Commit Graph

5554 Commits

Author SHA1 Message Date
Brad Fitzpatrick
20e66c5b92 net/interfaces: reconcile interface filtering with address printing in logs
The interface.State logging tried to only log interfaces which had
interesting IPs, but the what-is-interesting checks differed between
the code that gathered the interface names to print and the printing
of their addresses.
2021-02-12 18:42:45 -08:00
Josh Bleecher Snyder
c7e5ab8094 wgengine/magicsock: retry and re-send packets in TestTwoDevicePing
When a handshake race occurs, a queued data packet can get lost.
TestTwoDevicePing expected that the very first data packet would arrive.
This caused occasional flakes.

Change TestTwoDevicePing to repeatedly re-send packets
and succeed when one of them makes it through.

This is acceptable (vs making WireGuard not drop the packets)
because this only affects communication with extremely old clients.
And those extremely old clients will eventually connect,
because the kernel will retry sends on timeout.

Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
2021-02-12 14:18:58 -08:00
Brad Fitzpatrick
ca51529b81 derp/derphttp: return nicer errors from Recv on Close 2021-02-12 12:04:16 -08:00
Brad Fitzpatrick
741d654aa3 derp/derphttp: add a context and infoLogger option to RunWatchConnectionLoop 2021-02-12 10:59:11 -08:00
Josh Bleecher Snyder
1632f9fd6b wgengine/magicsock: reduce log spam during tests
Only do the type assertion to *net.UDPAddr when addr is non-nil.
This prevents a bunch of log spam during tests.
2021-02-12 10:49:02 -08:00
Josh Bleecher Snyder
88586ec4a4 wgengine/magicsock: remove an alloc from ReceiveIPvN
We modified the standard net package to not allocate a *net.UDPAddr
during a call to (*net.UDPConn).ReadFromUDP if the caller's use
of the *net.UDPAddr does not cause it to escape.
That is https://golang.org/cl/291390.

This is the companion change to magicsock.
There are two changes required.
First, call ReadFromUDP instead of ReadFrom, if possible.
ReadFrom returns a net.Addr, which is an interface, which always allocates.
Second, reduce the lifetime of the returned *net.UDPAddr.
We do this by immediately converting it into a netaddr.IPPort.

We left the existing RebindingUDPConn.ReadFrom method in place,
as it is required to satisfy the net.PacketConn interface.

With the upstream change and both of these fixes in place,
we have removed one large allocation per packet received.

name           old time/op    new time/op    delta
ReceiveFrom-8    16.7µs ± 5%    16.4µs ± 8%     ~     (p=0.310 n=5+5)

name           old alloc/op   new alloc/op   delta
ReceiveFrom-8      112B ± 0%       64B ± 0%  -42.86%  (p=0.008 n=5+5)

name           old allocs/op  new allocs/op  delta
ReceiveFrom-8      3.00 ± 0%      2.00 ± 0%  -33.33%  (p=0.008 n=5+5)

Co-authored-by: Sonia Appasamy <sonia@tailscale.com>
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
2021-02-12 09:52:43 -08:00
Josh Bleecher Snyder
0c673c1344 wgengine/magicsock: unify on netaddr types in addrSet
addrSet maintained duplicate lists of netaddr.IPPorts and net.UDPAddrs.
Unify to use the netaddr type only.

This makes (*Conn).ReceiveIPvN a bit uglier,
but that'll be cleaned up in a subsequent commit.

This is preparatory work to remove an allocation from ReceiveIPv4.

Co-authored-by: Sonia Appasamy <sonia@tailscale.com>
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
2021-02-12 09:52:43 -08:00
Josh Bleecher Snyder
4cd9218351 wgengine/magicsock: prevent logging while running benchmarks
Co-authored-by: Sonia Appasamy <sonia@tailscale.com>
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
2021-02-12 09:52:43 -08:00
Brad Fitzpatrick
be906dabd4 version: bump date 2021-02-11 20:11:00 -08:00
Brad Fitzpatrick
6680976b50 cmd/tailscaled: pick automatic tun device name on darwin 2021-02-11 20:10:07 -08:00
Brad Fitzpatrick
88ab0173a7 wgengine/router: fix BSD router to support multiple local addrs, IPv6
Fixes #1201
2021-02-11 19:13:03 -08:00
Ross Zurowski
25321cbd01
cmd/hello: truncate long strings (#1328)
Signed-off-by: Ross Zurowski <ross@rosszurowski.com>
2021-02-11 20:56:22 -05:00
Brad Fitzpatrick
5378776043 cmd/hello: chop DNS name at first dot 2021-02-11 16:38:26 -08:00
Ross Zurowski
6075135e0a
cmd/hello: style welcome message (#1325)
Signed-off-by: Ross Zurowski <ross@rosszurowski.com>
2021-02-11 17:42:07 -05:00
Brad Fitzpatrick
917307a90c wgengine/tstun: reply to MagicDNS pings
Fixes #849

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
2021-02-11 11:55:38 -08:00
Brad Fitzpatrick
34ffd4f7c6 cmd/hello: serve fake data in dev mode on whois failure 2021-02-11 10:57:08 -08:00
Brad Fitzpatrick
de3001bc79 cmd/hello: in dev mode, live reload template 2021-02-11 10:53:33 -08:00
Josh Bleecher Snyder
11bbfbd8bb go.mod: update to latest wireguard-go
All changes are trivial.
2021-02-10 14:14:11 -08:00
Josh Bleecher Snyder
635e4c7435 wgengine/magicsock: increase legacy ping timeout again
I based my estimation of the required timeout based on locally
observed behavior. But CI machines are worse than my local machine.
16s was enough to reduce flakiness but not eliminate it. Bump it up again.

Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
2021-02-10 13:50:18 -08:00
Brad Fitzpatrick
1ec64bc94d wgengine/router: add another Windows firewall rule to allow incoming UDP
Based on @sailorfrag's research.

Fixes #1312

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
2021-02-10 13:12:17 -08:00
Brad Fitzpatrick
7e201806b1 wgengine/magicsock: reconnect to DERP home after network comes back up
Updates #1310
2021-02-10 10:29:03 -08:00
Josh Bleecher Snyder
1f0fa8b814 go.mod: pull in upstream wireguard-go bug fixes 2021-02-10 08:04:12 -08:00
moncho
e101d8396d portlist, version: update build tags for Go 1.16, Apple M1
Build tags have been updated to build native Apple M1 binaries, existing build
tags for ios have been changed from darwin,arm64 to ios,arm64.

With this change, running go build cmd/tailscale{,d}/tailscale{,d}.go on an Apple
machine with the new processor works and resulting binaries show the expected
architecture, e.g. tailscale: Mach-O 64-bit executable arm64.

Tested using go version go1.16beta1 darwin/arm64.

Updates #943

Signed-off-by: moncho <50428+moncho@users.noreply.github.com>
2021-02-09 21:10:12 -08:00
Brad Fitzpatrick
cbd6224ca4 wgengine/winnet: don't build on non-windows
It only affects 'go install ./...', etc, and only on darwin/arm64 (M1 Macs) where
the go-ole package doesn't compile.

No need to build it.

Updates #943
2021-02-09 21:09:24 -08:00
Josh Bleecher Snyder
4a82e36491 go.mod: bump to latest wireguard-go
Stabilization and performance improvements.
2021-02-09 14:20:01 -08:00
Brad Fitzpatrick
9b4e50cec0 wgengine/magicsock: fix typo in comment 2021-02-09 09:37:24 -08:00
Naman Sood
07c3df13c6 wgengine/tstun: inform userspaceEngine about injected outbound packets in tundev
Signed-off-by: Naman Sood <mail@nsood.in>
2021-02-09 08:08:01 -08:00
Josh Bleecher Snyder
e7caad61fb wgengine: remove IpcGetOperation filter
This was in place because retrieved allowed_ips was very expensive.
Upstream changed the data structure to make them cheaper to compute.

This commit is an experiment to find out whether they're now cheap enough.

Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
2021-02-08 20:22:55 -08:00
Brad Fitzpatrick
6b365b0239 wgengine/magicsock: fix DERP reader hang regression during concurrent reads
Fixes #1282

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
2021-02-08 14:30:15 -08:00
Josh Bleecher Snyder
e1f773ebba wgengine/magicsock: allow more time for pings to transit
We removed the "fast retry" code from our wireguard-go fork.
As a result, pings can take longer to transit when retries are required. 
Allow that.

Fixes #1277

Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
2021-02-08 13:54:37 -08:00
Brad Fitzpatrick
6d2b8df06d wgengine/magicsock: add disabled failing (deadlocking) test for #1282
The fix can make this test run unconditionally.

This moves code from 5c619882bc for
testability but doesn't fix it yet. The #1282 problem remains (when I
wrote its wake-up mechanism, I forgot there were N DERP readers
funneling into 1 UDP reader, and the code just isn't correct at all
for that case).

Also factor out some test helper code from BenchmarkReceiveFrom.

The refactoring in magicsock.go for testability should have no
behavior change.
2021-02-06 21:34:16 -08:00
David Anderson
e86b39b73f ipn/ipnlocal: don't short-circuit default route filtering.
If no exit node is specified, the filter must still run to remove
offered default routes from all peers.

Signed-off-by: David Anderson <danderson@tailscale.com>
2021-02-05 20:33:18 -08:00
Brad Fitzpatrick
1e7a35b225 types/netmap: split controlclient.NetworkMap off into its own leaf package
Updates #1278

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
2021-02-05 16:18:52 -08:00
Brad Fitzpatrick
ddfcc4326c types/persist: split controlclient.Persist into a small leaf package
This one alone doesn't modify the global dependency map much
(depaware.txt if anything looks slightly worse), but it leave
controlclient as only containing NetworkMap:

bradfitz@tsdev:~/src/tailscale.com/ipn$ grep -F "controlclient." *.go
backend.go:     NetMap        *controlclient.NetworkMap // new netmap received
fake_test.go:   b.notify(Notify{NetMap: &controlclient.NetworkMap{}})
fake_test.go:   b.notify(Notify{NetMap: &controlclient.NetworkMap{}})
handle.go:      netmapCache       *controlclient.NetworkMap
handle.go:func (h *Handle) NetMap() *controlclient.NetworkMap {

Once that goes into a leaf package, then ipn doesn't depend on
controlclient at all, and then the client gets smaller.

Updates #1278
2021-02-05 15:25:33 -08:00
David Anderson
a046b48593 cmd/tailscale/cli: display currently active exit node in tailscale status.
Signed-off-by: David Anderson <danderson@tailscale.com>
2021-02-05 14:53:17 -08:00
Brad Fitzpatrick
6064b6ff47 wgengine/wgcfg/nmcfg: split control/controlclient/netmap.go into own package
It couldn't move to ipnlocal due to test dependency cycles.

Updates #1278

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
2021-02-05 14:21:30 -08:00
Josh Bleecher Snyder
138055dd70 tstest/natlab: use net.ErrClosed instead of a new error
Upstream wireguard-go decided to use errors.Is(err, net.ErrClosed)
instead of checking the error string.

It also provided an unsafe linknamed version of net.ErrClosed
for clients running Go 1.15. Switch to that.

This reduces the time required for the wgengine/magicsock tests
on my machine from ~35s back to the ~13s it was before
456cf8a376.

Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
2021-02-05 13:44:46 -08:00
David Anderson
ace57d7627 wgengine/magicsock: set a dummy private key in benchmark.
Magicsock started dropping all traffic internally when Tailscale is
shut down, to avoid spurious wireguard logspam. This made the benchmark
not receive anything. Setting a dummy private key is sufficient to get
magicsock to pass traffic for benchmarking purposes.

Fixes #1270.

Signed-off-by: David Anderson <danderson@tailscale.com>
2021-02-05 13:36:05 -08:00
David Anderson
b9c2231fdf ipn: program exit node into the data plane according to user pref.
Part of #1153, #1154. Fixes #1224.

Signed-off-by: David Anderson <danderson@tailscale.com>
2021-02-05 13:07:11 -08:00
Brad Fitzpatrick
fb6b0e247c cmd/tailscaled: rename Windows service to just Tailscale
Updates #1232
2021-02-05 11:13:34 -08:00
Brad Fitzpatrick
98f9e82c62 logpolicy: on Windows, use tailscale-ipn log name if it already existed
For the migration to tailscaled.exe on Windows, don't create a new logid
if one existed under the old filename.

Updates #1232
2021-02-05 10:57:51 -08:00
Brad Fitzpatrick
e8d4afedd1 control/controlclient: don't call lite endpoint update path when logged out
This was the other half of the #1271 problem.

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
2021-02-05 10:00:35 -08:00
Brad Fitzpatrick
a7562be5e1 cmd/tailscaled: move more of the Windows server setup code into tailscaled
Updates #1232
2021-02-05 09:53:54 -08:00
Brad Fitzpatrick
6f7974b7f2 cmd/tailscaled: add missing depaware.txt update 2021-02-05 08:48:00 -08:00
Brad Fitzpatrick
6099ecf7f4 cmd/tailscaled: run as a service on Windows
Updates #1232
2021-02-05 08:46:12 -08:00
Brad Fitzpatrick
7529b74018 control/controlclient: avoid crash sending map request with zero node key
Fixes #1271

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
2021-02-04 16:23:50 -08:00
Josh Bleecher Snyder
aa6856a9eb wgengine: adapt to wireguard-go changes
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
2021-02-04 15:15:33 -08:00
Brad Fitzpatrick
d76334d2f0 ipn: split LocalBackend off into new ipn/ipnlocal package
And move a couple other types down into leafier packages.

Now cmd/tailscale doesn't bring in netlink, magicsock, wgengine, etc.

Fixes #1181

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
2021-02-04 14:04:23 -08:00
Brad Fitzpatrick
6254efb9ef cmd/tailscale{,d}: move debug subcommand to tailscaled
Work on reducing the size of the tailscale binary, which is
currently pulling in most of the same code as tailscaled.

Updates #1181
2021-02-04 12:23:06 -08:00
Brad Fitzpatrick
70eb05fd47 wgengine: access flow pending problem with lock held
Missed review feedback from just-submitted d37058af72.
2021-02-04 11:18:32 -08:00