[PATCH v2 net-next 1/1] tcp: Replace min_tso_segs() with tso_segs() CC callback for TCP Prague

[PATCH v2 net-next 1/1] tcp: Replace min_tso_segs() with tso_segs() CC callback for TCP Prague

[chia-yu.chang]

From: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>

This patch replaces existing min_tso_segs() with tso_segs() CC callbak
for CC algorithm to provides explicit tso segment number of each data
burst and overrides tcp_tso_autosize().

No functional change.

Signed-off-by: Ilpo Järvinen <ij@kernel.org>
Signed-off-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
---
 include/net/tcp.h                                |  7 +++++--
 net/ipv4/bpf_tcp_ca.c                            |  4 ++--
 net/ipv4/tcp_bbr.c                               | 14 +++++++++++---
 net/ipv4/tcp_output.c                            | 13 +++++++------
 tools/testing/selftests/bpf/progs/tcp_ca_kfunc.c |  8 ++++----
 5 files changed, 29 insertions(+), 17 deletions(-)

diff --git a/include/net/tcp.h b/include/net/tcp.h
index f063eccbbba3..34d370ea9ceb 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -824,6 +824,9 @@ unsigned int tcp_sync_mss(struct sock *sk, u32 pmtu);
 unsigned int tcp_current_mss(struct sock *sk);
 u32 tcp_clamp_probe0_to_user_timeout(const struct sock *sk, u32 when);
 
+u32 tcp_tso_autosize(const struct sock *sk, unsigned int mss_now,
+		     int min_tso_segs);
+
 /* Bound MSS / TSO packet size with the half of the window */
 static inline int tcp_bound_to_half_wnd(struct tcp_sock *tp, int pktsize)
 {
@@ -1361,8 +1364,8 @@ struct tcp_congestion_ops {
 	/* hook for packet ack accounting (optional) */
 	void (*pkts_acked)(struct sock *sk, const struct ack_sample *sample);
 
-	/* override sysctl_tcp_min_tso_segs (optional) */
-	u32 (*min_tso_segs)(struct sock *sk);
+	/* override tcp_tso_autosize (optional)*/
+	u32 (*tso_segs)(struct sock *sk, u32 mss_now);
 
 	/* new value of cwnd after loss (required) */
 	u32  (*undo_cwnd)(struct sock *sk);
diff --git a/net/ipv4/bpf_tcp_ca.c b/net/ipv4/bpf_tcp_ca.c
index 791e15063237..ed4fea98dfde 100644
--- a/net/ipv4/bpf_tcp_ca.c
+++ b/net/ipv4/bpf_tcp_ca.c
@@ -284,7 +284,7 @@ static void bpf_tcp_ca_pkts_acked(struct sock *sk, const struct ack_sample *samp
 {
 }
 
-static u32 bpf_tcp_ca_min_tso_segs(struct sock *sk)
+static u32 bpf_tcp_ca_tso_segs(struct sock *sk, u32 mss_now)
 {
 	return 0;
 }
@@ -320,7 +320,7 @@ static struct tcp_congestion_ops __bpf_ops_tcp_congestion_ops = {
 	.cwnd_event_tx_start = bpf_tcp_ca_cwnd_event_tx_start,
 	.in_ack_event = bpf_tcp_ca_in_ack_event,
 	.pkts_acked = bpf_tcp_ca_pkts_acked,
-	.min_tso_segs = bpf_tcp_ca_min_tso_segs,
+	.tso_segs = bpf_tcp_ca_tso_segs,
 	.cong_control = bpf_tcp_ca_cong_control,
 	.undo_cwnd = bpf_tcp_ca_undo_cwnd,
 	.sndbuf_expand = bpf_tcp_ca_sndbuf_expand,
diff --git a/net/ipv4/tcp_bbr.c b/net/ipv4/tcp_bbr.c
index 82378a2bfd1e..15536564246c 100644
--- a/net/ipv4/tcp_bbr.c
+++ b/net/ipv4/tcp_bbr.c
@@ -297,11 +297,19 @@ static void bbr_set_pacing_rate(struct sock *sk, u32 bw, int gain)
 }
 
 /* override sysctl_tcp_min_tso_segs */
-__bpf_kfunc static u32 bbr_min_tso_segs(struct sock *sk)
+static u32 bbr_min_tso_segs(struct sock *sk)
 {
 	return READ_ONCE(sk->sk_pacing_rate) < (bbr_min_tso_rate >> 3) ? 1 : 2;
 }
 
+__bpf_kfunc static u32 bbr_tso_segs(struct sock *sk, u32 mss_now)
+{
+	u32 min_tso;
+
+	min_tso = bbr_min_tso_segs(sk);
+	return tcp_tso_autosize(sk, mss_now, min_tso);
+}
+
 static u32 bbr_tso_segs_goal(struct sock *sk)
 {
 	struct tcp_sock *tp = tcp_sk(sk);
@@ -1151,7 +1159,7 @@ static struct tcp_congestion_ops tcp_bbr_cong_ops __read_mostly = {
 	.undo_cwnd	= bbr_undo_cwnd,
 	.cwnd_event_tx_start	= bbr_cwnd_event_tx_start,
 	.ssthresh	= bbr_ssthresh,
-	.min_tso_segs	= bbr_min_tso_segs,
+	.tso_segs	= bbr_tso_segs,
 	.get_info	= bbr_get_info,
 	.set_state	= bbr_set_state,
 };
@@ -1163,7 +1171,7 @@ BTF_ID_FLAGS(func, bbr_sndbuf_expand)
 BTF_ID_FLAGS(func, bbr_undo_cwnd)
 BTF_ID_FLAGS(func, bbr_cwnd_event_tx_start)
 BTF_ID_FLAGS(func, bbr_ssthresh)
-BTF_ID_FLAGS(func, bbr_min_tso_segs)
+BTF_ID_FLAGS(func, bbr_tso_segs)
 BTF_ID_FLAGS(func, bbr_set_state)
 BTF_KFUNCS_END(tcp_bbr_check_kfunc_ids)
 
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 26dd751ec72a..a09b00c5483e 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -2253,8 +2253,8 @@ static bool tcp_nagle_check(bool partial, const struct tcp_sock *tp,
  * for every 2^9 usec (aka 512 us) of RTT, so that the RTT-based allowance
  * is below 1500 bytes after 6 * ~500 usec = 3ms.
  */
-static u32 tcp_tso_autosize(const struct sock *sk, unsigned int mss_now,
-			    int min_tso_segs)
+u32 tcp_tso_autosize(const struct sock *sk, unsigned int mss_now,
+		     int min_tso_segs)
 {
 	unsigned long bytes;
 	u32 r;
@@ -2269,6 +2269,7 @@ static u32 tcp_tso_autosize(const struct sock *sk, unsigned int mss_now,
 
 	return max_t(u32, bytes / mss_now, min_tso_segs);
 }
+EXPORT_SYMBOL(tcp_tso_autosize);
 
 /* Return the number of segments we want in the skb we are transmitting.
  * See if congestion control module wants to decide; otherwise, autosize.
@@ -2278,11 +2279,11 @@ static u32 tcp_tso_segs(struct sock *sk, unsigned int mss_now)
 	const struct tcp_congestion_ops *ca_ops = inet_csk(sk)->icsk_ca_ops;
 	u32 min_tso, tso_segs;
 
-	min_tso = ca_ops->min_tso_segs ?
-			ca_ops->min_tso_segs(sk) :
-			READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_min_tso_segs);
+	min_tso = READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_min_tso_segs);
 
-	tso_segs = tcp_tso_autosize(sk, mss_now, min_tso);
+	tso_segs = ca_ops->tso_segs ?
+			ca_ops->tso_segs(sk, mss_now) :
+			tcp_tso_autosize(sk, mss_now, min_tso);
 	return min_t(u32, tso_segs, sk->sk_gso_max_segs);
 }
 
diff --git a/tools/testing/selftests/bpf/progs/tcp_ca_kfunc.c b/tools/testing/selftests/bpf/progs/tcp_ca_kfunc.c
index 0a3e9d35bf6f..58262e490336 100644
--- a/tools/testing/selftests/bpf/progs/tcp_ca_kfunc.c
+++ b/tools/testing/selftests/bpf/progs/tcp_ca_kfunc.c
@@ -10,7 +10,7 @@ extern u32 bbr_sndbuf_expand(struct sock *sk) __ksym;
 extern u32 bbr_undo_cwnd(struct sock *sk) __ksym;
 extern void bbr_cwnd_event_tx_start(struct sock *sk) __ksym;
 extern u32 bbr_ssthresh(struct sock *sk) __ksym;
-extern u32 bbr_min_tso_segs(struct sock *sk) __ksym;
+extern u32 bbr_tso_segs(struct sock *sk, u32 mss_now) __ksym;
 extern void bbr_set_state(struct sock *sk, u8 new_state) __ksym;
 
 extern void dctcp_init(struct sock *sk) __ksym;
@@ -90,9 +90,9 @@ u32 BPF_PROG(ssthresh, struct sock *sk)
 }
 
 SEC("struct_ops")
-u32 BPF_PROG(min_tso_segs, struct sock *sk)
+u32 BPF_PROG(tso_segs, struct sock *sk, u32 mss_now)
 {
-	return bbr_min_tso_segs(sk);
+	return bbr_tso_segs(sk, mss_now);
 }
 
 SEC("struct_ops")
@@ -120,7 +120,7 @@ struct tcp_congestion_ops tcp_ca_kfunc = {
 	.cwnd_event	= (void *)cwnd_event,
 	.cwnd_event_tx_start = (void *)cwnd_event_tx_start,
 	.ssthresh	= (void *)ssthresh,
-	.min_tso_segs	= (void *)min_tso_segs,
+	.tso_segs	= (void *)tso_segs,
 	.set_state	= (void *)set_state,
 	.pkts_acked     = (void *)pkts_acked,
 	.name		= "tcp_ca_kfunc",
-- 
2.34.1

Re: [PATCH v2 net-next 1/1] tcp: Replace min_tso_segs() with tso_segs() CC callback for TCP Prague

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] BPF programs can cause a kernel panic by passing `mss_now == 0` to the `bbr_tso_segs` kfunc, leading to a division by zero.
- [Medium] The `tcp_tso_autosize` function is not registered as a BPF kfunc, preventing BPF congestion control algorithms from reusing default TSO autosizing logic.
--

commit 39a9bf39a3860ba9dd1b2bed8669b97491df87c6
Author: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>

tcp: Replace min_tso_segs() with tso_segs() CC callback for TCP Prague

This patch replaces existing min_tso_segs() with tso_segs() CC callbak
for CC algorithm to provides explicit tso segment number of each data
burst and overrides tcp_tso_autosize().

> diff --git a/net/ipv4/tcp_bbr.c b/net/ipv4/tcp_bbr.c
> --- a/net/ipv4/tcp_bbr.c
> +++ b/net/ipv4/tcp_bbr.c
[ ... ]
> @@ -297,11 +297,19 @@ static void bbr_set_pacing_rate(struct sock *sk, u32 bw, int gain)
>  }
>  
>  /* override sysctl_tcp_min_tso_segs */
> -__bpf_kfunc static u32 bbr_min_tso_segs(struct sock *sk)
> +static u32 bbr_min_tso_segs(struct sock *sk)
>  {
>  	return READ_ONCE(sk->sk_pacing_rate) < (bbr_min_tso_rate >> 3) ? 1 : 2;
>  }
>  
> +__bpf_kfunc static u32 bbr_tso_segs(struct sock *sk, u32 mss_now)
> +{
> +	u32 min_tso;
> +
> +	min_tso = bbr_min_tso_segs(sk);
> +	return tcp_tso_autosize(sk, mss_now, min_tso);
> +}

[Severity: High]
Can a buggy BPF program trigger a divide-by-zero panic here?

Since bbr_tso_segs() is exposed as a BPF kfunc, the verifier does not
restrict the value ranges of scalar arguments like mss_now. If mss_now is 0,
it is passed directly to tcp_tso_autosize(), where it is used as a divisor:

tcp_tso_autosize()
    ...
    return max_t(u32, bytes / mss_now, min_tso_segs);

Does this need an explicit check for mss_now == 0 to prevent a local denial
of service?

> +
>  static u32 bbr_tso_segs_goal(struct sock *sk)
>  {
>  	struct tcp_sock *tp = tcp_sk(sk);
[ ... ]
> diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
> --- a/net/ipv4/tcp_output.c
> +++ b/net/ipv4/tcp_output.c
[ ... ]
> @@ -2269,6 +2269,7 @@ static u32 tcp_tso_autosize(const struct sock *sk, unsigned int mss_now,
>  
>  	return max_t(u32, bytes / mss_now, min_tso_segs);
>  }
> +EXPORT_SYMBOL(tcp_tso_autosize);

[Severity: Medium]
Is there a way for out-of-tree BPF congestion controls to reuse the default
autosizing logic?

With the removal of min_tso_segs, BPF congestion control algorithms that
want a custom minimum TSO size must implement tso_segs and handle the
autosizing themselves. However, while tcp_tso_autosize() is exported to C
modules here, it does not appear to be registered as a BPF kfunc (e.g., via
BTF_KFUNCS_START).

Does this force BPF developers to either lose custom minimum TSO sizing
capabilities or duplicate the tcp_tso_autosize() logic within their BPF
programs?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260614071756.239686-1-chia-yu.chang@nokia-bell-labs.com?part=1

Re: [PATCH v2 net-next 1/1] tcp: Replace min_tso_segs() with tso_segs() CC callback for TCP Prague

[Jakub Kicinski]

On Sun, 14 Jun 2026 09:17:56 +0200 chia-yu.chang@nokia-bell-labs.com
wrote:
> This patch replaces existing min_tso_segs() with tso_segs() CC callbak
> for CC algorithm to provides explicit tso segment number of each data
> burst and overrides tcp_tso_autosize().
> 
> No functional change.

Eric, Neal, looks good?

The min rtt thing in tcp_tso_autosize() helps a bit but if the sender
gets congested for a longer stretch min_rtts on new connections are
high and we're back to sending small TSO, keeping the sender overloaded.
Which is to say - I _hope_ this also solves some of Meta's problems :)

Re: [PATCH v2 net-next 1/1] tcp: Replace min_tso_segs() with tso_segs() CC callback for TCP Prague

[Jakub Kicinski]

On Mon, 15 Jun 2026 18:51:02 -0700 Jakub Kicinski wrote:
> On Sun, 14 Jun 2026 09:17:56 +0200 chia-yu.chang@nokia-bell-labs.com
> wrote:
> > This patch replaces existing min_tso_segs() with tso_segs() CC callbak
> > for CC algorithm to provides explicit tso segment number of each data
> > burst and overrides tcp_tso_autosize().
> > 
> > No functional change.  
> 
> Eric, Neal, looks good?
> 
> The min rtt thing in tcp_tso_autosize() helps a bit but if the sender
> gets congested for a longer stretch min_rtts on new connections are
> high and we're back to sending small TSO, keeping the sender overloaded.
> Which is to say - I _hope_ this also solves some of Meta's problems :)

Ugh, I didn't see the Sashiko report, it's only CCed to the author and
bpf@, not to netdev :/

The zero-check sounds legit. Let's revisit this after the merge window.

RE: [PATCH v2 net-next 1/1] tcp: Replace min_tso_segs() with tso_segs() CC callback for TCP Prague

[Chia-Yu Chang (Nokia)]

git send-email --to='jolsa@kernel.org' --to='yonghong.song@linux.dev' --to='song@kernel.org' --to='linux-kselftest@vger.kernel.org' --to='memxor@gmail.com' --to='shuah@kernel.org' --to='martin.lau@linux.dev' --to='ast@kernel.org' --to='daniel@iogearbox.net' --to='andrii@kernel.org' --to='eddyz87@gmail.com' --to='horms@kernel.org' --to='dsahern@kernel.org' --to='bpf@vger.kernel.org' --to='netdev@vger.kernel.org' --to='pabeni@redhat.com' --to='jhs@mojatatu.com' --to='kuba@kernel.org' --to='stephen@networkplumber.org' --to='davem@davemloft.net' --to='edumazet@google.com' --to='andrew+netdev@lunn.ch' --to='donald.hunter@gmail.com' --to='kuniyu@google.com' --to='ij@kernel.org' --to='ncardwell@google.com' --to='koen.de_schepper@nokia-bell-labs.com' --to='g.white@cablelabs.com' --to='ingemar.s.johansson@ericsson.com' --to='mirja.kuehlewind@ericsson.com' --to='cheshire@apple.com' --to='rs.ietf@gmx.at' --to='Jason_Livingood@comcast.com' --to='vidhi_goel@apple.com' ~/L4S_patches_20260613/linux_net_next_prague_series0/*.patch --from='chia-yu.chang@nokia-bell-labs.com' --smtp-server='mailrelay.int.nokia.com' --smtp-server-port=25


> -----Original Message-----
> From: Jakub Kicinski <kuba@kernel.org> 
> Sent: Tuesday, June 16, 2026 4:17 AM
> To: edumazet@google.com; ncardwell@google.com
> Cc: Chia-Yu Chang (Nokia) <chia-yu.chang@nokia-bell-labs.com>; jolsa@kernel.org; yonghong.song@linux.dev; song@kernel.org; linux-kselftest@vger.kernel.org; memxor@gmail.com; shuah@kernel.org; martin.lau@linux.dev; ast@kernel.org; daniel@iogearbox.net; andrii@kernel.org; eddyz87@gmail.com; horms@kernel.org; dsahern@kernel.org; bpf@vger.kernel.org; netdev@vger.kernel.org; pabeni@redhat.com; jhs@mojatatu.com; stephen@networkplumber.org; davem@davemloft.net; andrew+netdev@lunn.ch; donald.hunter@gmail.com; kuniyu@google.com; ij@kernel.org; Koen De Schepper (Nokia) <koen.de_schepper@nokia-bell-labs.com>; g.white@cablelabs.com; ingemar.s.johansson@ericsson.com; mirja.kuehlewind@ericsson.com; cheshire@apple.com; rs.ietf@gmx.at; Jason_Livingood@comcast.com; vidhi_goel@apple.com
> Subject: Re: [PATCH v2 net-next 1/1] tcp: Replace min_tso_segs() with tso_segs() CC callback for TCP Prague
> 
> 
> CAUTION: This is an external email. Please be very careful when clicking links or opening attachments. See the URL nok.it/ext for additional information.
> 
> 
> 
> On Mon, 15 Jun 2026 18:51:02 -0700 Jakub Kicinski wrote:
> > On Sun, 14 Jun 2026 09:17:56 +0200 chia-yu.chang@nokia-bell-labs.com
> > wrote:
> > > This patch replaces existing min_tso_segs() with tso_segs() CC 
> > > callbak for CC algorithm to provides explicit tso segment number of 
> > > each data burst and overrides tcp_tso_autosize().
> > >
> > > No functional change.
> >
> > Eric, Neal, looks good?
> >
> > The min rtt thing in tcp_tso_autosize() helps a bit but if the sender 
> > gets congested for a longer stretch min_rtts on new connections are 
> > high and we're back to sending small TSO, keeping the sender overloaded.
> > Which is to say - I _hope_ this also solves some of Meta's problems :)
> 
> Ugh, I didn't see the Sashiko report, it's only CCed to the author and bpf@, not to netdev :/
> 
> The zero-check sounds legit. Let's revisit this after the merge window.

Thanks for the comment, I will take action after the merge window.

And, please correct me if I am wrong, the next eligible submission is expected from 30-June, right?
Thanks!

Chia-Yu

Re: [PATCH v2 net-next 1/1] tcp: Replace min_tso_segs() with tso_segs() CC callback for TCP Prague

[Jakub Kicinski]

On Tue, 16 Jun 2026 12:23:11 +0000 Chia-Yu Chang (Nokia) wrote:
> > On Mon, 15 Jun 2026 18:51:02 -0700 Jakub Kicinski wrote:  
> > > On Sun, 14 Jun 2026 09:17:56 +0200 chia-yu.chang@nokia-bell-labs.com
> > > Eric, Neal, looks good?
> > >
> > > The min rtt thing in tcp_tso_autosize() helps a bit but if the sender 
> > > gets congested for a longer stretch min_rtts on new connections are 
> > > high and we're back to sending small TSO, keeping the sender overloaded.
> > > Which is to say - I _hope_ this also solves some of Meta's problems :)  
> > 
> > Ugh, I didn't see the Sashiko report, it's only CCed to the author and bpf@, not to netdev :/
> > 
> > The zero-check sounds legit. Let's revisit this after the merge window.  
> 
> Thanks for the comment, I will take action after the merge window.
> 
> And, please correct me if I am wrong, the next eligible submission is expected from 30-June, right?

It usually opens Monday morning (PST) so Jun 29th