Tailscale's wireguard-go has a hybrid handshake,
Noise_IKhfs+psk2_25519+MLKEM768, which both peers must agree to use,
with no fallback to the classic one. A kernel-mode node without it
would be cut off from any peer that asked for it.
The initiation now may carry an ephemeral ML-KEM-768 encapsulation
key, sealed under the es key (with the static key then using nonce
1), and the response a ciphertext for it, sealed under the ee key,
whose shared secret is mixed into the chaining key before se. That's
per peer, with the new TSWGPEER_F_HYBRID flag. Hybrid retries carry
wireguard-go's retry bit (bit 31 of the type), which we accept, set and
echo, and which the socket's message classifier now allows.
Linux has no ML-KEM, and its SHA-3 is too new for the kernels we build
for, so src/mlkem.c is a FIPS 203 ML-KEM-768 with its own Keccak.
tests/mlkem checks it against Go's crypto/mlkem: the same keys from the
same seeds, each decapsulating the other's ciphertexts, and the same
implicit rejection. The natlab tests in tailscale.com check the
handshake against wireguard-go, in both roles.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: Ic524483d6cd27d36900a730553e1b21e4530c399
Handing a packet to a crypt worker, the tx worker or a NAPI kthread on
another CPU can wake a halted vCPU, which in a VM is a VM exit and a
trip through the host's scheduler. tswg's idle round trip time was
1.5-2x wireguard-go's in VMs because of it.
So when the crypt workers look idle, a packet that isn't part of a
batch is encrypted or decrypted where it is, after going on the peer's
in-order queue as before. To keep TCP ACKs under load from moving
crypto onto the CPUs taking the NIC's interrupts (8% off a subnet
router), it's limited to 8 packets per CPU per jiffy, and never done
for encryption from softirq (forwarding). And peers' NAPI runs in
softirq again, not kthreads, unless napi_threaded is set.
Idle RTT p50 is now 0.51ms in the lab (wireguard-go 0.60) and 0.75ms
on EC2 (0.97), with subnet router throughput unchanged at the same
NAPI setting. Threaded NAPI is still 4-5% faster for a CPU-bound
router.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: Ib5ebd1edcda1a515dac0ae4e55540bcfe52d1d7f
With real tailscaled on the router and 37 clients (a c6in.32xlarge, MTU
1280), kernel mode forwards 69.2 Gbps, 10x TUN mode's 6.9, at 0.74x its
CPU per GB, matching the bare tswg tunnel. Getting there found two
kernel mode bugs in tailscale.com (a private key race, and the device
left behind on exit), now fixed, and exercised kernel mode over IPv6.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I6af1beb0c0b3bd72eeb76d3340de9a977ba5f09c
The plain routing baseline used jumbo frames (165 Gbps). Clamped to MTU
1280, the same size packets as through the tunnels, it gets 88.9 Gbps
through the same c6in.32xlarge router with 37 clients, and tswg 70.9
(0.80x), wireguard-go 6.2, kernel WireGuard 8.5. So the platform's
per-packet costs limit routing at normal MTUs with or without a tunnel.
Also record that CPU figures before now came from /proc/stat, which on
Ubuntu's AWS kernels counts softirq work on otherwise idle CPUs as idle,
understating plain routing and kernel WireGuard most. With idle-state
residency accounting, tswg uses 0.76x wireguard-go's CPU per GB here.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I104bec537a1dcc2c1f460d287245e5ab448442c5
rx_zerocopy (writable at /sys/module/tswg/parameters/rx_zerocopy) makes
tswg_decrypt_gso decrypt data messages in place in the received packet's
pages, when nothing else can see them, and hand out packets that
reference those pages, rather than copy each message into a new packet.
It's off because it doesn't pay: on a CPU-bound subnet router it was 7%
slower, scatterlist decryption costing more than linear decryption of a
cache-hot copy, and on a 200 Gbps c6in.32xlarge router, where the NIC
interrupt CPUs' page allocation is the limit, it took tswg's own page
allocations out but was still no faster (62.5-63.3 Gbps against
63.4-64.7) at 10% more CPU per byte, since ENA's page per received
packet is what binds. It stays for NICs and loads where it might.
The copying path, unchanged, moves into tswg_decrypt_message_copy.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: Ieb6aa3531f869dd47d768fe367c424fbd22aa3f1
On a c6in.32xlarge router (200 Gbps), with 60 clients at MTU 1280, tswg
forwards 74 Gbps, 12x wireguard-go (6.2) and 9x kernel WireGuard (8.1,
with all 128 CPUs busy). There, with no EC2 allowance exceeded and the
router at 45% CPU, the limit is the NIC's interrupt CPUs, which spend
much of their time on the page allocator's zone lock (ENA allocates a
page per received packet, freed on other CPUs). RPS, bigger per-CPU page
lists and capping GRO for hardware TSO didn't help.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I61acb155937b49375c23e584bad8c98a3d6f5f66
It has the current subnet router and point-to-point results against
wireguard-go and kernel WireGuard in tables, with tswg/wireguard-go
throughput and CPU-per-GB ratios, to keep up to date as results come in.
The TODO list notes that c6in instances are now allowed for the next
subnet router runs.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I1feefcf5d91d4baaf0bc3408d779fbb168643869
With 24 clients at MTU 1280, tswg forwards 40 Gbps through a c7i.48xlarge
(5.6x wireguard-go, 3x kernel WireGuard), using about 22 of its 192 CPUs,
and is then limited by EC2's packet rate allowance for the instance.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: Ifdb78f7d85d182a89a42c2f0f9f77bd14344fb59
ec2.sh gets "resize TYPE N..." (for stopped instances) and "start N..."
(some instances), and numbers new instances after all existing ones,
stopped ones included.
With a c7i.48xlarge router (50 Gbps) and 8 clients, tswg forwards 31.5
Gbps at MTU 1280 (wireguard-go 7.2, kernel WireGuard 12.3 using 137 of
192 CPUs), limited by EC2's packet rate per NIC receive queue (about
715k pps, 7.3 Gbps of tunnel packets) with only 8 tunnel flows to spread
over 16 queues; and at line rate with a jumbo tunnel MTU, a data point
only, since MTU 1280 is the target.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I70744651f3385343bf8890607911d9b4e0f70620
With 20 clients (160 flows), tswg as the subnet router forwards 2.2x (4
CPUs) and 3.6x (8 CPUs) what wireguard-go does, at 38-43% less CPU per
byte. wireguard-go stays at about 6.4 Gbps however many peers or CPUs it
has, so its limit is global, most likely its one UDP-reading goroutine.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I9274a49b47cdbdf9af06f1f9a1f6c8d04f126442
Profiling a CPU-bound tswg subnet router (EC2, 4 CPUs) found two costs
in decrypting merged packets:
Copying. Each message was copied out of the received packet and then
decrypted in place in the copy, two passes over it. When the message is
contiguous in the received packet (in its head or one page frag, as a
message received in one datagram is), decrypt straight from it into the
new packet instead (tswg_contig).
The Poly1305 key. lib/crypto makes it from a separate 32 bytes of
ChaCha20 keystream (block 0), and x86's ChaCha20 does anything up to a
block with its portable code: a scalar ChaCha20 block per message, 5.5%
of the router's CPU. tswg_aead.h has our own ChaCha20-Poly1305 for data
messages, from the library's primitives. When decrypting from one buffer
to another, it runs ChaCha20 from block 0 over the 64 bytes before the
message too (packet headers in the source, headroom in the destination),
and recovers block 0's keystream, the Poly1305 key, as the two XORed:
one SIMD pass per message (tswg_aead_decrypt_ahead). Otherwise, and for
encryption, it asks for two blocks of keystream, which takes the SIMD
path, and sets the counter back.
The 4-CPU router, 4 clients with 16 flows each, medians of 3 runs: 14.1
to 14.9 Gbps, 2.30 to 2.18 CPU-seconds per GB (wireguard-go: 6.6 Gbps,
3.36). Interop with wireguard-go is unchanged; KASAN is quiet.
Also tried, and not kept: zero-copy receive (decrypting in place in the
received pages, 7% worse), and a page pool for decrypted packets (its
pages didn't come back in time to save allocations).
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I98456d17b82d5ef640c55ca1a7fa7a018dc5a279
gokrazy-debug-kernel.sh hashed sha256sum's output for its cache key,
which includes the config's random mktemp path, so every run built
another 3.5GB debug kernel; seven of them filled the disk. Hash the
files' contents instead.
ec2.sh sorted hosts by name, which put bradfitz-tswg-10 before
bradfitz-tswg-2; sort by number.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: Ia758dae0077c54a2ef8f6112ae3c43d6604f66e1
subnetbench.sh runs tailscale.com's TestLeanSubnetRouter on the EC2
instances for each router CPU count given. ec2.sh now turns off source
and destination checks on its instances, so that one can forward as a
subnet router does, and installs nftables (for the router's masquerade).
DESIGN.md records the results: with 4 clients, tswg as the subnet
router forwards 2.1x (4 CPUs) to 3.2x (8 CPUs) what wireguard-go does,
which uses about 3 CPUs however many it has, at the least CPU per byte
of any tunnel, and reaches the 25 Gbps link at 16 CPUs. Zero-copy
receive was tried there and was 7% worse, so isn't in.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I6b1da8bb8eaa17f9b20cbe36896426411b8b905f
The QEMU lab runs on a host that's itself a VM, which distorted latency
in particular. ec2.sh manages benchmark instances in the sandbox AWS
account (us-west-2; the org's SCP allows only some regions and requires
IMDSv2): a key pair, a security group open to this host and between its
members, a cluster placement group, and Ubuntu 24.04 instances with IPv6
that install the build tools for their own kernel. Everything is named
bradfitz-tswg... and tagged owner=claude-tswg. It writes the hosts file
that tailscale.com's lean harness takes in remote mode.
To keep costs down, ec2-idle-stop.sh, run every 5 minutes from a systemd
user timer on the always-on dev host, stops the instances once nothing
has touched ~/.cache/tslinux/ec2/last-used for 30 minutes. ec2.sh and the
harness touch it while in use, and "ec2.sh start" brings stopped
instances back.
First EC2 results (c7i.xlarge pair) are in DESIGN.md: throughput is
capped by EC2 at about 7.25 Gbps for both tswg and wireguard-go, with
tswg's receiver using up to 17% less CPU per byte; 1.6x the small
packets; a quarter of the RTT under load at p50; twice the idle RTT.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: Ibe770a30bea31cb79ac514aeb545f468576b5f05
DESIGN.md's Performance section now describes what tswg changes in the
data path and why, the results against wireguard-go (parity to 1.05x on
bulk TCP at 6-10% less CPU per byte, much better small-packet delivery,
worse idle and p99-under-load latency in VMs) and the tools for this
work. The TODO list gets the open performance items.
leancompare.sh runs the lean harness's interleaved A/B comparison for
RTT and TCP with 1 and 4 streams.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: Icf8c7ca8bb3711a8172284382af9ce8c3e25ff3d
RTT under a TCP load had grown to about 11ms at p99 (wireguard-go: 3.7).
Instrumenting the queues found three causes:
Backpressure. Upstream encrypts in place, so a packet stays charged to
its socket (for TCP, counted by its small queues, TSQ) until the NIC has
sent it, and TCP is held to the NIC's pace. Our encrypted packets are
new (and earlier ones were orphaned), so TCP could fill the NIC's qdisc
as fast as we encrypt: on a slow uplink with a FIFO qdisc, that would be
massive bufferbloat. Now each encrypted packet takes over its packets'
socket ownership and charge (tswg_move_owner), for destructors that
just uncharge truesize (sock_wfree, and TCP's), settled to its real
size (segments' truesize is about twice their length). TSQ then allowed
TCP only about 1ms in flight, which cost 15-20% of throughput, so
sockets sending over tswg get sk_pacing_shift 9 (about 2ms), as Wi-Fi
drivers set 8 for their aggregation.
The tx worker. Upstream pins a peer's tx worker to one CPU by peer ID.
When that's the CPU taking the NIC's interrupts, softirq work starves
it, and encrypted packets waited up to 20ms to be sent. Queue it on the
CPU that finished encrypting instead (also saving an IPI).
Queue length. The peer queue's limit counted entries, but a GRO entry
is up to 64 packets; count packets, with a limit of 2048.
Medians of interleaved runs in the lean harness, tswg vs. wireguard-go:
TCP 1 stream 10.1 vs. 9.7 Gbps, 4 streams 12.3 vs. 10.8; RTT under load
p50 1.7 vs. 1.7ms. p99 under load is still about 10ms on this testbed:
at full load, the receiving VM's workers sometimes don't run for tens of
milliseconds. That's next to investigate.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: Ief80aac9a1e443727c89b990c86c7c8f71690954
The encrypt workers freed each list's original packets as soon as they
were encrypted. For TCP, freeing releases the socket's small queues
accounting (tcp_wfree), whose TSQ handlers then ran on all the encrypt
workers' CPUs at once and contended for the socket lock (about 5% of
each sender CPU in queued_spin_lock_slowpath). Free them instead on the
peer's tx worker once their encrypted packets are sent, which is also
when upstream releases them.
Bulk TCP in the lean harness (medians of interleaved runs): 1 stream
10.4 Gbps (wireguard-go 10.0), 4 streams 12.7 Gbps (wireguard-go 11.1).
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I904dea12ccfd7bc0534d616eb319fd692eefd832
Our UDP sockets now have gro_receive/gro_complete callbacks, as UDP
tunnels do. They merge consecutive data messages of a flow that use the
same receiving key and are all the same size but perhaps the last, and
UDP hands the merged packet (SKB_GSO_UDP_TUNNEL) to encap_rcv whole. It
goes through the decrypt ring and the peer's queue as one entry: one
worker decrypts each message into a new packet (copying it out and
decrypting in place there, rather than linearizing in place), and the
peer's NAPI poll delivers them in order. So the IP/UDP input path, the
ring and queue handoffs and the worker kick are paid once per merged
packet rather than per packet. Upstream gets no outer GRO at all.
skb_gro_receive isn't exported to modules, so socket.c has a copy
(tswg_gro_merge). gro_receive_network_flush and udp_tunnel_update_gro_rcv
are newer than 6.8, so they're probed. The latter matters: 7.x may make
the socket's gro_receive a static call, which detaching a user socket
must undo, or it would outlive the module.
Bulk TCP in the lean harness, now with guest USO off (so the receiving
VM sees individual datagrams, as from a physical network): about 9.8
Gbps, against 6.5 before, and within a few percent of wireguard-go
(about 10.0), at about 18% less receiver CPU. Making workers poll the
ring briefly before going idle (to save wakeup IPIs) was tried and
didn't help, so isn't included.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: Ic22482120c6ce75b9a4d896f16b6cf77aca28b4c
gokrazy-debug-kernel.sh builds the gokrazy kernel package's kernel again
with KASAN, lockdep, DEBUG_VM and DEBUG_NET, for tailscale.com's lean
harness to boot (TSWG_DEBUG_KERNEL=1). build-for-gokrazy.sh builds
against it when given TSWG_KDIR, since those options change kernel data
structures.
Its first find: our slab caches had upstream WireGuard's names (via
KMEM_CACHE), which clash with wireguard.ko or a built-in WireGuard, as
gokrazy has, and make slab_common warn at load. Give them tswg_ names.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I2e7d5448bcf8b2fb68810486f7e5d4bf2af4558d
Three changes to the data path, from profiling bulk TCP in the lean
harness:
Encryption. Upstream encrypts each packet in place, which first means
making it linear and writable (skb_cow_data, usually a copy), and the
previous commit then chained the encrypted packets as a GSO frag_list,
which segmentation copied again (skb_segment must unshare each clone's
head to write its headers). Now the encrypt worker copies each inner
packet (finishing its checksum on the way, if needed) into pages of its
own, encrypts it there, and gathers each run of equal-size messages into
a UDP GSO packet whose payload is those pages. Segmenting that needs only
page references, and a device with UDP segmentation offload takes it
whole: virtio with USO now sends 64KB superpackets. Sender CPU at the same
throughput went from about 65% to 43% per vCPU.
Worker kicks. Instead of kicking a crypt worker per packet (or per 16),
kick only when no worker is draining the ring, or when the backlog shows
the busy ones aren't keeping up, and prefer other CPUs than the one
taking packets from the NIC. A tswg_busy flag, paired with the ring
under full barriers, closes the race with a worker going idle.
Queue length. The peer's in-order queue overflowed at 1024 packets (1.5ms
at 7 Gbps) whenever a decrypt worker was held up, and TCP took those
silent drops (0.2% of packets) as congestion. Allow 4096 entries, and
count drops in rx_dropped and rx_fifo_errors, which upstream doesn't.
The receiver is now the bottleneck (about 6.5 Gbps): UDP packets arriving
coalesced are segmented back, with software checksums, before reaching
encap_rcv. That's next.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: Ib8d9cf9e5541b665e06d57a16b46f653d49bc3ea
Profiling bulk TCP in the lean harness (tailscale.com's tstest/tswgbench)
showed kernel WireGuard serializing on two per-packet costs:
On receive, the single softirq CPU spent half its time in queue_work_on:
upstream kicks the next CPU's crypt worker for every packet, round robin,
which usually means waking it with an IPI (a VM exit in a VM). All the
workers drain the same ring, so instead stay on one CPU for 16 packets at
a time, and skip queue_work_on when the work is already pending (a read,
rather than queue_work_on's atomic test-and-set on a cache line the
worker writes). The same pending check applies to the per-peer transmit
work.
On send, the per-peer transmit worker sent each encrypted packet through
the UDP/IP stack and the device on its own, which with virtio is a device
kick, a VM exit, per packet. Packets from one GSO packet encrypt to the
same size (WireGuard pads to 16 bytes) except perhaps the last, so chain
each such run, without copying, as the frag_list of its first packet and
send it as one UDP GSO (SKB_GSO_UDP_L4) packet, with its checksum set up
as for a UDP_SEGMENT socket. The stack (or a capable device) segments it,
and the segments go to the device in one batch.
The coalesced packets are orphaned first: segmenting a frag_list of
packets still owned by the inner TCP socket uncharged the socket more
than it was charged (refcount underflow in tcp_wfree). wireguard-go has
the same effect, since the TUN device frees the inner packet once read.
Bulk TCP over tswg goes from 3.6 to 5.8 Gbps (wireguard-go: about 10.5).
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I649a7f1790c35d1af584d2c14f588195232d5ef4
leanbench.sh runs tailscale.com's new lean data plane benchmark, which
compares wireguard-go, tswg, the kernel's WireGuard and no tunnel between
two QEMU VMs on a host bridge with vhost-net, without natlab, over IPv4
and IPv6. With -smoke it runs the quick correctness test instead.
The results overturn the natlab ones, which natlab's userspace network
had capped: kernel WireGuard (and tswg, identically) gets about a third
of wireguard-go's bulk TCP throughput, at 1.5x to 2.5x the CPU per byte,
without being CPU-bound. So outer UDP GSO/GRO and finding where the
kernel serializes are now the critical performance work.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I2a24de8d82370fc8dfb6a9af0971da20831ea307
bench.sh runs tailscale.com's TestTSWGBenchmark, which compares
tailscaled in kernel mode against tailscaled in TUN mode (wireguard-go)
on the same pair of natlab gokrazy VMs, and saves the results as JSON.
Record the first results: bulk TCP throughput and CPU per byte are about
the same (natlab caps throughput, and the kernel lacks outer UDP
GSO/GRO), while the kernel receives about 1.7x the small packets per
second at about 30% less CPU per packet and roughly halves latency under
load. Also record the plan for a lean benchmark harness without natlab,
whose userspace switch is expensive, and a flake seen when both peers
restart at once.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I69e696817f98e685576a30d766c42740d4d04810
Record how the kernel data plane now runs without polling: config
changes trigger full syncs, magicsock's kernel path hook (with a trust
expiry timer for idle sessions) triggers per-peer endpoint updates, and
the module's peer events drive disco key advertisements and activity
feedback, all applied by one goroutine.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I8c555a623e8bba7cf7561fe94e4bc9d17fd40003
tailscaled was dumping the device every second to notice new sessions
(to send its disco key advertisement, as wireguard-go does) and traffic
on direct paths (to keep magicsock heartbeating active sessions).
Instead, multicast TSWG_CMD_PEER_EVENT messages on a new "events"
genetlink group, in the device's network namespace: one when a new
session is established (from wg_timers_session_derived), and TX and RX
activity notices, each rate limited to one per peer every 2 seconds with
a single comparison on the packet path. Messages are only built when
someone is listening.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I08bdaa23a9f9b01a0bb3a4404296cae5961cde86
Record how tailscaled exchanges TSMP and its other generated packets with
peers in kernel mode (raw sockets on the tswg device, and a tstun
outbound inject hook), how it emulates wireguard-go's disco key
advertisement on session establishment, how it clears the kernel session
when a peer restarts, and that wireguard-go must never get peers in
kernel mode. Mark the packet filter and MagicDNS as deliberately
deferred.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I632c5a75b1c2e1dc74680faaa34bbd799e52fe74
When tailscaled learns that a peer restarted (its disco key changed), the
peer has lost its WireGuard keys, but the kernel would keep sending with
the dead session until WireGuard's own timers start a new handshake about
15 seconds later.
Add TSWGPEER_A_CLEAR_SESSION, an NLA_FLAG in the nested WGPEER_A_TSWG
attribute, which forgets the peer's keypairs and any handshake in
progress, as wg_pm_notification does for every peer before suspend, and
resets the handshake rate limit, so that the next packet to the peer is
staged and starts a new handshake right away.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: Ia9821a47e6c8114bbabd81d8d5a2be528295e74f
DERP via 127.0.0.0/8 loopback endpoints turned out to need no kernel
changes; record how the bridge works (including peer relays and the
untrusted-path spray) and how tailscaled now drives tswg on the
tailscale.com bradfitz/tswg branch, with its known gaps, the biggest
being that kernel mode has no packet filter yet.
Also record that Tailscale's wireguard-go is about to start using a
hybrid post-quantum handshake, which tswg doesn't implement, so the
assumption that tswg is wire-compatible with every Tailscale peer no
longer holds, and the options for dealing with that.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I2447db2494bfd1a572f8ce4d093a42a1949aa3d4
Upstream WireGuard moves a peer's endpoint to the source of each
authenticated packet. In Tailscale, disco picks the path, so that would
fight it: a packet that arrives via DERP would steal the endpoint from a
good direct path.
Add TSWGPEER_F_NO_ROAM, set through TSWGPEER_A_FLAGS nested in a new
WGPEER_A_TSWG attribute (numbered 64, like WGDEVICE_A_TSWG) and reported
in dumps. With it set, the endpoint learned from handshake initiations,
handshake responses and data packets is ignored once the peer has an
endpoint, and only userspace changes it. A peer with no endpoint yet
still gets one from its first authenticated packet, so it stays
reachable if it contacts us before userspace has configured one.
The gate, tswg_peer_may_roam, sits at the two learned-endpoint paths
rather than in wg_socket_set_peer_endpoint, which netlink also uses for
explicit updates. The flag is applied before WGPEER_A_ENDPOINT in the
same request, so pinning and setting an endpoint are one step.
Tested by TestTSWG{Gokrazy,Ubuntu}NoRoam in tailscale.com's natlab
vmtests, which fail as expected with the gate disabled.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I621a6e86d3947e154c4cfa582aa845bb49d42b2f
Performance is why we're doing this, so say so in the Goal section and
add a Performance section: what we inherit from upstream WireGuard (inner
GSO on send, inner GRO via NAPI on receive, parallel crypto), what's
missing (outer UDP GSO on send and outer UDP GRO on receive, the kernel
analogues of what makes wireguard-go fast), why tailscaled shouldn't set
UDP_GRO on the shared socket in kernel mode, and how to measure.
Group the performance work at the top of the TODO list.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: If844d30a81f5f2fe52bc2fbf2b31d7bbbb9359c7
Tailscale needs disco, STUN and WireGuard to share one UDP 4-tuple, so
that the NAT mappings disco discovers are the ones WireGuard uses. So let
tailscaled create and bind the socket itself and hand its fd to the
device, as L2TP allows for its tunnel sockets.
The new TSWGDEVICE_A_SOCK_FD4 and TSWGDEVICE_A_SOCK_FD6 attributes, nested
in a new WGDEVICE_A_TSWG attribute numbered 64 to stay clear of future
upstream attributes, take the fd of a bound UDP socket (IPV6_V6ONLY for
IPv6). The device hooks the socket's UDP encap_rcv, consumes WireGuard
messages (a first byte of 1 through 4 and three zero bytes), and returns
everything else to the normal UDP path for userspace to read. Userspace
can send its own datagrams on the socket as usual.
The device keeps a reference to the struct sock, not the file, and drops
the socket from its encap_destroy hook when userspace closes it, so the
socket lives exactly as long as userspace's fd. Passing -1 detaches the
socket, restoring it to plain UDP, and the device goes back to creating
its own sockets. User sockets persist across ifdown/ifup, and changing
the listen port while one is attached fails with EBUSY. Dumps report
which families have a user socket in TSWGDEVICE_A_FLAGS.
A global mutex serializes attach and detach, including detach from
encap_destroy, which runs from the user's close() with none of our locks
held.
Tested by TestTSWGGokrazySharedSocket (7.2.8) and
TestTSWGUbuntuSharedSocket (6.8) in tailscale.com's natlab vmtests.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I2e7a97952ced03df4bc954ba398a49daf9128d41
A gokrazy bot bumps github.com/gokrazy/kernel.amd64 to new kernel
releases and a tailscale.com cron bot bumps that go.mod dependency, so
the gokrazy vmtests follow upstream kernels automatically. Record that,
and that a failure after such a bump most likely means kernel API churn
needing a new compat shim.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I12428a124604edb65a61f101b72c87caed6c9332
Record the decision to test primarily on natlab's gokrazy image, which
boots a recent upstream kernel and has no userspace tools, as in the end
state where tailscaled drives the kernel entirely. Describe how its
module build tree is recreated and how the test agent loads and
configures tswg. Keep Ubuntu 24.04 as a secondary guest for distro and
DKMS-style coverage.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I89714b58c944be4f373f23471e3aeef11942b1f6
The natlab gokrazy image runs a recent upstream kernel (7.2.8 today) that
gokrazy rebuilds and tailscale.com bumps regularly, which makes it a
better primary test target than a distro kernel: it keeps us honest
against new kernel APIs, and gokrazy has no userspace tools, as in the
end state where tailscaled drives the kernel entirely.
gokrazy kernels are kernel.org releases plus the patches in
_build/series, built with an embedded config and without MODVERSIONS or
module signing. gokrazy-kdir.sh downloads the release, applies the
patches, installs the config pulled out of vmlinuz with the kernel's
extract-ikconfig (vendored), and runs modules_prepare, caching the
result by release and config hash. It takes about 20 seconds the first
time. The only config differences after olddefconfig with the host's gcc
13 (gokrazy uses Debian 12's gcc 12) are toolchain capability bits.
Factor the generic part of build-for-ubuntu.sh into build.sh, which now
copies only sources (not objects from an in-place build in src/), and
sets KBUILD_MODPOST_WARN=1 for trees without a Module.symvers. Add
build-for-gokrazy.sh.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: Id28d9c013bdfaa7cec28aeff846f7f6087e25518
Document the feature-probe compat framework, that the Go netlink client
and tswgctl live in tailscale.com on the bradfitz/tswg branch, and how
the natlab vmtests build, install and exercise tswg.ko. Update the TODO
list: TestTSWGManualTunnel now passes against Ubuntu 24.04's stock
6.8.0-106-generic kernel.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: Ie279a07e251d1ee81cdc91478f3eec9d0dd46afb
build-for-ubuntu.sh builds tswg.ko for a given Ubuntu kernel release in a
private copy of src/, so test builds for different guest kernels don't
share object files with each other or with a developer's in-place build.
The Makefile's SRC is now overridable to allow that.
vmtest.sh runs the tswg natlab vmtests, which live in a tailscale.com
checkout (on the bradfitz/tswg branch) because they use its natlab
framework and the Go tswg netlink client there.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I00eb5cbf8df5a203d6d39d541779060db74da9a1
The fork tracks the newest upstream WireGuard (7.4-rc1), but has to build
on distro kernels. Our first target is the natlab Ubuntu 24.04 guest
(6.8.0-106-generic) and the dev host is on 7.0.
Rather than LINUX_VERSION_CODE checks, which lie on heavily backported
kernels such as RHEL's, scripts/probe.sh compile-tests each
src/compat/probes/HAVE_*.c against the target kernel's build directory
(in parallel, about one second in total) and writes
src/compat/generated/autoconf.h. Kbuild force-includes src/compat/compat.h,
which shims whatever is missing: udp_get_len_short, the struct sock
udp_tunnel APIs, the ipcb_flags xmit arguments, the blake2s_ctx API,
WQ_PERCPU, netif_threaded_enable, rtnl_newlink_params, sockaddr_inet, the
kmalloc_obj family and timer_container_of.
Three spots in device.c can't be shimmed from a header (the lltx bit, the
old rtnl newlink signature, and pernet exit_rtnl) and get small #ifdefs
marked TSWG_COMPAT. device.c also always sets ndo_get_stats64 to
dev_get_tstats64, since older kernels don't fold tstats in
dev_get_stats(). That is equivalent on new kernels.
tswg.ko now builds against both 6.8.0-106-generic and 7.0.0-1009-aws.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I0ed279318b4f623688d1f178a9b8fbcc0b585219
Copy upstream/drivers/net/wireguard into src and rename only the
user-visible identifiers: the module (and thus the rtnl link kind and
device type, which come from KBUILD_MODNAME) is now tswg, and the
genetlink family is "tswg". The uapi header moves to
src/include/uapi/linux/tswg.h so it can't collide with the kernel's.
Internal wg_ symbols keep their names to keep upstream merges easy; they
aren't exported, so tswg.ko can coexist with wireguard.ko.
Add an out-of-tree Kbuild and top-level Makefile, plus
scripts/ubuntu-headers.sh, which fetches and unpacks an Ubuntu kernel's
headers so we can build for the natlab Ubuntu guest without installing
anything.
This still only builds against 7.4-era kernels; compat shims follow.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: Ia36ef82a53efbbcdb301ed069785162372efda55
Record the decisions made so far: an out-of-tree C fork of in-tree
WireGuard (no Rust, no upstreaming), a shared UDP socket passed in from
tailscaled, explicit endpoints with auto-roaming disabled, DERP via
127.0.0.0/8 loopback endpoints, and userspace spray. Also describe the
natlab vmtest testbed and the TODO list.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: Ie18d356974a0a0d8fce19674b590861144be8cea
The tswg kernel module, under kernel/linux, is derived from Linux's WireGuard,
and so is GPL-2.0, unlike the rest of this repository.
Updates #426
Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I8cab6152232f771364a06d79ecd8b406f3ef01dc