack currently only generates
EPOLLOUT from input path, when tp->snd_una has advanced
and skb(s) cleaned from rtx queue.

If a flow has a big RTT, and/or receives SACKs, it is possible
that the notsent part (tp->write_seq - tp->snd_nxt) reaches 0
and no more data can be sent until tp->snd_una finally advances.

What is needed is to also check if POLLOUT needs to be generated
whenever tp->snd_nxt is advanced, from output path.

This bug triggers more often after an idle period, as
we do not receive ACK for at least one RTT. tcp_notsent_lowat
could be a fraction of what CWND and pacing rate would allow to
send during this RTT.

In a followup patch, I will remove the bogus call
to tcp_chrono_stop(sk, TCP_CHRONO_SNDBUF_LIMITED)
from tcp_check_space(). Fact that we have decided to generate
an EPOLLOUT does not mean the application has immediately
refilled the transmit queue. This optimistic call
might have been the reason the bug seemed not too serious.

Tested:

200 ms rtt, 1% packet loss, 32 MB tcp_rmem[2] and tcp_wmem[2]

$ echo 500000 >/proc/sys/net/ipv4/tcp_notsent_lowat
$ cat bench_rr.sh
SUM=0
for i in {1..10}
do
 V=`netperf -H remote_host -l30 -t TCP_RR -- -r 10000000,10000 -o LOCAL_BYTES_SENT | egrep -v "MIGRATED|Bytes"`
 echo $V
 SUM=$(($SUM + $V))
done
echo SUM=$SUM

Before patch:
$ bench_rr.sh
130000000
80000000
140000000
140000000
140000000
140000000
130000000
40000000
90000000
110000000
SUM=1140000000

After patch:
$ bench_rr.sh
430000000
590000000
530000000
450000000
450000000
350000000
450000000
490000000
480000000
460000000
SUM=4680000000  # This is 410 % of the value before patch.

Fixes: c9bee3b7fdec ("tcp: TCP_NOTSENT_LOWAT socket option")
Signed-off-by: Eric Dumazet <edumazet@...gle.com>
Reported-by: Doug Porter <dsp@...com>
Cc: Soheil Hassas Yeganeh <soheil@...gle.com>
Cc: Neal Cardwell <ncardwell@...gle.com>
Acked-by: Soheil Hassas Yeganeh <soheil@...gle.com>
Signed-off-by: David S. Miller <davem@...emloft.net>
Signed-off-by: Sasha Levin <sashal@...nel.org>
---
 include/net/tcp.h     |  1 +
 net/ipv4/tcp_input.c  | 12 +++++++++++-
 net/ipv4/tcp_output.c |  1 +
 3 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/include/net/tcp.h b/include/net/tcp.h
index 3f0d654984cf..f0d2e2571f56 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -594,6 +594,7 @@ void tcp_synack_rtt_meas(struct sock *sk, struct request_sock *req);
 void tcp_reset(struct sock *sk);
 void tcp_skb_mark_lost_uncond_verify(struct tcp_sock *tp, struct sk_buff *skb);
 void tcp_fin(struct sock *sk);
+void tcp_check_space(struct sock *sk);
 
 /* tcp_timer.c */
 void tcp_init_xmit_timers(struct sock *);
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 757e1f60e00d..d71326f3777c 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -5167,7 +5167,17 @@ static void tcp_new_space(struct sock *sk)
 	sk->sk_write_space(sk);
 }
 
-static void tcp_check_space(struct sock *sk)
+/* Caller made space either from:
+ * 1) Freeing skbs in rtx queues (after tp->snd_una has advanced)
+ * 2) Sent skbs from output queue (and thus advancing tp->snd_nxt)
+ *
+ * We might be able to generate EPOLLOUT to the application if:
+ * 1) Space consumed in output/rtx queues is below sk->sk_sndbuf/2
+ * 2) notsent amount (tp->write_seq - tp->snd_nxt) became
+ *    small enough that tcp_stream_memory_free() decides it
+ *    is time to generate EPOLLOUT.
+ */
+void tcp_check_space(struct sock *sk)
 {
 	if (sock_flag(sk, SOCK_QUEUE_SHRUNK)) {
 		sock_reset_flag(sk, SOCK_QUEUE_SHRUNK);
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 97c3b616d594..8543cd724d54 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -69,6 +69,7 @@ static void tcp_event_new_data_sent(struct sock *sk, struct sk_buff *skb)
 
 	NET_ADD_STATS(sock_net(sk), LINUX_MIB_TCPORIGDATASENT,
 		      tcp_skb_pcount(skb));
+	tcp_check_space(sk);
 }
 
 /* SND.NXT, if window was not shrunk or the amount of shrunk was less than one
-- 
2.35.1