From a6ca391e6e62dc7043a331e542c3c07e7e21547f Mon Sep 17 00:00:00 2001
From: Brian Brazil <brian.brazil@robustperception.io>
Date: Fri, 16 Jun 2017 13:09:50 +0100
Subject: [PATCH] Reject scrapes with invalid utf-8 label values.

---
 pkg/textparse/lex.l         |  5 +++++
 pkg/textparse/lex.l.go      |  5 +++++
 pkg/textparse/parse_test.go | 31 +++++++++++++++++++++++++++++++
 retrieval/scrape_test.go    | 23 +++++++++++++++++++++++
 4 files changed, 64 insertions(+)

diff --git a/pkg/textparse/lex.l b/pkg/textparse/lex.l
index f0342d4d9..7f6aaf99d 100644
--- a/pkg/textparse/lex.l
+++ b/pkg/textparse/lex.l
@@ -18,6 +18,7 @@ import (
         "fmt"
         "math"
         "strconv"
+        "unicode/utf8"
 
         "github.com/prometheus/prometheus/pkg/value"
 )
@@ -82,6 +83,10 @@ M     [a-zA-Z_:]
                                       l.offsets = append(l.offsets, l.i-1)
 
 <lstateLValue>\"(\\.|[^\\"]|\0)*\"    s = lstateLabels
+                                      if !utf8.Valid(l.b[l.offsets[len(l.offsets)-1]+2:l.i-1]) {
+                                        l.err = fmt.Errorf("Invalid UTF-8 label value.")
+                                        return -1
+                                      }
                                       l.offsets = append(l.offsets, l.i-1)
 <lstateLValue>\'(\\.|[^\\']|\0)*\'    s = lstateLabels
                                       l.offsets = append(l.offsets, l.i-1)
diff --git a/pkg/textparse/lex.l.go b/pkg/textparse/lex.l.go
index 4742c20d8..37fec2b1b 100644
--- a/pkg/textparse/lex.l.go
+++ b/pkg/textparse/lex.l.go
@@ -19,6 +19,7 @@ import (
 	"fmt"
 	"math"
 	"strconv"
+	"unicode/utf8"
 
 	"github.com/prometheus/prometheus/pkg/value"
 )
@@ -413,6 +414,10 @@ yyrule9: // {S}({L}|{D})*=
 yyrule10: // \"(\\.|[^\\"]|\0)*\"
 	{
 		s = lstateLabels
+		if !utf8.Valid(l.b[l.offsets[len(l.offsets)-1]+2 : l.i-1]) {
+			l.err = fmt.Errorf("Invalid UTF-8 label value.")
+			return -1
+		}
 		l.offsets = append(l.offsets, l.i-1)
 		goto yystate0
 	}
diff --git a/pkg/textparse/parse_test.go b/pkg/textparse/parse_test.go
index 52284d282..8e338cdb7 100644
--- a/pkg/textparse/parse_test.go
+++ b/pkg/textparse/parse_test.go
@@ -104,6 +104,37 @@ go_goroutines 33  	123123`
 
 }
 
+func TestParseErrors(t *testing.T) {
+	cases := []struct {
+		input string
+		err   string
+	}{
+		{
+			input: "a",
+			err:   "no token found",
+		},
+		{
+			input: "a{\xff=\"foo\"} 1\n",
+			err:   "no token found",
+		},
+		{
+			input: "a{b=\"\xff\"} 1\n",
+			err:   "Invalid UTF-8 label value.",
+		},
+		{
+			input: "a true\n",
+			err:   "strconv.ParseFloat: parsing \"true\": invalid syntax",
+		},
+	}
+
+	for _, c := range cases {
+		p := New([]byte(c.input))
+		for p.Next() {
+		}
+		require.Equal(t, c.err, p.Err().Error())
+	}
+}
+
 const (
 	testdataSampleCount = 410
 )
diff --git a/retrieval/scrape_test.go b/retrieval/scrape_test.go
index dc507b931..adcd37fe8 100644
--- a/retrieval/scrape_test.go
+++ b/retrieval/scrape_test.go
@@ -844,6 +844,29 @@ func TestScrapeLoopRunReportsTargetDownOnScrapeError(t *testing.T) {
 	}
 }
 
+func TestScrapeLoopRunReportsTargetDownOnInvalidUTF8(t *testing.T) {
+	var (
+		scraper        = &testScraper{}
+		reportAppender = &collectResultAppender{}
+		reportApp      = func() storage.Appender { return reportAppender }
+	)
+
+	ctx, cancel := context.WithCancel(context.Background())
+	sl := newScrapeLoop(ctx, scraper, func() storage.Appender { return nopAppender{} }, reportApp, nil)
+
+	scraper.scrapeFunc = func(ctx context.Context, w io.Writer) error {
+		cancel()
+		w.Write([]byte("a{l=\"\xff\"} 0\n"))
+		return nil
+	}
+
+	sl.run(10*time.Millisecond, time.Hour, nil)
+
+	if reportAppender.result[0].v != 0 {
+		t.Fatalf("bad 'up' value; want 0, got %v", reportAppender.result[0].v)
+	}
+}
+
 type errorAppender struct {
 	collectResultAppender
 }