summaryrefslogtreecommitdiffstats
path: root/src/vnet/tcp
AgeCommit message (Collapse)AuthorFilesLines
2022-11-29session: transport endpt cleanup on owner threadFlorin Coras1-2/+2
Maintain a single writer multiple readers usage model for transport endpoints pool. Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I8555700ed725971341f145ea97f031042a298e83
2022-10-07tcp: cmake option VPP_TCP_DEBUG_ALWAYS=ON not taken by all filesSteven Luong1-0/+1
Some files include tcp_debug.h without including <vpp/vnet/config.h> As a result, those files do not get VPP_TCP_DEBUG_ALWAYS option set. The fix is to include <vpp/vnet/config.h> in tcp_debug.h Type: fix Signed-off-by: Steven Luong <sluong@cisco.com> Change-Id: I0c141147f1e8d1b49c5a1440fac1e97cbd96aaa7
2022-10-07tcp: build image with TCP_DEBUG_ALWAYS via makeSteven Luong1-0/+4
Add cmake option to enable TCP_DEBUG_ALWAYS. make rebuild VPP_EXTRA_CMAKE_ARGS=-DVPP_TCP_DEBUG_ALWAYS=ON make rebuild VPP_EXTRA_CMAKE_ARGS=-DVPP_TCP_DEBUG_ALWAYS=OFF Type: improvement Signed-off-by: Steven Luong <sluong@cisco.com> Change-Id: I911a8d615f76516ae0a988bc6135c3b0d8fcb3df
2022-10-03tcp: replace tcp_time_now with tcp_time_now_usSteven Luong1-28/+29
It looks like tcp_time_now has been deprecated for a while and the replacement is tcp_time_now_us Type: fix Signed-off-by: Steven Luong <sluong@cisco.com> Change-Id: Ifaed2632baa49d489d4e03f2623d8cc9a6f36e64
2022-08-30tcp: do not overcount ooo bytesFlorin Coras1-1/+3
Type: fix Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: Ic81702bffb5b3189db48efe1ab3b237fa2bf75f2
2022-08-11ip: only set rx_sw_if_index when connection found to avoid following crash ↵Xiaoming Jiang1-5/+5
like tcp punt Type: fix Signed-off-by: Xiaoming Jiang <jiangxiaoming@outlook.com> Change-Id: I894a881cec1888b392d26fdfb385f97c31113ef1
2022-04-12tcp: limit persist segment size if window availableFlorin Coras1-1/+3
Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: Ie1ae401d5105aa42761d3cac5dfbe523f3995c87
2022-04-04vppinfra: make _vec_len() read-onlyDamjan Marion2-4/+4
Use of _vec_len() to set vector length breaks address sanitizer. Users should use vec_set_len(), vec_inc_len(), vec_dec_len () instead. Type: improvement Change-Id: I441ae948771eb21c23a61f3ff9163bdad74a2cb8 Signed-off-by: Damjan Marion <damarion@cisco.com>
2022-04-04tcp: use safe realloc poolFlorin Coras1-3/+3
Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I8ba93643236aaa50eeb4755b74821a8c4ca6179b
2022-03-15tcp: update error counters in listen nodeFlorin Coras1-9/+7
Type: fix Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: Ib2e1d847607c9c7d928b174b87e5c21d53153ebe
2022-03-15tcp: update persist timer if data ackedFlorin Coras1-3/+7
Update persist timer if data sent during snd_wnd < snd_mss was acked. Type: fix Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I5c75ff8ddc0e49750b2088237d32afa4eda99e7f
2022-02-15tcp: Do not include the tcp_packet.h file in the ip4_packet.hNeale Ranns1-1/+96
Type: refactor IP4 does not depend on TCP (it's the other way around). This upside down dependency leads to some nasty circular includes when trying to use ip46_address.h in interface.h Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: I4a1bd21543b08b9c1cf1e5563da738414734a878
2022-02-10session: use transport endpoint cfg for listenFlorin Coras1-2/+2
Makes it similar to connects. Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I38c328670054e1a9ba4dc4ea8fe7519a5a09e8be
2022-01-12session: pass tx buffers in bulk to transportsFlorin Coras2-6/+35
Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I1025cccd784f80b557847f69c3ea1ada5c9de60d
2022-01-10session: deschedule sessions with no data to sendFlorin Coras1-1/+0
This ensures the scheduler always tracks sessions that are descheduled, i.e., do not have events in the old io events list. When app retries to send, clear descheduled flag and potentially the pacer. Consequently, transports no longer need to reset the pacer when sessions are rescheduled after a long app tx pause. This also fixes a tcp bug whereby the pacer was reset too often when snd_una was equal to snd_nxt as there was no way to distinguish betwen app tx breaks and congestion. Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: Id3cc6c98cd76299e15030e504380dcf3c04c5189
2022-01-07tcp: fix cubic cwnd accumulate use of bytes ackedFlorin Coras1-1/+1
Use what was provided instead of tc->bytes_acked Type: fix Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I0ed736d2ee247e231fccdf4a969fcf6bc15b7978
2022-01-06tcp: optimize newreno cong ack handlerFlorin Coras1-19/+20
Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I196ee5f4630cec637245493f370b5f83a939fe44
2022-01-06tcp: update snd_congestion only during congestionFlorin Coras3-31/+14
If running without sacks, if snd_una does not cover snd_congestion fast recovery can be missed but the two heuristics from RFC6582 should avoid that. Also snd_congestion was used as a means of inferring if the connection recently exited congestion while setting the persist timer but that does not always work correctly if not congested. Type: fix Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I94d4ac738cdd4f7f23f62e97dd63059de1cd4af9
2022-01-06tcp: mark lost first sack hole on timeoutFlorin Coras3-1/+27
Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I1abff943f3fe3ff0219126b5b8beded4ad859758
2022-01-06tcp: handle start tx event in cubicFlorin Coras1-0/+18
If app was idle update start time of current congestion avoidance phase unless tcp connection was not idle. Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: Idf6a03a9ef96c409462de9f9cb19df609f730afe
2022-01-06tcp: always exit recovery if not in fastrecoveryFlorin Coras1-3/+3
Stay in fast recovery only if it's already on. Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: Idcdbbacfed3e5f3c991fa293c532be1c671f5217
2022-01-06tcp: exit retransmit before processing feedbackFlorin Coras1-21/+17
Type: refactor Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I5838e4c370d0c02a21b5eadb4af3baae781df097
2022-01-06tcp: use bytes delivered to compute cwndFlorin Coras3-7/+9
Should estimated cwnd better with loss Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: Idd75d40dbab212ac0a5d533009c5540b1a58f4c4
2022-01-06tcp: cast timer ticks to u32Florin Coras3-7/+8
tc->rto * TCP_TO_TIMER_TICK can return garbage if not cast to u32 and that confuses clib_max Type: fix Signed-off-by: Florin Coras <fcoras@cisco.com> signed-off-by: Vipul Agrawal <Vipul.Agrawal@enea.com> Change-Id: Ief4d29b9625e2ef2e75e0c7e3d731ab147465f6d
2022-01-05interface: refactor interface capabilities codeDamjan Marion1-1/+1
Make it shorter to type, easier to debug, make adding callbacks in future simpler. Type: improvement Change-Id: I6cdd6375e36da23bd452a7c7273ff42789e94433 Signed-off-by: Damjan Marion <damarion@cisco.com>
2021-12-26tcp: set sw_if_index in tcp src-address cliMercury1-2/+2
the receive dpo added by tcp src-address cli do not have a valid sw_if_index , ip4_local_check_src() and tcp_input_lookup_buffer() will set ~0 to vnet_buffer(b)->sw_if_index[VLIB_RX], which will cause crash in tcp46_reset_inline, Type: fix Signed-off-by: Mercury <mercury124185@gmail.com> Change-Id: Ie01c31f3575e14187c6380ebcfff96fcb6098cde
2021-12-23tcp: fix endpoint lookup failed caused by key uninitializedMercury1-0/+1
fix ipv4 key uninitialized in local_endpoints_table, which will cause transport_endpoint_cleanup() failed to lookup the endpoint and can not delete it, as for ipv6, clib_memcpy_fast() will change all bytes of lcl_addr and there maybe no need to initalize, Type: fix Signed-off-by: Mercury <mercury124185@gmail.com> Change-Id: I56676493a393b1d64eaa438224e256094ca75d2f
2021-12-12tcp: fix the tcp src-address cliMercury1-0/+2
When the src-address to add match a route without a valid output interface(such as default route 0.0.0.0/0), fib_entry_get_resolving_interface() will return (u32)~0, which will cause crash in ip4_neighbor_proxy_enable(). Type: fix Signed-off-by: Mercury <mercury124185@gmail.com> Change-Id: I5aee5676a2ff43ec06745ebed4dba2b9e5b98c4d
2021-11-10ip: always set ip rx_sw_if_indexFlorin Coras1-4/+1
Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I43f6bfa028ee37227f27a2fe0303662bf2631b10
2021-11-08tcp: refactor reset nodeFlorin Coras1-49/+71
Use vectorized buffer translate function and refactor tracing. Type: refactor Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I5a014cea1a753eb10a564ac316c1727a18f884ff
2021-11-06tcp: fix reset with packet ack numberFlorin Coras1-0/+1
Type: fix Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I35c1fc75df065398821cc275e853a3caa9db94c2
2021-11-06tcp: in place tcp header for buffer resetsFlorin Coras1-2/+2
Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I2a5de36175bca1181ffd4a1865d41f0a1f6bc035
2021-11-03tcp: make buffer reset function publicFlorin Coras2-32/+28
Also does a bit of code cleanup. Type: refactor Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: Ifbff1e03a2e1fdca98925fdd3db7eb230ff24a37
2021-11-02tcp: fix uninitializad valueDamjan Marion1-1/+1
May not be valid error but gcc-11 complains. Type: fix Change-Id: I207b8b4966c5eadd534495e29c873663249a377f Signed-off-by: Damjan Marion <damarion@cisco.com>
2021-10-27tcp: fix use of invalid local sw_if_indexFlorin Coras1-1/+2
Type: fix Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: Id58ac44d45fb8b8a0d803f02e0242ec6f4b3db05
2021-10-25tcp: fix: TCP timewait port reuse rfc complianceMichal Kalderon2-4/+84
This patch provides a fix for early-kill of timewait sockets that is based on rfc's 1122, 6191. The following commits provided a solution for port re-use. However, they are not fully compliant with rfc 1122 4.2.2.13 ( Closing a connection ) and rfc 6191 (Reducing the TIME-WAIT State Using TCP Timestamps) commit b092b77cf238ba ("tcp: Enable TCP timewait port use") introduced a significant improvement by enabling TCP timewait port re-use. commit ee1cb469b2dd ("tcp: fix port reuse with multiple listeners") fixed usage of the wrong value for connection_index when searching for a listener, by storing the state in tcp.flags. Implementation details: When a SYN is received during time-wait state, the code checks whether all the requirements for accepting the SYN packet are met. If they aren't, the SYN can't be accepted and the packet is dropped, otherwise, connection is deleted and a new connection with same port is opened. Type: fix Signed-off-by: Ofer Heifetz <oferh@marvell.com> Signed-off-by: Yuval Caduri <cyuval@marvell.com> Signed-off-by: Michal Kalderon <mkalderon@marvell.com> Change-Id: I38a33c6e321c760d45ebec9154399e1c90dd0250
2021-10-21ip: Add ip46-local node for local swif[rx]Nathan Skrzypczak1-0/+7
Type: improvement This adds a new ip[46]-receive node, sibling of ip[46]-local. Its goal is to set vnet_buffer (b)->ip.rx_sw_if_index to the sw_if_index of the local interface. In dependant nodes further down the line (e.g. hoststack) we then set sw_if_idx[rx] to this value. So that we know which local interface did receive the packet. The TCP issue this fixes is that : On accepts, we were setting tc->sw_if_index to the source sw_if_index. We should use the dest sw_if_index, so that packets coming back on this connection have the right source sw_if_index. And also setting it in the tx-ed packet. Change-Id: I569ed673e15c21e71f365c3ad45439b05bd14a9f Signed-off-by: Nathan Skrzypczak <nathan.skrzypczak@gmail.com>
2021-10-06tcp: fix severity infoFilip Tehlar5-66/+66
Type: fix Signed-off-by: Filip Tehlar <ftehlar@cisco.com> Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: Ibe39bc045c3b154209a83b59ef95a37c61b32c0c
2021-09-13tcp session: initialize sw_if_index at connect timeFlorin Coras4-4/+5
Type: fix Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: Id0e67a9ce9c5d8bca2cd7a30493f03a02a46dc46
2021-07-15misc: replace CLIB_PREFETCH with clib_prefetch_{load,store}Damjan Marion1-4/+4
Type: refactor Change-Id: Id10cbf52e8f2dd809080a228d8fa282308be84ac Signed-off-by: Damjan Marion <damarion@cisco.com>
2021-07-13tcp session: next node config on connectsFlorin Coras1-0/+2
Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: Ief06b1509d31b55efc8d1436b6ff9e01c6037a32
2021-07-06tcp: use tcp output for half open connectionsFlorin Coras1-39/+12
Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I05be39671a9ed0688d4e006f0f9354aa6560a41b
2021-07-06tcp: use main thread pool for half-opensFlorin Coras5-146/+29
Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I4b3427b966f9ff1ba8895fed7db662d56650f3f5
2021-06-29tcp: add ACK flag to RST packetliuyacan1-1/+1
According to RFC 793, the ACK control bit is always sent once the connection is established. Type: fix Signed-off-by: liuyacan <liuyacan@corp.netease.com> Change-Id: Id0fe19114a0cc468dbce4c0938b345c2ac339e73
2021-06-22tcp: api cleanupFilip Tehlar2-43/+9
Use autogenerated code. Does not change API definitions. Type: improvement Signed-off-by: Filip Tehlar <ftehlar@cisco.com> Change-Id: I3c06e9cb3fabdcdce9c17e93cfedfd771295f589
2021-06-15tcp: prevent timer handler being called frequentlyliuyacan4-6/+18
In the current implement, tcp would start or up an one tick retransmit timer for that connection if vlib_buffer_alloc() return 0. Now the tick is 0.1ms, this means that if VPP is in a buffer shortage state, there would be a large number of burst timer expirations. This commit limits the minimum interval of the retransmission timer to 100ms. Type: fix Signed-off-by: liuyacan <liuyacan@corp.netease.com> Change-Id: Ia11d693fe46119c5dc16b24ca93c30c31109057a
2021-06-12tcp: fix proto in port reuse checkFlorin Coras1-2/+2
Type: fix Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I338e61654a62ed6308ecd8bb15e1a8b13cd859b9
2021-05-14tcp: remove ho lockFlorin Coras3-13/+0
Half-open sessions are allocated by main thread and cleaned up on main with timers. Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I37f000920a45908b62b5501ae9d54a88a9e4c609
2021-05-12session: support half-close connectionliuyacan1-1/+25
Some app(e.g. Envoy) may call shutdown() instead of close() when draining connection. Type: improvement Signed-off-by: liuyacan <liuyacan@corp.netease.com> Change-Id: I9543b9ca3caa87b10b134fd1fc4019124e41e4d2
2021-05-10session: use half-open sessions for vc establishmentFlorin Coras1-3/+15
Use half-open sessions to track virtual circuit connection establishment. These sesssions can only be allocated and freed by the thread that allocates half-open connections (main). Consequently, they can only be freed on half-open cleanup notifications from transports. Goal is to simplify state tracking within the session layer but it's also a first step towards allowing builtin apps to track and cleanup outstanding connects. Type: improvement Signed-off-by: Florin Coras <fcoras@cisco.com> Change-Id: I8a535906d13eb7f8966deb82333839de80f8049f