fix: drop negative samples (#3955)

Author: kolesnikovae

pkg/pprof/pprof.go

@@ -367,16 +367,31 @@ func (p *Profile) Normalize() {
 		})
 	}
 
+	// Remove samples.
+	var removedSamples []*profilev1.Sample
+	p.Sample = slices.RemoveInPlace(p.Sample, func(s *profilev1.Sample, i int) bool {
+		for j := 0; j < len(s.Value); j++ {
+			if s.Value[j] < 0 {
+				removedSamples = append(removedSamples, s)
+				return true
+			}
+		}
+		for j := 0; j < len(s.Value); j++ {
+			if s.Value[j] > 0 {
+				return false
+			}
+		}
+		removedSamples = append(removedSamples, s)
+		return true
+	})
+
 	// first we sort the samples.
 	hashes := p.hasher.Hashes(p.Sample)
 	ss := &sortedSample{samples: p.Sample, hashes: hashes}
 	sort.Sort(ss)
 	p.Sample = ss.samples
 	hashes = ss.hashes
 
-	// Remove samples.
-	var removedSamples []*profilev1.Sample
-
 	p.Sample = slices.RemoveInPlace(p.Sample, func(s *profilev1.Sample, i int) bool {
 		// if the next sample has the same hash and labels, we can remove this sample but add the value to the next sample.
 		if i < len(p.Sample)-1 && hashes[i] == hashes[i+1] {
@@ -387,15 +402,7 @@ func (p *Profile) Normalize() {
 			removedSamples = append(removedSamples, s)
 			return true
 		}
-		for j := 0; j < len(s.Value); j++ {
-			if s.Value[j] > 0 {
-				// we found a non-zero value, so we can keep this sample.
-				return false
-			}
-		}
-		// all values are 0, remove the sample.
-		removedSamples = append(removedSamples, s)
-		return true
+		return false
 	})
 
 	// Remove references to removed samples.

pkg/pprof/pprof_test.go

@@ -105,6 +105,74 @@ func TestNormalizeProfile(t *testing.T) {
 	}, pf.Profile)
 }
 
+func TestNormalizeProfile_NegativeSample(t *testing.T) {
+	currentTime = func() time.Time {
+		t, _ := time.Parse(time.RFC3339, "2020-01-01T00:00:00Z")
+		return t
+	}
+	defer func() {
+		currentTime = time.Now
+	}()
+
+	p := &profilev1.Profile{
+		SampleType: []*profilev1.ValueType{
+			{Type: 1, Unit: 2},
+		},
+		Sample: []*profilev1.Sample{
+			{LocationId: []uint64{2, 1}, Value: []int64{10}, Label: []*profilev1.Label{{Str: 5, Key: 5}, {Str: 5, Key: 6}}},
+			{LocationId: []uint64{2, 1}, Value: []int64{10}, Label: []*profilev1.Label{{Str: 5, Key: 5}, {Str: 5, Key: 7}}},
+			{LocationId: []uint64{2, 1}, Value: []int64{10}, Label: []*profilev1.Label{{Str: 5, Key: 5}, {Str: 5, Key: 7}}},
+			{LocationId: []uint64{2, 1}, Value: []int64{-10}, Label: []*profilev1.Label{{Str: 5, Key: 5}, {Str: 5, Key: 7}}},
+			{LocationId: []uint64{2, 1}, Value: []int64{0}, Label: []*profilev1.Label{{Str: 5, Key: 5}, {Str: 5, Key: 7}}},
+		},
+		Mapping: []*profilev1.Mapping{{Id: 1, HasFunctions: true, MemoryStart: 100, MemoryLimit: 200, FileOffset: 200}},
+		Location: []*profilev1.Location{
+			{Id: 1, MappingId: 1, Address: 5, Line: []*profilev1.Line{{FunctionId: 1, Line: 1}}},
+			{Id: 2, MappingId: 1, Address: 2, Line: []*profilev1.Line{{FunctionId: 2, Line: 1}}},
+		},
+		Function: []*profilev1.Function{
+			{Id: 1, Name: 3, SystemName: 3, Filename: 4, StartLine: 1},
+			{Id: 2, Name: 5, SystemName: 5, Filename: 4, StartLine: 1},
+		},
+		StringTable: []string{
+			"",
+			"cpu", "nanoseconds",
+			"main", "main.go",
+			"foo", "bar", "baz",
+		},
+		PeriodType: &profilev1.ValueType{Type: 1, Unit: 2},
+	}
+
+	pf := &Profile{Profile: p}
+	pf.Normalize()
+	require.Equal(t, pf.Profile, &profilev1.Profile{
+		SampleType: []*profilev1.ValueType{
+			{Type: 1, Unit: 2},
+		},
+		Sample: []*profilev1.Sample{
+			{LocationId: []uint64{2, 1}, Value: []int64{10}, Label: []*profilev1.Label{{Str: 5, Key: 5}, {Str: 5, Key: 6}}},
+			{LocationId: []uint64{2, 1}, Value: []int64{20}, Label: []*profilev1.Label{{Str: 5, Key: 5}, {Str: 5, Key: 7}}},
+		},
+		Mapping: []*profilev1.Mapping{{Id: 1, HasFunctions: true}},
+		Location: []*profilev1.Location{
+			{Id: 1, MappingId: 1, Line: []*profilev1.Line{{FunctionId: 1, Line: 1}}},
+			{Id: 2, MappingId: 1, Line: []*profilev1.Line{{FunctionId: 2, Line: 1}}},
+		},
+		Function: []*profilev1.Function{
+			{Id: 1, Name: 3, SystemName: 3, Filename: 4, StartLine: 1},
+			{Id: 2, Name: 5, SystemName: 5, Filename: 4, StartLine: 1},
+		},
+		StringTable: []string{
+			"",
+			"cpu", "nanoseconds",
+			"main", "main.go",
+			"foo", "bar", "baz",
+		},
+		PeriodType: &profilev1.ValueType{Type: 1, Unit: 2},
+		TimeNanos:  1577836800000000000,
+	})
+}
+
 func TestNormalizeProfile_SampleLabels(t *testing.T) {
 	currentTime = func() time.Time {
 		t, _ := time.Parse(time.RFC3339, "2020-01-01T00:00:00Z")