syz-cluster: remember LKML series tags

Sometimes the patch series directly hint at the kernel tree they should
be applied to. Extract and remember this information.

7 files changed, 27 insertions, 11 deletions

diff --git a/syz-cluster/pkg/api/api.go b/syz-cluster/pkg/api/api.go
index 128be02ba..5483513af 100644
--- a/syz-cluster/pkg/api/api.go
+++ b/syz-cluster/pkg/api/api.go
@@ -103,6 +103,7 @@ type Series struct {
 	Cc          []string      `json:"cc"`
 	Version     int           `json:"version"`
 	Link        string        `json:"link"`
+	SubjectTags []string      `json:"subject_tags"`
 	PublishedAt time.Time     `json:"published_at"`
 	Patches     []SeriesPatch `json:"patches"`
 }
diff --git a/syz-cluster/pkg/controller/api_test.go b/syz-cluster/pkg/controller/api_test.go
index 9e709e089..1c1c8de9f 100644
--- a/syz-cluster/pkg/controller/api_test.go
+++ b/syz-cluster/pkg/controller/api_test.go
@@ -5,11 +5,12 @@ package controller
 
 import (
 	"bytes"
+	"testing"
+	"time"
+
 	"github.com/google/syzkaller/syz-cluster/pkg/api"
 	"github.com/google/syzkaller/syz-cluster/pkg/app"
 	"github.com/stretchr/testify/assert"
-	"testing"
-	"time"
 )
 
 func TestAPIGetSeries(t *testing.T) {
@@ -108,6 +109,7 @@ var testSeries = &api.Series{
 	Version:     2,
 	PublishedAt: time.Date(2020, time.January, 1, 3, 0, 0, 0, time.UTC),
 	Cc:          []string{"email"},
+	SubjectTags: []string{"tag"},
 	Patches: []api.SeriesPatch{
 		{
 			Seq:  1,
diff --git a/syz-cluster/pkg/db/entities.go b/syz-cluster/pkg/db/entities.go
index 77c066b5e..b1ffc6949 100644
--- a/syz-cluster/pkg/db/entities.go
+++ b/syz-cluster/pkg/db/entities.go
@@ -4,18 +4,21 @@
 package db
 
 import (
-	"cloud.google.com/go/spanner"
 	"time"
+
+	"cloud.google.com/go/spanner"
 )
 
 type Series struct {
-	ID          string    `spanner:"ID"`
-	ExtID       string    `spanner:"ExtID"`
-	AuthorName  string    `spanner:"AuthorName"`
-	AuthorEmail string    `spanner:"AuthorEmail"`
-	Title       string    `spanner:"Title"`
-	Link        string    `spanner:"Link"`
-	Version     int64     `spanner:"Version"`
+	ID          string `spanner:"ID"`
+	ExtID       string `spanner:"ExtID"`
+	AuthorName  string `spanner:"AuthorName"`
+	AuthorEmail string `spanner:"AuthorEmail"`
+	Title       string `spanner:"Title"`
+	Link        string `spanner:"Link"`
+	Version     int64  `spanner:"Version"`
+	// In LKML patches, there are often hints at the target tree for the patch.
+	SubjectTags []string  `spanner:"SubjectTags"`
 	PublishedAt time.Time `spanner:"PublishedAt"`
 	// TODO: we could ger rid of the field by using slightly more complicated SQL queries.
 	LatestSessionID spanner.NullString `spanner:"LatestSessionID"`
diff --git a/syz-cluster/pkg/db/migrations/4_add_series_subject_tags.down.sql b/syz-cluster/pkg/db/migrations/4_add_series_subject_tags.down.sql
new file mode 100644
index 000000000..d63028aa2
--- /dev/null
+++ b/syz-cluster/pkg/db/migrations/4_add_series_subject_tags.down.sql
@@ -0,0 +1 @@
+ALTER TABLE Series DROP COLUMN SubjectTags;
diff --git a/syz-cluster/pkg/db/migrations/4_add_series_subject_tags.up.sql b/syz-cluster/pkg/db/migrations/4_add_series_subject_tags.up.sql
new file mode 100644
index 000000000..a049fdde6
--- /dev/null
+++ b/syz-cluster/pkg/db/migrations/4_add_series_subject_tags.up.sql
@@ -0,0 +1 @@
+ALTER TABLE Series ADD COLUMN SubjectTags ARRAY<STRING(512)>;
diff --git a/syz-cluster/pkg/service/series.go b/syz-cluster/pkg/service/series.go
index 6583306ce..d1ace9b09 100644
--- a/syz-cluster/pkg/service/series.go
+++ b/syz-cluster/pkg/service/series.go
@@ -63,6 +63,13 @@ func (s *SeriesService) UploadSeries(ctx context.Context, series *api.Series) (*
 		PublishedAt: series.PublishedAt,
 		Cc:          series.Cc,
 	}
+	for _, tag := range series.SubjectTags {
+		const tageSizeLimit = 511
+		if len(tag) > tageSizeLimit {
+			tag = tag[:tageSizeLimit]
+		}
+		seriesObj.SubjectTags = append(seriesObj.SubjectTags, tag)
+	}
 	err := s.seriesRepo.Insert(ctx, seriesObj, func() ([]*db.Patch, error) {
 		var ret []*db.Patch
 		for _, patch := range series.Patches {
@@ -121,6 +128,7 @@ func (s *SeriesService) getSeries(ctx context.Context,
 		Cc:          series.Cc,
 		PublishedAt: series.PublishedAt,
 		Link:        series.Link,
+		SubjectTags: series.SubjectTags,
 	}
 	for _, patch := range patches {
 		var body []byte
diff --git a/syz-cluster/series-tracker/main.go b/syz-cluster/series-tracker/main.go
index b8638d897..d6f28a497 100644
--- a/syz-cluster/series-tracker/main.go
+++ b/syz-cluster/series-tracker/main.go
@@ -155,9 +155,9 @@ func (sf *SeriesFetcher) handleSeries(ctx context.Context, series *lore.Series,
 	apiSeries := &api.Series{
 		ExtID:       series.MessageID,
 		AuthorEmail: first.Author,
-		// TODO: set Cc.
 		Title:       series.Subject,
 		Version:     series.Version,
+		SubjectTags: series.Tags,
 		Link:        "https://lore.kernel.org/all/" + series.MessageID,
 		PublishedAt: date,
 	}