d/freebsd-nq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
freebsd-nq/sys/netinet/cc/cc_newreno.c

660 lines
21 KiB
C
Raw Normal View History

This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
/*-
sys: general adoption of SPDX licensing ID tags. Mainly focus on files that use BSD 2-Clause license, however the tool I was using misidentified many licenses so this was mostly a manual - error prone - task. The Software Package Data Exchange (SPDX) group provides a specification to make it easier for automated tools to detect and summarize well known opensource licenses. We are gradually adopting the specification, noting that the tags are considered only advisory and do not, in any way, superceed or replace the license texts. No functional change intended.
2017-11-27 15:23:17 +00:00
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
* Copyright (c) 1982, 1986, 1988, 1990, 1993, 1994, 1995
* The Regents of the University of California.
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
* Copyright (c) 2007-2008,2010,2014
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
* Swinburne University of Technology, Melbourne, Australia.
* Copyright (c) 2009-2010 Lawrence Stewart <lstewart@freebsd.org>
* Copyright (c) 2010 The FreeBSD Foundation
* All rights reserved.
*
* This software was developed at the Centre for Advanced Internet
Use the full and proper company name for Swinburne University of Technology throughout the source tree. Requested by: Grenville Armitage, Director of CAIA at Swinburne University of Technology MFC after: 3 days
2011-04-12 08:13:18 +00:00
* Architectures, Swinburne University of Technology, by Lawrence Stewart, James
* Healy and David Hayes, made possible in part by a grant from the Cisco
* University Research Program Fund at Community Foundation Silicon Valley.
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
*
* Portions of this software were developed at the Centre for Advanced
* Internet Architectures, Swinburne University of Technology, Melbourne,
* Australia by David Hayes under sponsorship from the FreeBSD Foundation.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
/*
* This software was first released in 2007 by James Healy and Lawrence Stewart
Use the full and proper company name for Swinburne University of Technology throughout the source tree. Requested by: Grenville Armitage, Director of CAIA at Swinburne University of Technology MFC after: 3 days
2011-04-12 08:13:18 +00:00
* whilst working on the NewTCP research project at Swinburne University of
* Technology's Centre for Advanced Internet Architectures, Melbourne,
* Australia, which was made possible in part by a grant from the Cisco
* University Research Program Fund at Community Foundation Silicon Valley.
* More details are available at:
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
* http://caia.swin.edu.au/urp/newtcp/
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
*
* Dec 2014 garmitage@swin.edu.au
* Borrowed code fragments from cc_cdg.c to add modifiable beta
* via sysctls.
*
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/kernel.h>
General cleanup of the NewReno CC module (no functional changes): - Remove superfluous includes and unhelpful comments. - Alphabetically order functions. - Make functions static. Sponsored by: FreeBSD Foundation MFC after: 9 weeks X-MFC with: r215166
2010-12-02 02:32:46 +00:00
#include <sys/malloc.h>
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
#include <sys/module.h>
#include <sys/socket.h>
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
#include <sys/lock.h>
#include <sys/mutex.h>
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
#include <sys/socketvar.h>
#include <sys/sysctl.h>
General cleanup of the NewReno CC module (no functional changes): - Remove superfluous includes and unhelpful comments. - Alphabetically order functions. - Make functions static. Sponsored by: FreeBSD Foundation MFC after: 9 weeks X-MFC with: r215166
2010-12-02 02:32:46 +00:00
#include <sys/systm.h>
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
General cleanup of the NewReno CC module (no functional changes): - Remove superfluous includes and unhelpful comments. - Alphabetically order functions. - Make functions static. Sponsored by: FreeBSD Foundation MFC after: 9 weeks X-MFC with: r215166
2010-12-02 02:32:46 +00:00
#include <net/vnet.h>
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
#include <net/route.h>
#include <net/route/nhop.h>
#include <netinet/in_pcb.h>
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
#include <netinet/in.h>
#include <netinet/in_pcb.h>
- Rename cc.h to more meaningful tcp_cc.h. - Declare it a kernel only include, which it already is. - Don't include tcp.h implicitly from tcp_cc.h
2016-01-21 22:34:51 +00:00
#include <netinet/tcp.h>
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
#include <netinet/tcp_seq.h>
#include <netinet/tcp_var.h>
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
#include <netinet/tcp_log_buf.h>
#include <netinet/tcp_hpts.h>
Rename netinet/tcp_cc.h to netinet/cc/cc.h. Discussed with: lstewart
2016-01-27 17:59:39 +00:00
#include <netinet/cc/cc.h>
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
#include <netinet/cc/cc_module.h>
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
#include <netinet/cc/cc_newreno.h>
Plug a memory leak and potential NULL-pointer dereference introduced in r331214. Each TCP connection that uses the system default cc_newreno(4) congestion control algorithm module leaks a "struct newreno" (8 bytes of memory) at connection initialisation time. The NULL-pointer dereference is only germane when using the ABE feature, which is disabled by default. While at it: - Defer the allocation of memory until it is actually needed given that ABE is optional and disabled by default. - Document the ENOMEM errno in getsockopt(2)/setsockopt(2). - Document ENOMEM and ENOBUFS in tcp(4) as being synonymous given that they are used interchangeably throughout the code. - Fix a few other nits also accidentally omitted from the original patch. Reported by: Harsh Jain on freebsd-net@ Tested by: tjh@ Differential Revision: https://reviews.freebsd.org/D15358
2018-05-17 02:46:27 +00:00
static void newreno_cb_destroy(struct cc_var *ccv);
General cleanup of the NewReno CC module (no functional changes): - Remove superfluous includes and unhelpful comments. - Alphabetically order functions. - Make functions static. Sponsored by: FreeBSD Foundation MFC after: 9 weeks X-MFC with: r215166
2010-12-02 02:32:46 +00:00
static void newreno_ack_received(struct cc_var *ccv, uint16_t type);
static void newreno_after_idle(struct cc_var *ccv);
static void newreno_cong_signal(struct cc_var *ccv, uint32_t type);
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
static int newreno_ctl_output(struct cc_var *ccv, struct sockopt *sopt, void *buf);
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
static void newreno_newround(struct cc_var *ccv, uint32_t round_cnt);
static void newreno_rttsample(struct cc_var *ccv, uint32_t usec_rtt, uint32_t rxtcnt, uint32_t fas);
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
static int newreno_cb_init(struct cc_var *ccv, void *);
static size_t newreno_data_sz(void);
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
VNET_DECLARE(uint32_t, newreno_beta);
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
#define V_newreno_beta VNET(newreno_beta)
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
VNET_DEFINE(uint32_t, newreno_beta_ecn) = 80;
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
#define V_newreno_beta_ecn VNET(newreno_beta_ecn)
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
struct cc_algo newreno_cc_algo = {
.name = "newreno",
Plug a memory leak and potential NULL-pointer dereference introduced in r331214. Each TCP connection that uses the system default cc_newreno(4) congestion control algorithm module leaks a "struct newreno" (8 bytes of memory) at connection initialisation time. The NULL-pointer dereference is only germane when using the ABE feature, which is disabled by default. While at it: - Defer the allocation of memory until it is actually needed given that ABE is optional and disabled by default. - Document the ENOMEM errno in getsockopt(2)/setsockopt(2). - Document ENOMEM and ENOBUFS in tcp(4) as being synonymous given that they are used interchangeably throughout the code. - Fix a few other nits also accidentally omitted from the original patch. Reported by: Harsh Jain on freebsd-net@ Tested by: tjh@ Differential Revision: https://reviews.freebsd.org/D15358
2018-05-17 02:46:27 +00:00
.cb_destroy = newreno_cb_destroy,
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
.ack_received = newreno_ack_received,
General cleanup of the NewReno CC module (no functional changes): - Remove superfluous includes and unhelpful comments. - Alphabetically order functions. - Make functions static. Sponsored by: FreeBSD Foundation MFC after: 9 weeks X-MFC with: r215166
2010-12-02 02:32:46 +00:00
.after_idle = newreno_after_idle,
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
.cong_signal = newreno_cong_signal,
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
.post_recovery = newreno_cc_post_recovery,
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
.ctl_output = newreno_ctl_output,
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
.newround = newreno_newround,
.rttsample = newreno_rttsample,
.cb_init = newreno_cb_init,
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
.cc_data_sz = newreno_data_sz,
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
};
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
static uint32_t hystart_lowcwnd = 16;
static uint32_t hystart_minrtt_thresh = 4000;
static uint32_t hystart_maxrtt_thresh = 16000;
static uint32_t hystart_n_rttsamples = 8;
static uint32_t hystart_css_growth_div = 4;
static uint32_t hystart_css_rounds = 5;
static uint32_t hystart_bblogs = 0;
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
static void
newreno_log_hystart_event(struct cc_var *ccv, struct newreno *nreno, uint8_t mod, uint32_t flex1)
{
/*
* Types of logs (mod value)
* 1 - rtt_thresh in flex1, checking to see if RTT is to great.
* 2 - rtt is too great, rtt_thresh in flex1.
* 3 - CSS is active incr in flex1
* 4 - A new round is beginning flex1 is round count
* 5 - A new RTT measurement flex1 is the new measurement.
* 6 - We enter CA ssthresh is also in flex1.
* 7 - Socket option to change hystart executed opt.val in flex1.
* 8 - Back out of CSS into SS, flex1 is the css_baseline_minrtt
*/
struct tcpcb *tp;
if (hystart_bblogs == 0)
return;
tp = ccv->ccvc.tcp;
if (tp->t_logstate != TCP_LOG_STATE_OFF) {
union tcp_log_stackspecific log;
struct timeval tv;
memset(&log, 0, sizeof(log));
log.u_bbr.flex1 = flex1;
log.u_bbr.flex2 = nreno->css_current_round_minrtt;
log.u_bbr.flex3 = nreno->css_lastround_minrtt;
log.u_bbr.flex4 = nreno->css_rttsample_count;
log.u_bbr.flex5 = nreno->css_entered_at_round;
log.u_bbr.flex6 = nreno->css_baseline_minrtt;
/* We only need bottom 16 bits of flags */
log.u_bbr.flex7 = nreno->newreno_flags & 0x0000ffff;
log.u_bbr.flex8 = mod;
log.u_bbr.epoch = nreno->css_current_round;
log.u_bbr.timeStamp = tcp_get_usecs(&tv);
log.u_bbr.lt_epoch = nreno->css_fas_at_css_entry;
log.u_bbr.pkts_out = nreno->css_last_fas;
log.u_bbr.delivered = nreno->css_lowrtt_fas;
TCP_LOG_EVENTP(tp, NULL,
&tp->t_inpcb->inp_socket->so_rcv,
&tp->t_inpcb->inp_socket->so_snd,
TCP_HYSTART, 0,
0, &log, false, &tv);
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
}
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
}
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
static size_t
newreno_data_sz(void)
{
return (sizeof(struct newreno));
}
tcp: allow new reno functions to be called from other CC modules Some new reno functions use the internal data, but are also called from functions of other CC modules. Ensure that in this case, the internal data is not accessed. Reported by: syzbot+1d219ea351caa5109d4b@syzkaller.appspotmail.com Reported by: syzbot+b08144f8cad9c67258c5@syzkaller.appspotmail.com Reviewed by: rrs Sponsored by: Netflix, Inc. Differential Revision: https://reviews.freebsd.org/D32649
2021-10-25 22:48:36 +02:00
static int
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
newreno_cb_init(struct cc_var *ccv, void *ptr)
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
{
struct newreno *nreno;
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
INP_WLOCK_ASSERT(ccv->ccvc.tcp->t_inpcb);
if (ptr == NULL) {
ccv->cc_data = malloc(sizeof(struct newreno), M_CC_MEM, M_NOWAIT);
if (ccv->cc_data == NULL)
return (ENOMEM);
} else
ccv->cc_data = ptr;
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
nreno = (struct newreno *)ccv->cc_data;
/* NB: nreno is not zeroed, so initialise all fields. */
nreno->beta = V_newreno_beta;
nreno->beta_ecn = V_newreno_beta_ecn;
/*
* We set the enabled flag so that if
* the socket option gets strobed and
* we have not hit a loss
*/
nreno->newreno_flags = CC_NEWRENO_HYSTART_ENABLED;
/* At init set both to infinity */
nreno->css_lastround_minrtt = 0xffffffff;
nreno->css_current_round_minrtt = 0xffffffff;
nreno->css_current_round = 0;
nreno->css_baseline_minrtt = 0xffffffff;
nreno->css_rttsample_count = 0;
nreno->css_entered_at_round = 0;
nreno->css_fas_at_css_entry = 0;
nreno->css_lowrtt_fas = 0;
nreno->css_last_fas = 0;
return (0);
Plug a memory leak and potential NULL-pointer dereference introduced in r331214. Each TCP connection that uses the system default cc_newreno(4) congestion control algorithm module leaks a "struct newreno" (8 bytes of memory) at connection initialisation time. The NULL-pointer dereference is only germane when using the ABE feature, which is disabled by default. While at it: - Defer the allocation of memory until it is actually needed given that ABE is optional and disabled by default. - Document the ENOMEM errno in getsockopt(2)/setsockopt(2). - Document ENOMEM and ENOBUFS in tcp(4) as being synonymous given that they are used interchangeably throughout the code. - Fix a few other nits also accidentally omitted from the original patch. Reported by: Harsh Jain on freebsd-net@ Tested by: tjh@ Differential Revision: https://reviews.freebsd.org/D15358
2018-05-17 02:46:27 +00:00
}
static void
newreno_cb_destroy(struct cc_var *ccv)
{
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
free(ccv->cc_data, M_CC_MEM);
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
}
General cleanup of the NewReno CC module (no functional changes): - Remove superfluous includes and unhelpful comments. - Alphabetically order functions. - Make functions static. Sponsored by: FreeBSD Foundation MFC after: 9 weeks X-MFC with: r215166
2010-12-02 02:32:46 +00:00
static void
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
newreno_ack_received(struct cc_var *ccv, uint16_t type)
{
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
struct newreno *nreno;
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
nreno = ccv->cc_data;
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
if (type == CC_ACK && !IN_RECOVERY(CCV(ccv, t_flags)) &&
(ccv->flags & CCF_CWND_LIMITED)) {
u_int cw = CCV(ccv, snd_cwnd);
u_int incr = CCV(ccv, t_maxseg);
/*
* Regular in-order ACK, open the congestion window.
* Method depends on which congestion control state we're
* in (slow start or cong avoid) and if ABC (RFC 3465) is
* enabled.
*
* slow start: cwnd <= ssthresh
* cong avoid: cwnd > ssthresh
*
* slow start and ABC (RFC 3465):
* Grow cwnd exponentially by the amount of data
* ACKed capping the max increment per ACK to
* (abc_l_var * maxseg) bytes.
*
* slow start without ABC (RFC 5681):
* Grow cwnd exponentially by maxseg per ACK.
*
* cong avoid and ABC (RFC 3465):
* Grow cwnd linearly by maxseg per RTT for each
* cwnd worth of ACKed data.
*
* cong avoid without ABC (RFC 5681):
* Grow cwnd linearly by approximately maxseg per RTT using
* maxseg^2 / cwnd per ACK as the increment.
* If cwnd > maxseg^2, fix the cwnd increment at 1 byte to
* avoid capping cwnd.
*/
if (cw > CCV(ccv, snd_ssthresh)) {
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
if (nreno->newreno_flags & CC_NEWRENO_HYSTART_IN_CSS) {
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
/*
* We have slipped into CA with
* CSS active. Deactivate all.
*/
/* Turn off the CSS flag */
nreno->newreno_flags &= ~CC_NEWRENO_HYSTART_IN_CSS;
/* Disable use of CSS in the future except long idle */
nreno->newreno_flags &= ~CC_NEWRENO_HYSTART_ENABLED;
}
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
if (V_tcp_do_rfc3465) {
if (ccv->flags & CCF_ABC_SENTAWND)
ccv->flags &= ~CCF_ABC_SENTAWND;
else
incr = 0;
} else
incr = max((incr * incr / cw), 1);
} else if (V_tcp_do_rfc3465) {
/*
* In slow-start with ABC enabled and no RTO in sight?
* (Must not use abc_l_var > 1 if slow starting after
* an RTO. On RTO, snd_nxt = snd_una, so the
* snd_nxt == snd_max check is sufficient to
* handle this).
*
* XXXLAS: Find a way to signal SS after RTO that
* doesn't rely on tcpcb vars.
*/
This brings into sync FreeBSD with the netflix versions of rack and bbr. This fixes several breakages (panics) since the tcp_lro code was committed that have been reported. Quite a few new features are now in rack (prefecting of DGP -- Dynamic Goodput Pacing among the largest). There is also support for ack-war prevention. Documents comming soon on rack.. Sponsored by: Netflix Reviewed by: rscheff, mtuexen Differential Revision: https://reviews.freebsd.org/D30036
2021-05-06 11:22:26 -04:00
uint16_t abc_val;
if (ccv->flags & CCF_USE_LOCAL_ABC)
abc_val = ccv->labc;
else
abc_val = V_tcp_abc_l_var;
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
if ((nreno->newreno_flags & CC_NEWRENO_HYSTART_ALLOWED) &&
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
(nreno->newreno_flags & CC_NEWRENO_HYSTART_ENABLED) &&
((nreno->newreno_flags & CC_NEWRENO_HYSTART_IN_CSS) == 0)) {
/*
* Hystart is allowed and still enabled and we are not yet
* in CSS. Lets check to see if we can make a decision on
* if we need to go into CSS.
*/
if ((nreno->css_rttsample_count >= hystart_n_rttsamples) &&
(CCV(ccv, snd_cwnd) >
(hystart_lowcwnd * tcp_fixed_maxseg(ccv->ccvc.tcp)))) {
uint32_t rtt_thresh;
/* Clamp (minrtt_thresh, lastround/8, maxrtt_thresh) */
rtt_thresh = (nreno->css_lastround_minrtt >> 3);
if (rtt_thresh < hystart_minrtt_thresh)
rtt_thresh = hystart_minrtt_thresh;
if (rtt_thresh > hystart_maxrtt_thresh)
rtt_thresh = hystart_maxrtt_thresh;
newreno_log_hystart_event(ccv, nreno, 1, rtt_thresh);
if (nreno->css_current_round_minrtt >= (nreno->css_lastround_minrtt + rtt_thresh)) {
/* Enter CSS */
nreno->newreno_flags |= CC_NEWRENO_HYSTART_IN_CSS;
nreno->css_fas_at_css_entry = nreno->css_lowrtt_fas;
nreno->css_baseline_minrtt = nreno->css_current_round_minrtt;
nreno->css_entered_at_round = nreno->css_current_round;
newreno_log_hystart_event(ccv, nreno, 2, rtt_thresh);
}
}
}
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
if (CCV(ccv, snd_nxt) == CCV(ccv, snd_max))
incr = min(ccv->bytes_this_ack,
This brings into sync FreeBSD with the netflix versions of rack and bbr. This fixes several breakages (panics) since the tcp_lro code was committed that have been reported. Quite a few new features are now in rack (prefecting of DGP -- Dynamic Goodput Pacing among the largest). There is also support for ack-war prevention. Documents comming soon on rack.. Sponsored by: Netflix Reviewed by: rscheff, mtuexen Differential Revision: https://reviews.freebsd.org/D30036
2021-05-06 11:22:26 -04:00
ccv->nsegs * abc_val *
Pass the number of segments coalesced by LRO up the stack by repurposing the tso_segsz pkthdr field during RX processing, and use the information in TCP for more correct accounting and as a congestion control input. This is only a start, and an audit of other uses for the data is left as future work. Reviewed by: gallatin, rrs Sponsored by: Netflix, Inc. Differential Revision: https://reviews.freebsd.org/D7564
2016-08-25 13:33:32 +00:00
CCV(ccv, t_maxseg));
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
else
incr = min(ccv->bytes_this_ack, CCV(ccv, t_maxseg));
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
/* Only if Hystart is enabled will the flag get set */
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
if (nreno->newreno_flags & CC_NEWRENO_HYSTART_IN_CSS) {
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
incr /= hystart_css_growth_div;
newreno_log_hystart_event(ccv, nreno, 3, incr);
}
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
}
/* ABC is on by default, so incr equals 0 frequently. */
if (incr > 0)
CCV(ccv, snd_cwnd) = min(cw + incr,
TCP_MAXWIN << CCV(ccv, snd_scale));
}
}
General cleanup of the NewReno CC module (no functional changes): - Remove superfluous includes and unhelpful comments. - Alphabetically order functions. - Make functions static. Sponsored by: FreeBSD Foundation MFC after: 9 weeks X-MFC with: r215166
2010-12-02 02:32:46 +00:00
static void
newreno_after_idle(struct cc_var *ccv)
{
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
struct newreno *nreno;
TCP newreno: improve after_idle ssthresh Adjust ssthresh in after_idle to the maximum of the prior ssthresh, or 3/4 of the prior cwnd. See RFC2861 section 2 for an in depth explanation for the rationale around this. As newreno is the default "fall-through" reaction, most tcp variants will benefit from this. Reviewed by: tuexen MFC after: 2 weeks Sponsored by: NetApp, Inc. Differential Revision: https://reviews.freebsd.org/D22438
2020-09-25 10:23:14 +00:00
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
nreno = ccv->cc_data;
newreno_cc_after_idle(ccv);
if ((nreno->newreno_flags & CC_NEWRENO_HYSTART_ENABLED) == 0) {
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
if (CCV(ccv, snd_cwnd) <= (hystart_lowcwnd * tcp_fixed_maxseg(ccv->ccvc.tcp))) {
/*
* Re-enable hystart if our cwnd has fallen below
* the hystart lowcwnd point.
*/
nreno->newreno_flags &= ~CC_NEWRENO_HYSTART_IN_CSS;
nreno->newreno_flags |= CC_NEWRENO_HYSTART_ENABLED;
}
}
General cleanup of the NewReno CC module (no functional changes): - Remove superfluous includes and unhelpful comments. - Alphabetically order functions. - Make functions static. Sponsored by: FreeBSD Foundation MFC after: 9 weeks X-MFC with: r215166
2010-12-02 02:32:46 +00:00
}
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
/*
General cleanup of the NewReno CC module (no functional changes): - Remove superfluous includes and unhelpful comments. - Alphabetically order functions. - Make functions static. Sponsored by: FreeBSD Foundation MFC after: 9 weeks X-MFC with: r215166
2010-12-02 02:32:46 +00:00
* Perform any necessary tasks before we enter congestion recovery.
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
*/
General cleanup of the NewReno CC module (no functional changes): - Remove superfluous includes and unhelpful comments. - Alphabetically order functions. - Make functions static. Sponsored by: FreeBSD Foundation MFC after: 9 weeks X-MFC with: r215166
2010-12-02 02:32:46 +00:00
static void
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
newreno_cong_signal(struct cc_var *ccv, uint32_t type)
{
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
struct newreno *nreno;
Plug a memory leak and potential NULL-pointer dereference introduced in r331214. Each TCP connection that uses the system default cc_newreno(4) congestion control algorithm module leaks a "struct newreno" (8 bytes of memory) at connection initialisation time. The NULL-pointer dereference is only germane when using the ABE feature, which is disabled by default. While at it: - Defer the allocation of memory until it is actually needed given that ABE is optional and disabled by default. - Document the ENOMEM errno in getsockopt(2)/setsockopt(2). - Document ENOMEM and ENOBUFS in tcp(4) as being synonymous given that they are used interchangeably throughout the code. - Fix a few other nits also accidentally omitted from the original patch. Reported by: Harsh Jain on freebsd-net@ Tested by: tjh@ Differential Revision: https://reviews.freebsd.org/D15358
2018-05-17 02:46:27 +00:00
uint32_t beta, beta_ecn, cwin, factor;
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
u_int mss;
cwin = CCV(ccv, snd_cwnd);
This brings into sync FreeBSD with the netflix versions of rack and bbr. This fixes several breakages (panics) since the tcp_lro code was committed that have been reported. Quite a few new features are now in rack (prefecting of DGP -- Dynamic Goodput Pacing among the largest). There is also support for ack-war prevention. Documents comming soon on rack.. Sponsored by: Netflix Reviewed by: rscheff, mtuexen Differential Revision: https://reviews.freebsd.org/D30036
2021-05-06 11:22:26 -04:00
mss = tcp_fixed_maxseg(ccv->ccvc.tcp);
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
nreno = ccv->cc_data;
tcp: allow new reno functions to be called from other CC modules Some new reno functions use the internal data, but are also called from functions of other CC modules. Ensure that in this case, the internal data is not accessed. Reported by: syzbot+1d219ea351caa5109d4b@syzkaller.appspotmail.com Reported by: syzbot+b08144f8cad9c67258c5@syzkaller.appspotmail.com Reviewed by: rrs Sponsored by: Netflix, Inc. Differential Revision: https://reviews.freebsd.org/D32649
2021-10-25 22:48:36 +02:00
beta = (nreno == NULL) ? V_newreno_beta : nreno->beta;;
beta_ecn = (nreno == NULL) ? V_newreno_beta_ecn : nreno->beta_ecn;
This brings into sync FreeBSD with the netflix versions of rack and bbr. This fixes several breakages (panics) since the tcp_lro code was committed that have been reported. Quite a few new features are now in rack (prefecting of DGP -- Dynamic Goodput Pacing among the largest). There is also support for ack-war prevention. Documents comming soon on rack.. Sponsored by: Netflix Reviewed by: rscheff, mtuexen Differential Revision: https://reviews.freebsd.org/D30036
2021-05-06 11:22:26 -04:00
/*
* Note that we only change the backoff for ECN if the
* global sysctl V_cc_do_abe is set <or> the stack itself
* has set a flag in our newreno_flags (due to pacing) telling
* us to use the lower valued back-off.
*/
tcp: fix two bugs in new reno * Completely initialise the CC module specific data * Use beta_ecn in case of an ECN event whenever ABE is enabled or it is requested by the stack. Reviewed by: rscheff, rrs MFC after: 3 days Sponsored by: Netflix, Inc.
2021-06-11 15:37:39 +02:00
if ((type == CC_ECN) &&
(V_cc_do_abe ||
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
((nreno != NULL) && (nreno->newreno_flags & CC_NEWRENO_BETA_ECN_ENABLED))))
Plug a memory leak and potential NULL-pointer dereference introduced in r331214. Each TCP connection that uses the system default cc_newreno(4) congestion control algorithm module leaks a "struct newreno" (8 bytes of memory) at connection initialisation time. The NULL-pointer dereference is only germane when using the ABE feature, which is disabled by default. While at it: - Defer the allocation of memory until it is actually needed given that ABE is optional and disabled by default. - Document the ENOMEM errno in getsockopt(2)/setsockopt(2). - Document ENOMEM and ENOBUFS in tcp(4) as being synonymous given that they are used interchangeably throughout the code. - Fix a few other nits also accidentally omitted from the original patch. Reported by: Harsh Jain on freebsd-net@ Tested by: tjh@ Differential Revision: https://reviews.freebsd.org/D15358
2018-05-17 02:46:27 +00:00
factor = beta_ecn;
else
factor = beta;
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
Algorithm modules can define their own private congestion signal types in the top 8 bits of the 32 bit signal bit field space for internal use. These private signals should not be leaked outside of a module. Given that many algorithm modules use the NewReno hook functions to simplify their implementation, the obvious place such a leak would show up is in the NewReno cong_signal hook function. - Show the full number of significant bits in the signal type definitions in <netinet/cc.h>. - Add a bitmask to simplify figuring out if a given signal is in the private or public bit range. - Add a sanity check in newreno_cong_signal() to ensure private signals are not being leaked into the hook function. Sponsored by: FreeBSD Foundation Discussed with: David Hayes <dahayes at swin edu au> MFC after: 1 week X-MFC with: r215166
2011-02-01 13:32:27 +00:00
/* Catch algos which mistakenly leak private signal types. */
KASSERT((type & CC_SIGPRIVMASK) == 0,
("%s: congestion signal type 0x%08x is private\n", __func__, type));
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
cwin = max(((uint64_t)cwin * (uint64_t)factor) / (100ULL * (uint64_t)mss),
2) * mss;
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
switch (type) {
case CC_NDUPACK:
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
if (nreno->newreno_flags & CC_NEWRENO_HYSTART_ENABLED) {
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
/* Make sure the flags are all off we had a loss */
nreno->newreno_flags &= ~CC_NEWRENO_HYSTART_ENABLED;
nreno->newreno_flags &= ~CC_NEWRENO_HYSTART_IN_CSS;
}
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
if (!IN_FASTRECOVERY(CCV(ccv, t_flags))) {
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
if (IN_CONGRECOVERY(CCV(ccv, t_flags) &&
V_cc_do_abe && V_cc_abe_frlossreduce)) {
CCV(ccv, snd_ssthresh) =
((uint64_t)CCV(ccv, snd_ssthresh) *
This brings into sync FreeBSD with the netflix versions of rack and bbr. This fixes several breakages (panics) since the tcp_lro code was committed that have been reported. Quite a few new features are now in rack (prefecting of DGP -- Dynamic Goodput Pacing among the largest). There is also support for ack-war prevention. Documents comming soon on rack.. Sponsored by: Netflix Reviewed by: rscheff, mtuexen Differential Revision: https://reviews.freebsd.org/D30036
2021-05-06 11:22:26 -04:00
(uint64_t)beta) / (uint64_t)beta_ecn;
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
}
Revert r307901 - Inform CC modules about loss events. This was discussed between various transport@ members and it was requested to be reverted and discussed. Submitted by: Kevin Bowling <kevin.bowling@kev009.com> Reported by: lawrence Reviewed by: hiren Sponsored by: Limelight Networks
2017-07-25 15:08:52 +00:00
if (!IN_CONGRECOVERY(CCV(ccv, t_flags)))
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
CCV(ccv, snd_ssthresh) = cwin;
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
ENTER_RECOVERY(CCV(ccv, t_flags));
}
break;
case CC_ECN:
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
if (nreno->newreno_flags & CC_NEWRENO_HYSTART_ENABLED) {
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
/* Make sure the flags are all off we had a loss */
nreno->newreno_flags &= ~CC_NEWRENO_HYSTART_ENABLED;
nreno->newreno_flags &= ~CC_NEWRENO_HYSTART_IN_CSS;
}
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
if (!IN_CONGRECOVERY(CCV(ccv, t_flags))) {
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
CCV(ccv, snd_ssthresh) = cwin;
CCV(ccv, snd_cwnd) = cwin;
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
ENTER_CONGRECOVERY(CCV(ccv, t_flags));
}
break;
tcp: move cwnd and ssthresh updates into cc modules This will pave the way of setting ssthresh differently in TCP CUBIC, according to RFC8312 section 4.7. No functional change, only code movement. Submitted by: chengc_netapp.com Reviewed by: rrs, tuexen, rscheff MFC after: 2 weeks Sponsored by: NetApp, Inc. Differential Revision: https://reviews.freebsd.org/D26807
2020-10-24 16:09:18 +00:00
case CC_RTO:
CCV(ccv, snd_ssthresh) = max(min(CCV(ccv, snd_wnd),
CCV(ccv, snd_cwnd)) / 2 / mss,
2) * mss;
CCV(ccv, snd_cwnd) = mss;
break;
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
}
}
Plug a memory leak and potential NULL-pointer dereference introduced in r331214. Each TCP connection that uses the system default cc_newreno(4) congestion control algorithm module leaks a "struct newreno" (8 bytes of memory) at connection initialisation time. The NULL-pointer dereference is only germane when using the ABE feature, which is disabled by default. While at it: - Defer the allocation of memory until it is actually needed given that ABE is optional and disabled by default. - Document the ENOMEM errno in getsockopt(2)/setsockopt(2). - Document ENOMEM and ENOBUFS in tcp(4) as being synonymous given that they are used interchangeably throughout the code. - Fix a few other nits also accidentally omitted from the original patch. Reported by: Harsh Jain on freebsd-net@ Tested by: tjh@ Differential Revision: https://reviews.freebsd.org/D15358
2018-05-17 02:46:27 +00:00
static int
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
newreno_ctl_output(struct cc_var *ccv, struct sockopt *sopt, void *buf)
{
struct newreno *nreno;
struct cc_newreno_opts *opt;
if (sopt->sopt_valsize != sizeof(struct cc_newreno_opts))
return (EMSGSIZE);
tcp: Use local CC data only in the correct context Most CC algos do use local data, and when calling newreno_cong_signal from there, the latter misinterprets the data as its own struct, leading to incorrect behavior. Reported by: chengc_netapp.com Reviewed By: chengc_netapp.com, tuexen, #transport MFC after: 3 days Sponsored By: NetApp, Inc. Differential Revision: https://reviews.freebsd.org/D30470
2021-05-26 19:45:06 +02:00
if (CC_ALGO(ccv->ccvc.tcp) != &newreno_cc_algo)
return (ENOPROTOOPT);
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
nreno = (struct newreno *)ccv->cc_data;
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
opt = buf;
switch (sopt->sopt_dir) {
case SOPT_SET:
switch (opt->name) {
case CC_NEWRENO_BETA:
nreno->beta = opt->val;
break;
case CC_NEWRENO_BETA_ECN:
This brings into sync FreeBSD with the netflix versions of rack and bbr. This fixes several breakages (panics) since the tcp_lro code was committed that have been reported. Quite a few new features are now in rack (prefecting of DGP -- Dynamic Goodput Pacing among the largest). There is also support for ack-war prevention. Documents comming soon on rack.. Sponsored by: Netflix Reviewed by: rscheff, mtuexen Differential Revision: https://reviews.freebsd.org/D30036
2021-05-06 11:22:26 -04:00
if ((!V_cc_do_abe) && ((nreno->newreno_flags & CC_NEWRENO_BETA_ECN) == 0))
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
return (EACCES);
nreno->beta_ecn = opt->val;
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
nreno->newreno_flags |= CC_NEWRENO_BETA_ECN_ENABLED;
break;
case CC_NEWRENO_ENABLE_HYSTART:
/* Allow hystart on this connection */
if (opt->val != 0) {
nreno->newreno_flags |= CC_NEWRENO_HYSTART_ALLOWED;
if (opt->val > 1)
nreno->newreno_flags |= CC_NEWRENO_HYSTART_CAN_SH_CWND;
if (opt->val > 2)
nreno->newreno_flags |= CC_NEWRENO_HYSTART_CONS_SSTH;
} else
nreno->newreno_flags &= ~(CC_NEWRENO_HYSTART_ALLOWED|CC_NEWRENO_HYSTART_CAN_SH_CWND|CC_NEWRENO_HYSTART_CONS_SSTH);
newreno_log_hystart_event(ccv, nreno, 7, opt->val);
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
break;
default:
return (ENOPROTOOPT);
}
Plug a memory leak and potential NULL-pointer dereference introduced in r331214. Each TCP connection that uses the system default cc_newreno(4) congestion control algorithm module leaks a "struct newreno" (8 bytes of memory) at connection initialisation time. The NULL-pointer dereference is only germane when using the ABE feature, which is disabled by default. While at it: - Defer the allocation of memory until it is actually needed given that ABE is optional and disabled by default. - Document the ENOMEM errno in getsockopt(2)/setsockopt(2). - Document ENOMEM and ENOBUFS in tcp(4) as being synonymous given that they are used interchangeably throughout the code. - Fix a few other nits also accidentally omitted from the original patch. Reported by: Harsh Jain on freebsd-net@ Tested by: tjh@ Differential Revision: https://reviews.freebsd.org/D15358
2018-05-17 02:46:27 +00:00
break;
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
case SOPT_GET:
switch (opt->name) {
case CC_NEWRENO_BETA:
Plug a memory leak and potential NULL-pointer dereference introduced in r331214. Each TCP connection that uses the system default cc_newreno(4) congestion control algorithm module leaks a "struct newreno" (8 bytes of memory) at connection initialisation time. The NULL-pointer dereference is only germane when using the ABE feature, which is disabled by default. While at it: - Defer the allocation of memory until it is actually needed given that ABE is optional and disabled by default. - Document the ENOMEM errno in getsockopt(2)/setsockopt(2). - Document ENOMEM and ENOBUFS in tcp(4) as being synonymous given that they are used interchangeably throughout the code. - Fix a few other nits also accidentally omitted from the original patch. Reported by: Harsh Jain on freebsd-net@ Tested by: tjh@ Differential Revision: https://reviews.freebsd.org/D15358
2018-05-17 02:46:27 +00:00
opt->val = (nreno == NULL) ?
V_newreno_beta : nreno->beta;
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
break;
case CC_NEWRENO_BETA_ECN:
Plug a memory leak and potential NULL-pointer dereference introduced in r331214. Each TCP connection that uses the system default cc_newreno(4) congestion control algorithm module leaks a "struct newreno" (8 bytes of memory) at connection initialisation time. The NULL-pointer dereference is only germane when using the ABE feature, which is disabled by default. While at it: - Defer the allocation of memory until it is actually needed given that ABE is optional and disabled by default. - Document the ENOMEM errno in getsockopt(2)/setsockopt(2). - Document ENOMEM and ENOBUFS in tcp(4) as being synonymous given that they are used interchangeably throughout the code. - Fix a few other nits also accidentally omitted from the original patch. Reported by: Harsh Jain on freebsd-net@ Tested by: tjh@ Differential Revision: https://reviews.freebsd.org/D15358
2018-05-17 02:46:27 +00:00
opt->val = (nreno == NULL) ?
V_newreno_beta_ecn : nreno->beta_ecn;
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
break;
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
case CC_NEWRENO_ENABLE_HYSTART:
if (nreno->newreno_flags & CC_NEWRENO_HYSTART_ALLOWED) {
if (nreno->newreno_flags & CC_NEWRENO_HYSTART_CONS_SSTH)
opt->val = 3;
else if (nreno->newreno_flags & CC_NEWRENO_HYSTART_CAN_SH_CWND)
opt->val = 2;
else
opt->val = 1;
} else
opt->val = 0;
break;
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
default:
return (ENOPROTOOPT);
}
Plug a memory leak and potential NULL-pointer dereference introduced in r331214. Each TCP connection that uses the system default cc_newreno(4) congestion control algorithm module leaks a "struct newreno" (8 bytes of memory) at connection initialisation time. The NULL-pointer dereference is only germane when using the ABE feature, which is disabled by default. While at it: - Defer the allocation of memory until it is actually needed given that ABE is optional and disabled by default. - Document the ENOMEM errno in getsockopt(2)/setsockopt(2). - Document ENOMEM and ENOBUFS in tcp(4) as being synonymous given that they are used interchangeably throughout the code. - Fix a few other nits also accidentally omitted from the original patch. Reported by: Harsh Jain on freebsd-net@ Tested by: tjh@ Differential Revision: https://reviews.freebsd.org/D15358
2018-05-17 02:46:27 +00:00
break;
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
default:
return (EINVAL);
}
return (0);
}
static int
newreno_beta_handler(SYSCTL_HANDLER_ARGS)
{
Fix bugs in plugable CC algorithm and siftr sysctls. Use the sysctl_handle_int() handler to write out the old value and read the new value into a temporary variable. Use the temporary variable for any checks of values rather than using the CAST_PTR_INT() macro on req->newptr. The prior usage read directly from userspace memory if the sysctl() was called correctly. This is unsafe and doesn't work at all on some architectures (at least i386.) In some cases, the code could also be tricked into reading from kernel memory and leaking limited information about the contents or crashing the system. This was true for CDG, newreno, and siftr on all platforms and true for i386 in all cases. The impact of this bug is largest in VIMAGE jails which have been configured to allow writing to these sysctls. Per discussion with the security officer, we will not be issuing an advisory for this issue as root access and a non-default config are required to be impacted. Reviewed by: markj, bz Discussed with: gordon (security officer) MFC after: 3 days Security: kernel information leak, local DoS (both require root) Differential Revision: https://reviews.freebsd.org/D18443
2018-12-15 15:06:22 +00:00
int error;
uint32_t new;
Plug a memory leak and potential NULL-pointer dereference introduced in r331214. Each TCP connection that uses the system default cc_newreno(4) congestion control algorithm module leaks a "struct newreno" (8 bytes of memory) at connection initialisation time. The NULL-pointer dereference is only germane when using the ABE feature, which is disabled by default. While at it: - Defer the allocation of memory until it is actually needed given that ABE is optional and disabled by default. - Document the ENOMEM errno in getsockopt(2)/setsockopt(2). - Document ENOMEM and ENOBUFS in tcp(4) as being synonymous given that they are used interchangeably throughout the code. - Fix a few other nits also accidentally omitted from the original patch. Reported by: Harsh Jain on freebsd-net@ Tested by: tjh@ Differential Revision: https://reviews.freebsd.org/D15358
2018-05-17 02:46:27 +00:00
Fix bugs in plugable CC algorithm and siftr sysctls. Use the sysctl_handle_int() handler to write out the old value and read the new value into a temporary variable. Use the temporary variable for any checks of values rather than using the CAST_PTR_INT() macro on req->newptr. The prior usage read directly from userspace memory if the sysctl() was called correctly. This is unsafe and doesn't work at all on some architectures (at least i386.) In some cases, the code could also be tricked into reading from kernel memory and leaking limited information about the contents or crashing the system. This was true for CDG, newreno, and siftr on all platforms and true for i386 in all cases. The impact of this bug is largest in VIMAGE jails which have been configured to allow writing to these sysctls. Per discussion with the security officer, we will not be issuing an advisory for this issue as root access and a non-default config are required to be impacted. Reviewed by: markj, bz Discussed with: gordon (security officer) MFC after: 3 days Security: kernel information leak, local DoS (both require root) Differential Revision: https://reviews.freebsd.org/D18443
2018-12-15 15:06:22 +00:00
new = *(uint32_t *)arg1;
error = sysctl_handle_int(oidp, &new, 0, req);
if (error == 0 && req->newptr != NULL ) {
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
if (arg1 == &VNET_NAME(newreno_beta_ecn) && !V_cc_do_abe)
Fix bugs in plugable CC algorithm and siftr sysctls. Use the sysctl_handle_int() handler to write out the old value and read the new value into a temporary variable. Use the temporary variable for any checks of values rather than using the CAST_PTR_INT() macro on req->newptr. The prior usage read directly from userspace memory if the sysctl() was called correctly. This is unsafe and doesn't work at all on some architectures (at least i386.) In some cases, the code could also be tricked into reading from kernel memory and leaking limited information about the contents or crashing the system. This was true for CDG, newreno, and siftr on all platforms and true for i386 in all cases. The impact of this bug is largest in VIMAGE jails which have been configured to allow writing to these sysctls. Per discussion with the security officer, we will not be issuing an advisory for this issue as root access and a non-default config are required to be impacted. Reviewed by: markj, bz Discussed with: gordon (security officer) MFC after: 3 days Security: kernel information leak, local DoS (both require root) Differential Revision: https://reviews.freebsd.org/D18443
2018-12-15 15:06:22 +00:00
error = EACCES;
else if (new == 0 || new > 100)
error = EINVAL;
else
*(uint32_t *)arg1 = new;
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
}
Fix bugs in plugable CC algorithm and siftr sysctls. Use the sysctl_handle_int() handler to write out the old value and read the new value into a temporary variable. Use the temporary variable for any checks of values rather than using the CAST_PTR_INT() macro on req->newptr. The prior usage read directly from userspace memory if the sysctl() was called correctly. This is unsafe and doesn't work at all on some architectures (at least i386.) In some cases, the code could also be tricked into reading from kernel memory and leaking limited information about the contents or crashing the system. This was true for CDG, newreno, and siftr on all platforms and true for i386 in all cases. The impact of this bug is largest in VIMAGE jails which have been configured to allow writing to these sysctls. Per discussion with the security officer, we will not be issuing an advisory for this issue as root access and a non-default config are required to be impacted. Reviewed by: markj, bz Discussed with: gordon (security officer) MFC after: 3 days Security: kernel information leak, local DoS (both require root) Differential Revision: https://reviews.freebsd.org/D18443
2018-12-15 15:06:22 +00:00
return (error);
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
}
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
static void
newreno_newround(struct cc_var *ccv, uint32_t round_cnt)
{
struct newreno *nreno;
nreno = (struct newreno *)ccv->cc_data;
/* We have entered a new round */
nreno->css_lastround_minrtt = nreno->css_current_round_minrtt;
nreno->css_current_round_minrtt = 0xffffffff;
nreno->css_rttsample_count = 0;
nreno->css_current_round = round_cnt;
if ((nreno->newreno_flags & CC_NEWRENO_HYSTART_IN_CSS) &&
((round_cnt - nreno->css_entered_at_round) >= hystart_css_rounds)) {
/* Enter CA */
if (nreno->newreno_flags & CC_NEWRENO_HYSTART_CAN_SH_CWND) {
/*
* We engage more than snd_ssthresh, engage
* the brakes!! Though we will stay in SS to
* creep back up again, so lets leave CSS active
* and give us hystart_css_rounds more rounds.
*/
if (nreno->newreno_flags & CC_NEWRENO_HYSTART_CONS_SSTH) {
CCV(ccv, snd_ssthresh) = ((nreno->css_lowrtt_fas + nreno->css_fas_at_css_entry) / 2);
} else {
CCV(ccv, snd_ssthresh) = nreno->css_lowrtt_fas;
}
CCV(ccv, snd_cwnd) = nreno->css_fas_at_css_entry;
nreno->css_entered_at_round = round_cnt;
} else {
CCV(ccv, snd_ssthresh) = CCV(ccv, snd_cwnd);
/* Turn off the CSS flag */
nreno->newreno_flags &= ~CC_NEWRENO_HYSTART_IN_CSS;
/* Disable use of CSS in the future except long idle */
nreno->newreno_flags &= ~CC_NEWRENO_HYSTART_ENABLED;
}
newreno_log_hystart_event(ccv, nreno, 6, CCV(ccv, snd_ssthresh));
}
newreno_log_hystart_event(ccv, nreno, 4, round_cnt);
}
static void
newreno_rttsample(struct cc_var *ccv, uint32_t usec_rtt, uint32_t rxtcnt, uint32_t fas)
{
struct newreno *nreno;
nreno = (struct newreno *)ccv->cc_data;
if (rxtcnt > 1) {
/*
* Only look at RTT's that are non-ambiguous.
*/
return;
}
nreno->css_rttsample_count++;
nreno->css_last_fas = fas;
if (nreno->css_current_round_minrtt > usec_rtt) {
nreno->css_current_round_minrtt = usec_rtt;
nreno->css_lowrtt_fas = nreno->css_last_fas;
}
if ((nreno->newreno_flags & CC_NEWRENO_HYSTART_IN_CSS) &&
(nreno->css_rttsample_count >= hystart_n_rttsamples) &&
(nreno->css_baseline_minrtt > nreno->css_current_round_minrtt)) {
/*
* We were in CSS and the RTT is now less, we
* entered CSS erroneously.
*/
nreno->newreno_flags &= ~CC_NEWRENO_HYSTART_IN_CSS;
newreno_log_hystart_event(ccv, nreno, 8, nreno->css_baseline_minrtt);
nreno->css_baseline_minrtt = 0xffffffff;
}
newreno_log_hystart_event(ccv, nreno, 5, usec_rtt);
}
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
SYSCTL_DECL(_net_inet_tcp_cc_newreno);
Mark more nodes as CTLFLAG_MPSAFE or CTLFLAG_NEEDGIANT (17 of many) r357614 added CTLFLAG_NEEDGIANT to make it easier to find nodes that are still not MPSAFE (or already are but aren’t properly marked). Use it in preparation for a general review of all nodes. This is non-functional change that adds annotations to SYSCTL_NODE and SYSCTL_PROC nodes using one of the soon-to-be-required flags. Mark all obvious cases as MPSAFE. All entries that haven't been marked as MPSAFE before are by default marked as NEEDGIANT Approved by: kib (mentor, blanket) Commented by: kib, gallatin, melifaro Differential Revision: https://reviews.freebsd.org/D23718
2020-02-26 14:26:36 +00:00
SYSCTL_NODE(_net_inet_tcp_cc, OID_AUTO, newreno,
CTLFLAG_RW | CTLFLAG_MPSAFE, NULL,
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
"New Reno related settings");
SYSCTL_PROC(_net_inet_tcp_cc_newreno, OID_AUTO, beta,
Mark more nodes as CTLFLAG_MPSAFE or CTLFLAG_NEEDGIANT (17 of many) r357614 added CTLFLAG_NEEDGIANT to make it easier to find nodes that are still not MPSAFE (or already are but aren’t properly marked). Use it in preparation for a general review of all nodes. This is non-functional change that adds annotations to SYSCTL_NODE and SYSCTL_PROC nodes using one of the soon-to-be-required flags. Mark all obvious cases as MPSAFE. All entries that haven't been marked as MPSAFE before are by default marked as NEEDGIANT Approved by: kib (mentor, blanket) Commented by: kib, gallatin, melifaro Differential Revision: https://reviews.freebsd.org/D23718
2020-02-26 14:26:36 +00:00
CTLFLAG_VNET | CTLTYPE_UINT | CTLFLAG_RW | CTLFLAG_NEEDGIANT,
&VNET_NAME(newreno_beta), 3, &newreno_beta_handler, "IU",
"New Reno beta, specified as number between 1 and 100");
Add support for the experimental Internet-Draft "TCP Alternative Backoff with ECN (ABE)" proposal to the New Reno congestion control algorithm module. ABE reduces the amount of congestion window reduction in response to ECN-signalled congestion relative to the loss-inferred congestion response. More details about ABE can be found in the Internet-Draft: https://tools.ietf.org/html/draft-ietf-tcpm-alternativebackoff-ecn The implementation introduces four new sysctls: - net.inet.tcp.cc.abe defaults to 0 (disabled) and can be set to non-zero to enable ABE for ECN-enabled TCP connections. - net.inet.tcp.cc.newreno.beta and net.inet.tcp.cc.newreno.beta_ecn set the multiplicative window decrease factor, specified as a percentage, applied to the congestion window in response to a loss-based or ECN-based congestion signal respectively. They default to the values specified in the draft i.e. beta=50 and beta_ecn=80. - net.inet.tcp.cc.abe_frlossreduce defaults to 0 (disabled) and can be set to non-zero to enable the use of standard beta (50% by default) when repairing loss during an ECN-signalled congestion recovery episode. It enables a more conservative congestion response and is provided for the purposes of experimentation as a result of some discussion at IETF 100 in Singapore. The values of beta and beta_ecn can also be set per-connection by way of the TCP_CCALGOOPT TCP-level socket option and the new CC_NEWRENO_BETA or CC_NEWRENO_BETA_ECN CC algo sub-options. Submitted by: Tom Jones <tj@enoti.me> Tested by: Tom Jones <tj@enoti.me>, Grenville Armitage <garmitage@swin.edu.au> Relnotes: Yes Differential Revision: https://reviews.freebsd.org/D11616
2018-03-19 16:37:47 +00:00
SYSCTL_PROC(_net_inet_tcp_cc_newreno, OID_AUTO, beta_ecn,
Mark more nodes as CTLFLAG_MPSAFE or CTLFLAG_NEEDGIANT (17 of many) r357614 added CTLFLAG_NEEDGIANT to make it easier to find nodes that are still not MPSAFE (or already are but aren’t properly marked). Use it in preparation for a general review of all nodes. This is non-functional change that adds annotations to SYSCTL_NODE and SYSCTL_PROC nodes using one of the soon-to-be-required flags. Mark all obvious cases as MPSAFE. All entries that haven't been marked as MPSAFE before are by default marked as NEEDGIANT Approved by: kib (mentor, blanket) Commented by: kib, gallatin, melifaro Differential Revision: https://reviews.freebsd.org/D23718
2020-02-26 14:26:36 +00:00
CTLFLAG_VNET | CTLTYPE_UINT | CTLFLAG_RW | CTLFLAG_NEEDGIANT,
&VNET_NAME(newreno_beta_ecn), 3, &newreno_beta_handler, "IU",
"New Reno beta ecn, specified as number between 1 and 100");
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
tcp: Add hystart-plus to cc_newreno and rack. TCP Hystart draft version -03: https://datatracker.ietf.org/doc/html/draft-ietf-tcpm-hystartplusplus Is a new version of hystart that allows one to carefully exit slow start if the RTT spikes too much. The newer version has a slower-slow-start so to speak that then kicks in for five round trips. To see if you exited too early, if not into congestion avoidance. This commit will add that feature to our newreno CC and add the needed bits in rack to be able to enable it. Reviewed by: tuexen Sponsored by: Netflix Inc. Differential Revision: https://reviews.freebsd.org/D32373
2021-10-22 07:10:28 -04:00
SYSCTL_NODE(_net_inet_tcp_cc_newreno, OID_AUTO, hystartplusplus,
CTLFLAG_RW | CTLFLAG_MPSAFE, NULL,
"New Reno related HyStart++ settings");
SYSCTL_UINT(_net_inet_tcp_cc_newreno_hystartplusplus, OID_AUTO, lowcwnd,
CTLFLAG_RW,
&hystart_lowcwnd, 16,
"The number of MSS in the CWND before HyStart++ is active");
SYSCTL_UINT(_net_inet_tcp_cc_newreno_hystartplusplus, OID_AUTO, minrtt_thresh,
CTLFLAG_RW,
&hystart_minrtt_thresh, 4000,
"HyStarts++ minimum RTT thresh used in clamp (in microseconds)");
SYSCTL_UINT(_net_inet_tcp_cc_newreno_hystartplusplus, OID_AUTO, maxrtt_thresh,
CTLFLAG_RW,
&hystart_maxrtt_thresh, 16000,
"HyStarts++ maximum RTT thresh used in clamp (in microseconds)");
SYSCTL_UINT(_net_inet_tcp_cc_newreno_hystartplusplus, OID_AUTO, n_rttsamples,
CTLFLAG_RW,
&hystart_n_rttsamples, 8,
"The number of RTT samples that must be seen to consider HyStart++");
SYSCTL_UINT(_net_inet_tcp_cc_newreno_hystartplusplus, OID_AUTO, css_growth_div,
CTLFLAG_RW,
&hystart_css_growth_div, 4,
"The divisor to the growth when in Hystart++ CSS");
SYSCTL_UINT(_net_inet_tcp_cc_newreno_hystartplusplus, OID_AUTO, css_rounds,
CTLFLAG_RW,
&hystart_css_rounds, 5,
"The number of rounds HyStart++ lasts in CSS before falling to CA");
SYSCTL_UINT(_net_inet_tcp_cc_newreno_hystartplusplus, OID_AUTO, bblogs,
CTLFLAG_RW,
&hystart_bblogs, 0,
"Do we enable HyStart++ Black Box logs to be generated if BB logging is on");
This commit marks the first formal contribution of the "Five New TCP Congestion Control Algorithms for FreeBSD" FreeBSD Foundation funded project. More details about the project are available at: http://caia.swin.edu.au/freebsd/5cc/ - Add a KPI and supporting infrastructure to allow modular congestion control algorithms to be used in the net stack. Algorithms can maintain per-connection state if required, and connections maintain their own algorithm pointer, which allows different connections to concurrently use different algorithms. The TCP_CONGESTION socket option can be used with getsockopt()/setsockopt() to programmatically query or change the congestion control algorithm respectively from within an application at runtime. - Integrate the framework with the TCP stack in as least intrusive a manner as possible. Care was also taken to develop the framework in a way that should allow integration with other congestion aware transport protocols (e.g. SCTP) in the future. The hope is that we will one day be able to share a single set of congestion control algorithm modules between all congestion aware transport protocols. - Introduce a new congestion recovery (TF_CONGRECOVERY) state into the TCP stack and use it to decouple the meaning of recovery from a congestion event and recovery from packet loss (TF_FASTRECOVERY) a la RFC2581. ECN and delay based congestion control protocols don't generally need to recover from packet loss and need a different way to note a congestion recovery episode within the stack. - Remove the net.inet.tcp.newreno sysctl, which simplifies some portions of code and ensures the stack always uses the appropriate mechanisms for recovering from packet loss during a congestion recovery episode. - Extract the NewReno congestion control algorithm from the TCP stack and massage it into module form. NewReno is always built into the kernel and will remain the default algorithm for the forseeable future. Implementations of additional different algorithms will become available in the near future. - Bump __FreeBSD_version to 900025 and note in UPDATING that rebuilding code that relies on the size of "struct tcpcb" is required. Many thanks go to the Cisco University Research Program Fund at Community Foundation Silicon Valley and the FreeBSD Foundation. Their support of our work at the Centre for Advanced Internet Architectures, Swinburne University of Technology is greatly appreciated. In collaboration with: David Hayes <dahayes at swin edu au> and Grenville Armitage <garmitage at swin edu au> Sponsored by: Cisco URP, FreeBSD Foundation Reviewed by: rpaulo Tested by: David Hayes (and many others over the years) MFC after: 3 months
2010-11-12 06:41:55 +00:00
DECLARE_CC_MODULE(newreno, &newreno_cc_algo);
tcp: Congestion control cleanup. NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!! This patch does a bit of cleanup on TCP congestion control modules. There were some rather interesting surprises that one could get i.e. where you use a socket option to change from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get a memory failure and end up on cc_newreno. This is not what one would expect. The new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK and pass in to the init function preallocated memory. The CC init is expected in this case *not* to fail but if it does and a module does break the "no fail with memory given" contract we do fall back to the CC that was in place at the time. This also fixes up a set of common newreno utilities that can be shared amongst other CC modules instead of the other CC modules reaching into newreno and executing what they think is a "common and understood" function. Lets put these functions in cc.c and that way we have a common place that is easily findable by future developers or bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE and HYSTART++ without having to dance through hoops for other CC modules, instead both newreno and the other modules just call into the common functions if they desire that behavior or roll there own if that makes more sense. Note: This commit changes the kernel configuration!! If you are not using GENERIC in some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC, CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined as well if you desire. Note that if you create a kernel configuration that does not define a congestion control module and includes INET or INET6 the kernel compile will break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\" but you can specify any string that represents the name of the CC module (same names that show up in the CC module list under net.inet.tcp.cc). If you fail to add the options CC_DEFAULT in your kernel configuration the kernel build will also break. Reviewed by: Michael Tuexen Sponsored by: Netflix Inc. RELNOTES:YES Differential Revision: https://reviews.freebsd.org/D32693
2021-11-11 06:28:18 -05:00
MODULE_VERSION(newreno, 2);
Reference in New Issue Copy Permalink