cmd/tailscale: add --json-docs flag

This prints all command and flag docs as JSON. To be used for generating the contents of https://tailscale.com/kb/1080/cli. Updates https://github.com/tailscale/tailscale-www/issues/4722
wgengine/magicsock: refactor batchingUDPConn to batchingConn interface (#13042 )
2024-08-08 08:06:18 -07:00 · 2024-08-06 09:00:28 -07:00 · 2024-08-06 11:42:06 +01:00 · 2024-08-06 11:27:59 +01:00 · 2024-08-06 11:27:59 +01:00 · 2024-08-06 11:27:59 +01:00
57 changed files with 2886 additions and 911 deletions
--- a/cmd/derpprobe/derpprobe.go
+++ b/cmd/derpprobe/derpprobe.go
@@ -7,8 +7,6 @@ package main
 import (
 	"flag"
 	"fmt"
-	"html"
-	"io"
 	"log"
 	"net/http"
 	"sort"
@@ -70,8 +68,13 @@ func main() {
 	}

 	mux := http.NewServeMux()
-	tsweb.Debugger(mux)
-	mux.HandleFunc("/", http.HandlerFunc(serveFunc(p)))
+	d := tsweb.Debugger(mux)
+	d.Handle("probe-run", "Run a probe", tsweb.StdHandler(tsweb.ReturnHandlerFunc(p.RunHandler), tsweb.HandlerOptions{Logf: log.Printf}))
+	mux.Handle("/", tsweb.StdHandler(p.StatusHandler(
+		prober.WithTitle("DERP Prober"),
+		prober.WithPageLink("Prober metrics", "/debug/varz"),
+		prober.WithProbeLink("Run Probe", "/debug/probe-run?name={{.Name}}"),
+	), tsweb.HandlerOptions{Logf: log.Printf}))
 	log.Printf("Listening on %s", *listen)
 	log.Fatal(http.ListenAndServe(*listen, mux))
 }
@@ -105,26 +108,3 @@ func getOverallStatus(p *prober.Prober) (o overallStatus) {
 	sort.Strings(o.good)
 	return
 }
-
-func serveFunc(p *prober.Prober) func(w http.ResponseWriter, r *http.Request) {
-	return func(w http.ResponseWriter, r *http.Request) {
-		st := getOverallStatus(p)
-		summary := "All good"
-		if (float64(len(st.bad)) / float64(len(st.bad)+len(st.good))) > 0.25 {
-			// Returning a 500 allows monitoring this server externally and configuring
-			// an alert on HTTP response code.
-			w.WriteHeader(500)
-			summary = fmt.Sprintf("%d problems", len(st.bad))
-		}
-
-		io.WriteString(w, "<html><head><style>.bad { font-weight: bold; color: #700; }</style></head>\n")
-		fmt.Fprintf(w, "<body><h1>derp probe</h1>\n%s:<ul>", summary)
-		for _, s := range st.bad {
-			fmt.Fprintf(w, "<li class=bad>%s</li>\n", html.EscapeString(s))
-		}
-		for _, s := range st.good {
-			fmt.Fprintf(w, "<li>%s</li>\n", html.EscapeString(s))
-		}
-		io.WriteString(w, "</ul></body></html>\n")
-	}
-}
--- a/cmd/k8s-operator/depaware.txt
+++ b/cmd/k8s-operator/depaware.txt
@@ -310,6 +310,7 @@ tailscale.com/cmd/k8s-operator dependencies: (generated by github.com/tailscale/
        gvisor.dev/gvisor/pkg/tcpip/ports                            from gvisor.dev/gvisor/pkg/tcpip/stack+
        gvisor.dev/gvisor/pkg/tcpip/seqnum                           from gvisor.dev/gvisor/pkg/tcpip/header+
     💣 gvisor.dev/gvisor/pkg/tcpip/stack                            from gvisor.dev/gvisor/pkg/tcpip/adapters/gonet+
+        gvisor.dev/gvisor/pkg/tcpip/stack/gro                        from tailscale.com/wgengine/netstack
        gvisor.dev/gvisor/pkg/tcpip/transport                        from gvisor.dev/gvisor/pkg/tcpip/transport/icmp+
        gvisor.dev/gvisor/pkg/tcpip/transport/icmp                   from tailscale.com/wgengine/netstack
        gvisor.dev/gvisor/pkg/tcpip/transport/internal/network       from gvisor.dev/gvisor/pkg/tcpip/transport/icmp+
--- a/cmd/tailscale/cli/cli.go
+++ b/cmd/tailscale/cli/cli.go
@@ -7,6 +7,7 @@ package cli

 import (
 	"context"
+	"encoding/json"
 	"errors"
 	"flag"
 	"fmt"
@@ -159,8 +160,10 @@ func newRootCmd() *ffcli.Command {
 		return nil
 	})
 	rootfs.Lookup("socket").DefValue = localClient.Socket
+	jsonDocs := rootfs.Bool("json-docs", false, hidden+"print JSON-encoded docs for all subcommands and flags")

-	rootCmd := &ffcli.Command{
+	var rootCmd *ffcli.Command
+	rootCmd = &ffcli.Command{
 		Name:       "tailscale",
 		ShortUsage: "tailscale [flags] <subcommand> [command flags]",
 		ShortHelp:  "The easiest, most secure way to use WireGuard.",
@@ -202,6 +205,9 @@ change in the future.
 		},
 		FlagSet: rootfs,
 		Exec: func(ctx context.Context, args []string) error {
+			if *jsonDocs {
+				return printJSONDocs(rootCmd)
+			}
 			if len(args) > 0 {
 				return fmt.Errorf("tailscale: unknown subcommand: %s", args[0])
 			}
@@ -401,3 +407,54 @@ func colorableOutput() (w io.Writer, ok bool) {
 	}
 	return colorable.NewColorableStdout(), true
 }
+
+type commandDoc struct {
+	Name        string
+	Desc        string
+	Subcommands []commandDoc `json:",omitempty"`
+	Flags       []flagDoc    `json:",omitempty"`
+}
+
+type flagDoc struct {
+	Name string
+	Desc string
+}
+
+func printJSONDocs(root *ffcli.Command) error {
+	docs := jsonDocsWalk(root)
+	return json.NewEncoder(os.Stdout).Encode(docs)
+}
+
+func jsonDocsWalk(cmd *ffcli.Command) *commandDoc {
+	res := &commandDoc{
+		Name: cmd.Name,
+	}
+	if cmd.LongHelp != "" {
+		res.Desc = cmd.LongHelp
+	} else if cmd.ShortHelp != "" {
+		res.Desc = cmd.ShortHelp
+	} else {
+		res.Desc = cmd.ShortUsage
+	}
+	if strings.HasPrefix(res.Desc, hidden) {
+		return nil
+	}
+	if cmd.FlagSet != nil {
+		cmd.FlagSet.VisitAll(func(f *flag.Flag) {
+			if strings.HasPrefix(f.Usage, hidden) {
+				return
+			}
+			res.Flags = append(res.Flags, flagDoc{
+				Name: f.Name,
+				Desc: f.Usage,
+			})
+		})
+	}
+	for _, sub := range cmd.Subcommands {
+		subj := jsonDocsWalk(sub)
+		if subj != nil {
+			res.Subcommands = append(res.Subcommands, *subj)
+		}
+	}
+	return res
+}
--- a/cmd/tailscale/cli/network-lock.go
+++ b/cmd/tailscale/cli/network-lock.go
@@ -789,7 +789,7 @@ func runNetworkLockRevokeKeys(ctx context.Context, args []string) error {
 		}

 		fmt.Printf(`Run the following command on another machine with a trusted tailnet lock key:
-	%s lock recover-compromised-key --cosign %X
+	%s lock revoke-keys --cosign %X
 `, os.Args[0], aumBytes)
 		return nil
 	}
@@ -813,10 +813,10 @@ func runNetworkLockRevokeKeys(ctx context.Context, args []string) error {
 		fmt.Printf(`Co-signing completed successfully.

 To accumulate an additional signature, run the following command on another machine with a trusted tailnet lock key:
-	%s lock recover-compromised-key --cosign %X
+	%s lock revoke-keys --cosign %X

 Alternatively if you are done with co-signing, complete recovery by running the following command:
-	%s lock recover-compromised-key --finish %X
+	%s lock revoke-keys --finish %X
 `, os.Args[0], aumBytes, os.Args[0], aumBytes)
 	}

--- a/cmd/tailscaled/depaware.txt
+++ b/cmd/tailscaled/depaware.txt
@@ -221,6 +221,7 @@ tailscale.com/cmd/tailscaled dependencies: (generated by github.com/tailscale/de
        gvisor.dev/gvisor/pkg/tcpip/ports                            from gvisor.dev/gvisor/pkg/tcpip/stack+
        gvisor.dev/gvisor/pkg/tcpip/seqnum                           from gvisor.dev/gvisor/pkg/tcpip/header+
     💣 gvisor.dev/gvisor/pkg/tcpip/stack                            from gvisor.dev/gvisor/pkg/tcpip/adapters/gonet+
+        gvisor.dev/gvisor/pkg/tcpip/stack/gro                        from tailscale.com/wgengine/netstack
        gvisor.dev/gvisor/pkg/tcpip/transport                        from gvisor.dev/gvisor/pkg/tcpip/transport/icmp+
        gvisor.dev/gvisor/pkg/tcpip/transport/icmp                   from tailscale.com/wgengine/netstack
        gvisor.dev/gvisor/pkg/tcpip/transport/internal/network       from gvisor.dev/gvisor/pkg/tcpip/transport/icmp+
--- a/control/controlclient/direct.go
+++ b/control/controlclient/direct.go
@@ -333,6 +333,9 @@ func (c *Direct) Close() error {
 		}
 	}
 	c.noiseClient = nil
+	if tr, ok := c.httpc.Transport.(*http.Transport); ok {
+		tr.CloseIdleConnections()
+	}
 	return nil
 }

--- a/control/controlhttp/client.go
+++ b/control/controlhttp/client.go
@@ -46,6 +46,7 @@ import (
 	"tailscale.com/net/sockstats"
 	"tailscale.com/net/tlsdial"
 	"tailscale.com/net/tshttpproxy"
+	"tailscale.com/syncs"
 	"tailscale.com/tailcfg"
 	"tailscale.com/tstime"
 	"tailscale.com/util/multierr"
@@ -497,11 +498,9 @@ func (a *Dialer) tryURLUpgrade(ctx context.Context, u *url.URL, addr netip.Addr,
 	tr.DisableCompression = true

 	// (mis)use httptrace to extract the underlying net.Conn from the
-	// transport. We make exactly 1 request using this transport, so
-	// there will be exactly 1 GotConn call. Additionally, the
-	// transport handles 101 Switching Protocols correctly, such that
-	// the Conn will not be reused or kept alive by the transport once
-	// the response has been handed back from RoundTrip.
+	// transport. The transport handles 101 Switching Protocols correctly,
+	// such that the Conn will not be reused or kept alive by the transport
+	// once the response has been handed back from RoundTrip.
 	//
 	// In theory, the machinery of net/http should make it such that
 	// the trace callback happens-before we get the response, but
@@ -517,10 +516,16 @@ func (a *Dialer) tryURLUpgrade(ctx context.Context, u *url.URL, addr netip.Addr,
 	// unexpected EOFs...), and we're bound to forget someday and
 	// introduce a protocol optimization at a higher level that starts
 	// eagerly transmitting from the server.
-	connCh := make(chan net.Conn, 1)
+	var lastConn syncs.AtomicValue[net.Conn]
 	trace := httptrace.ClientTrace{
+		// Even though we only make a single HTTP request which should
+		// require a single connection, the context (with the attached
+		// trace configuration) might be used by our custom dialer to
+		// make other HTTP requests (e.g. BootstrapDNS). We only care
+		// about the last connection made, which should be the one to
+		// the control server.
 		GotConn: func(info httptrace.GotConnInfo) {
-			connCh <- info.Conn
+			lastConn.Store(info.Conn)
 		},
 	}
 	ctx = httptrace.WithClientTrace(ctx, &trace)
@@ -548,11 +553,7 @@ func (a *Dialer) tryURLUpgrade(ctx context.Context, u *url.URL, addr netip.Addr,
 	// is still a read buffer attached to it within resp.Body. So, we
 	// must direct I/O through resp.Body, but we can still use the
 	// underlying net.Conn for stuff like deadlines.
-	var switchedConn net.Conn
-	select {
-	case switchedConn = <-connCh:
-	default:
-	}
+	switchedConn := lastConn.Load()
 	if switchedConn == nil {
 		resp.Body.Close()
 		return nil, fmt.Errorf("httptrace didn't provide a connection")
--- a/control/controlhttp/http_test.go
+++ b/control/controlhttp/http_test.go
@@ -11,10 +11,12 @@ import (
 	"log"
 	"net"
 	"net/http"
+	"net/http/httptest"
 	"net/http/httputil"
 	"net/netip"
 	"net/url"
 	"runtime"
+	"slices"
 	"strconv"
 	"sync"
 	"testing"
@@ -41,6 +43,8 @@ type httpTestParam struct {
 	makeHTTPHangAfterUpgrade bool

 	doEarlyWrite bool
+
+	httpInDial bool
 }

 func TestControlHTTP(t *testing.T) {
@@ -120,6 +124,12 @@ func TestControlHTTP(t *testing.T) {
 			name:         "early_write",
 			doEarlyWrite: true,
 		},
+		// Dialer needed to make another HTTP request along the way (e.g. to
+		// resolve the hostname via BootstrapDNS).
+		{
+			name:       "http_request_in_dial",
+			httpInDial: true,
+		},
 	}

 	for _, test := range tests {
@@ -217,6 +227,29 @@ func testControlHTTP(t *testing.T, param httpTestParam) {
 		Clock:                clock,
 	}

+	if param.httpInDial {
+		// Spin up a separate server to get a different port on localhost.
+		secondServer := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) { return }))
+		defer secondServer.Close()
+
+		prev := a.Dialer
+		a.Dialer = func(ctx context.Context, network, addr string) (net.Conn, error) {
+			ctx, cancel := context.WithTimeout(ctx, time.Second)
+			defer cancel()
+			req, err := http.NewRequestWithContext(ctx, "GET", secondServer.URL, nil)
+			if err != nil {
+				t.Errorf("http.NewRequest: %v", err)
+			}
+			r, err := http.DefaultClient.Do(req)
+			if err != nil {
+				t.Errorf("http.Get: %v", err)
+			}
+			r.Body.Close()
+
+			return prev(ctx, network, addr)
+		}
+	}
+
 	if proxy != nil {
 		proxyEnv := proxy.Start(t)
 		defer proxy.Close()
@@ -238,6 +271,7 @@ func testControlHTTP(t *testing.T, param httpTestParam) {
 		t.Fatalf("dialing controlhttp: %v", err)
 	}
 	defer conn.Close()
+
 	si := <-sch
 	if si.conn != nil {
 		defer si.conn.Close()
@@ -266,6 +300,19 @@ func testControlHTTP(t *testing.T, param httpTestParam) {
 			t.Errorf("early write = %q; want %q", buf, earlyWriteMsg)
 		}
 	}
+
+	// When no proxy is used, the RemoteAddr of the returned connection should match
+	// one of the listeners of the test server.
+	if proxy == nil {
+		var expectedAddrs []string
+		for _, ln := range []net.Listener{httpLn, httpsLn} {
+			expectedAddrs = append(expectedAddrs, fmt.Sprintf("127.0.0.1:%d", ln.Addr().(*net.TCPAddr).Port))
+			expectedAddrs = append(expectedAddrs, fmt.Sprintf("[::1]:%d", ln.Addr().(*net.TCPAddr).Port))
+		}
+		if !slices.Contains(expectedAddrs, conn.RemoteAddr().String()) {
+			t.Errorf("unexpected remote addr: %s, want %s", conn.RemoteAddr(), expectedAddrs)
+		}
+	}
 }

 type serverResult struct {
--- a/flake.nix
+++ b/flake.nix
@@ -120,4 +120,4 @@
  in
    flake-utils.lib.eachDefaultSystem (system: flakeForSystem nixpkgs system);
 }
-# nix-direnv cache busting line: sha256-N0TZ1JuDqh6bZjOHcMfoEDOsiUlrC/tR72fBns1GwrM=
+# nix-direnv cache busting line: sha256-1hekcJr1jEJFu4ZnapNkbAAv+8phTQuMloULIZ0f018=
--- a/go.mod
+++ b/go.mod
@@ -80,7 +80,7 @@ require (
 	github.com/tailscale/peercred v0.0.0-20240214030740-b535050b2aa4
 	github.com/tailscale/web-client-prebuilt v0.0.0-20240226180453-5db17b287bf1
 	github.com/tailscale/wf v0.0.0-20240214030419-6fbb0a674ee6
-	github.com/tailscale/wireguard-go v0.0.0-20240724015428-60eeedfd624b
+	github.com/tailscale/wireguard-go v0.0.0-20240731203015-71393c576b98
 	github.com/tailscale/xnet v0.0.0-20240729143630-8497ac4dab2e
 	github.com/tc-hib/winres v0.2.1
 	github.com/tcnksm/go-httpstat v0.2.0
@@ -104,7 +104,7 @@ require (
 	golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2
 	golang.zx2c4.com/wireguard/windows v0.5.3
 	gopkg.in/square/go-jose.v2 v2.6.0
-	gvisor.dev/gvisor v0.0.0-20240306221502-ee1e1f6070e3
+	gvisor.dev/gvisor v0.0.0-20240722211153-64c016c92987
 	honnef.co/go/tools v0.4.6
 	k8s.io/api v0.30.3
 	k8s.io/apimachinery v0.30.3
--- a/go.mod.sri
+++ b/go.mod.sri
@@ -1 +1 @@
-sha256-N0TZ1JuDqh6bZjOHcMfoEDOsiUlrC/tR72fBns1GwrM=
+sha256-1hekcJr1jEJFu4ZnapNkbAAv+8phTQuMloULIZ0f018=
--- a/go.sum
+++ b/go.sum
@@ -934,8 +934,8 @@ github.com/tailscale/web-client-prebuilt v0.0.0-20240226180453-5db17b287bf1 h1:t
 github.com/tailscale/web-client-prebuilt v0.0.0-20240226180453-5db17b287bf1/go.mod h1:agQPE6y6ldqCOui2gkIh7ZMztTkIQKH049tv8siLuNQ=
 github.com/tailscale/wf v0.0.0-20240214030419-6fbb0a674ee6 h1:l10Gi6w9jxvinoiq15g8OToDdASBni4CyJOdHY1Hr8M=
 github.com/tailscale/wf v0.0.0-20240214030419-6fbb0a674ee6/go.mod h1:ZXRML051h7o4OcI0d3AaILDIad/Xw0IkXaHM17dic1Y=
-github.com/tailscale/wireguard-go v0.0.0-20240724015428-60eeedfd624b h1:8U9NaPB32iFoNjJ+H/yPkAVqXw/dudtj+fLTE4edF+Q=
-github.com/tailscale/wireguard-go v0.0.0-20240724015428-60eeedfd624b/go.mod h1:BOm5fXUBFM+m9woLNBoxI9TaBXXhGNP50LX/TGIvGb4=
+github.com/tailscale/wireguard-go v0.0.0-20240731203015-71393c576b98 h1:RNpJrXfI5u6e+uzyIzvmnXbhmhdRkVf//90sMBH3lso=
+github.com/tailscale/wireguard-go v0.0.0-20240731203015-71393c576b98/go.mod h1:BOm5fXUBFM+m9woLNBoxI9TaBXXhGNP50LX/TGIvGb4=
 github.com/tailscale/xnet v0.0.0-20240729143630-8497ac4dab2e h1:zOGKqN5D5hHhiYUp091JqK7DPCqSARyUfduhGUY8Bek=
 github.com/tailscale/xnet v0.0.0-20240729143630-8497ac4dab2e/go.mod h1:orPd6JZXXRyuDusYilywte7k094d7dycXXU5YnWsrwg=
 github.com/tc-hib/winres v0.2.1 h1:YDE0FiP0VmtRaDn7+aaChp1KiF4owBiJa5l964l5ujA=
@@ -1491,8 +1491,8 @@ gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
 gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
 gotest.tools/v3 v3.4.0 h1:ZazjZUfuVeZGLAmlKKuyv3IKP5orXcwtOwDQH6YVr6o=
 gotest.tools/v3 v3.4.0/go.mod h1:CtbdzLSsqVhDgMtKsx03ird5YTGB3ar27v0u/yKBW5g=
-gvisor.dev/gvisor v0.0.0-20240306221502-ee1e1f6070e3 h1:/8/t5pz/mgdRXhYOIeqqYhFAQLE4DDGegc0Y4ZjyFJM=
-gvisor.dev/gvisor v0.0.0-20240306221502-ee1e1f6070e3/go.mod h1:NQHVAzMwvZ+Qe3ElSiHmq9RUm1MdNHpUZ52fiEqvn+0=
+gvisor.dev/gvisor v0.0.0-20240722211153-64c016c92987 h1:TU8z2Lh3Bbq77w0t1eG8yRlLcNHzZu3x6mhoH2Mk0c8=
+gvisor.dev/gvisor v0.0.0-20240722211153-64c016c92987/go.mod h1:sxc3Uvk/vHcd3tj7/DHVBoR5wvWT/MmRq2pj7HRJnwU=
 honnef.co/go/tools v0.0.0-20190102054323-c2f93a96b099/go.mod h1:rf3lG4BRIbNafJWhAfAdb/ePZxsR/4RtNHQocxwk9r4=
 honnef.co/go/tools v0.0.0-20190106161140-3f1c8253044a/go.mod h1:rf3lG4BRIbNafJWhAfAdb/ePZxsR/4RtNHQocxwk9r4=
 honnef.co/go/tools v0.0.0-20190418001031-e561f6794a2a/go.mod h1:rf3lG4BRIbNafJWhAfAdb/ePZxsR/4RtNHQocxwk9r4=
--- a/health/health.go
+++ b/health/health.go
@@ -6,6 +6,7 @@
 package health

 import (
+	"context"
 	"errors"
 	"fmt"
 	"maps"
@@ -987,8 +988,12 @@ func (t *Tracker) updateBuiltinWarnablesLocked() {
 	}

 	if t.lastLoginErr != nil {
+		var errMsg string
+		if !errors.Is(t.lastLoginErr, context.Canceled) {
+			errMsg = t.lastLoginErr.Error()
+		}
 		t.setUnhealthyLocked(LoginStateWarnable, Args{
-			ArgError: t.lastLoginErr.Error(),
+			ArgError: errMsg,
 		})
 		return
 	} else {
--- a/licenses/apple.md
+++ b/licenses/apple.md
@@ -65,8 +65,8 @@ See also the dependencies in the [Tailscale CLI][].
 - [github.com/tailscale/hujson](https://pkg.go.dev/github.com/tailscale/hujson) ([BSD-3-Clause](https://github.com/tailscale/hujson/blob/20486734a56a/LICENSE))
 - [github.com/tailscale/netlink](https://pkg.go.dev/github.com/tailscale/netlink) ([Apache-2.0](https://github.com/tailscale/netlink/blob/cabfb018fe85/LICENSE))
 - [github.com/tailscale/peercred](https://pkg.go.dev/github.com/tailscale/peercred) ([BSD-3-Clause](https://github.com/tailscale/peercred/blob/b535050b2aa4/LICENSE))
- - [github.com/tailscale/wireguard-go](https://pkg.go.dev/github.com/tailscale/wireguard-go) ([MIT](https://github.com/tailscale/wireguard-go/blob/2f5d148bcfe1/LICENSE))
- - [github.com/tailscale/xnet/webdav](https://pkg.go.dev/github.com/tailscale/xnet/webdav) ([BSD-3-Clause](https://github.com/tailscale/xnet/blob/62b9a7c569f9/LICENSE))
+ - [github.com/tailscale/wireguard-go](https://pkg.go.dev/github.com/tailscale/wireguard-go) ([MIT](https://github.com/tailscale/wireguard-go/blob/71393c576b98/LICENSE))
+ - [github.com/tailscale/xnet/webdav](https://pkg.go.dev/github.com/tailscale/xnet/webdav) ([BSD-3-Clause](https://github.com/tailscale/xnet/blob/8497ac4dab2e/LICENSE))
 - [github.com/tcnksm/go-httpstat](https://pkg.go.dev/github.com/tcnksm/go-httpstat) ([MIT](https://github.com/tcnksm/go-httpstat/blob/v0.2.0/LICENSE))
 - [github.com/u-root/uio](https://pkg.go.dev/github.com/u-root/uio) ([BSD-3-Clause](https://github.com/u-root/uio/blob/a3c409a6018e/LICENSE))
 - [github.com/vishvananda/netlink/nl](https://pkg.go.dev/github.com/vishvananda/netlink/nl) ([Apache-2.0](https://github.com/vishvananda/netlink/blob/v1.2.1-beta.2/LICENSE))
@@ -82,7 +82,7 @@ See also the dependencies in the [Tailscale CLI][].
 - [golang.org/x/term](https://pkg.go.dev/golang.org/x/term) ([BSD-3-Clause](https://cs.opensource.google/go/x/term/+/v0.22.0:LICENSE))
 - [golang.org/x/text](https://pkg.go.dev/golang.org/x/text) ([BSD-3-Clause](https://cs.opensource.google/go/x/text/+/v0.16.0:LICENSE))
 - [golang.org/x/time/rate](https://pkg.go.dev/golang.org/x/time/rate) ([BSD-3-Clause](https://cs.opensource.google/go/x/time/+/v0.5.0:LICENSE))
- - [gvisor.dev/gvisor/pkg](https://pkg.go.dev/gvisor.dev/gvisor/pkg) ([Apache-2.0](https://github.com/google/gvisor/blob/ee1e1f6070e3/LICENSE))
+ - [gvisor.dev/gvisor/pkg](https://pkg.go.dev/gvisor.dev/gvisor/pkg) ([Apache-2.0](https://github.com/google/gvisor/blob/64c016c92987/LICENSE))
 - [nhooyr.io/websocket](https://pkg.go.dev/nhooyr.io/websocket) ([ISC](https://github.com/nhooyr/websocket/blob/v1.8.10/LICENSE.txt))
 - [tailscale.com](https://pkg.go.dev/tailscale.com) ([BSD-3-Clause](https://github.com/tailscale/tailscale/blob/HEAD/LICENSE))

--- a/licenses/tailscale.md
+++ b/licenses/tailscale.md
@@ -84,8 +84,8 @@ Some packages may only be included on certain architectures or operating systems
 - [github.com/tailscale/peercred](https://pkg.go.dev/github.com/tailscale/peercred) ([BSD-3-Clause](https://github.com/tailscale/peercred/blob/b535050b2aa4/LICENSE))
 - [github.com/tailscale/web-client-prebuilt](https://pkg.go.dev/github.com/tailscale/web-client-prebuilt) ([BSD-3-Clause](https://github.com/tailscale/web-client-prebuilt/blob/5db17b287bf1/LICENSE))
 - [github.com/tailscale/wf](https://pkg.go.dev/github.com/tailscale/wf) ([BSD-3-Clause](https://github.com/tailscale/wf/blob/6fbb0a674ee6/LICENSE))
- - [github.com/tailscale/wireguard-go](https://pkg.go.dev/github.com/tailscale/wireguard-go) ([MIT](https://github.com/tailscale/wireguard-go/blob/2f5d148bcfe1/LICENSE))
- - [github.com/tailscale/xnet/webdav](https://pkg.go.dev/github.com/tailscale/xnet/webdav) ([BSD-3-Clause](https://github.com/tailscale/xnet/blob/62b9a7c569f9/LICENSE))
+ - [github.com/tailscale/wireguard-go](https://pkg.go.dev/github.com/tailscale/wireguard-go) ([MIT](https://github.com/tailscale/wireguard-go/blob/71393c576b98/LICENSE))
+ - [github.com/tailscale/xnet/webdav](https://pkg.go.dev/github.com/tailscale/xnet/webdav) ([BSD-3-Clause](https://github.com/tailscale/xnet/blob/8497ac4dab2e/LICENSE))
 - [github.com/tcnksm/go-httpstat](https://pkg.go.dev/github.com/tcnksm/go-httpstat) ([MIT](https://github.com/tcnksm/go-httpstat/blob/v0.2.0/LICENSE))
 - [github.com/toqueteos/webbrowser](https://pkg.go.dev/github.com/toqueteos/webbrowser) ([MIT](https://github.com/toqueteos/webbrowser/blob/v1.2.0/LICENSE.md))
 - [github.com/u-root/u-root/pkg/termios](https://pkg.go.dev/github.com/u-root/u-root/pkg/termios) ([BSD-3-Clause](https://github.com/u-root/u-root/blob/v0.12.0/LICENSE))
@@ -95,19 +95,19 @@ Some packages may only be included on certain architectures or operating systems
 - [github.com/x448/float16](https://pkg.go.dev/github.com/x448/float16) ([MIT](https://github.com/x448/float16/blob/v0.8.4/LICENSE))
 - [go4.org/mem](https://pkg.go.dev/go4.org/mem) ([Apache-2.0](https://github.com/go4org/mem/blob/4f986261bf13/LICENSE))
 - [go4.org/netipx](https://pkg.go.dev/go4.org/netipx) ([BSD-3-Clause](https://github.com/go4org/netipx/blob/fdeea329fbba/LICENSE))
- - [golang.org/x/crypto](https://pkg.go.dev/golang.org/x/crypto) ([BSD-3-Clause](https://cs.opensource.google/go/x/crypto/+/v0.24.0:LICENSE))
+ - [golang.org/x/crypto](https://pkg.go.dev/golang.org/x/crypto) ([BSD-3-Clause](https://cs.opensource.google/go/x/crypto/+/v0.25.0:LICENSE))
 - [golang.org/x/exp](https://pkg.go.dev/golang.org/x/exp) ([BSD-3-Clause](https://cs.opensource.google/go/x/exp/+/1b970713:LICENSE))
- - [golang.org/x/net](https://pkg.go.dev/golang.org/x/net) ([BSD-3-Clause](https://cs.opensource.google/go/x/net/+/v0.26.0:LICENSE))
+ - [golang.org/x/net](https://pkg.go.dev/golang.org/x/net) ([BSD-3-Clause](https://cs.opensource.google/go/x/net/+/v0.27.0:LICENSE))
 - [golang.org/x/oauth2](https://pkg.go.dev/golang.org/x/oauth2) ([BSD-3-Clause](https://cs.opensource.google/go/x/oauth2/+/v0.16.0:LICENSE))
 - [golang.org/x/sync](https://pkg.go.dev/golang.org/x/sync) ([BSD-3-Clause](https://cs.opensource.google/go/x/sync/+/v0.7.0:LICENSE))
- - [golang.org/x/sys](https://pkg.go.dev/golang.org/x/sys) ([BSD-3-Clause](https://cs.opensource.google/go/x/sys/+/v0.21.0:LICENSE))
- - [golang.org/x/term](https://pkg.go.dev/golang.org/x/term) ([BSD-3-Clause](https://cs.opensource.google/go/x/term/+/v0.21.0:LICENSE))
+ - [golang.org/x/sys](https://pkg.go.dev/golang.org/x/sys) ([BSD-3-Clause](https://cs.opensource.google/go/x/sys/+/v0.22.0:LICENSE))
+ - [golang.org/x/term](https://pkg.go.dev/golang.org/x/term) ([BSD-3-Clause](https://cs.opensource.google/go/x/term/+/v0.22.0:LICENSE))
 - [golang.org/x/text](https://pkg.go.dev/golang.org/x/text) ([BSD-3-Clause](https://cs.opensource.google/go/x/text/+/v0.16.0:LICENSE))
 - [golang.org/x/time/rate](https://pkg.go.dev/golang.org/x/time/rate) ([BSD-3-Clause](https://cs.opensource.google/go/x/time/+/v0.5.0:LICENSE))
 - [golang.zx2c4.com/wintun](https://pkg.go.dev/golang.zx2c4.com/wintun) ([MIT](https://git.zx2c4.com/wintun-go/tree/LICENSE?id=0fa3db229ce2))
 - [golang.zx2c4.com/wireguard/windows/tunnel/winipcfg](https://pkg.go.dev/golang.zx2c4.com/wireguard/windows/tunnel/winipcfg) ([MIT](https://git.zx2c4.com/wireguard-windows/tree/COPYING?h=v0.5.3))
- - [gvisor.dev/gvisor/pkg](https://pkg.go.dev/gvisor.dev/gvisor/pkg) ([Apache-2.0](https://github.com/google/gvisor/blob/ee1e1f6070e3/LICENSE))
- - [k8s.io/client-go/util/homedir](https://pkg.go.dev/k8s.io/client-go/util/homedir) ([Apache-2.0](https://github.com/kubernetes/client-go/blob/v0.30.1/LICENSE))
+ - [gvisor.dev/gvisor/pkg](https://pkg.go.dev/gvisor.dev/gvisor/pkg) ([Apache-2.0](https://github.com/google/gvisor/blob/64c016c92987/LICENSE))
+ - [k8s.io/client-go/util/homedir](https://pkg.go.dev/k8s.io/client-go/util/homedir) ([Apache-2.0](https://github.com/kubernetes/client-go/blob/v0.30.3/LICENSE))
 - [nhooyr.io/websocket](https://pkg.go.dev/nhooyr.io/websocket) ([ISC](https://github.com/nhooyr/websocket/blob/v1.8.10/LICENSE.txt))
 - [sigs.k8s.io/yaml](https://pkg.go.dev/sigs.k8s.io/yaml) ([Apache-2.0](https://github.com/kubernetes-sigs/yaml/blob/v1.4.0/LICENSE))
 - [sigs.k8s.io/yaml/goyaml.v2](https://pkg.go.dev/sigs.k8s.io/yaml/goyaml.v2) ([Apache-2.0](https://github.com/kubernetes-sigs/yaml/blob/v1.4.0/goyaml.v2/LICENSE))
--- a/licenses/windows.md
+++ b/licenses/windows.md
@@ -57,9 +57,9 @@ Windows][].  See also the dependencies in the [Tailscale CLI][].
 - [github.com/tailscale/go-winio](https://pkg.go.dev/github.com/tailscale/go-winio) ([MIT](https://github.com/tailscale/go-winio/blob/c4f33415bf55/LICENSE))
 - [github.com/tailscale/hujson](https://pkg.go.dev/github.com/tailscale/hujson) ([BSD-3-Clause](https://github.com/tailscale/hujson/blob/20486734a56a/LICENSE))
 - [github.com/tailscale/netlink](https://pkg.go.dev/github.com/tailscale/netlink) ([Apache-2.0](https://github.com/tailscale/netlink/blob/cabfb018fe85/LICENSE))
- - [github.com/tailscale/walk](https://pkg.go.dev/github.com/tailscale/walk) ([BSD-3-Clause](https://github.com/tailscale/walk/blob/7601212d8e23/LICENSE))
+ - [github.com/tailscale/walk](https://pkg.go.dev/github.com/tailscale/walk) ([BSD-3-Clause](https://github.com/tailscale/walk/blob/4327221bd339/LICENSE))
 - [github.com/tailscale/win](https://pkg.go.dev/github.com/tailscale/win) ([BSD-3-Clause](https://github.com/tailscale/win/blob/6580b55d49ca/LICENSE))
- - [github.com/tailscale/xnet/webdav](https://pkg.go.dev/github.com/tailscale/xnet/webdav) ([BSD-3-Clause](https://github.com/tailscale/xnet/blob/62b9a7c569f9/LICENSE))
+ - [github.com/tailscale/xnet/webdav](https://pkg.go.dev/github.com/tailscale/xnet/webdav) ([BSD-3-Clause](https://github.com/tailscale/xnet/blob/8497ac4dab2e/LICENSE))
 - [github.com/tc-hib/winres](https://pkg.go.dev/github.com/tc-hib/winres) ([0BSD](https://github.com/tc-hib/winres/blob/v0.2.1/LICENSE))
 - [github.com/vishvananda/netlink/nl](https://pkg.go.dev/github.com/vishvananda/netlink/nl) ([Apache-2.0](https://github.com/vishvananda/netlink/blob/v1.2.1-beta.2/LICENSE))
 - [github.com/vishvananda/netns](https://pkg.go.dev/github.com/vishvananda/netns) ([Apache-2.0](https://github.com/vishvananda/netns/blob/v0.0.4/LICENSE))
@@ -69,7 +69,7 @@ Windows][].  See also the dependencies in the [Tailscale CLI][].
 - [golang.org/x/crypto](https://pkg.go.dev/golang.org/x/crypto) ([BSD-3-Clause](https://cs.opensource.google/go/x/crypto/+/v0.25.0:LICENSE))
 - [golang.org/x/exp/constraints](https://pkg.go.dev/golang.org/x/exp/constraints) ([BSD-3-Clause](https://cs.opensource.google/go/x/exp/+/fe59bbe5:LICENSE))
 - [golang.org/x/image/bmp](https://pkg.go.dev/golang.org/x/image/bmp) ([BSD-3-Clause](https://cs.opensource.google/go/x/image/+/v0.18.0:LICENSE))
- - [golang.org/x/mod](https://pkg.go.dev/golang.org/x/mod) ([BSD-3-Clause](https://cs.opensource.google/go/x/mod/+/v0.18.0:LICENSE))
+ - [golang.org/x/mod](https://pkg.go.dev/golang.org/x/mod) ([BSD-3-Clause](https://cs.opensource.google/go/x/mod/+/v0.19.0:LICENSE))
 - [golang.org/x/net](https://pkg.go.dev/golang.org/x/net) ([BSD-3-Clause](https://cs.opensource.google/go/x/net/+/v0.27.0:LICENSE))
 - [golang.org/x/sync](https://pkg.go.dev/golang.org/x/sync) ([BSD-3-Clause](https://cs.opensource.google/go/x/sync/+/v0.7.0:LICENSE))
 - [golang.org/x/sys](https://pkg.go.dev/golang.org/x/sys) ([BSD-3-Clause](https://cs.opensource.google/go/x/sys/+/v0.22.0:LICENSE))
--- a/net/captivedetection/captivedetection.go
+++ b/net/captivedetection/captivedetection.go
@@ -106,13 +106,19 @@ func (d *Detector) detectCaptivePortalWithGOOS(ctx context.Context, netMon *netm
 	return false
 }

+// interfaceNameDoesNotNeedCaptiveDetection returns true if an interface does not require captive portal detection
+// based on its name. This is useful to avoid making unnecessary HTTP requests on interfaces that are known to not
+// require it. We also avoid making requests on the interface prefixes "pdp" and "rmnet", which are cellular data
+// interfaces on iOS and Android, respectively, and would be needlessly battery-draining.
 func interfaceNameDoesNotNeedCaptiveDetection(ifName string, goos string) bool {
 	ifName = strings.ToLower(ifName)
 	excludedPrefixes := []string{"tailscale", "tun", "tap", "docker", "kube", "wg"}
 	if goos == "windows" {
 		excludedPrefixes = append(excludedPrefixes, "loopback", "tunnel", "ppp", "isatap", "teredo", "6to4")
 	} else if goos == "darwin" || goos == "ios" {
-		excludedPrefixes = append(excludedPrefixes, "awdl", "bridge", "ap", "utun", "tap", "llw", "anpi", "lo", "stf", "gif", "xhc")
+		excludedPrefixes = append(excludedPrefixes, "pdp", "awdl", "bridge", "ap", "utun", "tap", "llw", "anpi", "lo", "stf", "gif", "xhc", "pktap")
+	} else if goos == "android" {
+		excludedPrefixes = append(excludedPrefixes, "rmnet", "p2p", "dummy", "sit")
 	}
 	for _, prefix := range excludedPrefixes {
 		if strings.HasPrefix(ifName, prefix) {
--- a/net/captivedetection/captivedetection_test.go
+++ b/net/captivedetection/captivedetection_test.go
@@ -9,6 +9,7 @@ import (
 	"sync"
 	"testing"

+	"tailscale.com/cmd/testwrapper/flakytest"
 	"tailscale.com/net/netmon"
 )

@@ -36,6 +37,7 @@ func TestDetectCaptivePortalReturnsFalse(t *testing.T) {
 }

 func TestAllEndpointsAreUpAndReturnExpectedResponse(t *testing.T) {
+	flakytest.Mark(t, "https://github.com/tailscale/tailscale/issues/13019")
 	d := NewDetector(t.Logf)
 	endpoints := availableEndpoints(nil, 0, t.Logf, runtime.GOOS)

--- a/net/dnsfallback/dnsfallback.go
+++ b/net/dnsfallback/dnsfallback.go
@@ -281,6 +281,7 @@ func lookup(ctx context.Context, host string, logf logger.Logf, ht *health.Track
 func bootstrapDNSMap(ctx context.Context, serverName string, serverIP netip.Addr, queryName string, logf logger.Logf, ht *health.Tracker, netMon *netmon.Monitor) (dnsMap, error) {
 	dialer := netns.NewDialer(logf, netMon)
 	tr := http.DefaultTransport.(*http.Transport).Clone()
+	tr.DisableKeepAlives = true // This transport is meant to be used once.
 	tr.Proxy = tshttpproxy.ProxyFromEnvironment
 	tr.DialContext = func(ctx context.Context, netw, addr string) (net.Conn, error) {
 		return dialer.DialContext(ctx, "tcp", net.JoinHostPort(serverIP.String(), "443"))
--- a/net/packet/checksum/checksum.go
+++ b/net/packet/checksum/checksum.go
@@ -61,7 +61,7 @@ func UpdateDstAddr(q *packet.Parsed, dst netip.Addr) {
 	b := q.Buffer()
 	if dst.Is6() {
 		v6 := dst.As16()
-		copy(b[24:36], v6[:])
+		copy(b[24:40], v6[:])
 		updateV6PacketChecksums(q, old, dst)
 	} else {
 		v4 := dst.As4()
--- a/net/packet/checksum/checksum_test.go
+++ b/net/packet/checksum/checksum_test.go
@@ -5,6 +5,7 @@ package checksum

 import (
 	"encoding/binary"
+	"math/rand/v2"
 	"net/netip"
 	"testing"

@@ -94,7 +95,7 @@ func TestHeaderChecksumsV4(t *testing.T) {
 }

 func TestNatChecksumsV6UDP(t *testing.T) {
-	a1, a2 := netip.MustParseAddr("a::1"), netip.MustParseAddr("b::1")
+	a1, a2 := randV6Addr(), randV6Addr()

 	// Make a fake UDP packet with 32 bytes of zeros as the datagram payload.
 	b := header.IPv6(make([]byte, header.IPv6MinimumSize+header.UDPMinimumSize+32))
@@ -124,25 +125,43 @@ func TestNatChecksumsV6UDP(t *testing.T) {
 	}

 	// Parse the packet.
-	var p packet.Parsed
+	var p, p2 packet.Parsed
 	p.Decode(b)
 	t.Log(p.String())

 	// Update the source address of the packet to be the same as the dest.
 	UpdateSrcAddr(&p, a2)
+	p2.Decode(p.Buffer())
+	if p2.Src.Addr() != a2 {
+		t.Fatalf("got %v, want %v", p2.Src, a2)
+	}
 	if !udp.IsChecksumValid(tcpip.AddrFrom16Slice(a2.AsSlice()), tcpip.AddrFrom16Slice(a2.AsSlice()), checksum.Checksum(b.Payload()[header.UDPMinimumSize:], 0)) {
 		t.Fatal("incorrect checksum after updating source address")
 	}

 	// Update the dest address of the packet to be the original source address.
 	UpdateDstAddr(&p, a1)
+	p2.Decode(p.Buffer())
+	if p2.Dst.Addr() != a1 {
+		t.Fatalf("got %v, want %v", p2.Dst, a1)
+	}
 	if !udp.IsChecksumValid(tcpip.AddrFrom16Slice(a2.AsSlice()), tcpip.AddrFrom16Slice(a1.AsSlice()), checksum.Checksum(b.Payload()[header.UDPMinimumSize:], 0)) {
 		t.Fatal("incorrect checksum after updating destination address")
 	}
 }

+func randV6Addr() netip.Addr {
+	a1, a2 := rand.Int64(), rand.Int64()
+	return netip.AddrFrom16([16]byte{
+		byte(a1 >> 56), byte(a1 >> 48), byte(a1 >> 40), byte(a1 >> 32),
+		byte(a1 >> 24), byte(a1 >> 16), byte(a1 >> 8), byte(a1),
+		byte(a2 >> 56), byte(a2 >> 48), byte(a2 >> 40), byte(a2 >> 32),
+		byte(a2 >> 24), byte(a2 >> 16), byte(a2 >> 8), byte(a2),
+	})
+}
+
 func TestNatChecksumsV6TCP(t *testing.T) {
-	a1, a2 := netip.MustParseAddr("a::1"), netip.MustParseAddr("b::1")
+	a1, a2 := randV6Addr(), randV6Addr()

 	// Make a fake TCP packet with no payload.
 	b := header.IPv6(make([]byte, header.IPv6MinimumSize+header.TCPMinimumSize))
@@ -178,18 +197,26 @@ func TestNatChecksumsV6TCP(t *testing.T) {
 	}

 	// Parse the packet.
-	var p packet.Parsed
+	var p, p2 packet.Parsed
 	p.Decode(b)
 	t.Log(p.String())

 	// Update the source address of the packet to be the same as the dest.
 	UpdateSrcAddr(&p, a2)
+	p2.Decode(p.Buffer())
+	if p2.Src.Addr() != a2 {
+		t.Fatalf("got %v, want %v", p2.Src, a2)
+	}
 	if !tcp.IsChecksumValid(tcpip.AddrFrom16Slice(a2.AsSlice()), tcpip.AddrFrom16Slice(a2.AsSlice()), 0, 0) {
 		t.Fatal("incorrect checksum after updating source address")
 	}

 	// Update the dest address of the packet to be the original source address.
 	UpdateDstAddr(&p, a1)
+	p2.Decode(p.Buffer())
+	if p2.Dst.Addr() != a1 {
+		t.Fatalf("got %v, want %v", p2.Dst, a1)
+	}
 	if !tcp.IsChecksumValid(tcpip.AddrFrom16Slice(a2.AsSlice()), tcpip.AddrFrom16Slice(a1.AsSlice()), 0, 0) {
 		t.Fatal("incorrect checksum after updating destination address")
 	}
--- a/net/socks5/socks5.go
+++ b/net/socks5/socks5.go
@@ -13,8 +13,10 @@
 package socks5

 import (
+	"bytes"
 	"context"
 	"encoding/binary"
+	"errors"
 	"fmt"
 	"io"
 	"log"
@@ -121,7 +123,7 @@ func (s *Server) Serve(l net.Listener) error {
 		}
 		go func() {
 			defer c.Close()
-			conn := &Conn{clientConn: c, srv: s}
+			conn := &Conn{logf: s.Logf, clientConn: c, srv: s}
 			err := conn.Run()
 			if err != nil {
 				s.logf("client connection failed: %v", err)
@@ -136,9 +138,12 @@ type Conn struct {
 	// The struct is filled by each of the internal
 	// methods in turn as the transaction progresses.

+	logf       logger.Logf
 	srv        *Server
 	clientConn net.Conn
 	request    *request
+
+	udpClientAddr net.Addr
 }

 // Run starts the new connection.
@@ -172,58 +177,59 @@ func (c *Conn) Run() error {
 func (c *Conn) handleRequest() error {
 	req, err := parseClientRequest(c.clientConn)
 	if err != nil {
-		res := &response{reply: generalFailure}
+		res := errorResponse(generalFailure)
 		buf, _ := res.marshal()
 		c.clientConn.Write(buf)
 		return err
 	}
-	if req.command != connect {
-		res := &response{reply: commandNotSupported}
+
+	c.request = req
+	switch req.command {
+	case connect:
+		return c.handleTCP()
+	case udpAssociate:
+		return c.handleUDP()
+	default:
+		res := errorResponse(commandNotSupported)
 		buf, _ := res.marshal()
 		c.clientConn.Write(buf)
 		return fmt.Errorf("unsupported command %v", req.command)
 	}
-	c.request = req
+}

+func (c *Conn) handleTCP() error {
 	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
 	defer cancel()
 	srv, err := c.srv.dial(
 		ctx,
 		"tcp",
-		net.JoinHostPort(c.request.destination, strconv.Itoa(int(c.request.port))),
+		c.request.destination.hostPort(),
 	)
 	if err != nil {
-		res := &response{reply: generalFailure}
+		res := errorResponse(generalFailure)
 		buf, _ := res.marshal()
 		c.clientConn.Write(buf)
 		return err
 	}
 	defer srv.Close()
-	serverAddr, serverPortStr, err := net.SplitHostPort(srv.LocalAddr().String())
+
+	localAddr := srv.LocalAddr().String()
+	serverAddr, serverPort, err := splitHostPort(localAddr)
 	if err != nil {
 		return err
 	}
-	serverPort, _ := strconv.Atoi(serverPortStr)

-	var bindAddrType addrType
-	if ip := net.ParseIP(serverAddr); ip != nil {
-		if ip.To4() != nil {
-			bindAddrType = ipv4
-		} else {
-			bindAddrType = ipv6
-		}
-	} else {
-		bindAddrType = domainName
-	}
 	res := &response{
-		reply:        success,
-		bindAddrType: bindAddrType,
-		bindAddr:     serverAddr,
-		bindPort:     uint16(serverPort),
+		reply: success,
+		bindAddr: socksAddr{
+			addrType: getAddrType(serverAddr),
+			addr:     serverAddr,
+			port:     serverPort,
+		},
 	}
 	buf, err := res.marshal()
 	if err != nil {
-		res = &response{reply: generalFailure}
+		res = errorResponse(generalFailure)
 		buf, _ = res.marshal()
 	}
 	c.clientConn.Write(buf)
@@ -246,6 +252,208 @@ func (c *Conn) handleRequest() error {
 	return <-errc
 }

+func (c *Conn) handleUDP() error {
+	// The DST.ADDR and DST.PORT fields contain the address and port that
+	// the client expects to use to send UDP datagrams on for the
+	// association. The server MAY use this information to limit access
+	// to the association.
+	// @see Page 6, https://datatracker.ietf.org/doc/html/rfc1928.
+	//
+	// We do NOT limit the access from the client currently in this implementation.
+	_ = c.request.destination
+
+	addr := c.clientConn.LocalAddr()
+	host, _, err := net.SplitHostPort(addr.String())
+	if err != nil {
+		return err
+	}
+	clientUDPConn, err := net.ListenPacket("udp", net.JoinHostPort(host, "0"))
+	if err != nil {
+		res := errorResponse(generalFailure)
+		buf, _ := res.marshal()
+		c.clientConn.Write(buf)
+		return err
+	}
+	defer clientUDPConn.Close()
+
+	serverUDPConn, err := net.ListenPacket("udp", "[::]:0")
+	if err != nil {
+		res := errorResponse(generalFailure)
+		buf, _ := res.marshal()
+		c.clientConn.Write(buf)
+		return err
+	}
+	defer serverUDPConn.Close()
+
+	bindAddr, bindPort, err := splitHostPort(clientUDPConn.LocalAddr().String())
+	if err != nil {
+		return err
+	}
+
+	res := &response{
+		reply: success,
+		bindAddr: socksAddr{
+			addrType: getAddrType(bindAddr),
+			addr:     bindAddr,
+			port:     bindPort,
+		},
+	}
+	buf, err := res.marshal()
+	if err != nil {
+		res = errorResponse(generalFailure)
+		buf, _ = res.marshal()
+	}
+	c.clientConn.Write(buf)
+
+	return c.transferUDP(c.clientConn, clientUDPConn, serverUDPConn)
+}
+
+func (c *Conn) transferUDP(associatedTCP net.Conn, clientConn net.PacketConn, targetConn net.PacketConn) error {
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+	const bufferSize = 8 * 1024
+	const readTimeout = 5 * time.Second
+
+	// client -> target
+	go func() {
+		defer cancel()
+		buf := make([]byte, bufferSize)
+		for {
+			select {
+			case <-ctx.Done():
+				return
+			default:
+				err := c.handleUDPRequest(clientConn, targetConn, buf, readTimeout)
+				if err != nil {
+					if isTimeout(err) {
+						continue
+					}
+					if errors.Is(err, net.ErrClosed) {
+						return
+					}
+					c.logf("udp transfer: handle udp request fail: %v", err)
+				}
+			}
+		}
+	}()
+
+	// target -> client
+	go func() {
+		defer cancel()
+		buf := make([]byte, bufferSize)
+		for {
+			select {
+			case <-ctx.Done():
+				return
+			default:
+				err := c.handleUDPResponse(targetConn, clientConn, buf, readTimeout)
+				if err != nil {
+					if isTimeout(err) {
+						continue
+					}
+					if errors.Is(err, net.ErrClosed) {
+						return
+					}
+					c.logf("udp transfer: handle udp response fail: %v", err)
+				}
+			}
+		}
+	}()
+
+	// A UDP association terminates when the TCP connection that the UDP
+	// ASSOCIATE request arrived on terminates. RFC1928
+	_, err := io.Copy(io.Discard, associatedTCP)
+	if err != nil {
+		err = fmt.Errorf("udp associated tcp conn: %w", err)
+	}
+	return err
+}
+
+func (c *Conn) handleUDPRequest(
+	clientConn net.PacketConn,
+	targetConn net.PacketConn,
+	buf []byte,
+	readTimeout time.Duration,
+) error {
+	// add a deadline for the read to avoid blocking forever
+	_ = clientConn.SetReadDeadline(time.Now().Add(readTimeout))
+	n, addr, err := clientConn.ReadFrom(buf)
+	if err != nil {
+		return fmt.Errorf("read from client: %w", err)
+	}
+	c.udpClientAddr = addr
+	req, data, err := parseUDPRequest(buf[:n])
+	if err != nil {
+		return fmt.Errorf("parse udp request: %w", err)
+	}
+	targetAddr, err := net.ResolveUDPAddr("udp", req.addr.hostPort())
+	if err != nil {
+		c.logf("resolve target addr fail: %v", err)
+	}
+
+	nn, err := targetConn.WriteTo(data, targetAddr)
+	if err != nil {
+		return fmt.Errorf("write to target %s fail: %w", targetAddr, err)
+	}
+	if nn != len(data) {
+		return fmt.Errorf("write to target %s fail: %w", targetAddr, io.ErrShortWrite)
+	}
+	return nil
+}
+
+func (c *Conn) handleUDPResponse(
+	targetConn net.PacketConn,
+	clientConn net.PacketConn,
+	buf []byte,
+	readTimeout time.Duration,
+) error {
+	// add a deadline for the read to avoid blocking forever
+	_ = targetConn.SetReadDeadline(time.Now().Add(readTimeout))
+	n, addr, err := targetConn.ReadFrom(buf)
+	if err != nil {
+		return fmt.Errorf("read from target: %w", err)
+	}
+	host, port, err := splitHostPort(addr.String())
+	if err != nil {
+		return fmt.Errorf("split host port: %w", err)
+	}
+	hdr := udpRequest{addr: socksAddr{addrType: getAddrType(host), addr: host, port: port}}
+	pkt, err := hdr.marshal()
+	if err != nil {
+		return fmt.Errorf("marshal udp request: %w", err)
+	}
+	data := append(pkt, buf[:n]...)
+	// use addr from client to send back
+	nn, err := clientConn.WriteTo(data, c.udpClientAddr)
+	if err != nil {
+		return fmt.Errorf("write to client: %w", err)
+	}
+	if nn != len(data) {
+		return fmt.Errorf("write to client: %w", io.ErrShortWrite)
+	}
+	return nil
+}
+
+func isTimeout(err error) bool {
+	terr, ok := errors.Unwrap(err).(interface{ Timeout() bool })
+	return ok && terr.Timeout()
+}
+
+func splitHostPort(hostport string) (host string, port uint16, err error) {
+	host, portStr, err := net.SplitHostPort(hostport)
+	if err != nil {
+		return "", 0, err
+	}
+	portInt, err := strconv.Atoi(portStr)
+	if err != nil {
+		return "", 0, err
+	}
+	if portInt < 0 || portInt > 65535 {
+		return "", 0, fmt.Errorf("invalid port number %d", portInt)
+	}
+	return host, uint16(portInt), nil
+}
+
 // parseClientGreeting parses a request initiation packet.
 func parseClientGreeting(r io.Reader, authMethod byte) error {
 	var hdr [2]byte
@@ -295,114 +503,118 @@ func parseClientAuth(r io.Reader) (usr, pwd string, err error) {
 	return string(usrBytes), string(pwdBytes), nil
 }

+func getAddrType(addr string) addrType {
+	if ip := net.ParseIP(addr); ip != nil {
+		if ip.To4() != nil {
+			return ipv4
+		}
+		return ipv6
+	}
+	return domainName
+}
+
 // request represents data contained within a SOCKS5
 // connection request packet.
 type request struct {
-	command      commandType
-	destination  string
-	port         uint16
-	destAddrType addrType
+	command     commandType
+	destination socksAddr
 }

 // parseClientRequest converts raw packet bytes into a
 // SOCKS5Request struct.
 func parseClientRequest(r io.Reader) (*request, error) {
-	var hdr [4]byte
+	var hdr [3]byte
 	_, err := io.ReadFull(r, hdr[:])
 	if err != nil {
 		return nil, fmt.Errorf("could not read packet header")
 	}
 	cmd := hdr[1]
-	destAddrType := addrType(hdr[3])

+	destination, err := parseSocksAddr(r)
+	return &request{
+		command:     commandType(cmd),
+		destination: destination,
+	}, err
+}
+
+type socksAddr struct {
+	addrType addrType
+	addr     string
+	port     uint16
+}
+
+var zeroSocksAddr = socksAddr{addrType: ipv4, addr: "0.0.0.0", port: 0}
+
+func parseSocksAddr(r io.Reader) (addr socksAddr, err error) {
+	var addrTypeData [1]byte
+	_, err = io.ReadFull(r, addrTypeData[:])
+	if err != nil {
+		return socksAddr{}, fmt.Errorf("could not read address type")
+	}
+
+	dstAddrType := addrType(addrTypeData[0])
 	var destination string
-	var port uint16
-
-	if destAddrType == ipv4 {
+	switch dstAddrType {
+	case ipv4:
 		var ip [4]byte
 		_, err = io.ReadFull(r, ip[:])
 		if err != nil {
-			return nil, fmt.Errorf("could not read IPv4 address")
+			return socksAddr{}, fmt.Errorf("could not read IPv4 address")
 		}
 		destination = net.IP(ip[:]).String()
-	} else if destAddrType == domainName {
+	case domainName:
 		var dstSizeByte [1]byte
 		_, err = io.ReadFull(r, dstSizeByte[:])
 		if err != nil {
-			return nil, fmt.Errorf("could not read domain name size")
+			return socksAddr{}, fmt.Errorf("could not read domain name size")
 		}
 		dstSize := int(dstSizeByte[0])
 		domainName := make([]byte, dstSize)
 		_, err = io.ReadFull(r, domainName)
 		if err != nil {
-			return nil, fmt.Errorf("could not read domain name")
+			return socksAddr{}, fmt.Errorf("could not read domain name")
 		}
 		destination = string(domainName)
-	} else if destAddrType == ipv6 {
+	case ipv6:
 		var ip [16]byte
 		_, err = io.ReadFull(r, ip[:])
 		if err != nil {
-			return nil, fmt.Errorf("could not read IPv6 address")
+			return socksAddr{}, fmt.Errorf("could not read IPv6 address")
 		}
 		destination = net.IP(ip[:]).String()
-	} else {
-		return nil, fmt.Errorf("unsupported address type")
+	default:
+		return socksAddr{}, fmt.Errorf("unsupported address type")
 	}
 	var portBytes [2]byte
 	_, err = io.ReadFull(r, portBytes[:])
 	if err != nil {
-		return nil, fmt.Errorf("could not read port")
+		return socksAddr{}, fmt.Errorf("could not read port")
 	}
-	port = binary.BigEndian.Uint16(portBytes[:])
-
-	return &request{
-		command:      commandType(cmd),
-		destination:  destination,
-		port:         port,
-		destAddrType: destAddrType,
+	port := binary.BigEndian.Uint16(portBytes[:])
+	return socksAddr{
+		addrType: dstAddrType,
+		addr:     destination,
+		port:     port,
 	}, nil
 }

-// response contains the contents of
-// a response packet sent from the proxy
-// to the client.
-type response struct {
-	reply        replyCode
-	bindAddrType addrType
-	bindAddr     string
-	bindPort     uint16
-}
-
-// marshal converts a SOCKS5Response struct into
-// a packet. If res.reply == Success, it may throw an error on
-// receiving an invalid bind address. Otherwise, it will not throw.
-func (res *response) marshal() ([]byte, error) {
-	pkt := make([]byte, 4)
-	pkt[0] = socks5Version
-	pkt[1] = byte(res.reply)
-	pkt[2] = 0 // null reserved byte
-	pkt[3] = byte(res.bindAddrType)
-
-	if res.reply != success {
-		return pkt, nil
-	}
-
+func (s socksAddr) marshal() ([]byte, error) {
 	var addr []byte
-	switch res.bindAddrType {
+	switch s.addrType {
 	case ipv4:
-		addr = net.ParseIP(res.bindAddr).To4()
+		addr = net.ParseIP(s.addr).To4()
 		if addr == nil {
 			return nil, fmt.Errorf("invalid IPv4 address for binding")
 		}
 	case domainName:
-		if len(res.bindAddr) > 255 {
+		if len(s.addr) > 255 {
 			return nil, fmt.Errorf("invalid domain name for binding")
 		}
-		addr = make([]byte, 0, len(res.bindAddr)+1)
-		addr = append(addr, byte(len(res.bindAddr)))
-		addr = append(addr, []byte(res.bindAddr)...)
+		addr = make([]byte, 0, len(s.addr)+1)
+		addr = append(addr, byte(len(s.addr)))
+		addr = append(addr, []byte(s.addr)...)
 	case ipv6:
-		addr = net.ParseIP(res.bindAddr).To16()
+		addr = net.ParseIP(s.addr).To16()
 		if addr == nil {
 			return nil, fmt.Errorf("invalid IPv6 address for binding")
 		}
@@ -410,8 +622,86 @@ func (res *response) marshal() ([]byte, error) {
 		return nil, fmt.Errorf("unsupported address type")
 	}

+	pkt := []byte{byte(s.addrType)}
 	pkt = append(pkt, addr...)
-	pkt = binary.BigEndian.AppendUint16(pkt, uint16(res.bindPort))
-
+	pkt = binary.BigEndian.AppendUint16(pkt, s.port)
 	return pkt, nil
 }
+func (s socksAddr) hostPort() string {
+	return net.JoinHostPort(s.addr, strconv.Itoa(int(s.port)))
+}
+
+// response contains the contents of
+// a response packet sent from the proxy
+// to the client.
+type response struct {
+	reply    replyCode
+	bindAddr socksAddr
+}
+
+func errorResponse(code replyCode) *response {
+	return &response{reply: code, bindAddr: zeroSocksAddr}
+}
+
+// marshal converts a SOCKS5Response struct into
+// a packet. If res.reply == Success, it may throw an error on
+// receiving an invalid bind address. Otherwise, it will not throw.
+func (res *response) marshal() ([]byte, error) {
+	pkt := make([]byte, 3)
+	pkt[0] = socks5Version
+	pkt[1] = byte(res.reply)
+	pkt[2] = 0 // null reserved byte
+
+	addrPkt, err := res.bindAddr.marshal()
+	if err != nil {
+		return nil, err
+	}
+
+	return append(pkt, addrPkt...), nil
+}
+
+type udpRequest struct {
+	frag byte
+	addr socksAddr
+}
+
+// +----+------+------+----------+----------+----------+
+// |RSV | FRAG | ATYP | DST.ADDR | DST.PORT |   DATA   |
+// +----+------+------+----------+----------+----------+
+// | 2  |  1   |  1   | Variable |    2     | Variable |
+// +----+------+------+----------+----------+----------+
+func parseUDPRequest(data []byte) (*udpRequest, []byte, error) {
+	if len(data) < 4 {
+		return nil, nil, fmt.Errorf("invalid packet length")
+	}
+
+	// reserved bytes
+	if !(data[0] == 0 && data[1] == 0) {
+		return nil, nil, fmt.Errorf("invalid udp request header")
+	}
+
+	frag := data[2]
+
+	reader := bytes.NewReader(data[3:])
+	addr, err := parseSocksAddr(reader)
+	bodyLen := reader.Len() // (*bytes.Reader).Len() return unread data length
+	body := data[len(data)-bodyLen:]
+	return &udpRequest{
+		frag: frag,
+		addr: addr,
+	}, body, err
+}
+
+func (u *udpRequest) marshal() ([]byte, error) {
+	pkt := make([]byte, 3)
+	pkt[0] = 0
+	pkt[1] = 0
+	pkt[2] = u.frag
+
+	addrPkt, err := u.addr.marshal()
+	if err != nil {
+		return nil, err
+	}
+
+	return append(pkt, addrPkt...), nil
+}
--- a/net/socks5/socks5_test.go
+++ b/net/socks5/socks5_test.go
@@ -4,6 +4,7 @@
 package socks5

 import (
+	"bytes"
 	"errors"
 	"fmt"
 	"io"
@@ -32,6 +33,19 @@ func backendServer(listener net.Listener) {
 	listener.Close()
 }

+func udpEchoServer(conn net.PacketConn) {
+	var buf [1024]byte
+	n, addr, err := conn.ReadFrom(buf[:])
+	if err != nil {
+		panic(err)
+	}
+	_, err = conn.WriteTo(buf[:n], addr)
+	if err != nil {
+		panic(err)
+	}
+	conn.Close()
+}
+
 func TestRead(t *testing.T) {
 	// backend server which we'll use SOCKS5 to connect to
 	listener, err := net.Listen("tcp", ":0")
@@ -152,3 +166,102 @@ func TestReadPassword(t *testing.T) {
 		t.Fatal(err)
 	}
 }
+
+func TestUDP(t *testing.T) {
+	// backend UDP server which we'll use SOCKS5 to connect to
+	listener, err := net.ListenPacket("udp", ":0")
+	if err != nil {
+		t.Fatal(err)
+	}
+	backendServerPort := listener.LocalAddr().(*net.UDPAddr).Port
+	go udpEchoServer(listener)
+
+	// SOCKS5 server
+	socks5, err := net.Listen("tcp", ":0")
+	if err != nil {
+		t.Fatal(err)
+	}
+	socks5Port := socks5.Addr().(*net.TCPAddr).Port
+	go socks5Server(socks5)
+
+	// net/proxy don't support UDP, so we need to manually send the SOCKS5 UDP request
+	conn, err := net.Dial("tcp", fmt.Sprintf("localhost:%d", socks5Port))
+	if err != nil {
+		t.Fatal(err)
+	}
+	_, err = conn.Write([]byte{0x05, 0x01, 0x00}) // client hello with no auth
+	if err != nil {
+		t.Fatal(err)
+	}
+	buf := make([]byte, 1024)
+	n, err := conn.Read(buf) // server hello
+	if err != nil {
+		t.Fatal(err)
+	}
+	if n != 2 || buf[0] != 0x05 || buf[1] != 0x00 {
+		t.Fatalf("got: %q want: 0x05 0x00", buf[:n])
+	}
+
+	targetAddr := socksAddr{
+		addrType: domainName,
+		addr:     "localhost",
+		port:     uint16(backendServerPort),
+	}
+	targetAddrPkt, err := targetAddr.marshal()
+	if err != nil {
+		t.Fatal(err)
+	}
+	_, err = conn.Write(append([]byte{0x05, 0x03, 0x00}, targetAddrPkt...)) // client reqeust
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	n, err = conn.Read(buf) // server response
+	if err != nil {
+		t.Fatal(err)
+	}
+	if n < 3 || !bytes.Equal(buf[:3], []byte{0x05, 0x00, 0x00}) {
+		t.Fatalf("got: %q want: 0x05 0x00 0x00", buf[:n])
+	}
+	udpProxySocksAddr, err := parseSocksAddr(bytes.NewReader(buf[3:n]))
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	udpProxyAddr, err := net.ResolveUDPAddr("udp", udpProxySocksAddr.hostPort())
+	if err != nil {
+		t.Fatal(err)
+	}
+	udpConn, err := net.DialUDP("udp", nil, udpProxyAddr)
+	if err != nil {
+		t.Fatal(err)
+	}
+	udpPayload, err := (&udpRequest{addr: targetAddr}).marshal()
+	if err != nil {
+		t.Fatal(err)
+	}
+	udpPayload = append(udpPayload, []byte("Test")...)
+	_, err = udpConn.Write(udpPayload) // send udp package
+	if err != nil {
+		t.Fatal(err)
+	}
+	n, _, err = udpConn.ReadFrom(buf)
+	if err != nil {
+		t.Fatal(err)
+	}
+	_, responseBody, err := parseUDPRequest(buf[:n]) // read udp response
+	if err != nil {
+		t.Fatal(err)
+	}
+	if string(responseBody) != "Test" {
+		t.Fatalf("got: %q want: Test", responseBody)
+	}
+	err = udpConn.Close()
+	if err != nil {
+		t.Fatal(err)
+	}
+	err = conn.Close()
+	if err != nil {
+		t.Fatal(err)
+	}
+}
--- a/net/tsdial/tsdial.go
+++ b/net/tsdial/tsdial.go
@@ -166,6 +166,7 @@ func (d *Dialer) Close() error {
 		c.Close()
 	}
 	d.activeSysConns = nil
+	d.PeerAPITransport().CloseIdleConnections()
 	return nil
 }

--- a/net/tstun/wrap.go
+++ b/net/tstun/wrap.go
@@ -162,6 +162,10 @@ type Wrapper struct {
 	PreFilterPacketInboundFromWireGuard FilterFunc
 	// PostFilterPacketInboundFromWireGuard is the inbound filter function that runs after the main filter.
 	PostFilterPacketInboundFromWireGuard FilterFunc
+	// EndPacketVectorInboundFromWireGuardFlush is a function that runs after all packets in a given vector
+	// have been handled by all filters. Filters may queue packets for the purposes of GRO, requiring an
+	// explicit flush.
+	EndPacketVectorInboundFromWireGuardFlush func()
 	// PreFilterPacketOutboundToWireGuardNetstackIntercept is a filter function that runs before the main filter
 	// for packets from the local system. This filter is populated by netstack to hook
 	// packets that should be handled by netstack. If set, this filter runs before
@@ -1179,6 +1183,9 @@ func (t *Wrapper) Write(buffs [][]byte, offset int) (int, error) {
 			}
 		}
 	}
+	if t.EndPacketVectorInboundFromWireGuardFlush != nil {
+		t.EndPacketVectorInboundFromWireGuardFlush()
+	}
 	if t.disableFilter {
 		i = len(buffs)
 	}
--- a/prober/prober.go
+++ b/prober/prober.go
@@ -7,19 +7,26 @@
 package prober

 import (
+	"container/ring"
 	"context"
-	"errors"
+	"encoding/json"
 	"fmt"
 	"hash/fnv"
 	"log"
 	"maps"
 	"math/rand"
+	"net/http"
 	"sync"
 	"time"

 	"github.com/prometheus/client_golang/prometheus"
+	"tailscale.com/tsweb"
 )

+// recentHistSize is the number of recent probe results and latencies to keep
+// in memory.
+const recentHistSize = 10
+
 // ProbeClass defines a probe of a specific type: a probing function that will
 // be regularly ran, and metric labels that will be added automatically to all
 // probes using this class.
@@ -106,6 +113,14 @@ func (p *Prober) Run(name string, interval time.Duration, labels Labels, pc Prob
 		l[k] = v
 	}

+	probe := newProbe(p, name, interval, l, pc)
+	p.probes[name] = probe
+	go probe.loop()
+	return probe
+}
+
+// newProbe creates a new Probe with the given parameters, but does not start it.
+func newProbe(p *Prober, name string, interval time.Duration, l prometheus.Labels, pc ProbeClass) *Probe {
 	ctx, cancel := context.WithCancel(context.Background())
 	probe := &Probe{
 		prober:  p,
@@ -117,6 +132,9 @@ func (p *Prober) Run(name string, interval time.Duration, labels Labels, pc Prob
 		probeClass:   pc,
 		interval:     interval,
 		initialDelay: initialDelay(name, interval),
+		successHist:  ring.New(recentHistSize),
+		latencyHist:  ring.New(recentHistSize),
+
 		metrics:      prometheus.NewRegistry(),
 		metricLabels: l,
 		mInterval:    prometheus.NewDesc("interval_secs", "Probe interval in seconds", nil, l),
@@ -131,15 +149,14 @@ func (p *Prober) Run(name string, interval time.Duration, labels Labels, pc Prob
 			Name: "seconds_total", Help: "Total amount of time spent executing the probe", ConstLabels: l,
 		}, []string{"status"}),
 	}
-
-	prometheus.WrapRegistererWithPrefix(p.namespace+"_", p.metrics).MustRegister(probe.metrics)
+	if p.metrics != nil {
+		prometheus.WrapRegistererWithPrefix(p.namespace+"_", p.metrics).MustRegister(probe.metrics)
+	}
 	probe.metrics.MustRegister(probe)
-
-	p.probes[name] = probe
-	go probe.loop()
 	return probe
 }

+// unregister removes a probe from the prober's internal state.
 func (p *Prober) unregister(probe *Probe) {
 	p.mu.Lock()
 	defer p.mu.Unlock()
@@ -206,6 +223,7 @@ type Probe struct {
 	ctx     context.Context
 	cancel  context.CancelFunc // run to initiate shutdown
 	stopped chan struct{}      // closed when shutdown is complete
+	runMu   sync.Mutex         // ensures only one probe runs at a time

 	name         string
 	probeClass   ProbeClass
@@ -232,6 +250,10 @@ type Probe struct {
 	latency   time.Duration // last successful probe latency
 	succeeded bool          // whether the last doProbe call succeeded
 	lastErr   error
+
+	// History of recent probe results and latencies.
+	successHist *ring.Ring
+	latencyHist *ring.Ring
 }

 // Close shuts down the Probe and unregisters it from its Prober.
@@ -278,13 +300,17 @@ func (p *Probe) loop() {
 	}
 }

-// run invokes fun and records the results.
+// run invokes the probe function and records the result. It returns the probe
+// result and an error if the probe failed.
 //
-// fun is invoked with a timeout slightly less than interval, so that
-// the probe either succeeds or fails before the next cycle is
-// scheduled to start.
-func (p *Probe) run() {
-	start := p.recordStart()
+// The probe function is invoked with a timeout slightly less than interval, so
+// that the probe either succeeds or fails before the next cycle is scheduled to
+// start.
+func (p *Probe) run() (pi ProbeInfo, err error) {
+	p.runMu.Lock()
+	defer p.runMu.Unlock()
+
+	p.recordStart()
 	defer func() {
 		// Prevent a panic within one probe function from killing the
 		// entire prober, so that a single buggy probe doesn't destroy
@@ -293,29 +319,30 @@ func (p *Probe) run() {
 		// alert for debugging.
 		if r := recover(); r != nil {
 			log.Printf("probe %s panicked: %v", p.name, r)
-			p.recordEnd(start, errors.New("panic"))
+			err = fmt.Errorf("panic: %v", r)
+			p.recordEnd(err)
 		}
 	}()
 	timeout := time.Duration(float64(p.interval) * 0.8)
 	ctx, cancel := context.WithTimeout(p.ctx, timeout)
 	defer cancel()

-	err := p.probeClass.Probe(ctx)
-	p.recordEnd(start, err)
+	err = p.probeClass.Probe(ctx)
+	p.recordEnd(err)
 	if err != nil {
 		log.Printf("probe %s: %v", p.name, err)
 	}
+	pi = p.probeInfoLocked()
+	return
 }

-func (p *Probe) recordStart() time.Time {
-	st := p.prober.now()
+func (p *Probe) recordStart() {
 	p.mu.Lock()
-	defer p.mu.Unlock()
-	p.start = st
-	return st
+	p.start = p.prober.now()
+	p.mu.Unlock()
 }

-func (p *Probe) recordEnd(start time.Time, err error) {
+func (p *Probe) recordEnd(err error) {
 	end := p.prober.now()
 	p.mu.Lock()
 	defer p.mu.Unlock()
@@ -327,22 +354,55 @@ func (p *Probe) recordEnd(start time.Time, err error) {
 		p.latency = latency
 		p.mAttempts.WithLabelValues("ok").Inc()
 		p.mSeconds.WithLabelValues("ok").Add(latency.Seconds())
+		p.latencyHist.Value = latency
+		p.latencyHist = p.latencyHist.Next()
 	} else {
 		p.latency = 0
 		p.mAttempts.WithLabelValues("fail").Inc()
 		p.mSeconds.WithLabelValues("fail").Add(latency.Seconds())
 	}
+	p.successHist.Value = p.succeeded
+	p.successHist = p.successHist.Next()
 }

-// ProbeInfo is the state of a Probe.
+// ProbeInfo is a snapshot of the configuration and state of a Probe.
 type ProbeInfo struct {
-	Start   time.Time
-	End     time.Time
-	Latency string
-	Result  bool
-	Error   string
+	Name            string
+	Class           string
+	Interval        time.Duration
+	Labels          map[string]string
+	Start           time.Time
+	End             time.Time
+	Latency         time.Duration
+	Result          bool
+	Error           string
+	RecentResults   []bool
+	RecentLatencies []time.Duration
 }

+// RecentSuccessRatio returns the success ratio of the probe in the recent history.
+func (pb ProbeInfo) RecentSuccessRatio() float64 {
+	if len(pb.RecentResults) == 0 {
+		return 0
+	}
+	var sum int
+	for _, r := range pb.RecentResults {
+		if r {
+			sum++
+		}
+	}
+	return float64(sum) / float64(len(pb.RecentResults))
+}
+
+// RecentMedianLatency returns the median latency of the probe in the recent history.
+func (pb ProbeInfo) RecentMedianLatency() time.Duration {
+	if len(pb.RecentLatencies) == 0 {
+		return 0
+	}
+	return pb.RecentLatencies[len(pb.RecentLatencies)/2]
+}
+
+// ProbeInfo returns the state of all probes.
 func (p *Prober) ProbeInfo() map[string]ProbeInfo {
 	out := map[string]ProbeInfo{}

@@ -352,26 +412,100 @@ func (p *Prober) ProbeInfo() map[string]ProbeInfo {
 		probes = append(probes, probe)
 	}
 	p.mu.Unlock()
-
 	for _, probe := range probes {
 		probe.mu.Lock()
-		inf := ProbeInfo{
-			Start:  probe.start,
-			End:    probe.end,
-			Result: probe.succeeded,
-		}
-		if probe.lastErr != nil {
-			inf.Error = probe.lastErr.Error()
-		}
-		if probe.latency > 0 {
-			inf.Latency = probe.latency.String()
-		}
-		out[probe.name] = inf
+		out[probe.name] = probe.probeInfoLocked()
 		probe.mu.Unlock()
 	}
 	return out
 }

+// probeInfoLocked returns the state of the probe.
+func (probe *Probe) probeInfoLocked() ProbeInfo {
+	inf := ProbeInfo{
+		Name:     probe.name,
+		Class:    probe.probeClass.Class,
+		Interval: probe.interval,
+		Labels:   probe.metricLabels,
+		Start:    probe.start,
+		End:      probe.end,
+		Result:   probe.succeeded,
+	}
+	if probe.lastErr != nil {
+		inf.Error = probe.lastErr.Error()
+	}
+	if probe.latency > 0 {
+		inf.Latency = probe.latency
+	}
+	probe.latencyHist.Do(func(v any) {
+		if l, ok := v.(time.Duration); ok {
+			inf.RecentLatencies = append(inf.RecentLatencies, l)
+		}
+	})
+	probe.successHist.Do(func(v any) {
+		if r, ok := v.(bool); ok {
+			inf.RecentResults = append(inf.RecentResults, r)
+		}
+	})
+	return inf
+}
+
+// RunHandlerResponse is the JSON response format for the RunHandler.
+type RunHandlerResponse struct {
+	ProbeInfo             ProbeInfo
+	PreviousSuccessRatio  float64
+	PreviousMedianLatency time.Duration
+}
+
+// RunHandler runs a probe by name and returns the result as an HTTP response.
+func (p *Prober) RunHandler(w http.ResponseWriter, r *http.Request) error {
+	// Look up prober by name.
+	name := r.FormValue("name")
+	if name == "" {
+		return tsweb.Error(http.StatusBadRequest, "missing name parameter", nil)
+	}
+	p.mu.Lock()
+	probe, ok := p.probes[name]
+	p.mu.Unlock()
+	if !ok {
+		return tsweb.Error(http.StatusNotFound, fmt.Sprintf("unknown probe %q", name), nil)
+	}
+
+	probe.mu.Lock()
+	prevInfo := probe.probeInfoLocked()
+	probe.mu.Unlock()
+
+	info, err := probe.run()
+	respStatus := http.StatusOK
+	if err != nil {
+		respStatus = http.StatusFailedDependency
+	}
+
+	// Return serialized JSON response if the client requested JSON
+	if r.Header.Get("Accept") == "application/json" {
+		resp := &RunHandlerResponse{
+			ProbeInfo:             info,
+			PreviousSuccessRatio:  prevInfo.RecentSuccessRatio(),
+			PreviousMedianLatency: prevInfo.RecentMedianLatency(),
+		}
+		w.WriteHeader(respStatus)
+		w.Header().Set("Content-Type", "application/json")
+		if err := json.NewEncoder(w).Encode(resp); err != nil {
+			return tsweb.Error(http.StatusInternalServerError, "error encoding JSON response", err)
+		}
+		return nil
+	}
+
+	stats := fmt.Sprintf("Previous runs: success rate %d%%, median latency %v",
+		int(prevInfo.RecentSuccessRatio()*100), prevInfo.RecentMedianLatency())
+	if err != nil {
+		return tsweb.Error(respStatus, fmt.Sprintf("Probe failed: %s\n%s", err.Error(), stats), err)
+	}
+	w.WriteHeader(respStatus)
+	w.Write([]byte(fmt.Sprintf("Probe succeeded in %v\n%s", info.Latency, stats)))
+	return nil
+}
+
 // Describe implements prometheus.Collector.
 func (p *Probe) Describe(ch chan<- *prometheus.Desc) {
 	ch <- p.mInterval
--- a/prober/prober_test.go
+++ b/prober/prober_test.go
@@ -5,16 +5,22 @@ package prober

 import (
 	"context"
+	"encoding/json"
 	"errors"
 	"fmt"
+	"io"
+	"net/http/httptest"
 	"strings"
 	"sync"
 	"sync/atomic"
 	"testing"
 	"time"

+	"github.com/google/go-cmp/cmp"
+	"github.com/google/go-cmp/cmp/cmpopts"
 	"github.com/prometheus/client_golang/prometheus/testutil"
 	"tailscale.com/tstest"
+	"tailscale.com/tsweb"
 )

 const (
@@ -292,6 +298,254 @@ func TestOnceMode(t *testing.T) {
 	}
 }

+func TestProberProbeInfo(t *testing.T) {
+	clk := newFakeTime()
+	p := newForTest(clk.Now, clk.NewTicker).WithOnce(true)
+
+	p.Run("probe1", probeInterval, nil, FuncProbe(func(context.Context) error {
+		clk.Advance(500 * time.Millisecond)
+		return nil
+	}))
+	p.Run("probe2", probeInterval, nil, FuncProbe(func(context.Context) error { return fmt.Errorf("error2") }))
+	p.Wait()
+
+	info := p.ProbeInfo()
+	wantInfo := map[string]ProbeInfo{
+		"probe1": {
+			Name:            "probe1",
+			Interval:        probeInterval,
+			Labels:          map[string]string{"class": "", "name": "probe1"},
+			Latency:         500 * time.Millisecond,
+			Result:          true,
+			RecentResults:   []bool{true},
+			RecentLatencies: []time.Duration{500 * time.Millisecond},
+		},
+		"probe2": {
+			Name:            "probe2",
+			Interval:        probeInterval,
+			Labels:          map[string]string{"class": "", "name": "probe2"},
+			Error:           "error2",
+			RecentResults:   []bool{false},
+			RecentLatencies: nil, // no latency for failed probes
+		},
+	}
+
+	if diff := cmp.Diff(wantInfo, info, cmpopts.IgnoreFields(ProbeInfo{}, "Start", "End")); diff != "" {
+		t.Fatalf("unexpected ProbeInfo (-want +got):\n%s", diff)
+	}
+}
+
+func TestProbeInfoRecent(t *testing.T) {
+	type probeResult struct {
+		latency time.Duration
+		err     error
+	}
+	tests := []struct {
+		name                    string
+		results                 []probeResult
+		wantProbeInfo           ProbeInfo
+		wantRecentSuccessRatio  float64
+		wantRecentMedianLatency time.Duration
+	}{
+		{
+			name:                    "no_runs",
+			wantProbeInfo:           ProbeInfo{},
+			wantRecentSuccessRatio:  0,
+			wantRecentMedianLatency: 0,
+		},
+		{
+			name:    "single_success",
+			results: []probeResult{{latency: 100 * time.Millisecond, err: nil}},
+			wantProbeInfo: ProbeInfo{
+				Latency:         100 * time.Millisecond,
+				Result:          true,
+				RecentResults:   []bool{true},
+				RecentLatencies: []time.Duration{100 * time.Millisecond},
+			},
+			wantRecentSuccessRatio:  1,
+			wantRecentMedianLatency: 100 * time.Millisecond,
+		},
+		{
+			name:    "single_failure",
+			results: []probeResult{{latency: 100 * time.Millisecond, err: errors.New("error123")}},
+			wantProbeInfo: ProbeInfo{
+				Result:          false,
+				RecentResults:   []bool{false},
+				RecentLatencies: nil,
+				Error:           "error123",
+			},
+			wantRecentSuccessRatio:  0,
+			wantRecentMedianLatency: 0,
+		},
+		{
+			name: "recent_mix",
+			results: []probeResult{
+				{latency: 10 * time.Millisecond, err: errors.New("error1")},
+				{latency: 20 * time.Millisecond, err: nil},
+				{latency: 30 * time.Millisecond, err: nil},
+				{latency: 40 * time.Millisecond, err: errors.New("error4")},
+				{latency: 50 * time.Millisecond, err: nil},
+				{latency: 60 * time.Millisecond, err: nil},
+				{latency: 70 * time.Millisecond, err: errors.New("error7")},
+				{latency: 80 * time.Millisecond, err: nil},
+			},
+			wantProbeInfo: ProbeInfo{
+				Result:        true,
+				Latency:       80 * time.Millisecond,
+				RecentResults: []bool{false, true, true, false, true, true, false, true},
+				RecentLatencies: []time.Duration{
+					20 * time.Millisecond,
+					30 * time.Millisecond,
+					50 * time.Millisecond,
+					60 * time.Millisecond,
+					80 * time.Millisecond,
+				},
+			},
+			wantRecentSuccessRatio:  0.625,
+			wantRecentMedianLatency: 50 * time.Millisecond,
+		},
+		{
+			name: "only_last_10",
+			results: []probeResult{
+				{latency: 10 * time.Millisecond, err: errors.New("old_error")},
+				{latency: 20 * time.Millisecond, err: nil},
+				{latency: 30 * time.Millisecond, err: nil},
+				{latency: 40 * time.Millisecond, err: nil},
+				{latency: 50 * time.Millisecond, err: nil},
+				{latency: 60 * time.Millisecond, err: nil},
+				{latency: 70 * time.Millisecond, err: nil},
+				{latency: 80 * time.Millisecond, err: nil},
+				{latency: 90 * time.Millisecond, err: nil},
+				{latency: 100 * time.Millisecond, err: nil},
+				{latency: 110 * time.Millisecond, err: nil},
+			},
+			wantProbeInfo: ProbeInfo{
+				Result:        true,
+				Latency:       110 * time.Millisecond,
+				RecentResults: []bool{true, true, true, true, true, true, true, true, true, true},
+				RecentLatencies: []time.Duration{
+					20 * time.Millisecond,
+					30 * time.Millisecond,
+					40 * time.Millisecond,
+					50 * time.Millisecond,
+					60 * time.Millisecond,
+					70 * time.Millisecond,
+					80 * time.Millisecond,
+					90 * time.Millisecond,
+					100 * time.Millisecond,
+					110 * time.Millisecond,
+				},
+			},
+			wantRecentSuccessRatio:  1,
+			wantRecentMedianLatency: 70 * time.Millisecond,
+		},
+	}
+
+	clk := newFakeTime()
+	p := newForTest(clk.Now, clk.NewTicker).WithOnce(true)
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			probe := newProbe(p, "", probeInterval, nil, FuncProbe(func(context.Context) error { return nil }))
+			for _, r := range tt.results {
+				probe.recordStart()
+				clk.Advance(r.latency)
+				probe.recordEnd(r.err)
+			}
+			info := probe.probeInfoLocked()
+			if diff := cmp.Diff(tt.wantProbeInfo, info, cmpopts.IgnoreFields(ProbeInfo{}, "Start", "End", "Interval")); diff != "" {
+				t.Fatalf("unexpected ProbeInfo (-want +got):\n%s", diff)
+			}
+			if got := info.RecentSuccessRatio(); got != tt.wantRecentSuccessRatio {
+				t.Errorf("recentSuccessRatio() = %v, want %v", got, tt.wantRecentSuccessRatio)
+			}
+			if got := info.RecentMedianLatency(); got != tt.wantRecentMedianLatency {
+				t.Errorf("recentMedianLatency() = %v, want %v", got, tt.wantRecentMedianLatency)
+			}
+		})
+	}
+}
+
+func TestProberRunHandler(t *testing.T) {
+	clk := newFakeTime()
+
+	tests := []struct {
+		name                  string
+		probeFunc             func(context.Context) error
+		wantResponseCode      int
+		wantJSONResponse      RunHandlerResponse
+		wantPlaintextResponse string
+	}{
+		{
+			name:             "success",
+			probeFunc:        func(context.Context) error { return nil },
+			wantResponseCode: 200,
+			wantJSONResponse: RunHandlerResponse{
+				ProbeInfo: ProbeInfo{
+					Name:          "success",
+					Interval:      probeInterval,
+					Result:        true,
+					RecentResults: []bool{true, true},
+				},
+				PreviousSuccessRatio: 1,
+			},
+			wantPlaintextResponse: "Probe succeeded",
+		},
+		{
+			name:             "failure",
+			probeFunc:        func(context.Context) error { return fmt.Errorf("error123") },
+			wantResponseCode: 424,
+			wantJSONResponse: RunHandlerResponse{
+				ProbeInfo: ProbeInfo{
+					Name:          "failure",
+					Interval:      probeInterval,
+					Result:        false,
+					Error:         "error123",
+					RecentResults: []bool{false, false},
+				},
+			},
+			wantPlaintextResponse: "Probe failed",
+		},
+	}
+
+	for _, tt := range tests {
+		for _, reqJSON := range []bool{true, false} {
+			t.Run(fmt.Sprintf("%s_json-%v", tt.name, reqJSON), func(t *testing.T) {
+				p := newForTest(clk.Now, clk.NewTicker).WithOnce(true)
+				probe := p.Run(tt.name, probeInterval, nil, FuncProbe(tt.probeFunc))
+				defer probe.Close()
+				<-probe.stopped // wait for the first run.
+
+				w := httptest.NewRecorder()
+
+				req := httptest.NewRequest("GET", "/prober/run/?name="+tt.name, nil)
+				if reqJSON {
+					req.Header.Set("Accept", "application/json")
+				}
+				tsweb.StdHandler(tsweb.ReturnHandlerFunc(p.RunHandler), tsweb.HandlerOptions{}).ServeHTTP(w, req)
+				if w.Result().StatusCode != tt.wantResponseCode {
+					t.Errorf("unexpected response code: got %d, want %d", w.Code, tt.wantResponseCode)
+				}
+
+				if reqJSON {
+					var gotJSON RunHandlerResponse
+					if err := json.Unmarshal(w.Body.Bytes(), &gotJSON); err != nil {
+						t.Fatalf("failed to unmarshal JSON response: %v; body: %s", err, w.Body.String())
+					}
+					if diff := cmp.Diff(tt.wantJSONResponse, gotJSON, cmpopts.IgnoreFields(ProbeInfo{}, "Start", "End", "Labels", "RecentLatencies")); diff != "" {
+						t.Errorf("unexpected JSON response (-want +got):\n%s", diff)
+					}
+				} else {
+					body, _ := io.ReadAll(w.Result().Body)
+					if !strings.Contains(string(body), tt.wantPlaintextResponse) {
+						t.Errorf("unexpected response body: got %q, want to contain %q", body, tt.wantPlaintextResponse)
+					}
+				}
+			})
+		}
+	}
+
+}
+
 type fakeTicker struct {
 	ch       chan time.Time
 	interval time.Duration
--- a/prober/status.go
+++ b/prober/status.go
@@ -0,0 +1,124 @@
+// Copyright (c) Tailscale Inc & AUTHORS
+// SPDX-License-Identifier: BSD-3-Clause
+
+package prober
+
+import (
+	"embed"
+	"fmt"
+	"html/template"
+	"net/http"
+	"strings"
+	"time"
+
+	"tailscale.com/tsweb"
+	"tailscale.com/util/mak"
+)
+
+//go:embed status.html
+var statusFiles embed.FS
+var statusTpl = template.Must(template.ParseFS(statusFiles, "status.html"))
+
+type statusHandlerOpt func(*statusHandlerParams)
+type statusHandlerParams struct {
+	title string
+
+	pageLinks  map[string]string
+	probeLinks map[string]string
+}
+
+// WithTitle sets the title of the status page.
+func WithTitle(title string) statusHandlerOpt {
+	return func(opts *statusHandlerParams) {
+		opts.title = title
+	}
+}
+
+// WithPageLink adds a top-level link to the status page.
+func WithPageLink(text, url string) statusHandlerOpt {
+	return func(opts *statusHandlerParams) {
+		mak.Set(&opts.pageLinks, text, url)
+	}
+}
+
+// WithProbeLink adds a link to each probe on the status page.
+// The textTpl and urlTpl are Go templates that will be rendered
+// with the respective ProbeInfo struct as the data.
+func WithProbeLink(textTpl, urlTpl string) statusHandlerOpt {
+	return func(opts *statusHandlerParams) {
+		mak.Set(&opts.probeLinks, textTpl, urlTpl)
+	}
+}
+
+// StatusHandler is a handler for the probe overview HTTP endpoint.
+// It shows a list of probes and their current status.
+func (p *Prober) StatusHandler(opts ...statusHandlerOpt) tsweb.ReturnHandlerFunc {
+	params := &statusHandlerParams{
+		title: "Prober Status",
+	}
+	for _, opt := range opts {
+		opt(params)
+	}
+	return func(w http.ResponseWriter, r *http.Request) error {
+		type probeStatus struct {
+			ProbeInfo
+			TimeSinceLast time.Duration
+			Links         map[string]template.URL
+		}
+		vars := struct {
+			Title           string
+			Links           map[string]template.URL
+			TotalProbes     int64
+			UnhealthyProbes int64
+			Probes          map[string]probeStatus
+		}{
+			Title: params.title,
+		}
+
+		for text, url := range params.pageLinks {
+			mak.Set(&vars.Links, text, template.URL(url))
+		}
+
+		for name, info := range p.ProbeInfo() {
+			vars.TotalProbes++
+			if !info.Result {
+				vars.UnhealthyProbes++
+			}
+			s := probeStatus{ProbeInfo: info}
+			if !info.End.IsZero() {
+				s.TimeSinceLast = time.Since(info.End)
+			}
+			for textTpl, urlTpl := range params.probeLinks {
+				text, err := renderTemplate(textTpl, info)
+				if err != nil {
+					return tsweb.Error(500, err.Error(), err)
+				}
+				url, err := renderTemplate(urlTpl, info)
+				if err != nil {
+					return tsweb.Error(500, err.Error(), err)
+				}
+				mak.Set(&s.Links, text, template.URL(url))
+			}
+			mak.Set(&vars.Probes, name, s)
+		}
+
+		if err := statusTpl.ExecuteTemplate(w, "status", vars); err != nil {
+			return tsweb.HTTPError{Code: 500, Err: err, Msg: "error rendering status page"}
+		}
+		return nil
+	}
+}
+
+// renderTemplate renders the given Go template with the provided data
+// and returns the result as a string.
+func renderTemplate(tpl string, data any) (string, error) {
+	t, err := template.New("").Parse(tpl)
+	if err != nil {
+		return "", fmt.Errorf("error parsing template %q: %w", tpl, err)
+	}
+	var buf strings.Builder
+	if err := t.ExecuteTemplate(&buf, "", data); err != nil {
+		return "", fmt.Errorf("error rendering template %q with data %v: %w", tpl, data, err)
+	}
+	return buf.String(), nil
+}
--- a/prober/status.html
+++ b/prober/status.html
@@ -0,0 +1,132 @@
+{{define "status"}}
+<html>
+    <head><title>{{.Title}}</title></head>
+    <style>
+        body {
+            /* max-width: 60rem; */
+            margin-left: auto;
+            margin-right: auto;
+            padding: 3rem 1rem 8rem;
+            line-height: 1.4;
+            font-size: 1rem;
+            font-weight: 400;
+            font-family: Inter, ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, Segoe UI, Roboto, Helvetica Neue, Arial, Noto Sans, sans-serif, Apple Color Emoji, Segoe UI Emoji, Segoe UI Symbol, Noto Color Emoji;
+            text-rendering: optimizeLegibility;
+        }
+        .small {
+            font-size: 0.7rem;
+        }
+        h1 {
+            font-weight: 500;
+            letter-spacing: -.025em;
+        }
+        a { color: rgb(74 125 221); }
+        a:hover { color: rgb(73 100 149); }
+        ul {
+            list-style: none;
+            margin: 0;
+            padding: 0;
+        }
+        ul>li::before {
+            position: absolute;
+            top: .625rem;
+            left: .125rem;
+            height: .375rem;
+            width: .375rem;
+            border-radius: 9999px;
+            background-color: currentColor;
+            opacity: .4;
+            content: "";
+        }
+        ul>li {
+            position: relative;
+            padding-left: 1.25rem;
+        }
+        th, td {
+            padding: 5px;
+            text-align: left;
+            background: #eeeeee;
+        }
+        .error {
+            color: red;
+        }
+    </style>
+<body>
+    <h1>{{.Title}}</h1>
+    <ul>
+        <li>Prober Status:
+        {{if .UnhealthyProbes }}
+            <span class="error">{{.UnhealthyProbes}}</span>
+            out of {{.TotalProbes}} probes failed or never ran.
+        {{else}}
+            All {{.TotalProbes}} probes are healthy
+        {{end}}
+        </li>
+        {{ range $text, $url := .Links }}
+        <li><a href="{{$url}}">{{$text}}</a></li>
+        {{end}}
+    </ul>
+
+    <h1>Probes:</h1>
+    <table class="sortable">
+        <thead><tr>
+            <th>Name</th>
+            <th>Class & Labels</th>
+            <th>Interval</th>
+            <th>Result</th>
+            <th>Success</th>
+            <th>Latency</th>
+            <th>Error</th>
+        </tr></thead>
+        <tbody>
+        {{range $name, $probeInfo := .Probes}}
+        <tr>
+            <td>
+                {{$name}}
+                {{range $text, $url := $probeInfo.Links}}
+                <br/>
+                <button onclick="location.href='{{$url}}';" type="button">
+                    {{$text}}
+                </button>
+                {{end}}
+            </td>
+            <td>{{$probeInfo.Class}}<br/>
+                <div class="small">
+                {{range $label, $value := $probeInfo.Labels}}
+                    {{$label}}={{$value}}<br/>
+                {{end}}
+                </div>
+            </td>
+            <td>{{$probeInfo.Interval}}</td>
+            <td data-sort="{{$probeInfo.TimeSinceLast.Milliseconds}}">
+                {{if $probeInfo.TimeSinceLast}}
+                    {{$probeInfo.TimeSinceLast.String}}<br/>
+                    <span class="small">{{$probeInfo.End}}</span>
+                {{else}}
+                    Never
+                {{end}}
+            </td>
+            <td>
+                {{if $probeInfo.Result}}
+                    {{$probeInfo.Result}}
+                {{else}}
+                    <span class="error">{{$probeInfo.Result}}</span>
+                {{end}}<br/>
+                <div class="small">Recent: {{$probeInfo.RecentResults}}</div>
+                <div class="small">Mean: {{$probeInfo.RecentSuccessRatio}}</div>
+            </td>
+            <td data-sort="{{$probeInfo.Latency.Milliseconds}}">
+                {{$probeInfo.Latency.String}}
+                <div class="small">Recent: {{$probeInfo.RecentLatencies}}</div>
+                <div class="small">Median: {{$probeInfo.RecentMedianLatency}}</div>
+            </td>
+            <td class="small">{{$probeInfo.Error}}</td>
+        </tr>
+        {{end}}
+        </tbody>
+    </table>
+    <link href="https://cdn.jsdelivr.net/gh/tofsjonas/sortable@latest/sortable-base.min.css" rel="stylesheet" />
+    <script src="https://cdn.jsdelivr.net/gh/tofsjonas/sortable@latest/sortable.min.js"></script>
+</body>
+</html>
+{{end}}
--- a/shell.nix
+++ b/shell.nix
@@ -16,4 +16,4 @@
 ) {
  src =  ./.;
 }).shellNix
-# nix-direnv cache busting line: sha256-N0TZ1JuDqh6bZjOHcMfoEDOsiUlrC/tR72fBns1GwrM=
+# nix-direnv cache busting line: sha256-1hekcJr1jEJFu4ZnapNkbAAv+8phTQuMloULIZ0f018=
--- a/tailcfg/tailcfg.go
+++ b/tailcfg/tailcfg.go
@@ -146,7 +146,8 @@ type CapabilityVersion int
 //   - 101: 2024-07-01: Client supports SSH agent forwarding when handling connections with /bin/su
 //   - 102: 2024-07-12: NodeAttrDisableMagicSockCryptoRouting support
 //   - 103: 2024-07-24: Client supports NodeAttrDisableCaptivePortalDetection
-const CurrentCapabilityVersion CapabilityVersion = 103
+//   - 104: 2024-08-03: SelfNodeV6MasqAddrForThisPeer now works
+const CurrentCapabilityVersion CapabilityVersion = 104

 type StableID string

--- a/tstest/integration/integration_test.go
+++ b/tstest/integration/integration_test.go
@@ -842,6 +842,7 @@ func TestClientSideJailing(t *testing.T) {
 // TestNATPing creates two nodes, n1 and n2, sets up masquerades for both and
 // tries to do bi-directional pings between them.
 func TestNATPing(t *testing.T) {
+	flakytest.Mark(t, "https://github.com/tailscale/tailscale/issues/12169")
 	tstest.Shard(t)
 	tstest.Parallel(t)
 	for _, v6 := range []bool{false, true} {
--- a/tsweb/tsweb_test.go
+++ b/tsweb/tsweb_test.go
@@ -22,6 +22,7 @@ import (

 	"github.com/google/go-cmp/cmp"
 	"github.com/google/go-cmp/cmp/cmpopts"
+	"tailscale.com/cmd/testwrapper/flakytest"
 	"tailscale.com/metrics"
 	"tailscale.com/tstest"
 	"tailscale.com/util/httpm"
@@ -864,6 +865,7 @@ func TestStdHandler_CanceledAfterHeader(t *testing.T) {
 }

 func TestStdHandler_ConnectionClosedDuringBody(t *testing.T) {
+	flakytest.Mark(t, "https://github.com/tailscale/tailscale/issues/13017")
 	now := time.Now()

 	// Start a HTTP server that returns 1MB of data.
--- a/util/linuxfw/nftables_runner.go
+++ b/util/linuxfw/nftables_runner.go
@@ -592,9 +592,23 @@ func New(logf logger.Logf, prefHint string) (NetfilterRunner, error) {
 	mode := detectFirewallMode(logf, prefHint)
 	switch mode {
 	case FirewallModeIPTables:
-		return newIPTablesRunner(logf)
+		// Note that we don't simply return an newIPTablesRunner here because it
+		// would return a `nil` iptablesRunner which is different from returning
+		// a nil NetfilterRunner.
+		ipr, err := newIPTablesRunner(logf)
+		if err != nil {
+			return nil, err
+		}
+		return ipr, nil
 	case FirewallModeNfTables:
-		return newNfTablesRunner(logf)
+		// Note that we don't simply return an newNfTablesRunner here because it
+		// would return a `nil` nftablesRunner which is different from returning
+		// a nil NetfilterRunner.
+		nfr, err := newNfTablesRunner(logf)
+		if err != nil {
+			return nil, err
+		}
+		return nfr, nil
 	default:
 		return nil, fmt.Errorf("unknown firewall mode %v", mode)
 	}
--- a/util/winutil/gp/policylock_windows.go
+++ b/util/winutil/gp/policylock_windows.go
@@ -189,6 +189,7 @@ func (l *PolicyLock) lockSlow() (err error) {
 			select {
 			case resultCh <- policyLockResult{handle, err}:
 				// lockSlow has received the result.
+				break send_result
 			default:
 				select {
 				case <-closing:
--- a/wgengine/capture/ts-dissector.lua
+++ b/wgengine/capture/ts-dissector.lua
@@ -5,9 +5,9 @@ end
 tsdebug_ll = Proto("tsdebug", "Tailscale debug")
 PATH = ProtoField.string("tsdebug.PATH","PATH", base.ASCII)
 SNAT_IP_4 = ProtoField.ipv4("tsdebug.SNAT_IP_4", "Pre-NAT Source IPv4 address")
-SNAT_IP_6 = ProtoField.ipv4("tsdebug.SNAT_IP_6", "Pre-NAT Source IPv6 address")
+SNAT_IP_6 = ProtoField.ipv6("tsdebug.SNAT_IP_6", "Pre-NAT Source IPv6 address")
 DNAT_IP_4 = ProtoField.ipv4("tsdebug.DNAT_IP_4", "Pre-NAT Dest IPv4 address")
-DNAT_IP_6 = ProtoField.ipv4("tsdebug.DNAT_IP_6", "Pre-NAT Dest IPv6 address")
+DNAT_IP_6 = ProtoField.ipv6("tsdebug.DNAT_IP_6", "Pre-NAT Dest IPv6 address")
 tsdebug_ll.fields = {PATH, SNAT_IP_4, SNAT_IP_6, DNAT_IP_4, DNAT_IP_6}

 function tsdebug_ll.dissector(buffer, pinfo, tree)
@@ -63,7 +63,7 @@ local ts_dissectors = DissectorTable.new("ts.proto", "Tailscale-specific dissect
 tsdisco_meta = Proto("tsdisco", "Tailscale DISCO metadata")
 DISCO_IS_DERP = ProtoField.bool("tsdisco.IS_DERP","From DERP")
 DISCO_SRC_IP_4 = ProtoField.ipv4("tsdisco.SRC_IP_4", "Source IPv4 address")
-DISCO_SRC_IP_6 = ProtoField.ipv4("tsdisco.SRC_IP_6", "Source IPv6 address")
+DISCO_SRC_IP_6 = ProtoField.ipv6("tsdisco.SRC_IP_6", "Source IPv6 address")
 DISCO_SRC_PORT = ProtoField.uint16("tsdisco.SRC_PORT","Source port", base.DEC)
 DISCO_DERP_PUB = ProtoField.bytes("tsdisco.DERP_PUB", "DERP public key", base.SPACE)
 tsdisco_meta.fields = {DISCO_IS_DERP, DISCO_SRC_PORT, DISCO_DERP_PUB, DISCO_SRC_IP_4, DISCO_SRC_IP_6}
--- a/wgengine/magicsock/batching_conn.go
+++ b/wgengine/magicsock/batching_conn.go
@@ -4,200 +4,22 @@
 package magicsock

 import (
-	"errors"
-	"net"
 	"net/netip"
-	"sync"
-	"sync/atomic"
-	"syscall"
-	"time"

+	"golang.org/x/net/ipv4"
 	"golang.org/x/net/ipv6"
-	"tailscale.com/net/neterror"
 	"tailscale.com/types/nettype"
 )

-// xnetBatchReaderWriter defines the batching i/o methods of
-// golang.org/x/net/ipv4.PacketConn (and ipv6.PacketConn).
-// TODO(jwhited): This should eventually be replaced with the standard library
-// implementation of https://github.com/golang/go/issues/45886
-type xnetBatchReaderWriter interface {
-	xnetBatchReader
-	xnetBatchWriter
-}
-
-type xnetBatchReader interface {
-	ReadBatch([]ipv6.Message, int) (int, error)
-}
-
-type xnetBatchWriter interface {
-	WriteBatch([]ipv6.Message, int) (int, error)
-}
-
-// batchingUDPConn is a UDP socket that provides batched i/o.
-type batchingUDPConn struct {
-	pc                    nettype.PacketConn
-	xpc                   xnetBatchReaderWriter
-	rxOffload             bool                                  // supports UDP GRO or similar
-	txOffload             atomic.Bool                           // supports UDP GSO or similar
-	setGSOSizeInControl   func(control *[]byte, gsoSize uint16) // typically setGSOSizeInControl(); swappable for testing
-	getGSOSizeFromControl func(control []byte) (int, error)     // typically getGSOSizeFromControl(); swappable for testing
-	sendBatchPool         sync.Pool
-}
-
-func (c *batchingUDPConn) ReadFromUDPAddrPort(p []byte) (n int, addr netip.AddrPort, err error) {
-	if c.rxOffload {
-		// UDP_GRO is opt-in on Linux via setsockopt(). Once enabled you may
-		// receive a "monster datagram" from any read call. The ReadFrom() API
-		// does not support passing the GSO size and is unsafe to use in such a
-		// case. Other platforms may vary in behavior, but we go with the most
-		// conservative approach to prevent this from becoming a footgun in the
-		// future.
-		return 0, netip.AddrPort{}, errors.New("rx UDP offload is enabled on this socket, single packet reads are unavailable")
-	}
-	return c.pc.ReadFromUDPAddrPort(p)
-}
-
-func (c *batchingUDPConn) SetDeadline(t time.Time) error {
-	return c.pc.SetDeadline(t)
-}
-
-func (c *batchingUDPConn) SetReadDeadline(t time.Time) error {
-	return c.pc.SetReadDeadline(t)
-}
-
-func (c *batchingUDPConn) SetWriteDeadline(t time.Time) error {
-	return c.pc.SetWriteDeadline(t)
-}
-
-const (
-	// This was initially established for Linux, but may split out to
-	// GOOS-specific values later. It originates as UDP_MAX_SEGMENTS in the
-	// kernel's TX path, and UDP_GRO_CNT_MAX for RX.
-	udpSegmentMaxDatagrams = 64
+var (
+	// This acts as a compile-time check for our usage of ipv6.Message in
+	// batchingConn for both IPv6 and IPv4 operations.
+	_ ipv6.Message = ipv4.Message{}
 )

-const (
-	// Exceeding these values results in EMSGSIZE.
-	maxIPv4PayloadLen = 1<<16 - 1 - 20 - 8
-	maxIPv6PayloadLen = 1<<16 - 1 - 8
-)
-
-// coalesceMessages iterates msgs, coalescing them where possible while
-// maintaining datagram order. All msgs have their Addr field set to addr.
-func (c *batchingUDPConn) coalesceMessages(addr *net.UDPAddr, buffs [][]byte, msgs []ipv6.Message) int {
-	var (
-		base     = -1 // index of msg we are currently coalescing into
-		gsoSize  int  // segmentation size of msgs[base]
-		dgramCnt int  // number of dgrams coalesced into msgs[base]
-		endBatch bool // tracking flag to start a new batch on next iteration of buffs
-	)
-	maxPayloadLen := maxIPv4PayloadLen
-	if addr.IP.To4() == nil {
-		maxPayloadLen = maxIPv6PayloadLen
-	}
-	for i, buff := range buffs {
-		if i > 0 {
-			msgLen := len(buff)
-			baseLenBefore := len(msgs[base].Buffers[0])
-			freeBaseCap := cap(msgs[base].Buffers[0]) - baseLenBefore
-			if msgLen+baseLenBefore <= maxPayloadLen &&
-				msgLen <= gsoSize &&
-				msgLen <= freeBaseCap &&
-				dgramCnt < udpSegmentMaxDatagrams &&
-				!endBatch {
-				msgs[base].Buffers[0] = append(msgs[base].Buffers[0], make([]byte, msgLen)...)
-				copy(msgs[base].Buffers[0][baseLenBefore:], buff)
-				if i == len(buffs)-1 {
-					c.setGSOSizeInControl(&msgs[base].OOB, uint16(gsoSize))
-				}
-				dgramCnt++
-				if msgLen < gsoSize {
-					// A smaller than gsoSize packet on the tail is legal, but
-					// it must end the batch.
-					endBatch = true
-				}
-				continue
-			}
-		}
-		if dgramCnt > 1 {
-			c.setGSOSizeInControl(&msgs[base].OOB, uint16(gsoSize))
-		}
-		// Reset prior to incrementing base since we are preparing to start a
-		// new potential batch.
-		endBatch = false
-		base++
-		gsoSize = len(buff)
-		msgs[base].OOB = msgs[base].OOB[:0]
-		msgs[base].Buffers[0] = buff
-		msgs[base].Addr = addr
-		dgramCnt = 1
-	}
-	return base + 1
-}
-
-type sendBatch struct {
-	msgs []ipv6.Message
-	ua   *net.UDPAddr
-}
-
-func (c *batchingUDPConn) getSendBatch() *sendBatch {
-	batch := c.sendBatchPool.Get().(*sendBatch)
-	return batch
-}
-
-func (c *batchingUDPConn) putSendBatch(batch *sendBatch) {
-	for i := range batch.msgs {
-		batch.msgs[i] = ipv6.Message{Buffers: batch.msgs[i].Buffers, OOB: batch.msgs[i].OOB}
-	}
-	c.sendBatchPool.Put(batch)
-}
-
-func (c *batchingUDPConn) WriteBatchTo(buffs [][]byte, addr netip.AddrPort) error {
-	batch := c.getSendBatch()
-	defer c.putSendBatch(batch)
-	if addr.Addr().Is6() {
-		as16 := addr.Addr().As16()
-		copy(batch.ua.IP, as16[:])
-		batch.ua.IP = batch.ua.IP[:16]
-	} else {
-		as4 := addr.Addr().As4()
-		copy(batch.ua.IP, as4[:])
-		batch.ua.IP = batch.ua.IP[:4]
-	}
-	batch.ua.Port = int(addr.Port())
-	var (
-		n       int
-		retried bool
-	)
-retry:
-	if c.txOffload.Load() {
-		n = c.coalesceMessages(batch.ua, buffs, batch.msgs)
-	} else {
-		for i := range buffs {
-			batch.msgs[i].Buffers[0] = buffs[i]
-			batch.msgs[i].Addr = batch.ua
-			batch.msgs[i].OOB = batch.msgs[i].OOB[:0]
-		}
-		n = len(buffs)
-	}
-
-	err := c.writeBatch(batch.msgs[:n])
-	if err != nil && c.txOffload.Load() && neterror.ShouldDisableUDPGSO(err) {
-		c.txOffload.Store(false)
-		retried = true
-		goto retry
-	}
-	if retried {
-		return neterror.ErrUDPGSODisabled{OnLaddr: c.pc.LocalAddr().String(), RetryErr: err}
-	}
-	return err
-}
-
-func (c *batchingUDPConn) SyscallConn() (syscall.RawConn, error) {
-	sc, ok := c.pc.(syscall.Conn)
-	if !ok {
-		return nil, errUnsupportedConnType
-	}
-	return sc.SyscallConn()
+// batchingConn is a nettype.PacketConn that provides batched i/o.
+type batchingConn interface {
+	nettype.PacketConn
+	ReadBatch(msgs []ipv6.Message, flags int) (n int, err error)
+	WriteBatchTo(buffs [][]byte, addr netip.AddrPort) error
 }
--- a/wgengine/magicsock/batching_conn_default.go
+++ b/wgengine/magicsock/batching_conn_default.go
@@ -0,0 +1,14 @@
+// Copyright (c) Tailscale Inc & AUTHORS
+// SPDX-License-Identifier: BSD-3-Clause
+
+//go:build !linux
+
+package magicsock
+
+import (
+	"tailscale.com/types/nettype"
+)
+
+func tryUpgradeToBatchingConn(pconn nettype.PacketConn, _ string, _ int) nettype.PacketConn {
+	return pconn
+}
--- a/wgengine/magicsock/batching_conn_linux.go
+++ b/wgengine/magicsock/batching_conn_linux.go
@@ -0,0 +1,419 @@
+// Copyright (c) Tailscale Inc & AUTHORS
+// SPDX-License-Identifier: BSD-3-Clause
+
+package magicsock
+
+import (
+	"encoding/binary"
+	"errors"
+	"fmt"
+	"net"
+	"net/netip"
+	"strings"
+	"sync"
+	"sync/atomic"
+	"syscall"
+	"time"
+	"unsafe"
+
+	"golang.org/x/net/ipv4"
+	"golang.org/x/net/ipv6"
+	"golang.org/x/sys/unix"
+	"tailscale.com/hostinfo"
+	"tailscale.com/net/neterror"
+	"tailscale.com/types/nettype"
+)
+
+// xnetBatchReaderWriter defines the batching i/o methods of
+// golang.org/x/net/ipv4.PacketConn (and ipv6.PacketConn).
+// TODO(jwhited): This should eventually be replaced with the standard library
+// implementation of https://github.com/golang/go/issues/45886
+type xnetBatchReaderWriter interface {
+	xnetBatchReader
+	xnetBatchWriter
+}
+
+type xnetBatchReader interface {
+	ReadBatch([]ipv6.Message, int) (int, error)
+}
+
+type xnetBatchWriter interface {
+	WriteBatch([]ipv6.Message, int) (int, error)
+}
+
+// linuxBatchingConn is a UDP socket that provides batched i/o. It implements
+// batchingConn.
+type linuxBatchingConn struct {
+	pc                    nettype.PacketConn
+	xpc                   xnetBatchReaderWriter
+	rxOffload             bool                                  // supports UDP GRO or similar
+	txOffload             atomic.Bool                           // supports UDP GSO or similar
+	setGSOSizeInControl   func(control *[]byte, gsoSize uint16) // typically setGSOSizeInControl(); swappable for testing
+	getGSOSizeFromControl func(control []byte) (int, error)     // typically getGSOSizeFromControl(); swappable for testing
+	sendBatchPool         sync.Pool
+}
+
+func (c *linuxBatchingConn) ReadFromUDPAddrPort(p []byte) (n int, addr netip.AddrPort, err error) {
+	if c.rxOffload {
+		// UDP_GRO is opt-in on Linux via setsockopt(). Once enabled you may
+		// receive a "monster datagram" from any read call. The ReadFrom() API
+		// does not support passing the GSO size and is unsafe to use in such a
+		// case. Other platforms may vary in behavior, but we go with the most
+		// conservative approach to prevent this from becoming a footgun in the
+		// future.
+		return 0, netip.AddrPort{}, errors.New("rx UDP offload is enabled on this socket, single packet reads are unavailable")
+	}
+	return c.pc.ReadFromUDPAddrPort(p)
+}
+
+func (c *linuxBatchingConn) SetDeadline(t time.Time) error {
+	return c.pc.SetDeadline(t)
+}
+
+func (c *linuxBatchingConn) SetReadDeadline(t time.Time) error {
+	return c.pc.SetReadDeadline(t)
+}
+
+func (c *linuxBatchingConn) SetWriteDeadline(t time.Time) error {
+	return c.pc.SetWriteDeadline(t)
+}
+
+const (
+	// This was initially established for Linux, but may split out to
+	// GOOS-specific values later. It originates as UDP_MAX_SEGMENTS in the
+	// kernel's TX path, and UDP_GRO_CNT_MAX for RX.
+	udpSegmentMaxDatagrams = 64
+)
+
+const (
+	// Exceeding these values results in EMSGSIZE.
+	maxIPv4PayloadLen = 1<<16 - 1 - 20 - 8
+	maxIPv6PayloadLen = 1<<16 - 1 - 8
+)
+
+// coalesceMessages iterates msgs, coalescing them where possible while
+// maintaining datagram order. All msgs have their Addr field set to addr.
+func (c *linuxBatchingConn) coalesceMessages(addr *net.UDPAddr, buffs [][]byte, msgs []ipv6.Message) int {
+	var (
+		base     = -1 // index of msg we are currently coalescing into
+		gsoSize  int  // segmentation size of msgs[base]
+		dgramCnt int  // number of dgrams coalesced into msgs[base]
+		endBatch bool // tracking flag to start a new batch on next iteration of buffs
+	)
+	maxPayloadLen := maxIPv4PayloadLen
+	if addr.IP.To4() == nil {
+		maxPayloadLen = maxIPv6PayloadLen
+	}
+	for i, buff := range buffs {
+		if i > 0 {
+			msgLen := len(buff)
+			baseLenBefore := len(msgs[base].Buffers[0])
+			freeBaseCap := cap(msgs[base].Buffers[0]) - baseLenBefore
+			if msgLen+baseLenBefore <= maxPayloadLen &&
+				msgLen <= gsoSize &&
+				msgLen <= freeBaseCap &&
+				dgramCnt < udpSegmentMaxDatagrams &&
+				!endBatch {
+				msgs[base].Buffers[0] = append(msgs[base].Buffers[0], make([]byte, msgLen)...)
+				copy(msgs[base].Buffers[0][baseLenBefore:], buff)
+				if i == len(buffs)-1 {
+					c.setGSOSizeInControl(&msgs[base].OOB, uint16(gsoSize))
+				}
+				dgramCnt++
+				if msgLen < gsoSize {
+					// A smaller than gsoSize packet on the tail is legal, but
+					// it must end the batch.
+					endBatch = true
+				}
+				continue
+			}
+		}
+		if dgramCnt > 1 {
+			c.setGSOSizeInControl(&msgs[base].OOB, uint16(gsoSize))
+		}
+		// Reset prior to incrementing base since we are preparing to start a
+		// new potential batch.
+		endBatch = false
+		base++
+		gsoSize = len(buff)
+		msgs[base].OOB = msgs[base].OOB[:0]
+		msgs[base].Buffers[0] = buff
+		msgs[base].Addr = addr
+		dgramCnt = 1
+	}
+	return base + 1
+}
+
+type sendBatch struct {
+	msgs []ipv6.Message
+	ua   *net.UDPAddr
+}
+
+func (c *linuxBatchingConn) getSendBatch() *sendBatch {
+	batch := c.sendBatchPool.Get().(*sendBatch)
+	return batch
+}
+
+func (c *linuxBatchingConn) putSendBatch(batch *sendBatch) {
+	for i := range batch.msgs {
+		batch.msgs[i] = ipv6.Message{Buffers: batch.msgs[i].Buffers, OOB: batch.msgs[i].OOB}
+	}
+	c.sendBatchPool.Put(batch)
+}
+
+func (c *linuxBatchingConn) WriteBatchTo(buffs [][]byte, addr netip.AddrPort) error {
+	batch := c.getSendBatch()
+	defer c.putSendBatch(batch)
+	if addr.Addr().Is6() {
+		as16 := addr.Addr().As16()
+		copy(batch.ua.IP, as16[:])
+		batch.ua.IP = batch.ua.IP[:16]
+	} else {
+		as4 := addr.Addr().As4()
+		copy(batch.ua.IP, as4[:])
+		batch.ua.IP = batch.ua.IP[:4]
+	}
+	batch.ua.Port = int(addr.Port())
+	var (
+		n       int
+		retried bool
+	)
+retry:
+	if c.txOffload.Load() {
+		n = c.coalesceMessages(batch.ua, buffs, batch.msgs)
+	} else {
+		for i := range buffs {
+			batch.msgs[i].Buffers[0] = buffs[i]
+			batch.msgs[i].Addr = batch.ua
+			batch.msgs[i].OOB = batch.msgs[i].OOB[:0]
+		}
+		n = len(buffs)
+	}
+
+	err := c.writeBatch(batch.msgs[:n])
+	if err != nil && c.txOffload.Load() && neterror.ShouldDisableUDPGSO(err) {
+		c.txOffload.Store(false)
+		retried = true
+		goto retry
+	}
+	if retried {
+		return neterror.ErrUDPGSODisabled{OnLaddr: c.pc.LocalAddr().String(), RetryErr: err}
+	}
+	return err
+}
+
+func (c *linuxBatchingConn) SyscallConn() (syscall.RawConn, error) {
+	sc, ok := c.pc.(syscall.Conn)
+	if !ok {
+		return nil, errUnsupportedConnType
+	}
+	return sc.SyscallConn()
+}
+
+func (c *linuxBatchingConn) writeBatch(msgs []ipv6.Message) error {
+	var head int
+	for {
+		n, err := c.xpc.WriteBatch(msgs[head:], 0)
+		if err != nil || n == len(msgs[head:]) {
+			// Returning the number of packets written would require
+			// unraveling individual msg len and gso size during a coalesced
+			// write. The top of the call stack disregards partial success,
+			// so keep this simple for now.
+			return err
+		}
+		head += n
+	}
+}
+
+// splitCoalescedMessages splits coalesced messages from the tail of dst
+// beginning at index 'firstMsgAt' into the head of the same slice. It reports
+// the number of elements to evaluate in msgs for nonzero len (msgs[i].N). An
+// error is returned if a socket control message cannot be parsed or a split
+// operation would overflow msgs.
+func (c *linuxBatchingConn) splitCoalescedMessages(msgs []ipv6.Message, firstMsgAt int) (n int, err error) {
+	for i := firstMsgAt; i < len(msgs); i++ {
+		msg := &msgs[i]
+		if msg.N == 0 {
+			return n, err
+		}
+		var (
+			gsoSize    int
+			start      int
+			end        = msg.N
+			numToSplit = 1
+		)
+		gsoSize, err = c.getGSOSizeFromControl(msg.OOB[:msg.NN])
+		if err != nil {
+			return n, err
+		}
+		if gsoSize > 0 {
+			numToSplit = (msg.N + gsoSize - 1) / gsoSize
+			end = gsoSize
+		}
+		for j := 0; j < numToSplit; j++ {
+			if n > i {
+				return n, errors.New("splitting coalesced packet resulted in overflow")
+			}
+			copied := copy(msgs[n].Buffers[0], msg.Buffers[0][start:end])
+			msgs[n].N = copied
+			msgs[n].Addr = msg.Addr
+			start = end
+			end += gsoSize
+			if end > msg.N {
+				end = msg.N
+			}
+			n++
+		}
+		if i != n-1 {
+			// It is legal for bytes to move within msg.Buffers[0] as a result
+			// of splitting, so we only zero the source msg len when it is not
+			// the destination of the last split operation above.
+			msg.N = 0
+		}
+	}
+	return n, nil
+}
+
+func (c *linuxBatchingConn) ReadBatch(msgs []ipv6.Message, flags int) (n int, err error) {
+	if !c.rxOffload || len(msgs) < 2 {
+		return c.xpc.ReadBatch(msgs, flags)
+	}
+	// Read into the tail of msgs, split into the head.
+	readAt := len(msgs) - 2
+	numRead, err := c.xpc.ReadBatch(msgs[readAt:], 0)
+	if err != nil || numRead == 0 {
+		return 0, err
+	}
+	return c.splitCoalescedMessages(msgs, readAt)
+}
+
+func (c *linuxBatchingConn) LocalAddr() net.Addr {
+	return c.pc.LocalAddr().(*net.UDPAddr)
+}
+
+func (c *linuxBatchingConn) WriteToUDPAddrPort(b []byte, addr netip.AddrPort) (int, error) {
+	return c.pc.WriteToUDPAddrPort(b, addr)
+}
+
+func (c *linuxBatchingConn) Close() error {
+	return c.pc.Close()
+}
+
+// tryEnableUDPOffload attempts to enable the UDP_GRO socket option on pconn,
+// and returns two booleans indicating TX and RX UDP offload support.
+func tryEnableUDPOffload(pconn nettype.PacketConn) (hasTX bool, hasRX bool) {
+	if c, ok := pconn.(*net.UDPConn); ok {
+		rc, err := c.SyscallConn()
+		if err != nil {
+			return
+		}
+		err = rc.Control(func(fd uintptr) {
+			_, errSyscall := syscall.GetsockoptInt(int(fd), unix.IPPROTO_UDP, unix.UDP_SEGMENT)
+			hasTX = errSyscall == nil
+			errSyscall = syscall.SetsockoptInt(int(fd), unix.IPPROTO_UDP, unix.UDP_GRO, 1)
+			hasRX = errSyscall == nil
+		})
+		if err != nil {
+			return false, false
+		}
+	}
+	return hasTX, hasRX
+}
+
+// getGSOSizeFromControl returns the GSO size found in control. If no GSO size
+// is found or the len(control) < unix.SizeofCmsghdr, this function returns 0.
+// A non-nil error will be returned if len(control) > unix.SizeofCmsghdr but
+// its contents cannot be parsed as a socket control message.
+func getGSOSizeFromControl(control []byte) (int, error) {
+	var (
+		hdr  unix.Cmsghdr
+		data []byte
+		rem  = control
+		err  error
+	)
+
+	for len(rem) > unix.SizeofCmsghdr {
+		hdr, data, rem, err = unix.ParseOneSocketControlMessage(control)
+		if err != nil {
+			return 0, fmt.Errorf("error parsing socket control message: %w", err)
+		}
+		if hdr.Level == unix.SOL_UDP && hdr.Type == unix.UDP_GRO && len(data) >= 2 {
+			return int(binary.NativeEndian.Uint16(data[:2])), nil
+		}
+	}
+	return 0, nil
+}
+
+// setGSOSizeInControl sets a socket control message in control containing
+// gsoSize. If len(control) < controlMessageSize control's len will be set to 0.
+func setGSOSizeInControl(control *[]byte, gsoSize uint16) {
+	*control = (*control)[:0]
+	if cap(*control) < int(unsafe.Sizeof(unix.Cmsghdr{})) {
+		return
+	}
+	if cap(*control) < controlMessageSize {
+		return
+	}
+	*control = (*control)[:cap(*control)]
+	hdr := (*unix.Cmsghdr)(unsafe.Pointer(&(*control)[0]))
+	hdr.Level = unix.SOL_UDP
+	hdr.Type = unix.UDP_SEGMENT
+	hdr.SetLen(unix.CmsgLen(2))
+	binary.NativeEndian.PutUint16((*control)[unix.SizeofCmsghdr:], gsoSize)
+	*control = (*control)[:unix.CmsgSpace(2)]
+}
+
+// tryUpgradeToBatchingConn probes the capabilities of the OS and pconn, and
+// upgrades pconn to a *linuxBatchingConn if appropriate.
+func tryUpgradeToBatchingConn(pconn nettype.PacketConn, network string, batchSize int) nettype.PacketConn {
+	if network != "udp4" && network != "udp6" {
+		return pconn
+	}
+	if strings.HasPrefix(hostinfo.GetOSVersion(), "2.") {
+		// recvmmsg/sendmmsg were added in 2.6.33, but we support down to
+		// 2.6.32 for old NAS devices. See https://github.com/tailscale/tailscale/issues/6807.
+		// As a cheap heuristic: if the Linux kernel starts with "2", just
+		// consider it too old for mmsg. Nobody who cares about performance runs
+		// such ancient kernels. UDP offload was added much later, so no
+		// upgrades are available.
+		return pconn
+	}
+	uc, ok := pconn.(*net.UDPConn)
+	if !ok {
+		return pconn
+	}
+	b := &linuxBatchingConn{
+		pc:                    pconn,
+		getGSOSizeFromControl: getGSOSizeFromControl,
+		setGSOSizeInControl:   setGSOSizeInControl,
+		sendBatchPool: sync.Pool{
+			New: func() any {
+				ua := &net.UDPAddr{
+					IP: make([]byte, 16),
+				}
+				msgs := make([]ipv6.Message, batchSize)
+				for i := range msgs {
+					msgs[i].Buffers = make([][]byte, 1)
+					msgs[i].Addr = ua
+					msgs[i].OOB = make([]byte, controlMessageSize)
+				}
+				return &sendBatch{
+					ua:   ua,
+					msgs: msgs,
+				}
+			},
+		},
+	}
+	switch network {
+	case "udp4":
+		b.xpc = ipv4.NewPacketConn(uc)
+	case "udp6":
+		b.xpc = ipv6.NewPacketConn(uc)
+	default:
+		panic("bogus network")
+	}
+	var txOffload bool
+	txOffload, b.rxOffload = tryEnableUDPOffload(uc)
+	b.txOffload.Store(txOffload)
+	return b
+}
--- a/wgengine/magicsock/batching_conn_linux_test.go
+++ b/wgengine/magicsock/batching_conn_linux_test.go
@@ -0,0 +1,244 @@
+// Copyright (c) Tailscale Inc & AUTHORS
+// SPDX-License-Identifier: BSD-3-Clause
+
+package magicsock
+
+import (
+	"encoding/binary"
+	"net"
+	"testing"
+
+	"golang.org/x/net/ipv6"
+)
+
+func setGSOSize(control *[]byte, gsoSize uint16) {
+	*control = (*control)[:cap(*control)]
+	binary.LittleEndian.PutUint16(*control, gsoSize)
+}
+
+func getGSOSize(control []byte) (int, error) {
+	if len(control) < 2 {
+		return 0, nil
+	}
+	return int(binary.LittleEndian.Uint16(control)), nil
+}
+
+func Test_linuxBatchingConn_splitCoalescedMessages(t *testing.T) {
+	c := &linuxBatchingConn{
+		setGSOSizeInControl:   setGSOSize,
+		getGSOSizeFromControl: getGSOSize,
+	}
+
+	newMsg := func(n, gso int) ipv6.Message {
+		msg := ipv6.Message{
+			Buffers: [][]byte{make([]byte, 1024)},
+			N:       n,
+			OOB:     make([]byte, 2),
+		}
+		binary.LittleEndian.PutUint16(msg.OOB, uint16(gso))
+		if gso > 0 {
+			msg.NN = 2
+		}
+		return msg
+	}
+
+	cases := []struct {
+		name        string
+		msgs        []ipv6.Message
+		firstMsgAt  int
+		wantNumEval int
+		wantMsgLens []int
+		wantErr     bool
+	}{
+		{
+			name: "second last split last empty",
+			msgs: []ipv6.Message{
+				newMsg(0, 0),
+				newMsg(0, 0),
+				newMsg(3, 1),
+				newMsg(0, 0),
+			},
+			firstMsgAt:  2,
+			wantNumEval: 3,
+			wantMsgLens: []int{1, 1, 1, 0},
+			wantErr:     false,
+		},
+		{
+			name: "second last no split last empty",
+			msgs: []ipv6.Message{
+				newMsg(0, 0),
+				newMsg(0, 0),
+				newMsg(1, 0),
+				newMsg(0, 0),
+			},
+			firstMsgAt:  2,
+			wantNumEval: 1,
+			wantMsgLens: []int{1, 0, 0, 0},
+			wantErr:     false,
+		},
+		{
+			name: "second last no split last no split",
+			msgs: []ipv6.Message{
+				newMsg(0, 0),
+				newMsg(0, 0),
+				newMsg(1, 0),
+				newMsg(1, 0),
+			},
+			firstMsgAt:  2,
+			wantNumEval: 2,
+			wantMsgLens: []int{1, 1, 0, 0},
+			wantErr:     false,
+		},
+		{
+			name: "second last no split last split",
+			msgs: []ipv6.Message{
+				newMsg(0, 0),
+				newMsg(0, 0),
+				newMsg(1, 0),
+				newMsg(3, 1),
+			},
+			firstMsgAt:  2,
+			wantNumEval: 4,
+			wantMsgLens: []int{1, 1, 1, 1},
+			wantErr:     false,
+		},
+		{
+			name: "second last split last split",
+			msgs: []ipv6.Message{
+				newMsg(0, 0),
+				newMsg(0, 0),
+				newMsg(2, 1),
+				newMsg(2, 1),
+			},
+			firstMsgAt:  2,
+			wantNumEval: 4,
+			wantMsgLens: []int{1, 1, 1, 1},
+			wantErr:     false,
+		},
+		{
+			name: "second last no split last split overflow",
+			msgs: []ipv6.Message{
+				newMsg(0, 0),
+				newMsg(0, 0),
+				newMsg(1, 0),
+				newMsg(4, 1),
+			},
+			firstMsgAt:  2,
+			wantNumEval: 4,
+			wantMsgLens: []int{1, 1, 1, 1},
+			wantErr:     true,
+		},
+	}
+
+	for _, tt := range cases {
+		t.Run(tt.name, func(t *testing.T) {
+			got, err := c.splitCoalescedMessages(tt.msgs, 2)
+			if err != nil && !tt.wantErr {
+				t.Fatalf("err: %v", err)
+			}
+			if got != tt.wantNumEval {
+				t.Fatalf("got to eval: %d want: %d", got, tt.wantNumEval)
+			}
+			for i, msg := range tt.msgs {
+				if msg.N != tt.wantMsgLens[i] {
+					t.Fatalf("msg[%d].N: %d want: %d", i, msg.N, tt.wantMsgLens[i])
+				}
+			}
+		})
+	}
+}
+
+func Test_linuxBatchingConn_coalesceMessages(t *testing.T) {
+	c := &linuxBatchingConn{
+		setGSOSizeInControl:   setGSOSize,
+		getGSOSizeFromControl: getGSOSize,
+	}
+
+	cases := []struct {
+		name     string
+		buffs    [][]byte
+		wantLens []int
+		wantGSO  []int
+	}{
+		{
+			name: "one message no coalesce",
+			buffs: [][]byte{
+				make([]byte, 1, 1),
+			},
+			wantLens: []int{1},
+			wantGSO:  []int{0},
+		},
+		{
+			name: "two messages equal len coalesce",
+			buffs: [][]byte{
+				make([]byte, 1, 2),
+				make([]byte, 1, 1),
+			},
+			wantLens: []int{2},
+			wantGSO:  []int{1},
+		},
+		{
+			name: "two messages unequal len coalesce",
+			buffs: [][]byte{
+				make([]byte, 2, 3),
+				make([]byte, 1, 1),
+			},
+			wantLens: []int{3},
+			wantGSO:  []int{2},
+		},
+		{
+			name: "three messages second unequal len coalesce",
+			buffs: [][]byte{
+				make([]byte, 2, 3),
+				make([]byte, 1, 1),
+				make([]byte, 2, 2),
+			},
+			wantLens: []int{3, 2},
+			wantGSO:  []int{2, 0},
+		},
+		{
+			name: "three messages limited cap coalesce",
+			buffs: [][]byte{
+				make([]byte, 2, 4),
+				make([]byte, 2, 2),
+				make([]byte, 2, 2),
+			},
+			wantLens: []int{4, 2},
+			wantGSO:  []int{2, 0},
+		},
+	}
+
+	for _, tt := range cases {
+		t.Run(tt.name, func(t *testing.T) {
+			addr := &net.UDPAddr{
+				IP:   net.ParseIP("127.0.0.1"),
+				Port: 1,
+			}
+			msgs := make([]ipv6.Message, len(tt.buffs))
+			for i := range msgs {
+				msgs[i].Buffers = make([][]byte, 1)
+				msgs[i].OOB = make([]byte, 0, 2)
+			}
+			got := c.coalesceMessages(addr, tt.buffs, msgs)
+			if got != len(tt.wantLens) {
+				t.Fatalf("got len %d want: %d", got, len(tt.wantLens))
+			}
+			for i := range got {
+				if msgs[i].Addr != addr {
+					t.Errorf("msgs[%d].Addr != passed addr", i)
+				}
+				gotLen := len(msgs[i].Buffers[0])
+				if gotLen != tt.wantLens[i] {
+					t.Errorf("len(msgs[%d].Buffers[0]) %d != %d", i, gotLen, tt.wantLens[i])
+				}
+				gotGSO, err := getGSOSize(msgs[i].OOB)
+				if err != nil {
+					t.Fatalf("msgs[%d] getGSOSize err: %v", i, err)
+				}
+				if gotGSO != tt.wantGSO[i] {
+					t.Errorf("msgs[%d] gsoSize %d != %d", i, gotGSO, tt.wantGSO[i])
+				}
+			}
+		})
+	}
+}
--- a/wgengine/magicsock/cloudinfo.go
+++ b/wgengine/magicsock/cloudinfo.go
@@ -0,0 +1,182 @@
+// Copyright (c) Tailscale Inc & AUTHORS
+// SPDX-License-Identifier: BSD-3-Clause
+
+//go:build !(ios || android || js)
+
+package magicsock
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"io"
+	"net"
+	"net/http"
+	"net/netip"
+	"slices"
+	"strings"
+	"time"
+
+	"tailscale.com/types/logger"
+	"tailscale.com/util/cloudenv"
+)
+
+const maxCloudInfoWait = 2 * time.Second
+
+type cloudInfo struct {
+	client http.Client
+	logf   logger.Logf
+
+	// The following parameters are fixed for the lifetime of the cloudInfo
+	// object, but are used for testing.
+	cloud    cloudenv.Cloud
+	endpoint string
+}
+
+func newCloudInfo(logf logger.Logf) *cloudInfo {
+	tr := &http.Transport{
+		DisableKeepAlives: true,
+		Dial: (&net.Dialer{
+			Timeout: maxCloudInfoWait,
+		}).Dial,
+	}
+
+	return &cloudInfo{
+		client:   http.Client{Transport: tr},
+		logf:     logf,
+		cloud:    cloudenv.Get(),
+		endpoint: "http://" + cloudenv.CommonNonRoutableMetadataIP,
+	}
+}
+
+// GetPublicIPs returns any public IPs attached to the current cloud instance,
+// if the tailscaled process is running in a known cloud and there are any such
+// IPs present.
+func (ci *cloudInfo) GetPublicIPs(ctx context.Context) ([]netip.Addr, error) {
+	switch ci.cloud {
+	case cloudenv.AWS:
+		ret, err := ci.getAWS(ctx)
+		ci.logf("[v1] cloudinfo.GetPublicIPs: AWS: %v, %v", ret, err)
+		return ret, err
+	}
+
+	return nil, nil
+}
+
+// getAWSMetadata makes a request to the AWS metadata service at the given
+// path, authenticating with the provided IMDSv2 token. The returned metadata
+// is split by newline and returned as a slice.
+func (ci *cloudInfo) getAWSMetadata(ctx context.Context, token, path string) ([]string, error) {
+	req, err := http.NewRequestWithContext(ctx, "GET", ci.endpoint+path, nil)
+	if err != nil {
+		return nil, fmt.Errorf("creating request to %q: %w", path, err)
+	}
+	req.Header.Set("X-aws-ec2-metadata-token", token)
+
+	resp, err := ci.client.Do(req)
+	if err != nil {
+		return nil, fmt.Errorf("making request to metadata service %q: %w", path, err)
+	}
+	defer resp.Body.Close()
+
+	switch resp.StatusCode {
+	case http.StatusOK:
+		// Good
+	case http.StatusNotFound:
+		// Nothing found, but this isn't an error; just return
+		return nil, nil
+	default:
+		return nil, fmt.Errorf("unexpected status code: %d", resp.StatusCode)
+	}
+
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return nil, fmt.Errorf("reading response body for %q: %w", path, err)
+	}
+
+	return strings.Split(strings.TrimSpace(string(body)), "\n"), nil
+}
+
+// getAWS returns all public IPv4 and IPv6 addresses present in the AWS instance metadata.
+func (ci *cloudInfo) getAWS(ctx context.Context) ([]netip.Addr, error) {
+	ctx, cancel := context.WithTimeout(ctx, maxCloudInfoWait)
+	defer cancel()
+
+	// Get a token so we can query the metadata service.
+	req, err := http.NewRequestWithContext(ctx, "PUT", ci.endpoint+"/latest/api/token", nil)
+	if err != nil {
+		return nil, fmt.Errorf("creating token request: %w", err)
+	}
+	req.Header.Set("X-Aws-Ec2-Metadata-Token-Ttl-Seconds", "10")
+
+	resp, err := ci.client.Do(req)
+	if err != nil {
+		return nil, fmt.Errorf("making token request to metadata service: %w", err)
+	}
+	body, err := io.ReadAll(resp.Body)
+	resp.Body.Close()
+	if err != nil {
+		return nil, fmt.Errorf("reading token response body: %w", err)
+	}
+	token := string(body)
+
+	server := resp.Header.Get("Server")
+	if server != "EC2ws" {
+		return nil, fmt.Errorf("unexpected server header: %q", server)
+	}
+
+	// Iterate over all interfaces and get their public IP addresses, both IPv4 and IPv6.
+	macAddrs, err := ci.getAWSMetadata(ctx, token, "/latest/meta-data/network/interfaces/macs/")
+	if err != nil {
+		return nil, fmt.Errorf("getting interface MAC addresses: %w", err)
+	}
+
+	var (
+		addrs []netip.Addr
+		errs  []error
+	)
+
+	addAddr := func(addr string) {
+		ip, err := netip.ParseAddr(addr)
+		if err != nil {
+			errs = append(errs, fmt.Errorf("parsing IP address %q: %w", addr, err))
+			return
+		}
+		addrs = append(addrs, ip)
+	}
+	for _, mac := range macAddrs {
+		ips, err := ci.getAWSMetadata(ctx, token, "/latest/meta-data/network/interfaces/macs/"+mac+"/public-ipv4s")
+		if err != nil {
+			errs = append(errs, fmt.Errorf("getting IPv4 addresses for %q: %w", mac, err))
+			continue
+		}
+
+		for _, ip := range ips {
+			addAddr(ip)
+		}
+
+		// Try querying for IPv6 addresses.
+		ips, err = ci.getAWSMetadata(ctx, token, "/latest/meta-data/network/interfaces/macs/"+mac+"/ipv6s")
+		if err != nil {
+			errs = append(errs, fmt.Errorf("getting IPv6 addresses for %q: %w", mac, err))
+			continue
+		}
+		for _, ip := range ips {
+			addAddr(ip)
+		}
+	}
+
+	// Sort the returned addresses for determinism.
+	slices.SortFunc(addrs, func(a, b netip.Addr) int {
+		return a.Compare(b)
+	})
+
+	// Preferentially return any addresses we found, even if there were errors.
+	if len(addrs) > 0 {
+		return addrs, nil
+	}
+	if len(errs) > 0 {
+		return nil, fmt.Errorf("getting IP addresses: %w", errors.Join(errs...))
+	}
+	return nil, nil
+}
--- a/wgengine/magicsock/cloudinfo_nocloud.go
+++ b/wgengine/magicsock/cloudinfo_nocloud.go
@@ -0,0 +1,23 @@
+// Copyright (c) Tailscale Inc & AUTHORS
+// SPDX-License-Identifier: BSD-3-Clause
+
+//go:build ios || android || js
+
+package magicsock
+
+import (
+	"context"
+	"net/netip"
+
+	"tailscale.com/types/logger"
+)
+
+type cloudInfo struct{}
+
+func newCloudInfo(_ logger.Logf) *cloudInfo {
+	return &cloudInfo{}
+}
+
+func (ci *cloudInfo) GetPublicIPs(_ context.Context) ([]netip.Addr, error) {
+	return nil, nil
+}
--- a/wgengine/magicsock/cloudinfo_test.go
+++ b/wgengine/magicsock/cloudinfo_test.go
@@ -0,0 +1,123 @@
+// Copyright (c) Tailscale Inc & AUTHORS
+// SPDX-License-Identifier: BSD-3-Clause
+
+package magicsock
+
+import (
+	"context"
+	"net/http"
+	"net/http/httptest"
+	"net/netip"
+	"slices"
+	"testing"
+
+	"tailscale.com/util/cloudenv"
+)
+
+func TestCloudInfo_AWS(t *testing.T) {
+	const (
+		mac1      = "06:1d:00:00:00:00"
+		mac2      = "06:1d:00:00:00:01"
+		publicV4  = "1.2.3.4"
+		otherV4_1 = "5.6.7.8"
+		otherV4_2 = "11.12.13.14"
+		v6addr    = "2001:db8::1"
+
+		macsPrefix = "/latest/meta-data/network/interfaces/macs/"
+	)
+	// Launch a fake AWS IMDS server
+	fake := &fakeIMDS{
+		tb: t,
+		paths: map[string]string{
+			macsPrefix: mac1 + "\n" + mac2,
+			// This is the "main" public IP address for the instance
+			macsPrefix + mac1 + "/public-ipv4s": publicV4,
+
+			// There's another interface with two public IPs
+			// attached to it and an IPv6 address, all of which we
+			// should discover.
+			macsPrefix + mac2 + "/public-ipv4s": otherV4_1 + "\n" + otherV4_2,
+			macsPrefix + mac2 + "/ipv6s":        v6addr,
+		},
+	}
+
+	srv := httptest.NewServer(fake)
+	defer srv.Close()
+
+	ci := newCloudInfo(t.Logf)
+	ci.cloud = cloudenv.AWS
+	ci.endpoint = srv.URL
+
+	ips, err := ci.GetPublicIPs(context.Background())
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+
+	wantIPs := []netip.Addr{
+		netip.MustParseAddr(publicV4),
+		netip.MustParseAddr(otherV4_1),
+		netip.MustParseAddr(otherV4_2),
+		netip.MustParseAddr(v6addr),
+	}
+	if !slices.Equal(ips, wantIPs) {
+		t.Fatalf("got %v, want %v", ips, wantIPs)
+	}
+}
+
+func TestCloudInfo_AWSNotPublic(t *testing.T) {
+	returns404 := http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if r.Method == "PUT" && r.URL.Path == "/latest/api/token" {
+			w.Header().Set("Server", "EC2ws")
+			w.Write([]byte("fake-imds-token"))
+			return
+		}
+		http.NotFound(w, r)
+	})
+	srv := httptest.NewServer(returns404)
+	defer srv.Close()
+
+	ci := newCloudInfo(t.Logf)
+	ci.cloud = cloudenv.AWS
+	ci.endpoint = srv.URL
+
+	// If the IMDS server doesn't return any public IPs, it's not an error
+	// and we should just get an empty list.
+	ips, err := ci.GetPublicIPs(context.Background())
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(ips) != 0 {
+		t.Fatalf("got %v, want none", ips)
+	}
+}
+
+type fakeIMDS struct {
+	tb    testing.TB
+	paths map[string]string
+}
+
+func (f *fakeIMDS) ServeHTTP(w http.ResponseWriter, r *http.Request) {
+	f.tb.Logf("%s %s", r.Method, r.URL.Path)
+	path := r.URL.Path
+
+	// Handle the /latest/api/token case
+	const token = "fake-imds-token"
+	if r.Method == "PUT" && path == "/latest/api/token" {
+		w.Header().Set("Server", "EC2ws")
+		w.Write([]byte(token))
+		return
+	}
+
+	// Otherwise, require the IMDSv2 token to be set
+	if r.Header.Get("X-aws-ec2-metadata-token") != token {
+		f.tb.Errorf("missing or invalid IMDSv2 token")
+		http.Error(w, "missing or invalid IMDSv2 token", http.StatusForbidden)
+		return
+	}
+
+	if v, ok := f.paths[path]; ok {
+		w.Write([]byte(v))
+		return
+	}
+	http.NotFound(w, r)
+}
--- a/wgengine/magicsock/magicsock.go
+++ b/wgengine/magicsock/magicsock.go
@@ -25,7 +25,6 @@ import (

 	"github.com/tailscale/wireguard-go/conn"
 	"go4.org/mem"
-	"golang.org/x/net/ipv4"
 	"golang.org/x/net/ipv6"

 	"tailscale.com/control/controlknobs"
@@ -133,6 +132,9 @@ type Conn struct {
 	// bind is the wireguard-go conn.Bind for Conn.
 	bind *connBind

+	// cloudInfo is used to query cloud metadata services.
+	cloudInfo *cloudInfo
+
 	// ============================================================
 	// Fields that must be accessed via atomic load/stores.

@@ -425,9 +427,10 @@ func (o *Options) derpActiveFunc() func() {

 // newConn is the error-free, network-listening-side-effect-free based
 // of NewConn. Mostly for tests.
-func newConn() *Conn {
+func newConn(logf logger.Logf) *Conn {
 	discoPrivate := key.NewDisco()
 	c := &Conn{
+		logf:         logf,
 		derpRecvCh:   make(chan derpReadResult, 1), // must be buffered, see issue 3736
 		derpStarted:  make(chan struct{}),
 		peerLastDerp: make(map[key.NodePublic]int),
@@ -435,6 +438,7 @@ func newConn() *Conn {
 		discoInfo:    make(map[key.DiscoPublic]*discoInfo),
 		discoPrivate: discoPrivate,
 		discoPublic:  discoPrivate.Public(),
+		cloudInfo:    newCloudInfo(logf),
 	}
 	c.discoShort = c.discoPublic.ShortString()
 	c.bind = &connBind{Conn: c, closed: true}
@@ -462,10 +466,9 @@ func NewConn(opts Options) (*Conn, error) {
 		return nil, errors.New("magicsock.Options.NetMon must be non-nil")
 	}

-	c := newConn()
+	c := newConn(opts.logf())
 	c.port.Store(uint32(opts.Port))
 	c.controlKnobs = opts.ControlKnobs
-	c.logf = opts.logf()
 	c.epFunc = opts.endpointsFunc()
 	c.derpActiveFunc = opts.derpActiveFunc()
 	c.idleFunc = opts.IdleFunc
@@ -952,6 +955,27 @@ func (c *Conn) determineEndpoints(ctx context.Context) ([]tailcfg.Endpoint, erro
 		addAddr(ap, tailcfg.EndpointExplicitConf)
 	}

+	// If we're on a cloud instance, we might have a public IPv4 or IPv6
+	// address that we can be reached at. Find those, if they exist, and
+	// add them.
+	if addrs, err := c.cloudInfo.GetPublicIPs(ctx); err == nil {
+		var port4, port6 uint16
+		if addr := c.pconn4.LocalAddr(); addr != nil {
+			port4 = uint16(addr.Port)
+		}
+		if addr := c.pconn6.LocalAddr(); addr != nil {
+			port6 = uint16(addr.Port)
+		}
+
+		for _, addr := range addrs {
+			if addr.Is4() && port4 > 0 {
+				addAddr(netip.AddrPortFrom(addr, port4), tailcfg.EndpointLocal)
+			} else if addr.Is6() && port6 > 0 {
+				addAddr(netip.AddrPortFrom(addr, port6), tailcfg.EndpointLocal)
+			}
+		}
+	}
+
 	// Update our set of endpoints by adding any endpoints that we
 	// previously found but haven't expired yet. This also updates the
 	// cache with the set of endpoints discovered in this function.
@@ -1076,12 +1100,6 @@ var errNoUDP = errors.New("no UDP available on platform")

 var errUnsupportedConnType = errors.New("unsupported connection type")

-var (
-	// This acts as a compile-time check for our usage of ipv6.Message in
-	// batchingUDPConn for both IPv6 and IPv4 operations.
-	_ ipv6.Message = ipv4.Message{}
-)
-
 func (c *Conn) sendUDPBatch(addr netip.AddrPort, buffs [][]byte) (sent bool, err error) {
 	isIPv6 := false
 	switch {
@@ -2631,153 +2649,6 @@ func (c *Conn) ParseEndpoint(nodeKeyStr string) (conn.Endpoint, error) {
 	return ep, nil
 }

-func (c *batchingUDPConn) writeBatch(msgs []ipv6.Message) error {
-	var head int
-	for {
-		n, err := c.xpc.WriteBatch(msgs[head:], 0)
-		if err != nil || n == len(msgs[head:]) {
-			// Returning the number of packets written would require
-			// unraveling individual msg len and gso size during a coalesced
-			// write. The top of the call stack disregards partial success,
-			// so keep this simple for now.
-			return err
-		}
-		head += n
-	}
-}
-
-// splitCoalescedMessages splits coalesced messages from the tail of dst
-// beginning at index 'firstMsgAt' into the head of the same slice. It reports
-// the number of elements to evaluate in msgs for nonzero len (msgs[i].N). An
-// error is returned if a socket control message cannot be parsed or a split
-// operation would overflow msgs.
-func (c *batchingUDPConn) splitCoalescedMessages(msgs []ipv6.Message, firstMsgAt int) (n int, err error) {
-	for i := firstMsgAt; i < len(msgs); i++ {
-		msg := &msgs[i]
-		if msg.N == 0 {
-			return n, err
-		}
-		var (
-			gsoSize    int
-			start      int
-			end        = msg.N
-			numToSplit = 1
-		)
-		gsoSize, err = c.getGSOSizeFromControl(msg.OOB[:msg.NN])
-		if err != nil {
-			return n, err
-		}
-		if gsoSize > 0 {
-			numToSplit = (msg.N + gsoSize - 1) / gsoSize
-			end = gsoSize
-		}
-		for j := 0; j < numToSplit; j++ {
-			if n > i {
-				return n, errors.New("splitting coalesced packet resulted in overflow")
-			}
-			copied := copy(msgs[n].Buffers[0], msg.Buffers[0][start:end])
-			msgs[n].N = copied
-			msgs[n].Addr = msg.Addr
-			start = end
-			end += gsoSize
-			if end > msg.N {
-				end = msg.N
-			}
-			n++
-		}
-		if i != n-1 {
-			// It is legal for bytes to move within msg.Buffers[0] as a result
-			// of splitting, so we only zero the source msg len when it is not
-			// the destination of the last split operation above.
-			msg.N = 0
-		}
-	}
-	return n, nil
-}
-
-func (c *batchingUDPConn) ReadBatch(msgs []ipv6.Message, flags int) (n int, err error) {
-	if !c.rxOffload || len(msgs) < 2 {
-		return c.xpc.ReadBatch(msgs, flags)
-	}
-	// Read into the tail of msgs, split into the head.
-	readAt := len(msgs) - 2
-	numRead, err := c.xpc.ReadBatch(msgs[readAt:], 0)
-	if err != nil || numRead == 0 {
-		return 0, err
-	}
-	return c.splitCoalescedMessages(msgs, readAt)
-}
-
-func (c *batchingUDPConn) LocalAddr() net.Addr {
-	return c.pc.LocalAddr().(*net.UDPAddr)
-}
-
-func (c *batchingUDPConn) WriteToUDPAddrPort(b []byte, addr netip.AddrPort) (int, error) {
-	return c.pc.WriteToUDPAddrPort(b, addr)
-}
-
-func (c *batchingUDPConn) Close() error {
-	return c.pc.Close()
-}
-
-// tryUpgradeToBatchingUDPConn probes the capabilities of the OS and pconn, and
-// upgrades pconn to a *batchingUDPConn if appropriate.
-func tryUpgradeToBatchingUDPConn(pconn nettype.PacketConn, network string, batchSize int) nettype.PacketConn {
-	if network != "udp4" && network != "udp6" {
-		return pconn
-	}
-	if runtime.GOOS != "linux" {
-		return pconn
-	}
-	if strings.HasPrefix(hostinfo.GetOSVersion(), "2.") {
-		// recvmmsg/sendmmsg were added in 2.6.33, but we support down to
-		// 2.6.32 for old NAS devices. See https://github.com/tailscale/tailscale/issues/6807.
-		// As a cheap heuristic: if the Linux kernel starts with "2", just
-		// consider it too old for mmsg. Nobody who cares about performance runs
-		// such ancient kernels. UDP offload was added much later, so no
-		// upgrades are available.
-		return pconn
-	}
-	uc, ok := pconn.(*net.UDPConn)
-	if !ok {
-		return pconn
-	}
-	b := &batchingUDPConn{
-		pc:                    pconn,
-		getGSOSizeFromControl: getGSOSizeFromControl,
-		setGSOSizeInControl:   setGSOSizeInControl,
-		sendBatchPool: sync.Pool{
-			New: func() any {
-				ua := &net.UDPAddr{
-					IP: make([]byte, 16),
-				}
-				msgs := make([]ipv6.Message, batchSize)
-				for i := range msgs {
-					msgs[i].Buffers = make([][]byte, 1)
-					msgs[i].Addr = ua
-					msgs[i].OOB = make([]byte, controlMessageSize)
-				}
-				return &sendBatch{
-					ua:   ua,
-					msgs: msgs,
-				}
-			},
-		},
-	}
-	switch network {
-	case "udp4":
-		b.xpc = ipv4.NewPacketConn(uc)
-	case "udp6":
-		b.xpc = ipv6.NewPacketConn(uc)
-	default:
-		panic("bogus network")
-	}
-	var txOffload bool
-	txOffload, b.rxOffload = tryEnableUDPOffload(uc)
-	b.txOffload.Store(txOffload)
-	return b
-}
-
 func newBlockForeverConn() *blockForeverConn {
 	c := new(blockForeverConn)
 	c.cond = sync.NewCond(&c.mu)
--- a/wgengine/magicsock/magicsock_default.go
+++ b/wgengine/magicsock/magicsock_default.go
@@ -21,16 +21,6 @@ func trySetSocketBuffer(pconn nettype.PacketConn, logf logger.Logf) {
 	portableTrySetSocketBuffer(pconn, logf)
 }

-func tryEnableUDPOffload(pconn nettype.PacketConn) (hasTX bool, hasRX bool) {
-	return false, false
-}
-
-func getGSOSizeFromControl(control []byte) (int, error) {
-	return 0, nil
-}
-
-func setGSOSizeInControl(control *[]byte, gso uint16) {}
-
 const (
 	controlMessageSize = 0
 )
--- a/wgengine/magicsock/magicsock_linux.go
+++ b/wgengine/magicsock/magicsock_linux.go
@@ -318,70 +318,6 @@ func trySetSocketBuffer(pconn nettype.PacketConn, logf logger.Logf) {
 	}
 }

-// tryEnableUDPOffload attempts to enable the UDP_GRO socket option on pconn,
-// and returns two booleans indicating TX and RX UDP offload support.
-func tryEnableUDPOffload(pconn nettype.PacketConn) (hasTX bool, hasRX bool) {
-	if c, ok := pconn.(*net.UDPConn); ok {
-		rc, err := c.SyscallConn()
-		if err != nil {
-			return
-		}
-		err = rc.Control(func(fd uintptr) {
-			_, errSyscall := syscall.GetsockoptInt(int(fd), unix.IPPROTO_UDP, unix.UDP_SEGMENT)
-			hasTX = errSyscall == nil
-			errSyscall = syscall.SetsockoptInt(int(fd), unix.IPPROTO_UDP, unix.UDP_GRO, 1)
-			hasRX = errSyscall == nil
-		})
-		if err != nil {
-			return false, false
-		}
-	}
-	return hasTX, hasRX
-}
-
-// getGSOSizeFromControl returns the GSO size found in control. If no GSO size
-// is found or the len(control) < unix.SizeofCmsghdr, this function returns 0.
-// A non-nil error will be returned if len(control) > unix.SizeofCmsghdr but
-// its contents cannot be parsed as a socket control message.
-func getGSOSizeFromControl(control []byte) (int, error) {
-	var (
-		hdr  unix.Cmsghdr
-		data []byte
-		rem  = control
-		err  error
-	)
-
-	for len(rem) > unix.SizeofCmsghdr {
-		hdr, data, rem, err = unix.ParseOneSocketControlMessage(control)
-		if err != nil {
-			return 0, fmt.Errorf("error parsing socket control message: %w", err)
-		}
-		if hdr.Level == unix.SOL_UDP && hdr.Type == unix.UDP_GRO && len(data) >= 2 {
-			return int(binary.NativeEndian.Uint16(data[:2])), nil
-		}
-	}
-	return 0, nil
-}
-
-// setGSOSizeInControl sets a socket control message in control containing
-// gsoSize. If len(control) < controlMessageSize control's len will be set to 0.
-func setGSOSizeInControl(control *[]byte, gsoSize uint16) {
-	*control = (*control)[:0]
-	if cap(*control) < int(unsafe.Sizeof(unix.Cmsghdr{})) {
-		return
-	}
-	if cap(*control) < controlMessageSize {
-		return
-	}
-	*control = (*control)[:cap(*control)]
-	hdr := (*unix.Cmsghdr)(unsafe.Pointer(&(*control)[0]))
-	hdr.Level = unix.SOL_UDP
-	hdr.Type = unix.UDP_SEGMENT
-	hdr.SetLen(unix.CmsgLen(2))
-	binary.NativeEndian.PutUint16((*control)[unix.SizeofCmsghdr:], gsoSize)
-	*control = (*control)[:unix.CmsgSpace(2)]
-}
-
 var controlMessageSize = -1 // bomb if used for allocation before init

 func init() {
--- a/wgengine/magicsock/magicsock_test.go
+++ b/wgengine/magicsock/magicsock_test.go
@@ -35,7 +35,6 @@ import (
 	xmaps "golang.org/x/exp/maps"
 	"golang.org/x/net/icmp"
 	"golang.org/x/net/ipv4"
-	"golang.org/x/net/ipv6"
 	"tailscale.com/cmd/testwrapper/flakytest"
 	"tailscale.com/control/controlknobs"
 	"tailscale.com/derp"
@@ -452,7 +451,7 @@ func TestPickDERPFallback(t *testing.T) {
 	tstest.PanicOnLog()
 	tstest.ResourceCheck(t)

-	c := newConn()
+	c := newConn(t.Logf)
 	dm := &tailcfg.DERPMap{
 		Regions: map[int]*tailcfg.DERPRegion{
 			1: {},
@@ -483,7 +482,7 @@ func TestPickDERPFallback(t *testing.T) {
 	// distribution over nodes works.
 	got := map[int]int{}
 	for range 50 {
-		c = newConn()
+		c = newConn(t.Logf)
 		c.derpMap = dm
 		got[c.pickDERPFallback()]++
 	}
@@ -1185,8 +1184,7 @@ func testTwoDevicePing(t *testing.T, d *devices) {
 }

 func TestDiscoMessage(t *testing.T) {
-	c := newConn()
-	c.logf = t.Logf
+	c := newConn(t.Logf)
 	c.privateKey = key.NewNode()

 	peer1Pub := c.DiscoPublicKey()
@@ -2039,238 +2037,6 @@ func TestBufferedDerpWritesBeforeDrop(t *testing.T) {
 	t.Logf("bufferedDerpWritesBeforeDrop = %d", vv)
 }

-func setGSOSize(control *[]byte, gsoSize uint16) {
-	*control = (*control)[:cap(*control)]
-	binary.LittleEndian.PutUint16(*control, gsoSize)
-}
-
-func getGSOSize(control []byte) (int, error) {
-	if len(control) < 2 {
-		return 0, nil
-	}
-	return int(binary.LittleEndian.Uint16(control)), nil
-}
-
-func Test_batchingUDPConn_splitCoalescedMessages(t *testing.T) {
-	c := &batchingUDPConn{
-		setGSOSizeInControl:   setGSOSize,
-		getGSOSizeFromControl: getGSOSize,
-	}
-
-	newMsg := func(n, gso int) ipv6.Message {
-		msg := ipv6.Message{
-			Buffers: [][]byte{make([]byte, 1024)},
-			N:       n,
-			OOB:     make([]byte, 2),
-		}
-		binary.LittleEndian.PutUint16(msg.OOB, uint16(gso))
-		if gso > 0 {
-			msg.NN = 2
-		}
-		return msg
-	}
-
-	cases := []struct {
-		name        string
-		msgs        []ipv6.Message
-		firstMsgAt  int
-		wantNumEval int
-		wantMsgLens []int
-		wantErr     bool
-	}{
-		{
-			name: "second last split last empty",
-			msgs: []ipv6.Message{
-				newMsg(0, 0),
-				newMsg(0, 0),
-				newMsg(3, 1),
-				newMsg(0, 0),
-			},
-			firstMsgAt:  2,
-			wantNumEval: 3,
-			wantMsgLens: []int{1, 1, 1, 0},
-			wantErr:     false,
-		},
-		{
-			name: "second last no split last empty",
-			msgs: []ipv6.Message{
-				newMsg(0, 0),
-				newMsg(0, 0),
-				newMsg(1, 0),
-				newMsg(0, 0),
-			},
-			firstMsgAt:  2,
-			wantNumEval: 1,
-			wantMsgLens: []int{1, 0, 0, 0},
-			wantErr:     false,
-		},
-		{
-			name: "second last no split last no split",
-			msgs: []ipv6.Message{
-				newMsg(0, 0),
-				newMsg(0, 0),
-				newMsg(1, 0),
-				newMsg(1, 0),
-			},
-			firstMsgAt:  2,
-			wantNumEval: 2,
-			wantMsgLens: []int{1, 1, 0, 0},
-			wantErr:     false,
-		},
-		{
-			name: "second last no split last split",
-			msgs: []ipv6.Message{
-				newMsg(0, 0),
-				newMsg(0, 0),
-				newMsg(1, 0),
-				newMsg(3, 1),
-			},
-			firstMsgAt:  2,
-			wantNumEval: 4,
-			wantMsgLens: []int{1, 1, 1, 1},
-			wantErr:     false,
-		},
-		{
-			name: "second last split last split",
-			msgs: []ipv6.Message{
-				newMsg(0, 0),
-				newMsg(0, 0),
-				newMsg(2, 1),
-				newMsg(2, 1),
-			},
-			firstMsgAt:  2,
-			wantNumEval: 4,
-			wantMsgLens: []int{1, 1, 1, 1},
-			wantErr:     false,
-		},
-		{
-			name: "second last no split last split overflow",
-			msgs: []ipv6.Message{
-				newMsg(0, 0),
-				newMsg(0, 0),
-				newMsg(1, 0),
-				newMsg(4, 1),
-			},
-			firstMsgAt:  2,
-			wantNumEval: 4,
-			wantMsgLens: []int{1, 1, 1, 1},
-			wantErr:     true,
-		},
-	}
-
-	for _, tt := range cases {
-		t.Run(tt.name, func(t *testing.T) {
-			got, err := c.splitCoalescedMessages(tt.msgs, 2)
-			if err != nil && !tt.wantErr {
-				t.Fatalf("err: %v", err)
-			}
-			if got != tt.wantNumEval {
-				t.Fatalf("got to eval: %d want: %d", got, tt.wantNumEval)
-			}
-			for i, msg := range tt.msgs {
-				if msg.N != tt.wantMsgLens[i] {
-					t.Fatalf("msg[%d].N: %d want: %d", i, msg.N, tt.wantMsgLens[i])
-				}
-			}
-		})
-	}
-}
-
-func Test_batchingUDPConn_coalesceMessages(t *testing.T) {
-	c := &batchingUDPConn{
-		setGSOSizeInControl:   setGSOSize,
-		getGSOSizeFromControl: getGSOSize,
-	}
-
-	cases := []struct {
-		name     string
-		buffs    [][]byte
-		wantLens []int
-		wantGSO  []int
-	}{
-		{
-			name: "one message no coalesce",
-			buffs: [][]byte{
-				make([]byte, 1, 1),
-			},
-			wantLens: []int{1},
-			wantGSO:  []int{0},
-		},
-		{
-			name: "two messages equal len coalesce",
-			buffs: [][]byte{
-				make([]byte, 1, 2),
-				make([]byte, 1, 1),
-			},
-			wantLens: []int{2},
-			wantGSO:  []int{1},
-		},
-		{
-			name: "two messages unequal len coalesce",
-			buffs: [][]byte{
-				make([]byte, 2, 3),
-				make([]byte, 1, 1),
-			},
-			wantLens: []int{3},
-			wantGSO:  []int{2},
-		},
-		{
-			name: "three messages second unequal len coalesce",
-			buffs: [][]byte{
-				make([]byte, 2, 3),
-				make([]byte, 1, 1),
-				make([]byte, 2, 2),
-			},
-			wantLens: []int{3, 2},
-			wantGSO:  []int{2, 0},
-		},
-		{
-			name: "three messages limited cap coalesce",
-			buffs: [][]byte{
-				make([]byte, 2, 4),
-				make([]byte, 2, 2),
-				make([]byte, 2, 2),
-			},
-			wantLens: []int{4, 2},
-			wantGSO:  []int{2, 0},
-		},
-	}
-
-	for _, tt := range cases {
-		t.Run(tt.name, func(t *testing.T) {
-			addr := &net.UDPAddr{
-				IP:   net.ParseIP("127.0.0.1"),
-				Port: 1,
-			}
-			msgs := make([]ipv6.Message, len(tt.buffs))
-			for i := range msgs {
-				msgs[i].Buffers = make([][]byte, 1)
-				msgs[i].OOB = make([]byte, 0, 2)
-			}
-			got := c.coalesceMessages(addr, tt.buffs, msgs)
-			if got != len(tt.wantLens) {
-				t.Fatalf("got len %d want: %d", got, len(tt.wantLens))
-			}
-			for i := range got {
-				if msgs[i].Addr != addr {
-					t.Errorf("msgs[%d].Addr != passed addr", i)
-				}
-				gotLen := len(msgs[i].Buffers[0])
-				if gotLen != tt.wantLens[i] {
-					t.Errorf("len(msgs[%d].Buffers[0]) %d != %d", i, gotLen, tt.wantLens[i])
-				}
-				gotGSO, err := getGSOSize(msgs[i].OOB)
-				if err != nil {
-					t.Fatalf("msgs[%d] getGSOSize err: %v", i, err)
-				}
-				if gotGSO != tt.wantGSO[i] {
-					t.Errorf("msgs[%d] gsoSize %d != %d", i, gotGSO, tt.wantGSO[i])
-				}
-			}
-		})
-	}
-}
-
 // newWireguard starts up a new wireguard-go device attached to a test tun, and
 // returns the device, tun and endpoint port. To add peers call device.IpcSet with UAPI instructions.
 func newWireguard(t *testing.T, uapi string, aips []netip.Prefix) (*device.Device, *tuntest.ChannelTUN, uint16) {
@@ -3161,8 +2927,7 @@ func TestMaybeSetNearestDERP(t *testing.T) {
 	for _, tt := range testCases {
 		t.Run(tt.name, func(t *testing.T) {
 			ht := new(health.Tracker)
-			c := newConn()
-			c.logf = t.Logf
+			c := newConn(t.Logf)
 			c.myDerp = tt.old
 			c.derpMap = derpMap
 			c.health = ht
--- a/wgengine/magicsock/rebinding_conn.go
+++ b/wgengine/magicsock/rebinding_conn.go
@@ -35,12 +35,12 @@ type RebindingUDPConn struct {

 // setConnLocked sets the provided nettype.PacketConn. It should be called only
 // after acquiring RebindingUDPConn.mu. It upgrades the provided
-// nettype.PacketConn to a *batchingUDPConn when appropriate. This upgrade
-// is intentionally pushed closest to where read/write ops occur in order to
-// avoid disrupting surrounding code that assumes nettype.PacketConn is a
+// nettype.PacketConn to a batchingConn when appropriate. This upgrade is
+// intentionally pushed closest to where read/write ops occur in order to avoid
+// disrupting surrounding code that assumes nettype.PacketConn is a
 // *net.UDPConn.
 func (c *RebindingUDPConn) setConnLocked(p nettype.PacketConn, network string, batchSize int) {
-	upc := tryUpgradeToBatchingUDPConn(p, network, batchSize)
+	upc := tryUpgradeToBatchingConn(p, network, batchSize)
 	c.pconn = upc
 	c.pconnAtomic.Store(&upc)
 	c.port = uint16(c.localAddrLocked().Port)
@@ -74,7 +74,7 @@ func (c *RebindingUDPConn) ReadFromUDPAddrPort(b []byte) (int, netip.AddrPort, e
 func (c *RebindingUDPConn) WriteBatchTo(buffs [][]byte, addr netip.AddrPort) error {
 	for {
 		pconn := *c.pconnAtomic.Load()
-		b, ok := pconn.(*batchingUDPConn)
+		b, ok := pconn.(batchingConn)
 		if !ok {
 			for _, buf := range buffs {
 				_, err := c.writeToUDPAddrPortWithInitPconn(pconn, buf, addr)
@@ -101,7 +101,7 @@ func (c *RebindingUDPConn) WriteBatchTo(buffs [][]byte, addr netip.AddrPort) err
 func (c *RebindingUDPConn) ReadBatch(msgs []ipv6.Message, flags int) (int, error) {
 	for {
 		pconn := *c.pconnAtomic.Load()
-		b, ok := pconn.(*batchingUDPConn)
+		b, ok := pconn.(batchingConn)
 		if !ok {
 			n, ap, err := c.readFromWithInitPconn(pconn, msgs[0].Buffers[0])
 			if err == nil {
--- a/wgengine/netstack/gro_default.go
+++ b/wgengine/netstack/gro_default.go
@@ -0,0 +1,16 @@
+// Copyright (c) Tailscale Inc & AUTHORS
+// SPDX-License-Identifier: BSD-3-Clause
+
+//go:build !ios
+
+package netstack
+
+import (
+	nsgro "gvisor.dev/gvisor/pkg/tcpip/stack/gro"
+)
+
+// gro wraps a gVisor GRO implementation. It exists solely to prevent iOS from
+// importing said package (see _ios.go).
+type gro struct {
+	nsgro.GRO
+}
--- a/wgengine/netstack/gro_ios.go
+++ b/wgengine/netstack/gro_ios.go
@@ -0,0 +1,30 @@
+// Copyright (c) Tailscale Inc & AUTHORS
+// SPDX-License-Identifier: BSD-3-Clause
+
+//go:build ios
+
+package netstack
+
+import (
+	"gvisor.dev/gvisor/pkg/tcpip/stack"
+)
+
+// gro on iOS delivers packets to its Dispatcher, immediately. This type exists
+// to prevent importation of the gVisor GRO implementation as said package
+// increases binary size. This is a penalty we do not wish to pay since we
+// currently do not leverage GRO on iOS.
+type gro struct {
+	Dispatcher stack.NetworkDispatcher
+}
+
+func (g *gro) Init(v bool) {
+	if v {
+		panic("GRO is not supported on this platform")
+	}
+}
+
+func (g *gro) Flush() {}
+
+func (g *gro) Enqueue(pkt *stack.PacketBuffer) {
+	g.Dispatcher.DeliverNetworkPacket(pkt.NetworkProtocolNumber, pkt)
+}
--- a/wgengine/netstack/link_endpoint.go
+++ b/wgengine/netstack/link_endpoint.go
@@ -4,12 +4,18 @@
 package netstack

 import (
+	"bytes"
 	"context"
 	"sync"

+	"github.com/tailscale/wireguard-go/tun"
+	"gvisor.dev/gvisor/pkg/buffer"
 	"gvisor.dev/gvisor/pkg/tcpip"
 	"gvisor.dev/gvisor/pkg/tcpip/header"
+	"gvisor.dev/gvisor/pkg/tcpip/header/parse"
 	"gvisor.dev/gvisor/pkg/tcpip/stack"
+	"tailscale.com/net/packet"
+	"tailscale.com/types/ipproto"
 )

 type queue struct {
@@ -79,34 +85,53 @@ var _ stack.GSOEndpoint = (*linkEndpoint)(nil)

 // linkEndpoint implements stack.LinkEndpoint and stack.GSOEndpoint. Outbound
 // packets written by gVisor towards Tailscale are stored in a channel.
-// Inbound is fed to gVisor via InjectInbound. This is loosely modeled after
-// gvisor.dev/pkg/tcpip/link/channel.Endpoint.
+// Inbound is fed to gVisor via injectInbound or enqueueGRO. This is loosely
+// modeled after gvisor.dev/pkg/tcpip/link/channel.Endpoint.
 type linkEndpoint struct {
-	LinkEPCapabilities stack.LinkEndpointCapabilities
-	SupportedGSOKind   stack.SupportedGSO
+	SupportedGSOKind stack.SupportedGSO
+	initGRO          initGRO

 	mu         sync.RWMutex // mu guards the following fields
 	dispatcher stack.NetworkDispatcher
 	linkAddr   tcpip.LinkAddress
 	mtu        uint32
+	gro        gro // mu only guards access to gro.Dispatcher

 	q *queue // outbound
 }

-func newLinkEndpoint(size int, mtu uint32, linkAddr tcpip.LinkAddress) *linkEndpoint {
-	return &linkEndpoint{
+// TODO(jwhited): move to linkEndpointOpts struct or similar.
+type initGRO bool
+
+const (
+	disableGRO initGRO = false
+	enableGRO  initGRO = true
+)
+
+func newLinkEndpoint(size int, mtu uint32, linkAddr tcpip.LinkAddress, gro initGRO) *linkEndpoint {
+	le := &linkEndpoint{
 		q: &queue{
 			c: make(chan *stack.PacketBuffer, size),
 		},
 		mtu:      mtu,
 		linkAddr: linkAddr,
 	}
+	le.initGRO = gro
+	le.gro.Init(bool(gro))
+	return le
 }

 // Close closes l. Further packet injections will return an error, and all
 // pending packets are discarded. Close may be called concurrently with
 // WritePackets.
 func (l *linkEndpoint) Close() {
+	l.mu.Lock()
+	if l.gro.Dispatcher != nil {
+		l.gro.Flush()
+	}
+	l.dispatcher = nil
+	l.gro.Dispatcher = nil
+	l.mu.Unlock()
 	l.q.Close()
 	l.Drain()
 }
@@ -132,19 +157,149 @@ func (l *linkEndpoint) Drain() int {
 	return c
 }

-// NumQueued returns the number of packet queued for outbound.
+// NumQueued returns the number of packets queued for outbound.
 func (l *linkEndpoint) NumQueued() int {
 	return l.q.Num()
 }

-// InjectInbound injects an inbound packet. If the endpoint is not attached, the
-// packet is not delivered.
-func (l *linkEndpoint) InjectInbound(protocol tcpip.NetworkProtocolNumber, pkt *stack.PacketBuffer) {
+// rxChecksumOffload validates IPv4, TCP, and UDP header checksums in p,
+// returning an equivalent *stack.PacketBuffer if they are valid, otherwise nil.
+// The set of headers validated covers where gVisor would perform validation if
+// !stack.PacketBuffer.RXChecksumValidated, i.e. it satisfies
+// stack.CapabilityRXChecksumOffload. Other protocols with checksum fields,
+// e.g. ICMP{v6}, are still validated by gVisor regardless of rx checksum
+// offloading capabilities.
+func rxChecksumOffload(p *packet.Parsed) *stack.PacketBuffer {
+	var (
+		pn        tcpip.NetworkProtocolNumber
+		csumStart int
+	)
+	buf := p.Buffer()
+
+	switch p.IPVersion {
+	case 4:
+		if len(buf) < header.IPv4MinimumSize {
+			return nil
+		}
+		csumStart = int((buf[0] & 0x0F) * 4)
+		if csumStart < header.IPv4MinimumSize || csumStart > header.IPv4MaximumHeaderSize || len(buf) < csumStart {
+			return nil
+		}
+		if ^tun.Checksum(buf[:csumStart], 0) != 0 {
+			return nil
+		}
+		pn = header.IPv4ProtocolNumber
+	case 6:
+		if len(buf) < header.IPv6FixedHeaderSize {
+			return nil
+		}
+		csumStart = header.IPv6FixedHeaderSize
+		pn = header.IPv6ProtocolNumber
+		if p.IPProto != ipproto.ICMPv6 && p.IPProto != ipproto.TCP && p.IPProto != ipproto.UDP {
+			// buf could have extension headers before a UDP or TCP header, but
+			// packet.Parsed.IPProto will be set to the ext header type, so we
+			// have to look deeper. We are still responsible for validating the
+			// L4 checksum in this case. So, make use of gVisor's existing
+			// extension header parsing via parse.IPv6() in order to unpack the
+			// L4 csumStart index. This is not particularly efficient as we have
+			// to allocate a short-lived stack.PacketBuffer that cannot be
+			// re-used. parse.IPv6() "consumes" the IPv6 headers, so we can't
+			// inject this stack.PacketBuffer into the stack at a later point.
+			packetBuf := stack.NewPacketBuffer(stack.PacketBufferOptions{
+				Payload: buffer.MakeWithData(bytes.Clone(buf)),
+			})
+			defer packetBuf.DecRef()
+			// The rightmost bool returns false only if packetBuf is too short,
+			// which we've already accounted for above.
+			transportProto, _, _, _, _ := parse.IPv6(packetBuf)
+			if transportProto == header.TCPProtocolNumber || transportProto == header.UDPProtocolNumber {
+				csumLen := packetBuf.Data().Size()
+				if len(buf) < csumLen {
+					return nil
+				}
+				csumStart = len(buf) - csumLen
+				p.IPProto = ipproto.Proto(transportProto)
+			}
+		}
+	}
+
+	if p.IPProto == ipproto.TCP || p.IPProto == ipproto.UDP {
+		lenForPseudo := len(buf) - csumStart
+		csum := tun.PseudoHeaderChecksum(
+			uint8(p.IPProto),
+			p.Src.Addr().AsSlice(),
+			p.Dst.Addr().AsSlice(),
+			uint16(lenForPseudo))
+		csum = tun.Checksum(buf[csumStart:], csum)
+		if ^csum != 0 {
+			return nil
+		}
+	}
+
+	packetBuf := stack.NewPacketBuffer(stack.PacketBufferOptions{
+		Payload: buffer.MakeWithData(bytes.Clone(buf)),
+	})
+	packetBuf.NetworkProtocolNumber = pn
+	// Setting this is not technically required. gVisor overrides where
+	// stack.CapabilityRXChecksumOffload is advertised from Capabilities().
+	// https://github.com/google/gvisor/blob/64c016c92987cc04dfd4c7b091ddd21bdad875f8/pkg/tcpip/stack/nic.go#L763
+	// This is also why we offload for all packets since we cannot signal this
+	// per-packet.
+	packetBuf.RXChecksumValidated = true
+	return packetBuf
+}
+
+func (l *linkEndpoint) injectInbound(p *packet.Parsed) {
 	l.mu.RLock()
 	d := l.dispatcher
 	l.mu.RUnlock()
-	if d != nil {
-		d.DeliverNetworkPacket(protocol, pkt)
+	if d == nil {
+		return
+	}
+	pkt := rxChecksumOffload(p)
+	if pkt == nil {
+		return
+	}
+	d.DeliverNetworkPacket(pkt.NetworkProtocolNumber, pkt)
+	pkt.DecRef()
+}
+
+// enqueueGRO enqueues the provided packet for GRO. It may immediately deliver
+// it to the underlying stack.NetworkDispatcher depending on its contents and if
+// GRO was initialized via newLinkEndpoint. To explicitly flush previously
+// enqueued packets see flushGRO. enqueueGRO is not thread-safe and must not
+// be called concurrently with flushGRO.
+func (l *linkEndpoint) enqueueGRO(p *packet.Parsed) {
+	l.mu.RLock()
+	defer l.mu.RUnlock()
+	if l.gro.Dispatcher == nil {
+		return
+	}
+	pkt := rxChecksumOffload(p)
+	if pkt == nil {
+		return
+	}
+	// TODO(jwhited): gro.Enqueue() duplicates a lot of p.Decode().
+	//  We may want to push stack.PacketBuffer further up as a
+	//  replacement for packet.Parsed, or inversely push packet.Parsed
+	//  down into refactored GRO logic.
+	l.gro.Enqueue(pkt)
+	pkt.DecRef()
+}
+
+// flushGRO flushes previously enqueueGRO'd packets to the underlying
+// stack.NetworkDispatcher. flushGRO is not thread-safe, and must not be
+// called concurrently with enqueueGRO.
+func (l *linkEndpoint) flushGRO() {
+	if !l.initGRO {
+		// If GRO was not initialized fast path return to avoid scanning GRO
+		// buckets (see l.gro.Flush()) that will always be empty.
+		return
+	}
+	l.mu.RLock()
+	defer l.mu.RUnlock()
+	if l.gro.Dispatcher != nil {
+		l.gro.Flush()
 	}
 }

@@ -154,6 +309,7 @@ func (l *linkEndpoint) Attach(dispatcher stack.NetworkDispatcher) {
 	l.mu.Lock()
 	defer l.mu.Unlock()
 	l.dispatcher = dispatcher
+	l.gro.Dispatcher = dispatcher
 }

 // IsAttached implements stack.LinkEndpoint.IsAttached.
@@ -179,7 +335,9 @@ func (l *linkEndpoint) SetMTU(mtu uint32) {

 // Capabilities implements stack.LinkEndpoint.Capabilities.
 func (l *linkEndpoint) Capabilities() stack.LinkEndpointCapabilities {
-	return l.LinkEPCapabilities
+	// We are required to offload RX checksum validation for the purposes of
+	// GRO.
+	return stack.CapabilityRXChecksumOffload
 }

 // GSOMaxSize implements stack.GSOEndpoint.
--- a/wgengine/netstack/link_endpoint_test.go
+++ b/wgengine/netstack/link_endpoint_test.go
@@ -0,0 +1,112 @@
+// Copyright (c) Tailscale Inc & AUTHORS
+// SPDX-License-Identifier: BSD-3-Clause
+
+package netstack
+
+import (
+	"bytes"
+	"net/netip"
+	"testing"
+
+	"gvisor.dev/gvisor/pkg/tcpip"
+	"gvisor.dev/gvisor/pkg/tcpip/header"
+	"tailscale.com/net/packet"
+)
+
+func Test_rxChecksumOffload(t *testing.T) {
+	payloadLen := 100
+
+	tcpFields := &header.TCPFields{
+		SrcPort:    1,
+		DstPort:    1,
+		SeqNum:     1,
+		AckNum:     1,
+		DataOffset: 20,
+		Flags:      header.TCPFlagAck | header.TCPFlagPsh,
+		WindowSize: 3000,
+	}
+	tcp4 := make([]byte, 20+20+payloadLen)
+	ipv4H := header.IPv4(tcp4)
+	ipv4H.Encode(&header.IPv4Fields{
+		SrcAddr:     tcpip.AddrFromSlice(netip.MustParseAddr("192.0.2.1").AsSlice()),
+		DstAddr:     tcpip.AddrFromSlice(netip.MustParseAddr("192.0.2.2").AsSlice()),
+		Protocol:    uint8(header.TCPProtocolNumber),
+		TTL:         64,
+		TotalLength: uint16(len(tcp4)),
+	})
+	ipv4H.SetChecksum(^ipv4H.CalculateChecksum())
+	tcpH := header.TCP(tcp4[20:])
+	tcpH.Encode(tcpFields)
+	pseudoCsum := header.PseudoHeaderChecksum(header.TCPProtocolNumber, ipv4H.SourceAddress(), ipv4H.DestinationAddress(), uint16(20+payloadLen))
+	tcpH.SetChecksum(^tcpH.CalculateChecksum(pseudoCsum))
+
+	tcp6ExtHeader := make([]byte, 40+8+20+payloadLen)
+	ipv6H := header.IPv6(tcp6ExtHeader)
+	ipv6H.Encode(&header.IPv6Fields{
+		SrcAddr:           tcpip.AddrFromSlice(netip.MustParseAddr("2001:db8::1").AsSlice()),
+		DstAddr:           tcpip.AddrFromSlice(netip.MustParseAddr("2001:db8::2").AsSlice()),
+		TransportProtocol: 60, // really next header; destination options ext header
+		HopLimit:          64,
+		PayloadLength:     uint16(8 + 20 + payloadLen),
+	})
+	tcp6ExtHeader[40] = uint8(header.TCPProtocolNumber) // next header
+	tcp6ExtHeader[41] = 0                               // length of ext header in 8-octet units, exclusive of first 8 octets.
+	// 42-47 options and padding
+	tcpH = header.TCP(tcp6ExtHeader[48:])
+	tcpH.Encode(tcpFields)
+	pseudoCsum = header.PseudoHeaderChecksum(header.TCPProtocolNumber, ipv6H.SourceAddress(), ipv6H.DestinationAddress(), uint16(20+payloadLen))
+	tcpH.SetChecksum(^tcpH.CalculateChecksum(pseudoCsum))
+
+	tcp4InvalidCsum := make([]byte, len(tcp4))
+	copy(tcp4InvalidCsum, tcp4)
+	at := 20 + 16
+	tcp4InvalidCsum[at] = ^tcp4InvalidCsum[at]
+
+	tcp6ExtHeaderInvalidCsum := make([]byte, len(tcp6ExtHeader))
+	copy(tcp6ExtHeaderInvalidCsum, tcp6ExtHeader)
+	at = 40 + 8 + 16
+	tcp6ExtHeaderInvalidCsum[at] = ^tcp6ExtHeaderInvalidCsum[at]
+
+	tests := []struct {
+		name   string
+		input  []byte
+		wantPB bool
+	}{
+		{
+			"tcp4 packet valid csum",
+			tcp4,
+			true,
+		},
+		{
+			"tcp6 with ext header valid csum",
+			tcp6ExtHeader,
+			true,
+		},
+		{
+			"tcp4 packet invalid csum",
+			tcp4InvalidCsum,
+			false,
+		},
+		{
+			"tcp6 with ext header invalid csum",
+			tcp6ExtHeaderInvalidCsum,
+			false,
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			p := &packet.Parsed{}
+			p.Decode(tt.input)
+			got := rxChecksumOffload(p)
+			if tt.wantPB != (got != nil) {
+				t.Fatalf("wantPB = %v != (got != nil): %v", tt.wantPB, got != nil)
+			}
+			if tt.wantPB {
+				gotBuf := got.ToBuffer()
+				if !bytes.Equal(tt.input, gotBuf.Flatten()) {
+					t.Fatal("output packet unequal to input")
+				}
+			}
+		})
+	}
+}
--- a/wgengine/netstack/netstack.go
+++ b/wgengine/netstack/netstack.go
@@ -5,7 +5,6 @@
 package netstack

 import (
-	"bytes"
 	"context"
 	"errors"
 	"expvar"
@@ -21,7 +20,6 @@ import (
 	"sync/atomic"
 	"time"

-	"gvisor.dev/gvisor/pkg/buffer"
 	"gvisor.dev/gvisor/pkg/refs"
 	"gvisor.dev/gvisor/pkg/tcpip"
 	"gvisor.dev/gvisor/pkg/tcpip/adapters/gonet"
@@ -244,6 +242,44 @@ const nicID = 1
 // have a UDP packet as big as the MTU.
 const maxUDPPacketSize = tstun.MaxPacketSize

+func setTCPBufSizes(ipstack *stack.Stack) error {
+	// tcpip.TCP{Receive,Send}BufferSizeRangeOption is gVisor's version of
+	// Linux's tcp_{r,w}mem. Application within gVisor differs as some Linux
+	// features are not (yet) implemented, and socket buffer memory is not
+	// controlled within gVisor, e.g. we allocate *stack.PacketBuffer's for the
+	// write path within Tailscale. Therefore, we loosen our understanding of
+	// the relationship between these Linux and gVisor tunables. The chosen
+	// values are biased towards higher throughput on high bandwidth-delay
+	// product paths, except on memory-constrained platforms.
+	tcpRXBufOpt := tcpip.TCPReceiveBufferSizeRangeOption{
+		// Min is unused by gVisor at the time of writing, but partially plumbed
+		// for application by the TCP_WINDOW_CLAMP socket option.
+		Min: tcpRXBufMinSize,
+		// Default is used by gVisor at socket creation.
+		Default: tcpRXBufDefSize,
+		// Max is used by gVisor to cap the advertised receive window post-read.
+		// (tcp_moderate_rcvbuf=true, the default).
+		Max: tcpRXBufMaxSize,
+	}
+	tcpipErr := ipstack.SetTransportProtocolOption(tcp.ProtocolNumber, &tcpRXBufOpt)
+	if tcpipErr != nil {
+		return fmt.Errorf("could not set TCP RX buf size: %v", tcpipErr)
+	}
+	tcpTXBufOpt := tcpip.TCPSendBufferSizeRangeOption{
+		// Min in unused by gVisor at the time of writing.
+		Min: tcpTXBufMinSize,
+		// Default is used by gVisor at socket creation.
+		Default: tcpTXBufDefSize,
+		// Max is used by gVisor to cap the send window.
+		Max: tcpTXBufMaxSize,
+	}
+	tcpipErr = ipstack.SetTransportProtocolOption(tcp.ProtocolNumber, &tcpTXBufOpt)
+	if tcpipErr != nil {
+		return fmt.Errorf("could not set TCP TX buf size: %v", tcpipErr)
+	}
+	return nil
+}
+
 // Create creates and populates a new Impl.
 func Create(logf logger.Logf, tundev *tstun.Wrapper, e wgengine.Engine, mc *magicsock.Conn, dialer *tsdial.Dialer, dns *dns.Manager, pm *proxymap.Mapper, driveForLocal drive.FileSystemForLocal) (*Impl, error) {
 	if mc == nil {
@@ -284,10 +320,17 @@ func Create(logf logger.Logf, tundev *tstun.Wrapper, e wgengine.Engine, mc *magi
 			return nil, fmt.Errorf("could not disable TCP RACK: %v", tcpipErr)
 		}
 	}
-	linkEP := newLinkEndpoint(512, uint32(tstun.DefaultTUNMTU()), "")
+	err := setTCPBufSizes(ipstack)
+	if err != nil {
+		return nil, err
+	}
+	var linkEP *linkEndpoint
 	if runtime.GOOS == "linux" {
 		// TODO(jwhited): add Windows support https://github.com/tailscale/corp/issues/21874
+		linkEP = newLinkEndpoint(512, uint32(tstun.DefaultTUNMTU()), "", enableGRO)
 		linkEP.SupportedGSOKind = stack.HostGSOSupported
+	} else {
+		linkEP = newLinkEndpoint(512, uint32(tstun.DefaultTUNMTU()), "", disableGRO)
 	}
 	if tcpipProblem := ipstack.CreateNIC(nicID, linkEP); tcpipProblem != nil {
 		return nil, fmt.Errorf("could not create netstack NIC: %v", tcpipProblem)
@@ -336,6 +379,7 @@ func Create(logf logger.Logf, tundev *tstun.Wrapper, e wgengine.Engine, mc *magi
 	ns.ctx, ns.ctxCancel = context.WithCancel(context.Background())
 	ns.atomicIsLocalIPFunc.Store(ipset.FalseContainsIPFunc())
 	ns.tundev.PostFilterPacketInboundFromWireGuard = ns.injectInbound
+	ns.tundev.EndPacketVectorInboundFromWireGuardFlush = linkEP.flushGRO
 	ns.tundev.PreFilterPacketOutboundToWireGuardNetstackIntercept = ns.handleLocalPackets
 	stacksForMetrics.Store(ns, struct{}{})
 	return ns, nil
@@ -512,9 +556,7 @@ func (ns *Impl) Start(lb *ipnlocal.LocalBackend) error {
 		panic("nil LocalBackend")
 	}
 	ns.lb = lb
-	// size = 0 means use default buffer size
-	const tcpReceiveBufferSize = 0
-	tcpFwd := tcp.NewForwarder(ns.ipstack, tcpReceiveBufferSize, maxInFlightConnectionAttempts(), ns.acceptTCP)
+	tcpFwd := tcp.NewForwarder(ns.ipstack, tcpRXBufDefSize, maxInFlightConnectionAttempts(), ns.acceptTCP)
 	udpFwd := udp.NewForwarder(ns.ipstack, ns.acceptUDP)
 	ns.ipstack.SetTransportProtocolHandler(tcp.ProtocolNumber, ns.wrapTCPProtocolHandler(tcpFwd.HandlePacket))
 	ns.ipstack.SetTransportProtocolHandler(udp.ProtocolNumber, ns.wrapUDPProtocolHandler(udpFwd.HandlePacket))
@@ -737,23 +779,11 @@ func (ns *Impl) handleLocalPackets(p *packet.Parsed, t *tstun.Wrapper) filter.Re
 		// care about the packet; resume processing.
 		return filter.Accept
 	}
-
-	var pn tcpip.NetworkProtocolNumber
-	switch p.IPVersion {
-	case 4:
-		pn = header.IPv4ProtocolNumber
-	case 6:
-		pn = header.IPv6ProtocolNumber
-	}
 	if debugPackets {
 		ns.logf("[v2] service packet in (from %v): % x", p.Src, p.Buffer())
 	}

-	packetBuf := stack.NewPacketBuffer(stack.PacketBufferOptions{
-		Payload: buffer.MakeWithData(bytes.Clone(p.Buffer())),
-	})
-	ns.linkEP.InjectInbound(pn, packetBuf)
-	packetBuf.DecRef()
+	ns.linkEP.injectInbound(p)
 	return filter.DropSilently
 }

@@ -794,7 +824,7 @@ func (ns *Impl) DialContextUDP(ctx context.Context, ipp netip.AddrPort) (*gonet.
 func (ns *Impl) inject() {
 	for {
 		pkt := ns.linkEP.ReadContext(ns.ctx)
-		if pkt.IsNil() {
+		if pkt == nil {
 			if ns.ctx.Err() != nil {
 				// Return without logging.
 				return
@@ -1038,21 +1068,10 @@ func (ns *Impl) injectInbound(p *packet.Parsed, t *tstun.Wrapper) filter.Respons
 		return filter.DropSilently
 	}

-	var pn tcpip.NetworkProtocolNumber
-	switch p.IPVersion {
-	case 4:
-		pn = header.IPv4ProtocolNumber
-	case 6:
-		pn = header.IPv6ProtocolNumber
-	}
 	if debugPackets {
 		ns.logf("[v2] packet in (from %v): % x", p.Src, p.Buffer())
 	}
-	packetBuf := stack.NewPacketBuffer(stack.PacketBufferOptions{
-		Payload: buffer.MakeWithData(bytes.Clone(p.Buffer())),
-	})
-	ns.linkEP.InjectInbound(pn, packetBuf)
-	packetBuf.DecRef()
+	ns.linkEP.enqueueGRO(p)

 	// We've now delivered this to netstack, so we're done.
 	// Instead of returning a filter.Accept here (which would also
--- a/wgengine/netstack/netstack_tcpbuf_default.go
+++ b/wgengine/netstack/netstack_tcpbuf_default.go
@@ -0,0 +1,20 @@
+// Copyright (c) Tailscale Inc & AUTHORS
+// SPDX-License-Identifier: BSD-3-Clause
+
+//go:build !ios
+
+package netstack
+
+import (
+	"gvisor.dev/gvisor/pkg/tcpip/transport/tcp"
+)
+
+const (
+	tcpRXBufMinSize = tcp.MinBufferSize
+	tcpRXBufDefSize = tcp.DefaultSendBufferSize
+	tcpRXBufMaxSize = 8 << 20 // 8MiB
+
+	tcpTXBufMinSize = tcp.MinBufferSize
+	tcpTXBufDefSize = tcp.DefaultReceiveBufferSize
+	tcpTXBufMaxSize = 6 << 20 // 6MiB
+)
--- a/wgengine/netstack/netstack_tcpbuf_ios.go
+++ b/wgengine/netstack/netstack_tcpbuf_ios.go
@@ -0,0 +1,24 @@
+// Copyright (c) Tailscale Inc & AUTHORS
+// SPDX-License-Identifier: BSD-3-Clause
+
+//go:build ios
+
+package netstack
+
+import (
+	"gvisor.dev/gvisor/pkg/tcpip/transport/tcp"
+)
+
+const (
+	// tcp{RX,TX}Buf{Min,Def,Max}Size mirror gVisor defaults. We leave these
+	// unchanged on iOS for now as to not increase pressure towards the
+	// NetworkExtension memory limit.
+	// TODO(jwhited): test memory/throughput impact of collapsing to values in _default.go
+	tcpRXBufMinSize = tcp.MinBufferSize
+	tcpRXBufDefSize = tcp.DefaultSendBufferSize
+	tcpRXBufMaxSize = tcp.MaxBufferSize
+
+	tcpTXBufMinSize = tcp.MinBufferSize
+	tcpTXBufDefSize = tcp.DefaultReceiveBufferSize
+	tcpTXBufMaxSize = tcp.MaxBufferSize
+)
--- a/wgengine/userspace.go
+++ b/wgengine/userspace.go
@@ -374,7 +374,7 @@ func NewUserspaceEngine(logf logger.Logf, conf Config) (_ Engine, reterr error)
 		e.logf("onPortUpdate(port=%v, network=%s)", port, network)

 		if err := e.router.UpdateMagicsockPort(port, network); err != nil {
-			e.logf("UpdateMagicsockPort(port=%v, network=%s) failed: %w", port, network, err)
+			e.logf("UpdateMagicsockPort(port=%v, network=%s) failed: %v", port, network, err)
 		}
 	}
 	magicsockOpts := magicsock.Options{
--- a/words/tails.txt
+++ b/words/tails.txt
@@ -692,3 +692,5 @@ azules
 tabby
 ussuri
 kitty
+tanuki
+neko
Author	SHA1	Message	Date
Andrew Lytvynov	776ab357b1	cmd/tailscale: add --json-docs flag This prints all command and flag docs as JSON. To be used for generating the contents of https://tailscale.com/kb/1080/cli. Updates https://github.com/tailscale/tailscale-www/issues/4722	2024-08-08 08:06:18 -07:00
Jordan Whited	a93dc6cdb1	wgengine/magicsock: refactor batchingUDPConn to batchingConn interface (#13042 ) This commit adds a batchingConn interface, and renames batchingUDPConn to linuxBatchingConn. tryUpgradeToBatchingConn() may return a platform- specific implementation of batchingConn. So far only a Linux implementation of this interface exists, but this refactor is being done in anticipation of a Windows implementation. Updates tailscale/corp#21874 Signed-off-by: Jordan Whited <jordan@tailscale.com>	2024-08-06 09:00:28 -07:00
Anton Tolchanov	7bac5dffcb	control/controlhttp: extract the last network connection The same context we use for the HTTP request here might be re-used by the dialer, which could result in `GotConn` being called multiple times. We only care about the last one. Fixes #13009 Signed-off-by: Anton Tolchanov <anton@tailscale.com>	2024-08-06 11:42:06 +01:00
Anton Tolchanov	b3fc345aba	cmd/derpprobe: use a status page from the prober library Updates tailscale/corp#20583 Signed-off-by: Anton Tolchanov <anton@tailscale.com>	2024-08-06 11:27:59 +01:00
Anton Tolchanov	9106187a95	prober: support JSON response in RunHandler Updates tailscale/corp#20583 Signed-off-by: Anton Tolchanov <anton@tailscale.com>	2024-08-06 11:27:59 +01:00
Anton Tolchanov	9b08399d9e	prober: add a status page handler This change adds an HTTP handler with a table showing a list of all probes, their status, and a button that allows triggering a specific probe. Updates tailscale/corp#20583 Signed-off-by: Anton Tolchanov <anton@tailscale.com>	2024-08-06 11:27:59 +01:00
Anton Tolchanov	153a476957	prober: add an HTTP endpoint for triggering a probe - Keep track of the last 10 probe results and successful probe latencies; - Add an HTTP handler that triggers a given probe by name and returns it result as a plaintext HTML page, showing recent probe results as a baseline Updates tailscale/corp#20583 Signed-off-by: Anton Tolchanov <anton@tailscale.com>	2024-08-06 11:27:59 +01:00
Anton Tolchanov	227509547f	{control,net}: close idle connections of custom transports I noticed a few places with custom http.Transport where we are not closing idle connections when transport is no longer used. Updates tailscale/corp#21609 Signed-off-by: Anton Tolchanov <anton@tailscale.com>	2024-08-05 17:28:15 +01:00
VimT	e3f047618b	net/socks5: support UDP Updates #7581 Signed-off-by: VimT <me@vimt.me>	2024-08-05 09:25:24 -07:00
Kot C	91d2e1772d	words: raccoon dog, dog with the raccoon in 'im Signed-off-by: Kot C <kot@yukata.dev>	2024-08-05 09:24:33 -07:00
License Updater	3b6849e362	licenses: update license notices Signed-off-by: License Updater <noreply+license-updater@tailscale.com>	2024-08-05 08:45:07 -07:00
Anton Tolchanov	0fd73746dd	cmd/tailscale/cli: fix `revoke-keys` command name in CLI output During review of #8644 the `recover-compromised-key` command was renamed to `revoke-key`, but the old name remained in some messages printed by the command. Fixes tailscale/corp#19446 Signed-off-by: Anton Tolchanov <anton@tailscale.com>	2024-08-05 14:49:48 +01:00
Jordan Whited	17c88a19be	net/captivedetection: mark TestAllEndpointsAreUpAndReturnExpectedResponse flaky (#13021 ) Updates #13019 Signed-off-by: Jordan Whited <jordan@tailscale.com>	2024-08-03 22:08:55 +00:00
Jordan Whited	25f0a3fc8f	wgengine/netstack: use build tags to exclude gVisor GRO importation on iOS (#13015 ) Updates tailscale/corp#22125 Signed-off-by: Jordan Whited <jordan@tailscale.com>	2024-08-03 15:03:44 -07:00
Maisem Ali	a7a394e7d9	tstest/integration: mark TestNATPing flaky Updates #12169 Signed-off-by: Maisem Ali <maisem@tailscale.com>	2024-08-03 15:02:17 -07:00
Maisem Ali	07e2487c1d	wgengine/capture: fix v6 field typo in wireshark dissector It was using a v4 field for a v6 address. Updates tailscale/corp#8020 Signed-off-by: Maisem Ali <maisem@tailscale.com>	2024-08-03 14:56:17 -07:00
Maisem Ali	1dd9c44d51	tsweb: mark TestStdHandler_ConnectionClosedDuringBody flaky Updates #13107 Signed-off-by: Maisem Ali <maisem@tailscale.com>	2024-08-03 14:54:10 -07:00
Flakes Updater	0a6eb12f05	go.mod.sri: update SRI hash for go.mod changes Signed-off-by: Flakes Updater <noreply+flakes-updater@tailscale.com>	2024-08-03 11:45:38 -07:00
Maisem Ali	f205efcf18	net/packet/checksum: fix v6 NAT We were copying 12 out of the 16 bytes which meant that the 1:1 NAT required would only work if the last 4 bytes happened to match between the new and old address, something that our tests accidentally had. Fix it by copying the full 16 bytes and make the tests also verify the addr and use rand addresses. Updates #9511 Signed-off-by: Maisem Ali <maisem@tailscale.com>	2024-08-03 11:38:00 -07:00
Maisem Ali	a917718353	util/linuxfw: return nil interface not concrete type It was returning a nil `*iptablesRunner` instead of a nil `NetfilterRunner` interface which would then fail checks later. Fixes #13012 Signed-off-by: Maisem Ali <maisem@tailscale.com>	2024-08-03 09:53:46 -07:00
Nick Khyl	4099a36468	util/winutil/gp: fix a busy loop bug Updates #12687 Signed-off-by: Nick Khyl <nickk@tailscale.com>	2024-08-02 20:16:41 -05:00
Jordan Whited	d9d9d525d9	wgengine/netstack: increase gVisor's TCP send and receive buffer sizes (#12994 ) This commit increases gVisor's TCP max send (4->6MiB) and receive (4->8MiB) buffer sizes on all platforms except iOS. These values are biased towards higher throughput on high bandwidth-delay product paths. The iperf3 results below demonstrate the effect of this commit between two Linux computers with i5-12400 CPUs. 100ms of RTT latency is introduced via Linux's traffic control network emulator queue discipline. The first set of results are from commit `f0230ce` prior to TCP buffer resizing. gVisor write direction: Test Complete. Summary Results: [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.00 sec 180 MBytes 151 Mbits/sec 0 sender [ 5] 0.00-10.10 sec 179 MBytes 149 Mbits/sec receiver gVisor read direction: Test Complete. Summary Results: [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.10 sec 337 MBytes 280 Mbits/sec 20 sender [ 5] 0.00-10.00 sec 323 MBytes 271 Mbits/sec receiver The second set of results are from this commit with increased TCP buffer sizes. gVisor write direction: [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.00 sec 297 MBytes 249 Mbits/sec 0 sender [ 5] 0.00-10.10 sec 297 MBytes 247 Mbits/sec receiver gVisor read direction: [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.10 sec 501 MBytes 416 Mbits/sec 17 sender [ 5] 0.00-10.00 sec 485 MBytes 407 Mbits/sec receiver Updates #9707 Updates tailscale/corp#22119 Signed-off-by: Jordan Whited <jordan@tailscale.com>	2024-08-02 15:50:47 -07:00
Andrew Dunham	9939374c48	wgengine/magicsock: use cloud metadata to get public IPs Updates #12774 Signed-off-by: Andrew Dunham <andrew@du.nham.ca> Change-Id: I1661b6a2da7966ab667b075894837afd96f4742f	2024-08-02 16:05:14 -04:00
Andrea Gottardo	4055b63b9b	net/captivedetection: exclude cellular data interfaces (#13002 ) Updates tailscale/tailscale#1634 This PR optimizes captive portal detection on Android and iOS by excluding cellular data interfaces (`pdp*` and `rmnet`). As cellular networks do not present captive portals, frequent network switches between Wi-Fi and cellular would otherwise trigger captive detection unnecessarily, causing battery drain. Signed-off-by: Andrea Gottardo <andrea@gottardo.me>	2024-08-02 12:23:48 -07:00
Jordan Whited	f0230ce0b5	go.mod,net/tstun,wgengine/netstack: implement gVisor TCP GRO for Linux (#12921 ) This commit implements TCP GRO for packets being written to gVisor on Linux. Windows support will follow later. The wireguard-go dependency is updated in order to make use of newly exported IP checksum functions. gVisor is updated in order to make use of newly exported stack.PacketBuffer GRO logic. TCP throughput towards gVisor, i.e. TUN write direction, is dramatically improved as a result of this commit. Benchmarks show substantial improvement, sometimes as high as 2x. High bandwidth-delay product paths remain receive window limited, bottlenecked by gVisor's default TCP receive socket buffer size. This will be addressed in a follow-on commit. The iperf3 results below demonstrate the effect of this commit between two Linux computers with i5-12400 CPUs. There is roughly ~13us of round trip latency between them. The first result is from commit `57856fc` without TCP GRO. Starting Test: protocol: TCP, 1 streams, 131072 byte blocks - - - - - - - - - - - - - - - - - - - - - - - - - Test Complete. Summary Results: [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.00 sec 4.77 GBytes 4.10 Gbits/sec 20 sender [ 5] 0.00-10.00 sec 4.77 GBytes 4.10 Gbits/sec receiver The second result is from this commit with TCP GRO. Starting Test: protocol: TCP, 1 streams, 131072 byte blocks - - - - - - - - - - - - - - - - - - - - - - - - - Test Complete. Summary Results: [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.00 sec 10.6 GBytes 9.14 Gbits/sec 20 sender [ 5] 0.00-10.00 sec 10.6 GBytes 9.14 Gbits/sec receiver Updates #6816 Signed-off-by: Jordan Whited <jordan@tailscale.com>	2024-08-02 10:41:10 -07:00
Brad Fitzpatrick	cc370314e7	health: don't show login error details with context cancelations Fixes #12991 Change-Id: I2a5e109395761b720ecf1069d0167cf0caf72876 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	2024-08-01 08:29:27 -07:00