* [PATCH] tcp: use RTAX_CWND for outgoing connections properly @ 2010-12-17 22:14 Jiri Kosina 2010-12-22 10:00 ` Jiri Kosina 0 siblings, 1 reply; 10+ messages in thread From: Jiri Kosina @ 2010-12-17 22:14 UTC (permalink / raw) To: David Miller, netdev, Eric Dumazet; +Cc: linux-kernel, Vojtech Pavlik For outgoing connections, the initial value of cwnd is always set to 2 (in tcp_v4_init_sock()) regardless of setting of RTAX_CWND. For incoming connections, this is handled properly in tcp_init_metrics(). As a result of this, Linux TCP stack always uses cwnd == 2 at the beginning of outgoing TCP session (i.e. waits for ACK after 2 packets once the connection has been established) and grows it in accordance with slow-start algorithm only after it receives ACK for first two packets. When slow-start triggers later during the connection (e.g. after idle), cwnd is properly re-initialized to RTAX_CWND value (if specified) through tcp_cwnd_restart() -> tcp_init_cwnd(). Initialize tp->snd_cwnd properly so that RTAX_CWND value is being used also in the slow-start phase for the first packets in the connection. Signed-off-by: Jiri Kosina <jkosina@suse.cz> --- net/ipv4/tcp_output.c | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-) diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c index 61c2463..6dbc55b 100644 --- a/net/ipv4/tcp_output.c +++ b/net/ipv4/tcp_output.c @@ -2572,6 +2572,8 @@ static void tcp_connect_init(struct sock *sk) &rcv_wscale, dst_metric(dst, RTAX_INITRWND)); + tp->snd_cwnd = tcp_init_cwnd(tp, dst); + tp->rx_opt.rcv_wscale = rcv_wscale; tp->rcv_ssthresh = tp->rcv_wnd; -- 1.7.3.1 ^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [PATCH] tcp: use RTAX_CWND for outgoing connections properly 2010-12-17 22:14 [PATCH] tcp: use RTAX_CWND for outgoing connections properly Jiri Kosina @ 2010-12-22 10:00 ` Jiri Kosina 2010-12-22 11:29 ` Eric Dumazet 2010-12-22 19:57 ` David Miller 0 siblings, 2 replies; 10+ messages in thread From: Jiri Kosina @ 2010-12-22 10:00 UTC (permalink / raw) To: David Miller, netdev, Eric Dumazet; +Cc: linux-kernel, Vojtech Pavlik On Fri, 17 Dec 2010, Jiri Kosina wrote: > For outgoing connections, the initial value of cwnd is always set to 2 (in > tcp_v4_init_sock()) regardless of setting of RTAX_CWND. For incoming > connections, this is handled properly in tcp_init_metrics(). > > As a result of this, Linux TCP stack always uses cwnd == 2 at the beginning of > outgoing TCP session (i.e. waits for ACK after 2 packets once the connection > has been established) and grows it in accordance with slow-start algorithm > only after it receives ACK for first two packets. > > When slow-start triggers later during the connection (e.g. after idle), > cwnd is properly re-initialized to RTAX_CWND value (if specified) through > tcp_cwnd_restart() -> tcp_init_cwnd(). > > Initialize tp->snd_cwnd properly so that RTAX_CWND value is being used > also in the slow-start phase for the first packets in the connection. This should of course read RTAX_INITCWND instead of RTAX_CWND in the whole changelog, sorry. Besides that, any comments on this, please? Thanks. > > Signed-off-by: Jiri Kosina <jkosina@suse.cz> > --- > net/ipv4/tcp_output.c | 2 ++ > 1 files changed, 2 insertions(+), 0 deletions(-) > > diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c > index 61c2463..6dbc55b 100644 > --- a/net/ipv4/tcp_output.c > +++ b/net/ipv4/tcp_output.c > @@ -2572,6 +2572,8 @@ static void tcp_connect_init(struct sock *sk) > &rcv_wscale, > dst_metric(dst, RTAX_INITRWND)); > > + tp->snd_cwnd = tcp_init_cwnd(tp, dst); > + > tp->rx_opt.rcv_wscale = rcv_wscale; > tp->rcv_ssthresh = tp->rcv_wnd; > > -- > 1.7.3.1 > > -- Jiri Kosina SUSE Labs, Novell Inc. ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] tcp: use RTAX_CWND for outgoing connections properly 2010-12-22 10:00 ` Jiri Kosina @ 2010-12-22 11:29 ` Eric Dumazet 2010-12-22 11:38 ` Jiri Kosina 2010-12-22 19:57 ` David Miller 1 sibling, 1 reply; 10+ messages in thread From: Eric Dumazet @ 2010-12-22 11:29 UTC (permalink / raw) To: Jiri Kosina; +Cc: David Miller, netdev, linux-kernel, Vojtech Pavlik Le mercredi 22 décembre 2010 à 11:00 +0100, Jiri Kosina a écrit : > On Fri, 17 Dec 2010, Jiri Kosina wrote: > > > For outgoing connections, the initial value of cwnd is always set to 2 (in > > tcp_v4_init_sock()) regardless of setting of RTAX_CWND. For incoming > > connections, this is handled properly in tcp_init_metrics(). > > > > As a result of this, Linux TCP stack always uses cwnd == 2 at the beginning of > > outgoing TCP session (i.e. waits for ACK after 2 packets once the connection > > has been established) and grows it in accordance with slow-start algorithm > > only after it receives ACK for first two packets. > > > > When slow-start triggers later during the connection (e.g. after idle), > > cwnd is properly re-initialized to RTAX_CWND value (if specified) through > > tcp_cwnd_restart() -> tcp_init_cwnd(). > > > > Initialize tp->snd_cwnd properly so that RTAX_CWND value is being used > > also in the slow-start phase for the first packets in the connection. > > This should of course read RTAX_INITCWND instead of RTAX_CWND in the whole > changelog, sorry. > > Besides that, any comments on this, please? > Its a bit strange, here is what I have with two net-next-2.6 machines (without your patch) 192.168.20.0/24 dev eth1 scope link initcwnd 10 12:13:46.855786 IP 192.168.20.110.39146 > 192.168.20.108.59636: S 1615862982:1615862982(0) win 14600 <mss 1460,sackOK,timestamp 277530 0,nop,wscale 8> 12:13:46.855807 IP 192.168.20.108.59636 > 192.168.20.110.39146: S 1603053412:1603053412(0) ack 1615862983 win 14480 <mss 1460,sackOK,timestamp 744840 277530,nop,wscale 7> 12:13:46.855878 IP 192.168.20.110.39146 > 192.168.20.108.59636: . ack 1 win 58 <nop,nop,timestamp 277530 744840> 12:13:46.856779 IP 192.168.20.110.39146 > 192.168.20.108.59636: P 1:14481(14480) ack 1 win 58 <nop,nop,timestamp 277530 744840> 12:13:46.856794 IP 192.168.20.108.59636 > 192.168.20.110.39146: . ack 14481 win 136 <nop,nop,timestamp 744841 277530> 12:13:46.856901 IP 192.168.20.110.39146 > 192.168.20.108.59636: . 14481:15929(1448) ack 1 win 58 <nop,nop,timestamp 277531 744841> 12:13:46.856912 IP 192.168.20.108.59636 > 192.168.20.110.39146: . ack 15929 win 159 <nop,nop,timestamp 744841 277531> 12:13:46.856930 IP 192.168.20.110.39146 > 192.168.20.108.59636: . 15929:18825(2896) ack 1 win 58 <nop,nop,timestamp 277531 744841> We can see 192.168.20.110 sends 14480 bytes in its first frame. Are you sure your patch still needed after commit 356f039822b8d802138f (TCP: increase default initial receive window.) ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] tcp: use RTAX_CWND for outgoing connections properly 2010-12-22 11:29 ` Eric Dumazet @ 2010-12-22 11:38 ` Jiri Kosina 2010-12-22 14:07 ` Jiri Kosina 0 siblings, 1 reply; 10+ messages in thread From: Jiri Kosina @ 2010-12-22 11:38 UTC (permalink / raw) To: Eric Dumazet; +Cc: David Miller, netdev, linux-kernel, Vojtech Pavlik On Wed, 22 Dec 2010, Eric Dumazet wrote: > > > For outgoing connections, the initial value of cwnd is always set to 2 (in > > > tcp_v4_init_sock()) regardless of setting of RTAX_CWND. For incoming > > > connections, this is handled properly in tcp_init_metrics(). > > > > > > As a result of this, Linux TCP stack always uses cwnd == 2 at the beginning of > > > outgoing TCP session (i.e. waits for ACK after 2 packets once the connection > > > has been established) and grows it in accordance with slow-start algorithm > > > only after it receives ACK for first two packets. > > > > > > When slow-start triggers later during the connection (e.g. after idle), > > > cwnd is properly re-initialized to RTAX_CWND value (if specified) through > > > tcp_cwnd_restart() -> tcp_init_cwnd(). > > > > > > Initialize tp->snd_cwnd properly so that RTAX_CWND value is being used > > > also in the slow-start phase for the first packets in the connection. > > > > This should of course read RTAX_INITCWND instead of RTAX_CWND in the whole > > changelog, sorry. > > > > Besides that, any comments on this, please? > > > > Its a bit strange, here is what I have with two net-next-2.6 machines > (without your patch) > > 192.168.20.0/24 dev eth1 scope link initcwnd 10 > > 12:13:46.855786 IP 192.168.20.110.39146 > 192.168.20.108.59636: S 1615862982:1615862982(0) win 14600 <mss 1460,sackOK,timestamp 277530 0,nop,wscale 8> > 12:13:46.855807 IP 192.168.20.108.59636 > 192.168.20.110.39146: S 1603053412:1603053412(0) ack 1615862983 win 14480 <mss 1460,sackOK,timestamp 744840 277530,nop,wscale 7> > 12:13:46.855878 IP 192.168.20.110.39146 > 192.168.20.108.59636: . ack 1 win 58 <nop,nop,timestamp 277530 744840> > 12:13:46.856779 IP 192.168.20.110.39146 > 192.168.20.108.59636: P 1:14481(14480) ack 1 win 58 <nop,nop,timestamp 277530 744840> > 12:13:46.856794 IP 192.168.20.108.59636 > 192.168.20.110.39146: . ack 14481 win 136 <nop,nop,timestamp 744841 277530> > 12:13:46.856901 IP 192.168.20.110.39146 > 192.168.20.108.59636: . 14481:15929(1448) ack 1 win 58 <nop,nop,timestamp 277531 744841> > 12:13:46.856912 IP 192.168.20.108.59636 > 192.168.20.110.39146: . ack 15929 win 159 <nop,nop,timestamp 744841 277531> > 12:13:46.856930 IP 192.168.20.110.39146 > 192.168.20.108.59636: . 15929:18825(2896) ack 1 win 58 <nop,nop,timestamp 277531 744841> > > We can see 192.168.20.110 sends 14480 bytes in its first frame. So is this with 356f039822b8d802138f applied? I have had a testing environment in which I had forced the 'receiver' to advertise large receive window, but still the sender started with a very small initial congestion window, waiting for ack after 1 or 2 MSS-sized packets. I must admit that I was reproducing it with slightly older kernel, but haven't seen any changes in the code which would influence this behavior. Will look how most up-to-date kernel (plus 356f039822b8d802138f) behaves in that environment. Where in the code is the congestion window enlarged to accomodate 14480 bytes in your case? I don't see where the value of '2' coming from tcp_v4_init_sock() is being overriden for "outgoing" connections ... Thanks, -- Jiri Kosina SUSE Labs, Novell Inc. ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] tcp: use RTAX_CWND for outgoing connections properly 2010-12-22 11:38 ` Jiri Kosina @ 2010-12-22 14:07 ` Jiri Kosina 2010-12-22 14:37 ` Eric Dumazet 0 siblings, 1 reply; 10+ messages in thread From: Jiri Kosina @ 2010-12-22 14:07 UTC (permalink / raw) To: Eric Dumazet; +Cc: David Miller, netdev, linux-kernel, Vojtech Pavlik On Wed, 22 Dec 2010, Jiri Kosina wrote: > > 12:13:46.855786 IP 192.168.20.110.39146 > 192.168.20.108.59636: S 1615862982:1615862982(0) win 14600 <mss 1460,sackOK,timestamp 277530 0,nop,wscale 8> > > 12:13:46.855807 IP 192.168.20.108.59636 > 192.168.20.110.39146: S 1603053412:1603053412(0) ack 1615862983 win 14480 <mss 1460,sackOK,timestamp 744840 277530,nop,wscale 7> > > 12:13:46.855878 IP 192.168.20.110.39146 > 192.168.20.108.59636: . ack 1 win 58 <nop,nop,timestamp 277530 744840> > > 12:13:46.856779 IP 192.168.20.110.39146 > 192.168.20.108.59636: P 1:14481(14480) ack 1 win 58 <nop,nop,timestamp 277530 744840> > > 12:13:46.856794 IP 192.168.20.108.59636 > 192.168.20.110.39146: . ack 14481 win 136 <nop,nop,timestamp 744841 277530> > > 12:13:46.856901 IP 192.168.20.110.39146 > 192.168.20.108.59636: . 14481:15929(1448) ack 1 win 58 <nop,nop,timestamp 277531 744841> > > 12:13:46.856912 IP 192.168.20.108.59636 > 192.168.20.110.39146: . ack 15929 win 159 <nop,nop,timestamp 744841 277531> > > 12:13:46.856930 IP 192.168.20.110.39146 > 192.168.20.108.59636: . 15929:18825(2896) ack 1 win 58 <nop,nop,timestamp 277531 744841> > > > > We can see 192.168.20.110 sends 14480 bytes in its first frame. > > So is this with 356f039822b8d802138f applied? > > I have had a testing environment in which I had forced the 'receiver' to > advertise large receive window, but still the sender started with a very > small initial congestion window, waiting for ack after 1 or 2 MSS-sized > packets. OK, so current vanilla (not even net-next needed) seems to correctly handle initcwnd value and sends as many packets as fit into it, instead of waiting after one or two MSS-sized packets. So someone already fixed it in some other way apparently. Sorry for the noise. -- Jiri Kosina SUSE Labs, Novell Inc. ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] tcp: use RTAX_CWND for outgoing connections properly 2010-12-22 14:07 ` Jiri Kosina @ 2010-12-22 14:37 ` Eric Dumazet 0 siblings, 0 replies; 10+ messages in thread From: Eric Dumazet @ 2010-12-22 14:37 UTC (permalink / raw) To: Jiri Kosina; +Cc: David Miller, netdev, linux-kernel, Vojtech Pavlik Le mercredi 22 décembre 2010 à 15:07 +0100, Jiri Kosina a écrit : > > OK, so current vanilla (not even net-next needed) seems to correctly > handle initcwnd value and sends as many packets as fit into it, instead of > waiting after one or two MSS-sized packets. So someone already fixed it > in some other way apparently. Sorry for the noise. > Indeed ;) net-next was necessary at least for the other side, so that its SYNACK packet advertizes a large enough window ;) ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] tcp: use RTAX_CWND for outgoing connections properly 2010-12-22 10:00 ` Jiri Kosina 2010-12-22 11:29 ` Eric Dumazet @ 2010-12-22 19:57 ` David Miller 2010-12-22 23:26 ` Jiri Kosina 1 sibling, 1 reply; 10+ messages in thread From: David Miller @ 2010-12-22 19:57 UTC (permalink / raw) To: jkosina; +Cc: netdev, eric.dumazet, linux-kernel, vojtech From: Jiri Kosina <jkosina@suse.cz> Date: Wed, 22 Dec 2010 11:00:39 +0100 (CET) > Besides that, any comments on this, please? It's in my backlog, patience is a virtue. ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] tcp: use RTAX_CWND for outgoing connections properly 2010-12-22 19:57 ` David Miller @ 2010-12-22 23:26 ` Jiri Kosina 2010-12-22 23:27 ` David Miller 0 siblings, 1 reply; 10+ messages in thread From: Jiri Kosina @ 2010-12-22 23:26 UTC (permalink / raw) To: David Miller; +Cc: netdev, eric.dumazet, linux-kernel, Vojtech Pavlik On Wed, 22 Dec 2010, David Miller wrote: > > Besides that, any comments on this, please? > > It's in my backlog, patience is a virtue. Please just disregard it, it's not needed with newer kernels. Thanks, -- Jiri Kosina SUSE Labs, Novell Inc. ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] tcp: use RTAX_CWND for outgoing connections properly 2010-12-22 23:26 ` Jiri Kosina @ 2010-12-22 23:27 ` David Miller 2010-12-22 23:33 ` Jiri Kosina 0 siblings, 1 reply; 10+ messages in thread From: David Miller @ 2010-12-22 23:27 UTC (permalink / raw) To: jkosina; +Cc: netdev, eric.dumazet, linux-kernel, vojtech From: Jiri Kosina <jkosina@suse.cz> Date: Thu, 23 Dec 2010 00:26:21 +0100 (CET) > On Wed, 22 Dec 2010, David Miller wrote: > >> > Besides that, any comments on this, please? >> >> It's in my backlog, patience is a virtue. > > Please just disregard it, it's not needed with newer kernels. Thanks, Was it indeed fixed by commit 86bcebafc5e7f5? ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] tcp: use RTAX_CWND for outgoing connections properly 2010-12-22 23:27 ` David Miller @ 2010-12-22 23:33 ` Jiri Kosina 0 siblings, 0 replies; 10+ messages in thread From: Jiri Kosina @ 2010-12-22 23:33 UTC (permalink / raw) To: David Miller; +Cc: netdev, eric.dumazet, linux-kernel, vojtech On Wed, 22 Dec 2010, David Miller wrote: > >> > Besides that, any comments on this, please? > >> > >> It's in my backlog, patience is a virtue. > > > > Please just disregard it, it's not needed with newer kernels. Thanks, > > Was it indeed fixed by commit 86bcebafc5e7f5? Yes, exactly. It just wasn't obvious enough when looking at the code (hence my followup patch which hopefully makes the code more readable on a first sight). Thanks and sorry for the noise. -- Jiri Kosina SUSE Labs, Novell Inc. ^ permalink raw reply [flat|nested] 10+ messages in thread
end of thread, other threads:[~2010-12-22 23:33 UTC | newest] Thread overview: 10+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2010-12-17 22:14 [PATCH] tcp: use RTAX_CWND for outgoing connections properly Jiri Kosina 2010-12-22 10:00 ` Jiri Kosina 2010-12-22 11:29 ` Eric Dumazet 2010-12-22 11:38 ` Jiri Kosina 2010-12-22 14:07 ` Jiri Kosina 2010-12-22 14:37 ` Eric Dumazet 2010-12-22 19:57 ` David Miller 2010-12-22 23:26 ` Jiri Kosina 2010-12-22 23:27 ` David Miller 2010-12-22 23:33 ` Jiri Kosina
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox