Commit Graph

308 Commits (1ec64bc94decede7840d50f602cb3bfe90eda3ef)

Author SHA1 Message Date
Brad Fitzpatrick 7e201806b1 wgengine/magicsock: reconnect to DERP home after network comes back up
Updates #1310
4 years ago
Brad Fitzpatrick 9b4e50cec0 wgengine/magicsock: fix typo in comment 4 years ago
Brad Fitzpatrick 6b365b0239 wgengine/magicsock: fix DERP reader hang regression during concurrent reads
Fixes #1282

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
4 years ago
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>
4 years ago
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.
4 years ago
Brad Fitzpatrick 1e7a35b225 types/netmap: split controlclient.NetworkMap off into its own leaf package
Updates #1278

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
4 years ago
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>
4 years ago
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>
4 years ago
Brad Fitzpatrick f7eed25bb9 wgengine/magicsock: filter disco packets and packets when stopped from wireguard
Fixes #1167
Fixes tailscale/corp#219

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
4 years ago
Josh Bleecher Snyder e8cd7bb66f tstest: simplify goroutine leak tests
Use tb.Cleanup to simplify both the API and the implementation.

One behavior change: When the number of goroutines shrinks, don't log.
I've never found these logs to be useful, and they frequently add noise.

Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
4 years ago
Josh Bleecher Snyder dd10babaed wgenginer/magicsock: remove Addrs methods
They are now unused.

Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
4 years ago
Josh Bleecher Snyder fe7c3e9c17 all: move wgcfg from wireguard-go
This is mostly code movement from the wireguard-go repo.

Most of the new wgcfg package corresponds to the wireguard-go wgcfg package.

wgengine/wgcfg/device{_test}.go was device/config{_test}.go.
There were substantive but simple changes to device_test.go to remove
internal package device references.

The API of device.Config (now wgcfg.DeviceConfig) grew an error return;
we previously logged the error and threw it away.

Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
4 years ago
Josh Bleecher Snyder d5baeeed5c wgengine: use Tailscale-style peer identifiers in logs
Rewrite log lines on the fly, based on the set of known peers.

This enables us to use upstream wireguard-go logging,
but maintain the Tailscale-style peer public key identifiers
that the rest of our systems (and people) expect.

Fixes #1183

Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
4 years ago
Brad Fitzpatrick 9541886856 wgengine/magicsock: disable regular STUNs for all platforms by default
Reduces background CPU & network.

Updates #1034

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
4 years ago
Brad Fitzpatrick c55d26967b wgengine/magicsock: log more details of endpoints learned over disco
Also, don't try to use IPv6 LinkLocalUnicast addresses for now. Like endpoints
exchanged with control, we share them but don't yet use them.

Updates #1172
4 years ago
Brad Fitzpatrick 359055d3fa wgengine/magicsock: fix logging regression
c8c493f3d9 made it always say
`created=false` which scared me when I saw it, as that would've implied
things were broken much worse. Fortunately the logging was just wrong.
4 years ago
Brad Fitzpatrick edf64e0901 wgengine/magicsock: send, use endpoints in CallMeMaybe messages
Fixes #1172

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
4 years ago
Brad Fitzpatrick b5b4992eff disco: support parsing/encoding endpoints in call-me-maybe frames
Updates #1172

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
4 years ago
Josh Bleecher Snyder d3dd7c6270 wgengine/magicsock: make legacy DstToString match Addrs
DstToString is used in two places in wireguard-go: Logging and uapi.

We are switching to use uapi for wireguard-go config.
To preserve existing behavior, we need the full set of addrs.

And for logging, having the full set of addrs seems useful.

(The Addrs method itself is slated for removal. When that happens,
the implementation will move to DstToString.)


Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
4 years ago
Brad Fitzpatrick 187e22a756 wgengine/magicsock: don't run the DERP cleanup so often
To save CPU and wakeups, don't run the DERP cleanup timer regularly
unless there is a non-home DERP connection open.

Also eliminates the goroutine, moving to a time.AfterFunc.

Updates #1034

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
4 years ago
Josh Bleecher Snyder 5fe5402fcd Revert "wgengine/magicsock: shortcircuit discoEndpoint.heartbeat when its connection is closed"
This reverts commit 08baa17d9a.
It caused deadlocks due to lock ordering violations.
It was not the right fix, and thus should simply be reverted
while we look for the right fix (if we haven't already found it
in the interim; we've fixed other logging-after-test issues).

Fixes #1161
4 years ago
Josh Bleecher Snyder e4c075cd95 wgengine/magicsock: prevent log-after-test in TestTwoDevicePing 4 years ago
Brad Fitzpatrick edce91a8a6 wgengine/magicsock: fix a naked return bug/crash where we returned (nil, true)
The 'ok' from 'ipp, ok :=' above was the result parameter ok. Whoops.
4 years ago
Brad Fitzpatrick 51bd1feae4 wgengine/magicsock: add single element IPPort->endpoint cache in receive path
name           old time/op  new time/op  delta
ReceiveFrom-4  21.8µs ± 2%  20.9µs ± 2%  -4.27%  (p=0.000 n=10+10)

Updates #414

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
4 years ago
Brad Fitzpatrick 5c619882bc wgengine/magicsock: simplify ReceiveIPv4+DERP path
name           old time/op  new time/op  delta
ReceiveFrom-4  35.8µs ± 3%  21.9µs ± 5%  -38.92%  (p=0.008 n=5+5)

Fixes #1145
Updates #414

Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
4 years ago
Brad Fitzpatrick 3fa86a8b23 wgengine/magicsock: use relatively new netaddr.IPPort.IsZero method 4 years ago
Brad Fitzpatrick 4811236189 wgengine/magicsock: speed up BenchmarkReceiveFrom, store context.Done chan
context.cancelCtx.Done involves a mutex and isn't as cheap as I
previously assumed. Convert the donec method into a struct field and
store the channel value once. Our one magicsock.Conn gets one pointer
larger, but it cuts ~1% of the CPU time of the ReceiveFrom benchmark
and removes a bubble from the --svg output :)
4 years ago
Josh Bleecher Snyder ed2169ae99 wgengine/magicsock: prevent logging after TestActiveDiscovery completes 4 years ago
Josh Bleecher Snyder 63af950d8c wgengine/magicsock: adapt to wireguard-go without UpdateDst
22507adf54 stopped relying on
our fork of wireguard-go's UpdateDst callback.
As a result, we can unwind that code,
and the extra return value of ReceiveIPv{4,6}.

Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
4 years ago
Denton Gentry 23c2dc2165
magicksock: remove TestConnClosing. (#1140)
Test is flakey, remove it and figure out what to do differently later.

Signed-off-by: Denton Gentry <dgentry@tailscale.com>
4 years ago
David Anderson e23b4191c4 wgengine/magicsock: disable legacy networking everywhere except TwoDevicePing.
TwoDevicePing is explicitly testing the behavior of the legacy codepath, everything
else is happy to assume that code no longer exists.

Signed-off-by: David Anderson <danderson@tailscale.com>
4 years ago
David Anderson 0733c5d2e0 wgengine/magicsock: disable legacy behavior in a few more tests.
Signed-off-by: David Anderson <danderson@tailscale.com>
4 years ago
David Anderson 57d95dd005 wgengine/magicsock: default legacy networking to off for some tests.
Signed-off-by: David Anderson <danderson@tailscale.com>
4 years ago
David Anderson a2463e8948 wgengine/magicsock: add an option to disable legacy peer handling.
Used in tests to ensure we're not relying on behavior we're going
to remove eventually.

Signed-off-by: David Anderson <danderson@tailscale.com>
4 years ago
David Anderson d456bfdc6d wgengine/magicsock: fix BenchmarkReceiveFrom.
Previously, this benchmark relied on behavior of the legacy
receive codepath, which I changed in 22507adf. With this
change, the benchmark instead relies on the new active discovery
path.

Signed-off-by: David Anderson <danderson@tailscale.com>
4 years ago
Josh Bleecher Snyder 2d837f79dc wgengine/magicsock: close test loggers once we're done with them
This is a big hammer approach to helping with #1132.

Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
4 years ago
Josh Bleecher Snyder 08baa17d9a wgengine/magicsock: shortcircuit discoEndpoint.heartbeat when its connection is closed
This prevents us from continuing to do unnecessary work
(including logging) after the connection has closed.

Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
4 years ago
Josh Bleecher Snyder 7c76435bf7 wgengine/magicsock: simplify
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
4 years ago
Josh Bleecher Snyder d2529affa2 wgengine/magicsock: quiet wireguard-go logging in tests
We already do this in newUserspaceEngineAdvanced.
Apply it to newMagicStack as well.

Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
4 years ago
Josh Bleecher Snyder 654b5f1570 all: convert from []wgcfg.Endpoint to string
This eliminates a dependency on wgcfg.Endpoint,
as part of the effort to eliminate our wireguard-go fork.

Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
4 years ago
David Anderson 9abcb18061 wgengine/magicsock: import more of wireguard-go, update docstrings.
Signed-off-by: David Anderson <danderson@tailscale.com>
4 years ago
David Anderson 22507adf54 wgengine/magicsock: stop depending on UpdateDst in legacy codepaths.
This makes connectivity between ancient and new tailscale nodes slightly
worse in some cases, but only in cases where the ancient version would
likely have failed to get connectivity anyway.

Signed-off-by: David Anderson <danderson@tailscale.com>
4 years ago
Denton Gentry 8349e10907 magicsock: add description of testClosingContext
Signed-off-by: Denton Gentry <dgentry@tailscale.com>
4 years ago
Denton Gentry 2e9728023b magicsock: test error case in sendDiscoMessage
In sendDiscoMessage there is a check of whether the connection is
closed, which is not being reliably exercised by other tests.
This shows up in code coverage reports, the lines of code in
sendDiscoMessage are alternately added and subtracted from
code coverage.

Add a test to specifically exercise and verify this code path.

Signed-off-by: Denton Gentry <dgentry@tailscale.com>
4 years ago
Denton Gentry 0aa55bffce magicsock: test error case in derpWriteChanOfAddr
In derpWriteChanOfAddr when we call derphttp.NewRegionClient(),
there is a check of whether the connection is already errored and
if so it returns before grabbing the lock. The lock might already
be held and would be a deadlock.

This corner case is not being reliably exercised by other tests.
This shows up in code coverage reports, the lines of code in
derpWriteChanOfAddr are alternately added and subtracted from
code coverage.

Add a test to specifically exercise this code path, and verify that
it doesn't deadlock.

This is the best tradeoff I could come up with:
+ the moment code calls Err() to check if there is an error, we
  grab the lock to make sure it would deadlock if it tries to grab
  the lock itself.
+ if a new call to Err() is added in this code path, only the
  first one will be covered and the rest will not be tested.
+ this test doesn't verify whether code is checking for Err() in
  the right place, which ideally I guess it would.

Signed-off-by: Denton Gentry <dgentry@tailscale.com>
4 years ago
Brad Fitzpatrick 85e54af0d7 wgengine: on TCP connect fail/timeout, log some clues about why it failed
So users can see why things aren't working.

A start. More diagnostics coming.

Updates #1094
4 years ago
Brad Fitzpatrick f85769b1ed wgengine/magicsock: drop netaddr.IPPort cache
netaddr.IP no longer allocates, so don't need a cache or all its associated
code/complexity.

This totally removes groupcache/lru from the deps.

Also go mod tidy.
4 years ago
Brad Fitzpatrick 5aa5db89d6 cmd/tailscaled, wgengine/netstack: add start of gvisor userspace netstack work
Not usefully functional yet (mostly a proof of concept), but getting
it submitted for some work @namansood is going to do atop this.

Updates #707
Updates #634
Updates #48
Updates #835
4 years ago
Brad Fitzpatrick 5efb0a8bca cmd/tailscale: change formatting of "tailscale status"
* show DNS name over hostname, removing domain's common MagicDNS suffix.
  only show hostname if there's no DNS name.
  but still show shared devices' MagicDNS FQDN.

* remove nerdy low-level details by default: endpoints, DERP relay,
  public key.  They're available in JSON mode still for those who need
  them.

* only show endpoint or DERP relay when it's active with the goal of
  making debugging easier. (so it's easier for users to understand
  what's happening) The asterisks are gone.

* remove Tx/Rx numbers by default for idle peers; only show them when
  there's traffic.

* include peers' owner login names

* add CLI option to not show peers (matching --self=true, --peers= also
  defaults to true)

* sort by DNS/host name, not public key

* reorder columns
4 years ago
Brad Fitzpatrick b5b9866ba2 wgengine/magicsock: copy self DNS name to PeerStatus, re-fill OS
The OS used to be sent back from the server but that has since
been removed as being redundant.
4 years ago