Commit Graph

163 Commits (c3994fd77ce4e92d96cd75ffb62d324141ab88f5)

Author SHA1 Message Date
David Anderson f582eeabd1 wgengine/magicsock: add a test for active path discovery.
Uses natlab only, because the point of this active discovery test is going to be
that it should get through a lot of obstacles.

Signed-off-by: David Anderson <danderson@tailscale.com>
4 years ago
Brad Fitzpatrick 37903a9056 wgengine/magicsock: fix occasional deadlock on Conn.Close on c.derpStarted
The deadlock was:

* Conn.Close was called, which acquired c.mu
* Then this goroutine scheduled:

    if firstDerp {
        startGate = c.derpStarted
        go func() {
            dc.Connect(ctx)
            close(c.derpStarted)
        }()
    }

* The getRegion hook for that derphttp.Client then ran, which also
  tries to acquire c.mu.

This change makes that hook first see if we're already in a closing
state and then it can pretend that region doesn't exist.
4 years ago
Brad Fitzpatrick fe68841dc7 wgengine/magicsock: log better with less spam on transition to stopped state
Required a minor test update too, which now needs a private key to get far
enough to test the thing being tested.
4 years ago
Brad Fitzpatrick e298327ba8 wgengine/magicsock: remove overkill, slow reflect.DeepEqual of NetworkMap
No need to allocate or compare all the fields we don't care about.
4 years ago
Brad Fitzpatrick 16a9cfe2f4 wgengine: configure wireguard peers lazily, as needed
wireguard-go uses 3 goroutines per peer (with reasonably large stacks
& buffers).

Rather than tell wireguard-go about all our peers, only tell it about
peers we're actively communicating with. That means we need hooks into
magicsock's packet receiving path and tstun's packet sending path to
lazily create a wireguard peer on demand from the network map.

This frees up lots of memory for iOS (where we have almost nothing
left for larger domains with many users).

We should ideally do this in wireguard-go itself one day, but that'd
be a pretty big change.

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
4 years ago
Brad Fitzpatrick 5066b824a6 wgengine/magicsock: don't log about disco ping timeouts if we have a working address
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
4 years ago
Brad Fitzpatrick c06d2a8513 wgengine/magicsock: fix typo in comment 4 years ago
Brad Fitzpatrick bf195cd3d8 wgengine/magicsock: reduce log verbosity of discovery messages
Don't log heartbeat pings & pongs. Track the reason for pings and then
only log the ping/pong traffic if it was for initial path discovery.
4 years ago
Brad Fitzpatrick 10ac066013 all: fix vet warnings 4 years ago
Brad Fitzpatrick d74c9aa95b wgengine/magicsock: update comment, fix earlier commit
891898525c had a continue that meant the didCopy synchronization never ran.
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
4 years ago
Brad Fitzpatrick c976264bd1 wgengine/magicsock: gofmt 4 years ago
Dmytro Shynkevych f3e2b65637
wgengine/magicsock: time.Sleep -> time.After
Signed-off-by: Dmytro Shynkevych <dmytro@tailscale.com>
4 years ago
Dmytro Shynkevych 380ee76d00
wgengine/magicsock: make time.Sleep in runDerpReader respect cancellation.
Before this patch, the 250ms sleep would not be interrupted by context cancellation,
which would result in the goroutine sometimes lingering in tests (100ms grace period).

Signed-off-by: Dmytro Shynkevych <dmytro@tailscale.com>
4 years ago
Dmytro Shynkevych 891898525c
wgengine/magicsock: make receive from didCopy respect cancellation.
Very rarely, cancellation occurs between a successful send on derpRecvCh
and a call to copyBuf on the receiving side.
Without this patch, this situation results in <-copyBuf blocking indefinitely.

Signed-off-by: Dmytro Shynkevych <dmytro@tailscale.com>
4 years ago
Brad Fitzpatrick a2267aae99 wgengine: only launch pingers for peers predating the discovery protocol
Peers advertising a discovery key know how to speak the discovery
protocol and do their own heartbeats to get through NATs and keep NATs
open. No need for the pinger except for with legacy peers.
4 years ago
Dmytro Shynkevych 2f15894a10 wgengine/magicsock: wait for derphttp client goroutine to exit
Signed-off-by: Dmytro Shynkevych <dmytro@tailscale.com>
4 years ago
Brad Fitzpatrick 6c74065053 wgengine/magicsock, tstest/natlab: start hooking up natlab to magicsock
Also adds ephemeral port support to natlab.

Work in progress.

Pairing with @danderson.
4 years ago
Brad Fitzpatrick bd59bba8e6 wgengine/magicsock: stop discoEndpoint timers on Close
And add some defensive early returns on c.closed.
4 years ago
Brad Fitzpatrick de875a4d87 wgengine/magicsock: remove DisableSTUNForTesting 4 years ago
Brad Fitzpatrick 5c6d8e3053 netcheck, tailcfg, interfaces, magicsock: survey UPnP, NAT-PMP, PCP
Don't do anything with UPnP, NAT-PMP, PCP yet, but see how common they
are in the wild.

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
4 years ago
Brad Fitzpatrick 6196b7e658 wgengine/magicsock: change API to not permit disco key changes
Generate the disco key ourselves and give out the public half instead.

Fixes #525
4 years ago
Brad Fitzpatrick 5132edacf7 wgengine/magicsock: fix data race from undocumented wireguard-go requirement
Endpoints need to be Stringers apparently.

Fixes tailscale/corp#422
4 years ago
Brad Fitzpatrick 630379a1d0 cmd/tailscale: add tailscale status region name, last write, consistently star
There's a lot of confusion around what tailscale status shows, so make it better:
show region names, last write time, and put stars around DERP too if active.

Now stars are always present if activity, and always somewhere.
4 years ago
Brad Fitzpatrick 9a8700b02a wgengine/magicsock: add discoEndpoint heartbeat
Updates #483
4 years ago
Brad Fitzpatrick 9f930ef2bf wgengine/magicsock: remove the discoEndpoint.timers map
It ended up being more complicated than it was worth.
4 years ago
Brad Fitzpatrick f5f3885b5b wgengine/magicsock: bunch of misc discovery path cleanups
* fix tailscale status for peers using discovery
* as part of that, pull out disco address selection into reusable
  and testable discoEndpoint.addrForSendLocked
* truncate ping/pong logged hex txids in half to eliminate noise
* move a bunch of random time constants into named constants
  with docs
* track a history of per-endpoint pong replies for future use &
  status display
* add "send" and " got" prefix to discovery message logging
  immediately before the frame type so it's easier to read than
  searching for the "<-" or "->" arrows earlier in the line; but keep
  those as the more reasily machine readable part for later.

Updates #483
4 years ago
Brad Fitzpatrick 6c70cf7222 wgengine/magicsock: stop ping timeout timer on pong receipt, misc log cleanup
Updates #483
4 years ago
Brad Fitzpatrick c52905abaa wgengine/magicsock: log less on no-op disco route switches
Also, renew trustBestAddrUntil even if latency isn't better.
4 years ago
Brad Fitzpatrick 0f0ed3dca0 wgengine/magicsock: clean up discovery logging
Updates #483
4 years ago
Brad Fitzpatrick 056fbee4ef wgengine/magicsock: add TS_DEBUG_OMIT_LOCAL_ADDRS knob to force STUN use only
For debugging.
4 years ago
Brad Fitzpatrick e03cc2ef57 wgengine/magicsock: populate discoOfAddr upon receiving ping frames
Updates #483
4 years ago
Brad Fitzpatrick 275a20f817 wgengine/magicsock: keep discoOfAddr populated, use it for findEndpoint
Update the mapping from ip:port to discokey, so when we retrieve a
packet from the network, we can find the same conn.Endpoint that we
gave to wireguard-go previously, without making it think we've
roamed. (We did, but we're not using its roaming.)

Updates #483
4 years ago
Brad Fitzpatrick 77e89c4a72 wgengine/magicsock: handle CallMeMaybe discovery mesages
Roughly feature complete now. Testing and polish remains.

Updates #483
4 years ago
Brad Fitzpatrick 710ee88e94 wgengine/magicsock: add timeout on discovery pings, clean up state
Updates #483
4 years ago
Brad Fitzpatrick 77d3ef36f4 wgengine/magicsock: hook up discovery messages, upgrade to LAN works
Ping messages now go out somewhat regularly, pong replies are sent,
and pong replies are now partially handled enough to upgrade off DERP
to LAN.

CallMeMaybe packets are sent & received over DERP, but aren't yet
handled. That's next (and regular maintenance timers), and then WAN
should work.

Updates #483
4 years ago
Brad Fitzpatrick 9b8ca219a1 wgengine/magicsock: remove allocs in UDP write, use new netaddr.PutUDPAddr
The allocs were only introduced yesterday with a TODO. Now they're gone again.
4 years ago
Brad Fitzpatrick 7b3c0bb7f6 wgengine/magicsock: fix crash reading DERP packet
Starting at yesterday's e96f22e560 (convering some UDPAddrs to
IPPorts), Conn.ReceiveIPv4 could return a nil addr, which would make
its way through wireguard-go and blow up later. The DERP read path
wasn't initializing the addr result parameter any more, and wgRecvAddr
wasn't checking it either.

Fixes #515
4 years ago
Brad Fitzpatrick 47b4a19786 wgengine/magicsock: use netaddr.ParseIPPort instead of net.ResolveUDPAddr 4 years ago
Brad Fitzpatrick f7124c7f06 wgengine/magicsock: start of discoEndpoint state tracking
Updates #483
4 years ago
Brad Fitzpatrick 92252b0988 wgengine/magicsock: add a little LRU cache for netaddr.IPPort lookups
And while plumbing, a bit of discovery work I'll need: the
endpointOfAddr map to map from validated paths to the discoEndpoint.
Not being populated yet.

Updates #483
4 years ago
Brad Fitzpatrick 2d6e84e19e net/netcheck, wgengine/magicsock: replace more UDPAddr with netaddr.IPPort 4 years ago
Brad Fitzpatrick 9070aacdee wgengine/magicsock: minor comments & logging & TODO changes 4 years ago
Brad Fitzpatrick e96f22e560 wgengine/magicsock: start handling disco message, use netaddr.IPPort more
Updates #483
4 years ago
Brad Fitzpatrick a83ca9e734 wgengine/magicsock: cache precomputed nacl/box shared keys
Updates #483
4 years ago
Brad Fitzpatrick a975e86bb8 wgengine/magicsock: add new endpoint type used for discovery-supporting peers
This adds a new magicsock endpoint type only used when both sides
support discovery (that is, are advertising a discovery
key). Otherwise the old code is used.

So far the new code only communicates over DERP as proof that the new
code paths are wired up. None of the actually discovery messaging is
implemented yet.

Support for discovery (generating and advertising a key) are still
behind an environment variable for now.

Updates #483
4 years ago
Brad Fitzpatrick 103c06cc68 wgengine/magicsock: open discovery naclbox messages from known peers
And track known peers.

Doesn't yet do anything with the messages. (nor does it send any yet)

Start of docs on the message format. More will come in subsequent changes.

Updates #483
4 years ago
Brad Fitzpatrick 23e74a0f7a wgengine, magicsock, tstun: don't regularly STUN when idle (mobile only for now)
If there's been 5 minutes of inactivity, stop doing STUN lookups. That
means NAT mappings will expire, but they can resume later when there's
activity again.

We'll do this for all platforms later.

Updates tailscale/corp#320

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
4 years ago
Brad Fitzpatrick fe50cd0c48 ipn, wgengine: plumb NetworkMap down to magicsock
Now we can have magicsock make decisions based on tailcfg.Debug
settings sent by the server.

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
4 years ago
Brad Fitzpatrick 53fb25fc2f all: generate discovery key, plumb it around
Not actually used yet.

Updates #483
4 years ago
Brad Fitzpatrick abd79ea368 derp: reduce DERP memory use; don't require callers to pass in memory to use
The magicsock derpReader was holding onto 65KB for each DERP
connection forever, just in case.

Make the derp{,http}.Client be in charge of memory instead. It can
reuse its bufio.Reader buffer space.
4 years ago