rproxy: read the rest of a Connection: close visitor's body after an early answer (closes #373)

Go's server reads up to 256 KiB of a body left unread once the handler
returns, but not a body with a declared length from a visitor that sent
Connection: close. It closes that connection with the rest unread, and
the visitor, still sending, gets a reset. rproxy now reads that rest
itself, until the deadline already set on the connection, through a
MaxBytesReader made with Go's own ResponseWriter. With more than 256 KiB
left, that has Go's server tell the visitor it sends no more and wait
500 ms before it closes. rproxy reads the rest only after an answer with
a Content-Length, because Go's server ends any other answer only once
the handler returns.

Model: opus-5-5

internal/rproxy/limits_test.go

@@ -26,6 +26,10 @@ import (
 // post is the start of a POST whose headers end with what follows it.
 const post = "POST / HTTP/1.1\r\nHost: site.example\r\n"
 
+// limitPastFirstPart is a SWW_RPROXY_MAX_REQUEST_SIZE past the first part of a
+// body, the 64 KiB rproxy reads before it asks waf.
+const limitPastFirstPart = "128K"
+
 // announcedLength is the header that announces a body of size bytes, and the
 // end of the headers.
 func announcedLength(size int) string {
@@ -75,7 +79,10 @@ func TestABodyAnnouncedOverItsLimitIsRefusedOnceWAFAllowsIt(t *testing.T) {
 		size, asked int
 	}{
 		{"with the limit within the first part", "1K", 1025, 1024},
-		{"with the limit past the first part", "128K", 129 << 10, questionBodyMaxBytes},
+		{
+			"with the limit past the first part", limitPastFirstPart, 129 << 10,
+			questionBodyMaxBytes,
+		},
 	} {
 		t.Run(tc.name, func(t *testing.T) {
 			t.Parallel()
@@ -136,6 +143,134 @@ func TestADenyOutranksTheBodySizeLimit(t *testing.T) {
 	}
 }
 
+func TestAVisitorSendingConnectionCloseReadsAnAnswerBeforeTheRestOfItsBody(
+	t *testing.T,
+) {
+	t.Parallel()
+
+	for _, tc := range []struct {
+		name   string
+		waf    http.HandlerFunc
+		rest   int
+		status int
+	}{
+		{"over the size limit", allow, 65 << 10, http.StatusRequestEntityTooLarge},
+		{"denied", deny, 65 << 10, http.StatusForbidden},
+		// rproxy reads 256 KiB of this rest. What it leaves unread must fit in the
+		// connection's buffers. Otherwise the visitor is still sending when the
+		// connection closes, and gets a reset.
+		{
+			"with more than 256 KiB left", allow, 272 << 10,
+			http.StatusRequestEntityTooLarge,
+		},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			t.Parallel()
+
+			standIn := startWAF(t, tc.waf)
+			app := startApp(t, func(http.ResponseWriter, *http.Request) {
+				t.Error("the app was asked")
+			})
+			addr, _ := startRProxy(t, standIn.URL, app.URL,
+				map[string]string{maxRequestSize: limitPastFirstPart})
+
+			conn := dial(t, addr)
+			send(t, conn, post+"Connection: close\r\n"+
+				announcedLength(questionBodyMaxBytes+tc.rest)+
+				strings.Repeat("x", questionBodyMaxBytes))
+
+			reader := bufio.NewReader(conn)
+
+			res, err := http.ReadResponse(reader, nil)
+			if err != nil {
+				t.Fatalf("read the answer: %v", err)
+			}
+
+			_, _ = io.Copy(io.Discard, res.Body)
+			_ = res.Body.Close()
+
+			if res.StatusCode != tc.status {
+				t.Errorf("status %d, want %d", res.StatusCode, tc.status)
+			}
+
+			// The rest goes in small parts, so that a reset fails a later one.
+			for sent := 0; sent < tc.rest; sent += 1 << 10 {
+				send(t, conn, strings.Repeat("x", 1<<10))
+			}
+
+			got, err := io.ReadAll(reader)
+			if err != nil || len(got) != 0 {
+				t.Errorf(
+					"after the answer, read %q and %v, want the end",
+					got,
+					err,
+				)
+			}
+		})
+	}
+}
+
+func TestCloseDoesNotWaitForTheRestOfTheBodyOfAVisitorSendingConnectionClose(
+	t *testing.T,
+) {
+	t.Parallel()
+
+	standIn := startWAF(t, func(w http.ResponseWriter, _ *http.Request) {
+		w.Header().Set(dispositionHeader, dispositionClose)
+		w.WriteHeader(http.StatusOK)
+	})
+	app := startApp(t, func(http.ResponseWriter, *http.Request) {
+		t.Error("the app was asked")
+	})
+	// The request time limit is far longer than dial's deadline.
+	addr, _ := startRProxy(t, standIn.URL, app.URL, map[string]string{
+		maxRequestTime: "120000",
+	})
+
+	conn := dial(t, addr)
+	send(t, conn, post+"Connection: close\r\n"+
+		announcedLength(2*questionBodyMaxBytes)+
+		strings.Repeat("x", questionBodyMaxBytes))
+
+	// If the connection stays open, dial's deadline ends this read with a timeout.
+	text, err := io.ReadAll(conn)
+	if err != nil || len(text) != 0 {
+		t.Errorf(
+			"read %q (%v), want the connection closed without an answer",
+			text,
+			err,
+		)
+	}
+}
+
+func TestAVisitorSendingConnectionCloseGetsTheEndOfAnAnswerWithoutContentLengthAtOnce(
+	t *testing.T,
+) {
+	t.Parallel()
+
+	standIn := startWAF(t, allow)
+	// The flush has the app's server send greeting in chunks instead of adding a
+	// Content-Length.
+	app := startApp(t, func(w http.ResponseWriter, _ *http.Request) {
+		rc := http.NewResponseController(w)
+		_ = rc.EnableFullDuplex()
+
+		_, _ = io.WriteString(w, greeting)
+		_ = rc.Flush()
+	})
+	// The request time limit is far longer than dial's deadline.
+	addr, _ := startRProxy(t, standIn.URL, app.URL, map[string]string{
+		maxRequestTime: "120000",
+	})
+
+	conn := dial(t, addr)
+	send(t, conn, post+"Connection: close\r\n"+
+		announcedLength(2*questionBodyMaxBytes)+
+		strings.Repeat("x", questionBodyMaxBytes+1))
+
+	wantGreeting(t, readAnswer(t, conn))
+}
+
 func TestABodyNotAnnouncedOverItsLimitIsRefused(t *testing.T) {
 	t.Parallel()
 
@@ -148,7 +283,7 @@ func TestABodyNotAnnouncedOverItsLimitIsRefused(t *testing.T) {
 		mayReachApp bool
 	}{
 		{"within the first part", "1K", 2 << 10, false},
-		{"past the first part", "128K", 129 << 10, true},
+		{"past the first part", limitPastFirstPart, 129 << 10, true},
 	} {
 		t.Run(tc.name, func(t *testing.T) {
 			t.Parallel()