1604 Commits (2cc1100d242df512612781187eaa898d0de133dc)

Author	SHA1	Message	Date
Tim Walters	856ea2376b	wgengine/magicsock: log home DERP changes with latency ... This adds additional logging on DERP home changes to allow better troubleshooting. Updates tailscale/corp#18095 Signed-off-by: Tim Walters <tim@tailscale.com>	1 month ago
Anton Tolchanov	11e96760ff	wgengine/magicsock: fix stats packet counter on derp egress ... Updates tailscale/corp#22075 Signed-off-by: Anton Tolchanov <anton@tailscale.com>	1 month ago
Brad Fitzpatrick	6a885dbc36	wgengine/magicsock: fix CI-only test warning of missing health tracker ... While looking at deflaking TestTwoDevicePing/ping_1.0.0.2_via_SendPacket, there were a bunch of distracting: WARNING: (non-fatal) nil health.Tracker (being strict in CI): ... This pacifies those so it's easier to work on actually deflaking the test. Updates #11762 Updates #11874 Change-Id: I08dcb44511d4996b68d5f1ce5a2619b555a2a773 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	2 months ago
Kristoffer Dalby	e0d711c478	{net/connstats,wgengine/magicsock}: fix packet counting in connstats ... connstats currently increments the packet counter whenever it is called to store a length of data, however when udp batch sending was introduced we pass the length for a series of packages, and it is only incremented ones, making it count wrongly if we are on a platform supporting udp batches. Updates tailscale/corp#22075 Signed-off-by: Kristoffer Dalby <kristoffer@tailscale.com>	2 months ago
Kristoffer Dalby	40c991f6b8	wgengine: instrument with usermetrics ... Updates tailscale/corp#22075 Signed-off-by: Kristoffer Dalby <kristoffer@tailscale.com>	2 months ago
Brad Fitzpatrick	c763b7a7db	syncs: delete Map.Range, update callers to iterators ... Updates #11038 Change-Id: I2819fed896cc4035aba5e4e141b52c12637373b1 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	2 months ago
Brad Fitzpatrick	c48cc08de2	wgengine: stop conntrack log spam about Canonical net probes ... Like we do for the ones on iOS. As a bonus, this removes a caller of tsaddr.IsTailscaleIP which we want to revamp/remove soonish. Updates #13687 Change-Id: Iab576a0c48e9005c7844ab52a0aba5ba343b750e Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	2 months ago
Irbe Krumina	9bd158cc09	cmd/containerboot,util/linuxfw: create a SNAT rule for dst/src only once, clean up if needed (#13658) ... The AddSNATRuleForDst rule was adding a new rule each time it was called including: - if a rule already existed - if a rule matching the destination, but with different desired source already existed This was causing issues especially for the in-progress egress HA proxies work, where the rules are now refreshed more frequently, so more redundant rules were being created. This change: - only creates the rule if it doesn't already exist - if a rule for the same dst, but different source is found, delete it - also ensures that egress proxies refresh firewall rules if the node's tailnet IP changes Updates tailscale/tailscale#13406 Signed-off-by: Irbe Krumina <irbe@tailscale.com>	2 months ago
Brad Fitzpatrick	5f88b65764	wgengine/netstack: check userspace ping success on Windows ... Hacky temporary workaround until we do #13654 correctly. Updates #13654 Change-Id: I764eaedbb112fb3a34dddb89572fec1b2543fd4a Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	2 months ago
Brad Fitzpatrick	6f694da912	wgengine/magicsock: avoid log spam from ReceiveFunc on shutdown ... The new logging in 2dd71e64ac is spammy at shutdown: Receive func ReceiveIPv6 exiting with error: *net.OpError, read udp [::]:38869: raw-read udp6 [::]:38869: use of closed network connection Receive func ReceiveIPv4 exiting with error: *net.OpError, read udp 0.0.0.0:36123: raw-read udp4 0.0.0.0:36123: use of closed network connection Skip it if we're in the process of shutting down. Updates #10976 Change-Id: I4f6d1c68465557eb9ffe335d43d740e499ba9786 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	2 months ago
Irbe Krumina	096b090caf	cmd/containerboot,kube,util/linuxfw: configure kube egress proxies to route to 1+ tailnet targets (#13531) ... * cmd/containerboot,kube,util/linuxfw: configure kube egress proxies to route to 1+ tailnet targets This commit is first part of the work to allow running multiple replicas of the Kubernetes operator egress proxies per tailnet service + to allow exposing multiple tailnet services via each proxy replica. This expands the existing iptables/nftables-based proxy configuration mechanism. A proxy can now be configured to route to one or more tailnet targets via a (mounted) config file that, for each tailnet target, specifies: - the target's tailnet IP or FQDN - mappings of container ports to which cluster workloads will send traffic to tailnet target ports where the traffic should be forwarded. Example configfile contents: { "some-svc": {"tailnetTarget":{"fqdn":"foo.tailnetxyz.ts.net","ports"{"tcp:4006:80":{"protocol":"tcp","matchPort":4006,"targetPort":80},"tcp:4007:443":{"protocol":"tcp","matchPort":4007,"targetPort":443}}}} } A proxy that is configured with this config file will configure firewall rules to route cluster traffic to the tailnet targets. It will then watch the config file for updates as well as monitor relevant netmap updates and reconfigure firewall as needed. This adds a bunch of new iptables/nftables functionality to make it easier to dynamically update the firewall rules without needing to restart the proxy Pod as well as to make it easier to debug/understand the rules: - for iptables, each portmapping is a DNAT rule with a comment pointing at the 'service',i.e: -A PREROUTING ! -i tailscale0 -p tcp -m tcp --dport 4006 -m comment --comment "some-svc:tcp:4006 -> tcp:80" -j DNAT --to-destination 100.64.1.18:80 Additionally there is a SNAT rule for each tailnet target, to mask the source address. - for nftables, a separate prerouting chain is created for each tailnet target and all the portmapping rules are placed in that chain. This makes it easier to look up rules and delete services when no longer needed. (nftables allows hooking a custom chain to a prerouting hook, so no extra work is needed to ensure that the rules in the service chains are evaluated). The next steps will be to get the Kubernetes Operator to generate the configfile and ensure it is mounted to the relevant proxy nodes. Updates tailscale/tailscale#13406 Signed-off-by: Irbe Krumina <irbe@tailscale.com>	2 months ago
Kristoffer Dalby	5550a17391	wgengine: make opts.Metrics mandatory ... Fixes #13582 Signed-off-by: Kristoffer Dalby <kristoffer@tailscale.com>	2 months ago
James Tucker	9eb59c72c1	wgengine/magicsock: fix check for EPERM on macOS ... Like Linux, macOS will reply to sendto(2) with EPERM if the firewall is currently blocking writes, though this behavior is like Linux undocumented. This is often caused by a faulting network extension or content filter from EDR software. Updates #11710 Updates #12891 Updates #13511 Signed-off-by: James Tucker <james@tailscale.com>	2 months ago
Adrian Dewhurst	2fdbcbdf86	wgengine/magicsock: only used cached results for GetLastNetcheckReport ... When querying for an exit node suggestion, occasionally it triggers a new report concurrently with an existing report in progress. Generally, there should always be a recent report or one in progress, so it is redundant to start one there, and it causes concurrency issues. Fixes #12643 Change-Id: I66ab9003972f673e5d4416f40eccd7c6676272a5 Signed-off-by: Adrian Dewhurst <adrian@tailscale.com>	2 months ago
Kristoffer Dalby	0e0e53d3b3	util/usermetrics: make usermetrics non-global ... this commit changes usermetrics to be non-global, this is a building block for correct metrics if a go process runs multiple tsnets or in tests. Updates #13420 Updates tailscale/corp#22075 Signed-off-by: Kristoffer Dalby <kristoffer@tailscale.com>	2 months ago
Jordan Whited	951884b077	net/netcheck,wgengine/magicsock: plumb OnlyTCP443 controlknob through netcheck (#13491) ... Updates tailscale/corp#17879 Signed-off-by: Jordan Whited <jordan@tailscale.com>	2 months ago
Jordan Whited	5f4a4c6744	wgengine/magicsock: fix sendUDPStd docs (#13490) ... Updates #cleanup Signed-off-by: Jordan Whited <jordan@tailscale.com>	3 months ago
Jordan Whited	4084c6186d	wgengine/magicsock: add side-effect-free function for netcheck UDP sends (#13487) ... Updates #13484 Updates tailscale/corp#17879 Signed-off-by: Jordan Whited <jordan@tailscale.com>	3 months ago
Andrew Dunham	40833a7524	wgengine/magicsock: disable raw disco by default; add envknob to enable ... Updates #13140 Signed-off-by: Andrew Dunham <andrew@du.nham.ca> Change-Id: Ica85b2ac8ac7eab4ec5413b212f004aecc453279	3 months ago
Jordan Whited	afec2d41b4	wgengine/magicsock: remove redundant deadline from netcheck report call (#13395) ... netcheck.Client.GetReport() applies its own deadlines. This 2s deadline was causing GetReport() to never fall back to HTTPS/ICMP measurements as it was shorter than netcheck.stunProbeTimeout, leaving no time for fallbacks. Updates #13394 Updates #6187 Signed-off-by: Jordan Whited <jordan@tailscale.com>	3 months ago
Brad Fitzpatrick	9f9470fc10	ipnlocal,proxymap,wgengine/netstack: add optional WhoIs/proxymap debug ... Updates tailscale/corp#20600 Change-Id: I2bb17af0f40603ada1ba4cecc087443e00f9392a Co-authored-by: Maisem Ali <maisem@tailscale.com> Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	3 months ago
Jordan Whited	7aa766ee65	net/tstun: probe TCP GRO (#13376) ... Disable TCP & UDP GRO if the probe fails. torvalds/linux@e269d79c7d broke virtio_net TCP & UDP GRO causing GRO writes to return EINVAL. The bug was then resolved later in torvalds/linux@89add40066. The offending commit was pulled into various LTS releases. Updates #13041 Signed-off-by: Jordan Whited <jordan@tailscale.com>	3 months ago
Brad Fitzpatrick	3d401c11fa	all: use new Go 1.23 slices.Sorted more ... Updates #12912 Change-Id: If1294e5bc7b5d3cf0067535ae10db75e8b988d8b Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	3 months ago
Andrew Dunham	1c972bc7cb	wgengine/magicsock: actually use AF_PACKET socket for raw disco ... Previously, despite what the commit said, we were using a raw IP socket that was *not* an AF_PACKET socket, and thus was subject to the host firewall rules. Switch to using a real AF_PACKET socket to actually get the functionality we want. Updates #13140 Signed-off-by: Andrew Dunham <andrew@du.nham.ca> Change-Id: If657daeeda9ab8d967e75a4f049c66e2bca54b78	3 months ago
Brad Fitzpatrick	65fe0ba7b5	wgengine/magicsock: fix panic regression from cryptokey routing change ... Fixes #13332 Updates tailscale/corp#20732 Change-Id: I30f12746844bf77f5a664bf8e8d8ebf2511a2b27 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	3 months ago
Jordan Whited	0926954cf5	net/tstun,wgengine/netstack: implement TCP GRO for local services (#13315) ... Throughput improves substantially when measured via netstack loopback (TS_DEBUG_NETSTACK_LOOPBACK_PORT). Before (d21ebc2): jwhited@i5-12400-2:~$ iperf3 -V -c 100.100.100.100 Starting Test: protocol: TCP, 1 streams, 131072 byte blocks Test Complete. Summary Results: [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.00 sec 5.77 GBytes 4.95 Gbits/sec 0 sender [ 5] 0.00-10.01 sec 5.77 GBytes 4.95 Gbits/sec receiver After: jwhited@i5-12400-2:~$ iperf3 -V -c 100.100.100.100 Starting Test: protocol: TCP, 1 streams, 131072 byte blocks Test Complete. Summary Results: [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.00 sec 12.7 GBytes 10.9 Gbits/sec 0 sender [ 5] 0.00-10.00 sec 12.7 GBytes 10.9 Gbits/sec receiver Updates tailscale/corp#22754 Signed-off-by: Jordan Whited <jordan@tailscale.com>	3 months ago
Jordan Whited	d21ebc28af	wgengine/netstack: implement netstack loopback (#13301) ... When the TS_DEBUG_NETSTACK_LOOPBACK_PORT environment variable is set, netstack will loop back (dnat to addressFamilyLoopback:loopbackPort) TCP & UDP flows originally destined to localServicesIP:loopbackPort. localServicesIP is quad-100 or the IPv6 equivalent. Updates tailscale/corp#22713 Signed-off-by: Jordan Whited <jordan@tailscale.com>	3 months ago
Jordan Whited	31cdbd68b1	net/tstun: fix gvisor inbound GSO packet injection (#13283) ... buffs[0] was not sized to hold pkt with GSO, resulting in a panic. Updates tailscale/corp#22511 Signed-off-by: Jordan Whited <jordan@tailscale.com>	3 months ago
Jordan Whited	bfcb3562e6	wgengine/netstack: re-enable gVisor GSO on Linux (#13269) ... This was previously disabled in 8e42510 due to missing GSO-awareness in tstun, which was resolved in d097096. Updates tailscale/corp#22511 Signed-off-by: Jordan Whited <jordan@tailscale.com>	3 months ago
Jordan Whited	d097096ddc	net/tstun,wgengine/netstack: make inbound synthetic packet injection GSO-aware (#13266) ... Updates tailscale/corp#22511 Signed-off-by: Jordan Whited <jordan@tailscale.com>	3 months ago
Jordan Whited	6d4973e1e0	wgengine/netstack: use types/logger.Logf instead of stdlib log.Printf (#13267) ... Updates #cleanup Signed-off-by: Jordan Whited <jordan@tailscale.com>	3 months ago
Percy Wegmann	d00d6d6dc2	go.mod: update to github.com/tailscale/netlink library that doesn't require vishvananda/netlink ... After the upstream PR is merged, we can point directly at github.com/vishvananda/netlink and retire github.com/tailscale/netlink. See https://github.com/vishvananda/netlink/pull/1006 Updates #12298 Signed-off-by: Percy Wegmann <percy@tailscale.com>	3 months ago
Jordan Whited	8e42510a71	wgengine/netstack: disable gVisor GSO on Linux (#13215) ... net/tstun.Wrapper.InjectInboundPacketBuffer is not GSO-aware, which can break quad-100 TCP streams as a result. Linux is the only platform where gVisor GSO was previously enabled. Updates tailscale/corp#22511 Updates #13211 Signed-off-by: Jordan Whited <jordan@tailscale.com>	3 months ago
Jordan Whited	7675c3ebf2	wgengine/netstack/gro: exclude importation of gVisor GRO pkg on iOS (#13202) ... In df6014f1d7 we removed build tag gating preventing importation, which tripped a NetworkExtension limit test in corp. This was a reversal of 25f0a3fc8f which actually made the situation worse, hence the simplification. This commit goes back to the strategy in 25f0a3fc8f, and gets us back under the limit in my local testing. Admittedly, we don't fully understand the effects of importing or excluding importation of this package, and have seen mixed results, but this commit allows us to move forward again. Updates tailscale/corp#22125 Signed-off-by: Jordan Whited <jordan@tailscale.com>	3 months ago
Jordan Whited	df6014f1d7	net/tstun,wgengine{/netstack/gro}: refactor and re-enable gVisor GRO for Linux (#13172) ... In 2f27319baf we disabled GRO due to a data race around concurrent calls to tstun.Wrapper.Write(). This commit refactors GRO to be thread-safe, and re-enables it on Linux. This refactor now carries a GRO type across tstun and netstack APIs with a lifetime that is scoped to a single tstun.Wrapper.Write() call. In 25f0a3fc8f we used build tags to prevent importation of gVisor's GRO package on iOS as at the time we believed it was contributing to additional memory usage on that platform. It wasn't, so this commit simplifies and removes those build tags. Updates tailscale/corp#22353 Updates tailscale/corp#22125 Updates #6816 Signed-off-by: Jordan Whited <jordan@tailscale.com>	3 months ago
tomholford	16bb541adb	wgengine/magicsock: replace deprecated poly1305 (#13184) ... Signed-off-by: tomholford <tomholford@users.noreply.github.com>	3 months ago
Jordan Whited	ccf091e4a6	wgengine/magicsock: don't upgrade to linuxBatchingConn on Android (#13161) ... In a93dc6cdb1 tryUpgradeToBatchingConn() moved to build tag gated files, but the runtime.GOOS condition excluding Android was removed unintentionally from batching_conn_linux.go. Add it back. Updates tailscale/corp#22348 Signed-off-by: Jordan Whited <jordan@tailscale.com>	4 months ago
Andrew Dunham	e107977f75	wgengine/magicsock: disable SIO_UDP_NETRESET on Windows ... By default, Windows sets the SIO_UDP_CONNRESET and SIO_UDP_NETRESET options on created UDP sockets. These behaviours make the UDP socket ICMP-aware; when the system gets an ICMP message (e.g. an "ICMP Port Unreachable" message, in the case of SIO_UDP_CONNRESET), it will cause the underlying UDP socket to throw an error. Confusingly, this can occur even on reads, if the same UDP socket is used to write a packet that triggers this response. The Go runtime disabled the SIO_UDP_CONNRESET behavior in 3114bd6, but did not change SIO_UDP_NETRESET–probably because that socket option isn't documented particularly well. Various other networking code seem to disable this behaviour, such as the Godot game engine (godotengine/godot#22332) and the Eclipse TCF agent (link below). Others appear to work around this by ignoring the error returned (anacrolix/dht#16, among others). For now, until it's clear whether this ends up in the upstream Go implementation or not, let's also disable the SIO_UDP_NETRESET in a similar manner to SIO_UDP_CONNRESET. Eclipse TCF agent: https://gitlab.eclipse.org/eclipse/tcf/tcf.agent/-/blob/master/agent/tcf/framework/mdep.c Updates #10976 Updates golang/go#68614 Signed-off-by: Andrew Dunham <andrew@du.nham.ca> Change-Id: I70a2f19855f8dec1bfb82e63f6d14fc4a22ed5c3	4 months ago
Jordan Whited	2f27319baf	wgengine/netstack: disable gVisor TCP GRO for Linux (#13138) ... A SIGSEGV was observed around packet merging logic in gVisor's GRO package. Updates tailscale/corp#22353 Signed-off-by: Jordan Whited <jordan@tailscale.com>	4 months ago
Brad Fitzpatrick	2dd71e64ac	wgengine/magicsock: log when a ReceiveFunc fails ... Updates #10976 Change-Id: I86d30151a25c7d42ed36e273fb207873f4acfdb4 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	4 months ago
Brad Fitzpatrick	a61825c7b8	cmd/tta, vnet: add host firewall, env var support, more tests ... In particular, tests showing that #3824 works. But that test doesn't actually work yet; it only gets a DERP connection. (why?) Updates #13038 Change-Id: Ie1fd1b6a38d4e90fae7e72a0b9a142a95f0b2e8f Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	4 months ago
Jordan Whited	a93dc6cdb1	wgengine/magicsock: refactor batchingUDPConn to batchingConn interface (#13042) ... This commit adds a batchingConn interface, and renames batchingUDPConn to linuxBatchingConn. tryUpgradeToBatchingConn() may return a platform- specific implementation of batchingConn. So far only a Linux implementation of this interface exists, but this refactor is being done in anticipation of a Windows implementation. Updates tailscale/corp#21874 Signed-off-by: Jordan Whited <jordan@tailscale.com>	4 months ago
Jordan Whited	25f0a3fc8f	wgengine/netstack: use build tags to exclude gVisor GRO importation on iOS (#13015) ... Updates tailscale/corp#22125 Signed-off-by: Jordan Whited <jordan@tailscale.com>	4 months ago
Maisem Ali	07e2487c1d	wgengine/capture: fix v6 field typo in wireshark dissector ... It was using a v4 field for a v6 address. Updates tailscale/corp#8020 Signed-off-by: Maisem Ali <maisem@tailscale.com>	4 months ago
Maisem Ali	a917718353	util/linuxfw: return nil interface not concrete type ... It was returning a nil `*iptablesRunner` instead of a nil `NetfilterRunner` interface which would then fail checks later. Fixes #13012 Signed-off-by: Maisem Ali <maisem@tailscale.com>	4 months ago
Jordan Whited	d9d9d525d9	wgengine/netstack: increase gVisor's TCP send and receive buffer sizes (#12994) ... This commit increases gVisor's TCP max send (4->6MiB) and receive (4->8MiB) buffer sizes on all platforms except iOS. These values are biased towards higher throughput on high bandwidth-delay product paths. The iperf3 results below demonstrate the effect of this commit between two Linux computers with i5-12400 CPUs. 100ms of RTT latency is introduced via Linux's traffic control network emulator queue discipline. The first set of results are from commit f0230ce prior to TCP buffer resizing. gVisor write direction: Test Complete. Summary Results: [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.00 sec 180 MBytes 151 Mbits/sec 0 sender [ 5] 0.00-10.10 sec 179 MBytes 149 Mbits/sec receiver gVisor read direction: Test Complete. Summary Results: [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.10 sec 337 MBytes 280 Mbits/sec 20 sender [ 5] 0.00-10.00 sec 323 MBytes 271 Mbits/sec receiver The second set of results are from this commit with increased TCP buffer sizes. gVisor write direction: [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.00 sec 297 MBytes 249 Mbits/sec 0 sender [ 5] 0.00-10.10 sec 297 MBytes 247 Mbits/sec receiver gVisor read direction: [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.10 sec 501 MBytes 416 Mbits/sec 17 sender [ 5] 0.00-10.00 sec 485 MBytes 407 Mbits/sec receiver Updates #9707 Updates tailscale/corp#22119 Signed-off-by: Jordan Whited <jordan@tailscale.com>	4 months ago
Andrew Dunham	9939374c48	wgengine/magicsock: use cloud metadata to get public IPs ... Updates #12774 Signed-off-by: Andrew Dunham <andrew@du.nham.ca> Change-Id: I1661b6a2da7966ab667b075894837afd96f4742f	4 months ago
Jordan Whited	f0230ce0b5	go.mod,net/tstun,wgengine/netstack: implement gVisor TCP GRO for Linux (#12921) ... This commit implements TCP GRO for packets being written to gVisor on Linux. Windows support will follow later. The wireguard-go dependency is updated in order to make use of newly exported IP checksum functions. gVisor is updated in order to make use of newly exported stack.PacketBuffer GRO logic. TCP throughput towards gVisor, i.e. TUN write direction, is dramatically improved as a result of this commit. Benchmarks show substantial improvement, sometimes as high as 2x. High bandwidth-delay product paths remain receive window limited, bottlenecked by gVisor's default TCP receive socket buffer size. This will be addressed in a follow-on commit. The iperf3 results below demonstrate the effect of this commit between two Linux computers with i5-12400 CPUs. There is roughly ~13us of round trip latency between them. The first result is from commit 57856fc without TCP GRO. Starting Test: protocol: TCP, 1 streams, 131072 byte blocks - - - - - - - - - - - - - - - - - - - - - - - - - Test Complete. Summary Results: [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.00 sec 4.77 GBytes 4.10 Gbits/sec 20 sender [ 5] 0.00-10.00 sec 4.77 GBytes 4.10 Gbits/sec receiver The second result is from this commit with TCP GRO. Starting Test: protocol: TCP, 1 streams, 131072 byte blocks - - - - - - - - - - - - - - - - - - - - - - - - - Test Complete. Summary Results: [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.00 sec 10.6 GBytes 9.14 Gbits/sec 20 sender [ 5] 0.00-10.00 sec 10.6 GBytes 9.14 Gbits/sec receiver Updates #6816 Signed-off-by: Jordan Whited <jordan@tailscale.com>	4 months ago
Jordan Whited	7bc2ddaedc	go.mod,net/tstun,wgengine/netstack: implement gVisor TCP GSO for Linux (#12869) ... This commit implements TCP GSO for packets being read from gVisor on Linux. Windows support will follow later. The wireguard-go dependency is updated in order to make use of newly exported GSO logic from its tun package. A new gVisor stack.LinkEndpoint implementation has been established (linkEndpoint) that is loosely modeled after its predecessor (channel.Endpoint). This new implementation supports GSO of monster TCP segments up to 64K in size, whereas channel.Endpoint only supports up to 32K. linkEndpoint will also be required for GRO, which will be implemented in a follow-on commit. TCP throughput from gVisor, i.e. TUN read direction, is dramatically improved as a result of this commit. Benchmarks show substantial improvement through a wide range of RTT and loss conditions, sometimes as high as 5x. The iperf3 results below demonstrate the effect of this commit between two Linux computers with i5-12400 CPUs. There is roughly ~13us of round trip latency between them. The first result is from commit 57856fc without TCP GSO. Starting Test: protocol: TCP, 1 streams, 131072 byte blocks - - - - - - - - - - - - - - - - - - - - - - - - - Test Complete. Summary Results: [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.00 sec 2.51 GBytes 2.15 Gbits/sec 154 sender [ 5] 0.00-10.00 sec 2.49 GBytes 2.14 Gbits/sec receiver The second result is from this commit with TCP GSO. Starting Test: protocol: TCP, 1 streams, 131072 byte blocks - - - - - - - - - - - - - - - - - - - - - - - - - Test Complete. Summary Results: [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.00 sec 12.6 GBytes 10.8 Gbits/sec 6 sender [ 5] 0.00-10.00 sec 12.6 GBytes 10.8 Gbits/sec receiver Updates #6816 Signed-off-by: Jordan Whited <jordan@tailscale.com>	4 months ago
Brad Fitzpatrick	cf97cff33b	wgengine/netstack: simplify netaddrIPFromNetstackIP ... Updates #cleanup Change-Id: I66878b08a75d44170460cbf33c895277c187bd8d Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	4 months ago