From b39df73c1629cb4b2846cbb14a6a37e33d0060a1 Mon Sep 17 00:00:00 2001
From: Aleksandr Nogikh <nogikh@google.com>
Date: Tue, 31 Jan 2023 16:22:37 +0100
Subject: pkg/subsystem: remove unneeded paths from matching rules

We don't really care about Documentation/ and similar folders. Exclude
such path matching rules after parsing MAINTAINERS.
---
 pkg/subsystem/linux/maintainers.go | 16 ++++++++++++++++
 pkg/subsystem/linux/subsystems.go  |  8 ++++++++
 2 files changed, 24 insertions(+)

(limited to 'pkg/subsystem/linux')

diff --git a/pkg/subsystem/linux/maintainers.go b/pkg/subsystem/linux/maintainers.go
index 6a188b261..7c60a840a 100644
--- a/pkg/subsystem/linux/maintainers.go
+++ b/pkg/subsystem/linux/maintainers.go
@@ -173,6 +173,22 @@ func (r maintainersRecord) ToPathRule() entity.PathRule {
 	}
 }
 
+func removeMatchingPatterns(records []*maintainersRecord, rule *regexp.Regexp) {
+	filter := func(list []string) []string {
+		ret := []string{}
+		for _, item := range list {
+			if !rule.MatchString(item) {
+				ret = append(ret, item)
+			}
+		}
+		return ret
+	}
+	for _, record := range records {
+		record.includePatterns = filter(record.includePatterns)
+		record.excludePatterns = filter(record.excludePatterns)
+	}
+}
+
 var (
 	escapedSeparator = regexp.QuoteMeta(fmt.Sprintf("%c", filepath.Separator))
 	wildcardReplace  = map[byte]string{
diff --git a/pkg/subsystem/linux/subsystems.go b/pkg/subsystem/linux/subsystems.go
index abda295b4..43fea3b9e 100644
--- a/pkg/subsystem/linux/subsystems.go
+++ b/pkg/subsystem/linux/subsystems.go
@@ -7,6 +7,7 @@ import (
 	"fmt"
 	"io/fs"
 	"os"
+	"regexp"
 	"sort"
 
 	"github.com/google/syzkaller/pkg/subsystem/entity"
@@ -23,6 +24,7 @@ func listFromRepoInner(root fs.FS, rules *customRules) ([]*entity.Subsystem, err
 	if err != nil {
 		return nil, err
 	}
+	removeMatchingPatterns(records, dropPatterns)
 	ctx := &linuxCtx{
 		root:       root,
 		rawRecords: records,
@@ -59,6 +61,12 @@ type subsystemCandidate struct {
 	commonEmail string
 }
 
+var (
+	// Some of the patterns are not really needed for bug subsystem inteference and
+	// only complicate the manual review of the rules.
+	dropPatterns = regexp.MustCompile(`^(Documentation|scripts|samples|tools)|Makefile`)
+)
+
 func (ctx *linuxCtx) groupByList() []*subsystemCandidate {
 	perList := make(map[string][]*maintainersRecord)
 	for _, record := range ctx.rawRecords {
-- 
cgit mrf-deployment