mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Jiayuan Chen <jiayuan.chen@linux.dev>
To: netdev@vger.kernel.org
Cc: Jiayuan Chen <jiayuan.chen@linux.dev>,
	Eric Dumazet <edumazet@kernel.org>,
	Eric Dumazet <edumazet@google.com>,
	Neal Cardwell <ncardwell@google.com>,
	Kuniyuki Iwashima <kuniyu@google.com>,
	"David S. Miller" <davem@davemloft.net>,
	Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>,
	Simon Horman <horms@kernel.org>,
	Stephen Hemminger <stephen@networkplumber.org>,
	linux-kernel@vger.kernel.org
Subject: [PATCH net-next v3 3/3] tcp_cubic: fix divide by zero and endless loop on bad module params
Date: Thu,  8 Oct 2026 17:26:36 +0800	[thread overview]
Message-ID: <20261008092641.140777-4-jiayuan.chen@linux.dev> (raw)
In-Reply-To: <20261008092641.140777-1-jiayuan.chen@linux.dev>

beta_scale and cube_factor are computed once at module init, and
beta == 1024 or bic_scale == 0 is a divide by zero right there. Other
out of range values give garbage: negative beta can make beta_scale 0,
and bic_scale * 10 can overflow. Reject them. Read beta once, as it
can be written through sysfs before the init function runs.

A small beta also gives a small beta_scale, and (cwnd * beta_scale) >> 3
truncates to 0 for a tiny cwnd, so the TCP friendliness loop never
ends. Rather than testing delta on every ACK, make sure beta_scale is
at least 8 at init, so delta is >= 1 within the cwnd < 1 million
packets limit this code is designed for. This slows the TCP friendly
estimate for beta < 512, a backoff harder than Reno's 0.5, which is
not a setting anyone should use.

Fixes: df3271f3361b ("[TCP] BIC: CUBIC window growth (2.0)")
Suggested-by: Eric Dumazet <edumazet@kernel.org>
Reviewed-by: Eric Dumazet <edumazet@kernel.org>
Signed-off-by: Jiayuan Chen <jiayuan.chen@linux.dev>
---
(cwnd * beta_scale) can wrap for cwnd >= 33M packets, far beyond the
1 million packets this code is designed for, so no fast path check.
---
 net/ipv4/tcp_cubic.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/net/ipv4/tcp_cubic.c b/net/ipv4/tcp_cubic.c
index 119bf8cbb007c..a88e4bf86150f 100644
--- a/net/ipv4/tcp_cubic.c
+++ b/net/ipv4/tcp_cubic.c
@@ -500,16 +500,26 @@ static const struct btf_kfunc_id_set tcp_cubic_kfunc_set = {
 
 static int __init cubictcp_register(void)
 {
+	int b = READ_ONCE(beta);
 	int ret;
 
 	BUILD_BUG_ON(sizeof(struct bictcp) > ICSK_CA_PRIV_SIZE);
 
+	if (b < 0 || b >= BICTCP_BETA_SCALE ||
+	    bic_scale <= 0 || bic_scale > INT_MAX / 10) {
+		pr_err("tcp_cubic: invalid beta %d or bic_scale %d\n",
+		       b, bic_scale);
+		return -EINVAL;
+	}
+
 	/* Precompute a bunch of the scaling factors that are used per-packet
 	 * based on SRTT of 100ms
 	 */
 
-	beta_scale = 8*(BICTCP_BETA_SCALE+beta) / 3
-		/ (BICTCP_BETA_SCALE - beta);
+	beta_scale = 8 * (BICTCP_BETA_SCALE + b) / 3
+		/ (BICTCP_BETA_SCALE - b);
+	/* bictcp_update() needs (cwnd * beta_scale) >> 3 to be >= 1 */
+	beta_scale = max(beta_scale, 8U);
 
 	cube_rtt_scale = (bic_scale * 10);	/* 1024*c/rtt */
 
-- 
2.43.0


  parent reply	other threads:[~2026-10-08  9:27 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-10-08  9:26 [PATCH net-next v3 0/3] tcp: fix a few divide-by-zero in congestion control modules Jiayuan Chen
2026-10-08  9:26 ` [PATCH net-next v3 1/3] tcp_bic: fix divide by zero on max_increment == 0 Jiayuan Chen
2026-10-08  9:26 ` [PATCH net-next v3 2/3] tcp_hybla: fix divide by zero on rtt0 " Jiayuan Chen
2026-10-08  9:26 ` Jiayuan Chen [this message]
2026-10-08  9:30 ` [PATCH net-next v3 0/3] tcp: fix a few divide-by-zero in congestion control modules netdev-bot+sinfo

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20261008092641.140777-4-jiayuan.chen@linux.dev \
    --to=jiayuan.chen@linux.dev \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=edumazet@kernel.org \
    --cc=horms@kernel.org \
    --cc=kuba@kernel.org \
    --cc=kuniyu@google.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=ncardwell@google.com \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    --cc=stephen@networkplumber.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®