pkg/subsystem/linux: extract names for subsystems

Extract the short subsystem name from the mailing list email.

Stip the common prefixes and suffixes and make sure there are no
duplicates.

As a fallback, assign the whole list email address as a subsystem name.

4 files changed, 226 insertions, 0 deletions

diff --git a/pkg/subsystem/linux/names.go b/pkg/subsystem/linux/names.go
new file mode 100644
index 000000000..0f422ea05
--- /dev/null
+++ b/pkg/subsystem/linux/names.go
@@ -0,0 +1,103 @@
+// Copyright 2023 syzkaller project authors. All rights reserved.
+// Use of this source code is governed by Apache 2 LICENSE that can be found in the LICENSE file.
+
+package linux
+
+import (
+	"fmt"
+	"regexp"
+
+	"github.com/google/syzkaller/pkg/subsystem/entity"
+)
+
+// setSubsystemNames assigns unique names to the presented subsystems.
+// If it failed to assign a name to a subsystem, the Name field remains empty.
+func setSubsystemNames(list []*entity.Subsystem) error {
+	dups := map[string]bool{}
+	for _, item := range list {
+		// For now, we can only infer name from the list email.
+		if len(item.Lists) == 0 {
+			return fmt.Errorf("no lists for %#v", item)
+		}
+		email := item.Lists[0]
+		name := emailToName(email)
+		if !validateName(name) {
+			return fmt.Errorf("failed to extract a name from %s", email)
+		}
+		if dups[name] {
+			return fmt.Errorf("duplicate subsystem name: %s", item.Name)
+		}
+		item.Name = name
+		dups[name] = true
+	}
+	return nil
+}
+
+func validateName(name string) bool {
+	const (
+		minLen = 2
+		maxLen = 16 // otherwise the email subject can get too big
+	)
+	return len(name) >= minLen && len(name) <= maxLen
+}
+
+func emailToName(email string) string {
+	if name := emailExceptions[email]; name != "" {
+		return name
+	}
+	ret := emailStripRe.FindStringSubmatch(email)
+	if ret == nil {
+		return ""
+	}
+	return ret[1]
+}
+
+func buildEmailStripRe() *regexp.Regexp {
+	raw := `^(?:`
+	for i := 0; i < len(stripPrefixes); i++ {
+		if i > 0 {
+			raw += "|"
+		}
+		raw += regexp.QuoteMeta(stripPrefixes[i])
+	}
+	raw += ")*(.*?)(?:"
+	for i := 0; i < len(stripSuffixes); i++ {
+		if i > 0 {
+			raw += "|"
+		}
+		raw += regexp.QuoteMeta(stripSuffixes[i])
+	}
+	raw += ")*@.*$"
+	return regexp.MustCompile(raw)
+}
+
+var (
+	emailExceptions = map[string]string{
+		"patches@opensource.cirrus.com":             "cirrus",
+		"virtualization@lists.linux-foundation.org": "virt", // the name is too long
+		"dev@openvswitch.org":                       "openvswitch",
+		"devel@acpica.org":                          "acpica",
+		"kernel@dh-electronics.com":                 "dh-electr",
+		"devel@lists.orangefs.org":                  "orangefs",
+		"linux-arm-kernel@axis.com":                 "axis",
+		"Dell.Client.Kernel@dell.com":               "dell",
+		"sound-open-firmware@alsa-project.org":      "sof",
+		"platform-driver-x86@vger.kernel.org":       "x86-drivers",
+		"linux-trace-devel@vger.kernel.org":         "rt-tools",
+		"aws-nitro-enclaves-devel@amazon.com":       "nitro",
+		"brcm80211-dev-list.pdl@broadcom.com":       "brcm80211",
+		"osmocom-net-gprs@lists.osmocom.org":        "osmocom",
+		"netdev@vger.kernel.org":                    "net",
+		"megaraidlinux.pdl@broadcom.com":            "megaraid",
+		"mpi3mr-linuxdrv.pdl@broadcom.com":          "mpi3",
+		"MPT-FusionLinux.pdl@broadcom.com":          "mpt-fusion",
+	}
+	stripPrefixes = []string{"linux-"}
+	stripSuffixes = []string{
+		"-devel", "-dev", "-devs", "-developer", "devel",
+		"-user", "-users",
+		"-discussion", "-discuss", "-list", "-en",
+		"-kernel", "-linux", "-general",
+	}
+	emailStripRe = buildEmailStripRe()
+)
diff --git a/pkg/subsystem/linux/names_test.go b/pkg/subsystem/linux/names_test.go
new file mode 100644
index 000000000..0b2352411
--- /dev/null
+++ b/pkg/subsystem/linux/names_test.go
@@ -0,0 +1,115 @@
+// Copyright 2023 syzkaller project authors. All rights reserved.
+// Use of this source code is governed by Apache 2 LICENSE that can be found in the LICENSE file.
+
+package linux
+
+import (
+	"testing"
+
+	"github.com/google/syzkaller/pkg/subsystem/entity"
+)
+
+func TestEmailToName(t *testing.T) {
+	tests := map[string]string{
+		// These are following the general rules.
+		"linux-nilfs@vger.kernel.org":           "nilfs",
+		"tomoyo-dev-en@lists.osdn.me":           "tomoyo",
+		"tipc-discussion@lists.sourceforge.net": "tipc",
+		"v9fs-developer@lists.sourceforge.net":  "v9fs",
+		"zd1211-devs@lists.sourceforge.net":     "zd1211",
+		// Test that we can handle exceptions.
+		"virtualization@lists.linux-foundation.org": "virt",
+	}
+	for email, name := range tests {
+		result := emailToName(email)
+		if result != name {
+			t.Fatalf("%#v: expected %#v, got %#v", email, name, result)
+		}
+	}
+}
+
+type subsystemTestInput struct {
+	email   string
+	outName string
+}
+
+func (sti subsystemTestInput) ToSubsystem() *entity.Subsystem {
+	s := &entity.Subsystem{}
+	if sti.email != "" {
+		s.Lists = append(s.Lists, sti.email)
+	}
+	return s
+}
+
+func TestSetSubsystemNames(t *testing.T) {
+	tests := []struct {
+		name     string
+		inputs   []subsystemTestInput
+		mustFail bool
+	}{
+		{
+			name: "plan test",
+			inputs: []subsystemTestInput{
+				{
+					email:   "linux-ntfs-dev@lists.sourceforge.net",
+					outName: "ntfs",
+				},
+				{
+					email:   "llvm@lists.linux.dev",
+					outName: "llvm",
+				},
+			},
+		},
+		{
+			name: "has dup name",
+			inputs: []subsystemTestInput{
+				{
+					email:   "linux-ntfs-dev@lists.sourceforge.net",
+					outName: "ntfs",
+				},
+				{
+					email:   "ntfs@lists.sourceforge.net",
+					outName: "ntfs",
+				},
+			},
+			mustFail: true,
+		},
+		{
+			name: "has empty list",
+			inputs: []subsystemTestInput{
+				{
+					email:   "linux-ntfs-dev@lists.sourceforge.net",
+					outName: "ntfs",
+				},
+				{
+					email:   "",
+					outName: "",
+				},
+			},
+			mustFail: true,
+		},
+	}
+	for _, test := range tests {
+		curr := test
+		t.Run(curr.name, func(t *testing.T) {
+			list := []*entity.Subsystem{}
+			for _, i := range curr.inputs {
+				list = append(list, i.ToSubsystem())
+			}
+			err := setSubsystemNames(list)
+			if curr.mustFail != (err != nil) {
+				t.Fatalf("expected failure: %v, got: %v", curr.mustFail, err)
+			}
+			if curr.mustFail {
+				return
+			}
+			for i, item := range list {
+				if item.Name != curr.inputs[i].outName {
+					t.Fatalf("invalid name for #%d: expected %#v, got %#v",
+						i+1, curr.inputs[i].outName, item.Name,
+					)
+				}
+			}
+		})
+	}
+}
diff --git a/pkg/subsystem/linux/subsystems.go b/pkg/subsystem/linux/subsystems.go
index b5ffa924c..468ba5fbc 100644
--- a/pkg/subsystem/linux/subsystems.go
+++ b/pkg/subsystem/linux/subsystems.go
@@ -82,6 +82,9 @@ func (ctx *linuxCtx) getSubsystems() ([]*entity.Subsystem, error) {
 		mergeRawRecords(s, raw.records)
 		ret = append(ret, s)
 	}
+	if err := setSubsystemNames(ret); err != nil {
+		return nil, fmt.Errorf("failed to set names: %w", err)
+	}
 	return ret, nil
 }
 
@@ -95,6 +98,7 @@ func mergeRawRecords(subsystem *entity.Subsystem, records []*maintainersRecord)
 		for s := range m {
 			ret = append(ret, s)
 		}
+		sort.Strings(ret)
 		return ret
 	}
 	var lists, maintainers []string
diff --git a/pkg/subsystem/linux/subsystems_test.go b/pkg/subsystem/linux/subsystems_test.go
index f5699f26e..caf675244 100644
--- a/pkg/subsystem/linux/subsystems_test.go
+++ b/pkg/subsystem/linux/subsystems_test.go
@@ -25,17 +25,21 @@ func TestGroupLinuxSubsystems(t *testing.T) {
 	}
 	expected := []*entity.Subsystem{
 		{
+			Name:        "fs",
 			Lists:       []string{"linux-fsdevel@vger.kernel.org"},
 			Maintainers: []string{"email_vfs@email.com"},
 		},
 		{
+			Name:        "ext4",
 			Lists:       []string{"linux-ext4@vger.kernel.org"},
 			Maintainers: []string{"email_ext4@email.com", "email_ext4_2@email.com"},
 		},
 		{
+			Name:  "mm",
 			Lists: []string{"linux-mm@kvack.org"},
 		},
 		{
+			Name:        "kernel",
 			Lists:       []string{"linux-kernel@vger.kernel.org"},
 			Maintainers: []string{"email_rest@email.com"},
 		},