Not beautiful, but I'm debugging connectivity problems on
NEProvider.sleep+wake and need more clues.
Updates #1426
Updates tailscale/corp#1289
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Also change the type to netaddr.IP while here, because it made sorting
easier.
Updates tailscale/corp#1397
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
We often see things in logs like:
2021-03-02 17:52:45.2456258 +0800 +0800: winhttp: Open: The parameter is incorrect.
2021-03-02 17:52:45.2506261 +0800 +0800: tshttpproxy: winhttp: GetProxyForURL("https://log.tailscale.io/c/tailnode.log.tailscale.io/5037bb42f4bc330e2d6143e191a7ff7e837c6be538139231de69a439536e0d68"): ERROR_INVALID_PARAMETER [unexpected]
I have a hunch that WinHTTP has thread-local state. If so, this would fix it.
If not, this is pretty harmless.
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
DefaultRouteInterface was previously guarded by build tags such that
it was only accessible to tailscaled-on-macos, but there was no reason
for that. It runs fine in the sandbox and gives better default info,
so merge its file into interfaces_darwin.go.
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
And add a --socks5-server flag.
And fix a race in SOCKS5 replies where the response header was written
concurrently with the copy from the backend.
Co-authored with Naman Sood.
Updates #707
Updates #504
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
* move probing out of netcheck into new net/portmapper package
* use PCP ANNOUNCE op codes for PCP discovery, rather than causing
short-lived (sub-second) side effects with a 1-second-expiring map +
delete.
* track when we heard things from the router so we can be less wasteful
in querying the router's port mapping services in the future
* use portmapper from magicsock to map a public port
Fixes#1298Fixes#1080Fixes#1001
Updates #864
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
This allows proxy URLs such as:
http://azurediamond:hunter2@192.168.122.154:38274
to be used in order to dial out to control, logs or derp servers.
Signed-off-by: Christine Dodrill <xe@tailscale.com>
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.
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>
Unused for now, but I want to backport this commit to 1.4 so 1.6 can
start sending these and then at least 1.4 logs will stringify nicely.
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
The code was using a C "int", which is a signed 32-bit integer.
That means some valid IP addresses were negative numbers.
(In particular, the default router address handed out by AT&T
fiber: 192.168.1.254. No I don't know why they do that.)
A negative number is < 255, and so was treated by the Go code
as an error.
This fixes the unit test failure:
$ go test -v -run=TestLikelyHomeRouterIPSyscallExec ./net/interfaces
=== RUN TestLikelyHomeRouterIPSyscallExec
interfaces_darwin_cgo_test.go:15: syscall() = invalid IP, false, netstat = 192.168.1.254, true
--- FAIL: TestLikelyHomeRouterIPSyscallExec (0.00s)
Signed-off-by: David Crawshaw <crawshaw@tailscale.com>
Windows has a low resolution timer.
Some of the tests assumed that unblock takes effect immediately.
Consider:
t := time.Now()
elapsed := time.Now().After(t)
It seems plausible that elapsed should always be true.
However, with a low resolution timer, that might fail.
Change time.Now().After to !time.Now().Before,
so that unblocking always takes effect immediately.
Fixes#873.
This adds a new IP Protocol type, TSMP on protocol number 99 for
sending inter-tailscale messages over WireGuard, currently just for
why a peer rejects TCP SYNs (ACL rejection, shields up, and in the
future: nothing listening, something listening on that port but wrong
interface, etc)
Updates #1094
Updates tailscale/corp#1185
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Users in Amsterdam (as one example) were flipping back and forth
between equidistant London & Frankfurt relays too much.
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
In practice, we already provide IPv6 endpoint addresses via netcheck,
and that address is likely to match a local address anyway (i.e. no NAT66).
The comment at that piece of the code mentions needing to figure out a
good priority ordering, but that only applies to non-active-discovery
clients, who already don't do anything with IPv6 addresses.
Signed-off-by: David Anderson <danderson@tailscale.com>
likelyHomeRouterIPDarwinSyscall iterates through the list of routes,
looking for a private gateway, returning the first one it finds.
likelyHomeRouterIPDarwinExec does the same thing,
except that it returns the last one it finds.
As a result, when there are multiple gateways,
TestLikelyHomeRouterIPSyscallExec fails.
(At least, I think that that is what is happening;
I am going inferring from observed behavior.)
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
The compiler is failing to draw the connection between
slice cap and slice len, so is missing some obvious BCE opportunities.
Give it a hint by making the cap equal to the length.
The generated code is smaller and cleaner, and a bit faster.
name old time/op new time/op delta
Decode/tcp4-8 12.2ns ± 1% 11.6ns ± 3% -5.31% (p=0.000 n=28+29)
Decode/tcp6-8 12.5ns ± 2% 11.9ns ± 2% -4.84% (p=0.000 n=30+30)
Decode/udp4-8 11.5ns ± 1% 11.1ns ± 1% -3.11% (p=0.000 n=25+24)
Decode/udp6-8 11.8ns ± 3% 11.4ns ± 1% -3.08% (p=0.000 n=30+26)
Decode/icmp4-8 11.0ns ± 3% 10.6ns ± 1% -3.38% (p=0.000 n=25+30)
Decode/icmp6-8 11.4ns ± 1% 11.1ns ± 2% -2.29% (p=0.000 n=27+30)
Decode/igmp-8 10.3ns ± 0% 10.0ns ± 1% -3.26% (p=0.000 n=19+23)
Decode/unknown-8 8.68ns ± 1% 8.38ns ± 1% -3.55% (p=0.000 n=28+29)
Cache DNS results of earlier login.tailscale.com control dials, and use
them for future dials if DNS is slow or broken.
Fixes various issues with trickier setups with the domain's DNS server
behind a subnet router.
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
The packet filter still rejects all IPv6, but decodes enough from v6
packets to do something smarter in a followup.
name time/op
Decode/tcp4-8 28.8ns ± 2%
Decode/tcp6-8 20.6ns ± 1%
Decode/udp4-8 28.2ns ± 1%
Decode/udp6-8 20.0ns ± 6%
Decode/icmp4-8 21.7ns ± 2%
Decode/icmp6-8 14.1ns ± 2%
Decode/unknown-8 9.43ns ± 2%
Signed-off-by: David Anderson <danderson@tailscale.com>
Otherwise log upload HTTP requests generate proxy errrors which
generate logs which generate HTTP requests which generate proxy
errors which generate more logs, etc.
Fixes#879
Use golang.zx2c4.com/wireguard/windows/tunnel/winipcfg
instead of github.com/tailscale/winipcfg-go package.
Updates #760
Signed-off-by: Alex Brainman <alex.brainman@gmail.com>
If no interfaces are up, calm down and stop spamming so much. It was
noticed as especially bad on Windows, but probably was bad
everywhere. I just have the best network conditions testing on a
Windows VM.
Updates #604
Not used for anything yet (except logging), but populate the current
proxy autoconfig PAC URL in Interfaces.State.
A future change will do things based on it.
Otherwise when PAC server is down, we log, and each log entry is a new
HTTP request (from logtail) and a new GetProxyForURL call, which again
logs, non-stop. This is also nicer to the WinHTTP service.
Then also hook up link change notifications to the cache to reset it
if there's a chance the network might work sooner.
Also remove rebinding logic from the windows router. Magicsock will
instead rebind based on link change signals.
Signed-off-by: David Anderson <danderson@tailscale.com>
This will be used in a future change to do localhost connection
authentication. This lets us quickly map a localhost TCP connection to
a PID. (A future change will then map a pid to a user)
TODO: pull portlist's netstat code into this package. Then portlist
will be fast on Windows without requiring shelling out to netstat.exe.
We currently have a chickend-and-egg situation in some environments
where we can set up routes that WinHTTP's WPAD/PAC resolution service
needs to download the PAC file to evaluate GetProxyForURL, but the PAC
file is behind a route for which we need to call GetProxyForURL to
e.g. dial a DERP server.
As a short-term fix, just assume that the most recently returned proxy
is good enough for such situations.
iOS doesn't let you run subprocesses,
which means we can't use netstat to get routing information.
Instead, use syscalls and grub around in the results.
We keep the old netstat version around,
both for use in non-cgo builds,
and for use testing the syscall-based version.
Note that iOS doesn't ship route.h,
so we include a copy here from the macOS 10.15 SDK
(which is itself unchanged from the 10.14 SDK).
I have tested manually that this yields the correct
gateway IP address on my own macOS and iOS devices.
More coverage would be most welcome.
Signed-off-by: Josh Bleecher Snyder <josharian@gmail.com>
At least the Apple Airport Extreme doesn't allow hairpin
sends from a private socket until it's seen traffic from
that src IP:port to something else out on the internet.
See https://github.com/tailscale/tailscale/issues/188#issuecomment-600728643
And it seems that even sending to a likely-filtered RFC 5737
documentation-only IPv4 range is enough to set up the mapping.
So do that for now. In the future we might want to classify networks
that do and don't require this separately. But for now help it.
I've confirmed that this is enough to fix the hairpin check on Avery's
home network, even using the RFC 5737 IP.
Fixes#188
- Reuse IP length constants from net package.
- Remove beu16 to make endianness functions consistent.
Signed-off-by: Quoc-Viet Nguyen <afelion@gmail.com>
It'll be called a bunch, so worth a bit of effort. Could go further, but not yet.
(really, should hook into wgengine/monitor and only re-read on netlink changes?)
name old time/op new time/op delta
DefaultRouteInterface-8 60.8µs ±11% 44.6µs ± 5% -26.65% (p=0.000 n=20+19)
name old alloc/op new alloc/op delta
DefaultRouteInterface-8 3.29kB ± 0% 0.55kB ± 0% -83.21% (p=0.000 n=20+20)
name old allocs/op new allocs/op delta
DefaultRouteInterface-8 9.00 ± 0% 6.00 ± 0% -33.33% (p=0.000 n=20+20)
We'll use SO_BINDTODEVICE instead of fancy policy routing. This has
some limitations: for example, we will route all traffic through the
interface that has the main "default" (0.0.0.0/0) route, so machines
that have multiple physical interfaces might have to go through DERP to
get to some peers. But machines with multiple physical interfaces are
very likely to have policy routing (ip rule) support anyway.
So far, the only OS I know of that needs this feature is ChromeOS
(crostini). Fixes#245.
Signed-off-by: Avery Pennarun <apenwarr@tailscale.com>
My earlier 3fa58303d0 tried to implement
the net/http.Tranhsport.DialTLSContext hook, but I didn't return a
*tls.Conn, so we ended up sending a plaintext HTTP request to an HTTPS
port. The response ended up being Go telling as such, not the
/derp/latency-check handler's response (which is currently still a
404). But we didn't even get the 404.
This happened to work well enough because Go's built-in error response
was still a valid HTTP response that we can measure for timing
purposes, but it's not a great answer. Notably, it means we wouldn't
be able to get a future handler to run server-side and count those
latency requests.
tailscale netcheck was broken otherwise.
We can fix this a better way later; I'm just fixing a regression in
some way because I'm trying to work on netcheck at the moment.
This allows tailscaled's own traffic to bypass Tailscale-managed routes,
so that things like tailscale-provided default routes don't break
tailscaled itself.
Progress on #144.
Signed-off-by: David Anderson <danderson@tailscale.com>
Also:
* add -verbose flag to cmd/tailscale netcheck
* remove some API from the interfaces package
* convert some of the interfaces package to netaddr.IP
* don't even send IPv4 probes on machines with no IPv4 (or only v4
loopback)
* and once three regions have replied, stop waiting for other probes
at 2x the slowest duration.
Updates #376
Under some conditions, code would try to look things up in the maps
before the first call to updateLatency. I don't see any reason to delay
initialization of the maps, so let's just init them right away when
creating the Report instance.
Signed-off-by: Avery Pennarun <apenwarr@tailscale.com>
This is a lot like wiring up a local UDP socket, read and write
deadlines work. The big difference is the Block feature, which
lets you stop the packet flow without breaking the connection.
This lets you emulate broken sockets and test timeouts actually
work.
Signed-off-by: David Crawshaw <crawshaw@tailscale.com>
Basically, don't trust the OS-level link monitor to only tell you
interesting things. Sanity check it.
Also, move the interfaces package into the net directory now that we
have it.
I started to write a full DNS caching resolver and I realized it was
overkill and wouldn't work on Windows even in Go 1.14 yet, so I'm
doing this tiny one instead for now, just for all our netcheck STUN
derp lookups, and connections to DERP servers. (This will be caching a
exactly 8 DNS entries, all ours.)
Fixes#145 (can be better later, of course)
