[PATCH v2 net] net: fix the RTO timer retransmitting skb every 1ms if linear option is enabled

[PATCH v2 net] net: fix the RTO timer retransmitting skb every 1ms if linear option is enabled

[Jason Xing]

From: Jason Xing <kernelxing@tencent.com>

In the real workload, I encountered an issue which could cause the RTO
timer to retransmit the skb per 1ms with linear option enabled. The amount
of lost-retransmitted skbs can go up to 1000+ instantly.

The root cause is that if the icsk_rto happens to be zero in the 6th round
(which is the TCP_THIN_LINEAR_RETRIES value), then it will always be zero
due to the changed calculation method in tcp_retransmit_timer() as follows:

icsk->icsk_rto = min(icsk->icsk_rto << 1, TCP_RTO_MAX);

Above line could be converted to
icsk->icsk_rto = min(0 << 1, TCP_RTO_MAX) = 0

Therefore, the timer expires so quickly without any doubt.

I read through the RFC 6298 and found that the RTO value can be rounded
up to a certain value, in Linux, say TCP_RTO_MIN as default, which is
regarded as the lower bound in this patch as suggested by Eric.

Fixes: 36e31b0af587 ("net: TCP thin linear timeouts")
Suggested-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jason Xing <kernelxing@tencent.com>
---
v2:
1) nit: alway->always and the indentation style suggested by Simon.
---
 net/ipv4/tcp_timer.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c
index d45c96c7f5a4..69795b273419 100644
--- a/net/ipv4/tcp_timer.c
+++ b/net/ipv4/tcp_timer.c
@@ -599,7 +599,9 @@ void tcp_retransmit_timer(struct sock *sk)
 	    tcp_stream_is_thin(tp) &&
 	    icsk->icsk_retransmits <= TCP_THIN_LINEAR_RETRIES) {
 		icsk->icsk_backoff = 0;
-		icsk->icsk_rto = min(__tcp_set_rto(tp), TCP_RTO_MAX);
+		icsk->icsk_rto = clamp(__tcp_set_rto(tp),
+				       tcp_rto_min(sk),
+				       TCP_RTO_MAX);
 	} else if (sk->sk_state != TCP_SYN_SENT ||
 		   icsk->icsk_backoff >
 		   READ_ONCE(net->ipv4.sysctl_tcp_syn_linear_timeouts)) {
-- 
2.37.3

Re: [PATCH v2 net] net: fix the RTO timer retransmitting skb every 1ms if linear option is enabled

[Jason Xing]

On Fri, Aug 11, 2023 at 10:38 AM Jason Xing <kerneljasonxing@gmail.com> wrote:
>
> From: Jason Xing <kernelxing@tencent.com>
>
> In the real workload, I encountered an issue which could cause the RTO
> timer to retransmit the skb per 1ms with linear option enabled. The amount
> of lost-retransmitted skbs can go up to 1000+ instantly.
>
> The root cause is that if the icsk_rto happens to be zero in the 6th round
> (which is the TCP_THIN_LINEAR_RETRIES value), then it will always be zero
> due to the changed calculation method in tcp_retransmit_timer() as follows:
>
> icsk->icsk_rto = min(icsk->icsk_rto << 1, TCP_RTO_MAX);
>
> Above line could be converted to
> icsk->icsk_rto = min(0 << 1, TCP_RTO_MAX) = 0
>
> Therefore, the timer expires so quickly without any doubt.
>
> I read through the RFC 6298 and found that the RTO value can be rounded
> up to a certain value, in Linux, say TCP_RTO_MIN as default, which is
> regarded as the lower bound in this patch as suggested by Eric.
>
> Fixes: 36e31b0af587 ("net: TCP thin linear timeouts")
> Suggested-by: Eric Dumazet <edumazet@google.com>
> Signed-off-by: Jason Xing <kernelxing@tencent.com>

Hello maintainers,

I wonder why someone in the patchwork[1] changed this v2 patch into
Superseded status without comments? Should I convert it to a new
status or something else?

[1]: https://patchwork.kernel.org/project/netdevbpf/patch/20230811023747.12065-1-kerneljasonxing@gmail.com/

Thanks,
Jason

> ---
> v2:
> 1) nit: alway->always and the indentation style suggested by Simon.
> ---
>  net/ipv4/tcp_timer.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c
> index d45c96c7f5a4..69795b273419 100644
> --- a/net/ipv4/tcp_timer.c
> +++ b/net/ipv4/tcp_timer.c
> @@ -599,7 +599,9 @@ void tcp_retransmit_timer(struct sock *sk)
>             tcp_stream_is_thin(tp) &&
>             icsk->icsk_retransmits <= TCP_THIN_LINEAR_RETRIES) {
>                 icsk->icsk_backoff = 0;
> -               icsk->icsk_rto = min(__tcp_set_rto(tp), TCP_RTO_MAX);
> +               icsk->icsk_rto = clamp(__tcp_set_rto(tp),
> +                                      tcp_rto_min(sk),
> +                                      TCP_RTO_MAX);
>         } else if (sk->sk_state != TCP_SYN_SENT ||
>                    icsk->icsk_backoff >
>                    READ_ONCE(net->ipv4.sysctl_tcp_syn_linear_timeouts)) {
> --
> 2.37.3
>

Re: [PATCH v2 net] net: fix the RTO timer retransmitting skb every 1ms if linear option is enabled

[Jakub Kicinski]

On Tue, 15 Aug 2023 10:08:13 +0800 Jason Xing wrote:
> I wonder why someone in the patchwork[1] changed this v2 patch into
> Superseded status without comments? Should I convert it to a new
> status or something else?

Ack, looks like a mistake, let me bring it back.
-- 
pw-bot: under-review

Re: [PATCH v2 net] net: fix the RTO timer retransmitting skb every 1ms if linear option is enabled

[Jason Xing]

On Tue, Aug 15, 2023 at 10:24 AM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Tue, 15 Aug 2023 10:08:13 +0800 Jason Xing wrote:
> > I wonder why someone in the patchwork[1] changed this v2 patch into
> > Superseded status without comments? Should I convert it to a new
> > status or something else?
>
> Ack, looks like a mistake, let me bring it back.

Thanks.

Jason
> --
> pw-bot: under-review

Re: [PATCH v2 net] net: fix the RTO timer retransmitting skb every 1ms if linear option is enabled

[Eric Dumazet]

On Fri, Aug 11, 2023 at 4:38 AM Jason Xing <kerneljasonxing@gmail.com> wrote:
>
> From: Jason Xing <kernelxing@tencent.com>
>
> In the real workload, I encountered an issue which could cause the RTO
> timer to retransmit the skb per 1ms with linear option enabled. The amount
> of lost-retransmitted skbs can go up to 1000+ instantly.
>
> The root cause is that if the icsk_rto happens to be zero in the 6th round
> (which is the TCP_THIN_LINEAR_RETRIES value), then it will always be zero
> due to the changed calculation method in tcp_retransmit_timer() as follows:
>
> icsk->icsk_rto = min(icsk->icsk_rto << 1, TCP_RTO_MAX);
>
> Above line could be converted to
> icsk->icsk_rto = min(0 << 1, TCP_RTO_MAX) = 0
>
> Therefore, the timer expires so quickly without any doubt.
>
> I read through the RFC 6298 and found that the RTO value can be rounded
> up to a certain value, in Linux, say TCP_RTO_MIN as default, which is
> regarded as the lower bound in this patch as suggested by Eric.
>
> Fixes: 36e31b0af587 ("net: TCP thin linear timeouts")
> Suggested-by: Eric Dumazet <edumazet@google.com>
> Signed-off-by: Jason Xing <kernelxing@tencent.com>

Reviewed-by: Eric Dumazet <edumazet@google.com>

Re: [PATCH v2 net] net: fix the RTO timer retransmitting skb every 1ms if linear option is enabled

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to netdev/net.git (main)
by David S. Miller <davem@davemloft.net>:

On Fri, 11 Aug 2023 10:37:47 +0800 you wrote:
> From: Jason Xing <kernelxing@tencent.com>
> 
> In the real workload, I encountered an issue which could cause the RTO
> timer to retransmit the skb per 1ms with linear option enabled. The amount
> of lost-retransmitted skbs can go up to 1000+ instantly.
> 
> The root cause is that if the icsk_rto happens to be zero in the 6th round
> (which is the TCP_THIN_LINEAR_RETRIES value), then it will always be zero
> due to the changed calculation method in tcp_retransmit_timer() as follows:
> 
> [...]

Here is the summary with links:
  - [v2,net] net: fix the RTO timer retransmitting skb every 1ms if linear option is enabled
    https://git.kernel.org/netdev/net/c/e4dd0d3a2f64

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html