pkg: get and store Maintainers data

Create a struct on pkg/vcs to store data of syzkaller email recipients
and update its users. The struct contains default name, email, and a
label to divide user into To and Cc when sending the emails.

pkg/bisect/bisect.go

@@ -180,7 +180,8 @@ func runImpl(cfg *Config, repo vcs.Repo, inst instance.Env) (*Result, error) {
 	}
 	com := res.Commits[0]
 	env.log("first %v commit: %v %v", what, com.Hash, com.Title)
-	env.log("cc: %q", com.CC)
+	env.log("recipients (to): %q", com.Recipients.GetEmails(vcs.To))
+	env.log("recipients (cc): %q", com.Recipients.GetEmails(vcs.Cc))
 	if res.Report != nil {
 		env.log("crash: %v\n%s", res.Report.Title, res.Report.Report)
 	}

pkg/build/build.go

@@ -15,6 +15,7 @@ import (
 
 	"github.com/google/syzkaller/pkg/osutil"
 	"github.com/google/syzkaller/pkg/report"
+	"github.com/google/syzkaller/pkg/vcs"
 )
 
 // Params is input arguments for the Image function.
@@ -90,10 +91,10 @@ func Clean(targetOS, targetArch, vmType, kernelDir string) error {
 }
 
 type KernelError struct {
-	Report      []byte
-	Output      []byte
-	Maintainers []string
-	guiltyFile  string
+	Report     []byte
+	Output     []byte
+	Recipients vcs.Recipients
+	guiltyFile string
 }
 
 func (err *KernelError) Error() string {
@@ -195,7 +196,7 @@ func extractRootCause(err error, OS, kernelSrc string) error {
 		if err != nil {
 			kernelErr.Output = append(kernelErr.Output, err.Error()...)
 		}
-		kernelErr.Maintainers = maintainers
+		kernelErr.Recipients = maintainers
 	}
 	return kernelErr
 }

pkg/report/linux.go

@@ -7,7 +7,6 @@ import (
 	"bufio"
 	"bytes"
 	"fmt"
-	"net/mail"
 	"path/filepath"
 	"regexp"
 	"strconv"
@@ -16,6 +15,7 @@ import (
 
 	"github.com/google/syzkaller/pkg/osutil"
 	"github.com/google/syzkaller/pkg/symbolizer"
+	"github.com/google/syzkaller/pkg/vcs"
 )
 
 type linux struct {
@@ -351,7 +351,7 @@ func (ctx *linux) Symbolize(rep *Report) error {
 		if err != nil {
 			return err
 		}
-		rep.Maintainers = maintainers
+		rep.Recipients = maintainers
 	}
 	return nil
 }
@@ -466,14 +466,14 @@ nextFile:
 	return ""
 }
 
-func (ctx *linux) getMaintainers(file string) ([]string, error) {
+func (ctx *linux) getMaintainers(file string) (vcs.Recipients, error) {
 	if ctx.kernelSrc == "" {
 		return nil, nil
 	}
 	return GetLinuxMaintainers(ctx.kernelSrc, file)
 }
 
-func GetLinuxMaintainers(kernelSrc, file string) ([]string, error) {
+func GetLinuxMaintainers(kernelSrc, file string) (vcs.Recipients, error) {
 	mtrs, err := getMaintainersImpl(kernelSrc, file, false)
 	if err != nil {
 		return nil, err
@@ -487,9 +487,9 @@ func GetLinuxMaintainers(kernelSrc, file string) ([]string, error) {
 	return mtrs, nil
 }
 
-func getMaintainersImpl(kernelSrc, file string, blame bool) ([]string, error) {
+func getMaintainersImpl(kernelSrc, file string, blame bool) (vcs.Recipients, error) {
 	// See #1441 re --git-min-percent.
-	args := []string{"--no-n", "--no-rolestats", "--git-min-percent=15"}
+	args := []string{"--git-min-percent=15"}
 	if blame {
 		args = append(args, "--git-blame")
 	}
@@ -499,16 +499,7 @@ func getMaintainersImpl(kernelSrc, file string, blame bool) ([]string, error) {
 	if err != nil {
 		return nil, err
 	}
-	lines := strings.Split(string(output), "\n")
-	var mtrs []string
-	for _, line := range lines {
-		addr, err := mail.ParseAddress(line)
-		if err != nil {
-			continue
-		}
-		mtrs = append(mtrs, addr.Address)
-	}
-	return mtrs, nil
+	return vcs.ParseMaintainersLinux(output), nil
 }
 
 func (ctx *linux) extractFiles(report []byte) []string {

pkg/report/report.go

@@ -13,6 +13,7 @@ import (
 	"strings"
 
 	"github.com/google/syzkaller/pkg/mgrconfig"
+	"github.com/google/syzkaller/pkg/vcs"
 	"github.com/google/syzkaller/sys/targets"
 )
 
@@ -50,8 +51,8 @@ type Report struct {
 	Corrupted bool
 	// CorruptedReason contains reason why the report is marked as corrupted.
 	CorruptedReason string
-	// Maintainers is list of maintainer emails (filled in by Symbolize).
-	Maintainers []string
+	// Recipients is a list of RecipientInfo with Email, Display Name, and type.
+	Recipients vcs.Recipients
 	// guiltyFile is the source file that we think is to blame for the crash  (filled in by Symbolize).
 	guiltyFile string
 	// reportPrefixLen is length of additional prefix lines that we added before actual crash report.

pkg/vcs/git.go

@@ -196,8 +196,8 @@ func gitParseCommit(output, user, domain []byte, ignoreCC map[string]bool) (*Com
 	if err != nil {
 		return nil, fmt.Errorf("failed to parse date in git log output: %v\n%q", err, output)
 	}
-	cc := make(map[string]bool)
-	cc[strings.ToLower(string(lines[2]))] = true
+	recipients := make(map[string]bool)
+	recipients[strings.ToLower(string(lines[2]))] = true
 	var tags []string
 	// Use summary line + all description lines.
 	for _, line := range append([][]byte{lines[1]}, lines[6:]...) {
@@ -235,15 +235,15 @@ func gitParseCommit(output, user, domain []byte, ignoreCC map[string]bool) (*Com
 			if ignoreCC[email] {
 				continue
 			}
-			cc[email] = true
+			recipients[email] = true
 			break
 		}
 	}
-	sortedCC := make([]string, 0, len(cc))
-	for addr := range cc {
-		sortedCC = append(sortedCC, addr)
+	sortedRecipients := make(Recipients, 0, len(recipients))
+	for addr := range recipients {
+		sortedRecipients = append(sortedRecipients, RecipientInfo{mail.Address{Address: addr}, To})
 	}
-	sort.Strings(sortedCC)
+	sort.Sort(sortedRecipients)
 	parents := strings.Split(string(lines[5]), " ")
 	com := &Commit{
 		Hash:       string(lines[0]),
@@ -251,7 +251,7 @@ func gitParseCommit(output, user, domain []byte, ignoreCC map[string]bool) (*Com
 		Author:     string(lines[2]),
 		AuthorName: string(lines[3]),
 		Parents:    parents,
-		CC:         sortedCC,
+		Recipients: sortedRecipients,
 		Tags:       tags,
 		Date:       date,
 	}

pkg/vcs/git_repo_test.go

@@ -167,8 +167,8 @@ func checkCommit(t *testing.T, idx int, test testCommit, com *Commit, checkTags
 	if userName != com.AuthorName {
 		t.Errorf("#%v: want author name %q, got %q", idx, userName, com.Author)
 	}
-	if diff := cmp.Diff(test.cc, com.CC); diff != "" {
-		t.Logf("%#v", com.CC)
+	if diff := cmp.Diff(test.cc, com.Recipients.GetEmails(To)); diff != "" {
+		t.Logf("%#v", com.Recipients)
 		t.Error(diff)
 	}
 	if diff := cmp.Diff(test.tags, com.Tags); checkTags && diff != "" {

pkg/vcs/git_test.go

@@ -37,7 +37,7 @@ Signed-off-by: Linux Master <linux@linux-foundation.org>
 			Title:      "rbtree: include rcu.h",
 			Author:     "foobar@foobar.de",
 			AuthorName: "Foo Bar",
-			CC: []string{
+			Recipients: NewRecipients([]string{
 				"and@me.com",
 				"another@email.de",
 				"foobar@foobar.de",
@@ -46,7 +46,7 @@ Signed-off-by: Linux Master <linux@linux-foundation.org>
 				"name@name.com",
 				"subsystem@reviewer.com",
 				"yetanother@email.org",
-			},
+			}, To),
 			Date: time.Date(2018, 5, 11, 16, 02, 14, 0, time.FixedZone("", -7*60*60)),
 		},
 	}
@@ -70,7 +70,7 @@ Signed-off-by: Linux Master <linux@linux-foundation.org>
 		if com.Author != res.Author {
 			t.Fatalf("want author %q, got %q", com.Author, res.Author)
 		}
-		if diff := cmp.Diff(com.CC, res.CC); diff != "" {
+		if diff := cmp.Diff(com.Recipients, res.Recipients); diff != "" {
 			t.Fatalf("bad CC: %v", diff)
 		}
 		if !com.Date.Equal(res.Date) {

pkg/vcs/linux.go

@@ -12,12 +12,12 @@ import (
 	"net/mail"
 	"os"
 	"path/filepath"
+	"regexp"
 	"sort"
 	"strconv"
 	"strings"
 	"time"
 
-	"github.com/google/syzkaller/pkg/email"
 	"github.com/google/syzkaller/pkg/osutil"
 	"github.com/google/syzkaller/prog"
 )
@@ -214,21 +214,22 @@ func (ctx *linux) Bisect(bad, good string, trace io.Writer, pred func() (BisectR
 }
 
 func (ctx *linux) addMaintainers(com *Commit) {
-	if len(com.CC) > 2 {
+	if len(com.Recipients) > 2 {
 		return
 	}
-	list := ctx.getMaintainers(com.Hash, false)
-	if len(list) < 3 {
-		list = ctx.getMaintainers(com.Hash, true)
+	mtrs := ctx.getMaintainers(com.Hash, false)
+	if len(mtrs) < 3 {
+		mtrs = ctx.getMaintainers(com.Hash, true)
 	}
-	com.CC = email.MergeEmailLists(com.CC, list)
+	com.Recipients = append(com.Recipients, mtrs...)
+	sort.Sort(com.Recipients)
 }
 
-func (ctx *linux) getMaintainers(hash string, blame bool) []string {
+func (ctx *linux) getMaintainers(hash string, blame bool) Recipients {
 	// See #1441 re --git-min-percent.
 	args := "git show " + hash + " | " +
 		filepath.FromSlash("scripts/get_maintainer.pl") +
-		" --no-n --no-rolestats --git-min-percent=20"
+		" --git-min-percent=20"
 	if blame {
 		args += " --git-blame"
 	}
@@ -236,15 +237,42 @@ func (ctx *linux) getMaintainers(hash string, blame bool) []string {
 	if err != nil {
 		return nil
 	}
-	var list []string
-	for _, line := range strings.Split(string(output), "\n") {
-		addr, err := mail.ParseAddress(line)
+	return ParseMaintainersLinux(output)
+}
+
+func ParseMaintainersLinux(text []byte) Recipients {
+	lines := strings.Split(string(text), "\n")
+	reRole := regexp.MustCompile(` \([^)]+\)$`)
+	var mtrs Recipients
+	// LMKL is To by default, but it changes to Cc if there's also a subsystem list.
+	lkmlType := To
+	foundLkml := false
+	for _, line := range lines {
+		role := reRole.FindString(line)
+		address := strings.Replace(line, role, "", 1)
+		addr, err := mail.ParseAddress(address)
 		if err != nil {
 			continue
 		}
-		list = append(list, strings.ToLower(addr.Address))
+		var roleType RecipientType
+		if addr.Address == "linux-kernel@vger.kernel.org" {
+			foundLkml = true
+			continue
+		} else if strings.Contains(role, "list") {
+			lkmlType = Cc
+			roleType = To
+		} else if strings.Contains(role, "maintainer") || strings.Contains(role, "supporter") {
+			roleType = To
+		} else {
+			roleType = Cc // Reviewer or other role; default to Cc.
+		}
+		mtrs = append(mtrs, RecipientInfo{*addr, roleType})
 	}
-	return list
+	if foundLkml {
+		mtrs = append(mtrs, RecipientInfo{mail.Address{Address: "linux-kernel@vger.kernel.org"}, lkmlType})
+	}
+	sort.Sort(mtrs)
+	return mtrs
 }
 
 const configBisectTag = "# Minimized by syzkaller"

pkg/vcs/vcs.go

@@ -8,13 +8,66 @@ import (
 	"bytes"
 	"fmt"
 	"io"
+	"net/mail"
 	"regexp"
+	"sort"
 	"strings"
 	"time"
 
+	"github.com/google/syzkaller/dashboard/dashapi"
 	"github.com/google/syzkaller/pkg/osutil"
 )
 
+type RecipientType int
+
+const (
+	To RecipientType = iota
+	Cc
+)
+
+func (t RecipientType) String() string {
+	return [...]string{"To", "Cc"}[t]
+}
+
+type RecipientInfo struct {
+	Address mail.Address
+	Type    RecipientType
+}
+
+type Recipients []RecipientInfo
+
+func (r Recipients) GetEmails(filter RecipientType) []string {
+	emails := []string{}
+	for _, user := range r {
+		if user.Type == filter {
+			emails = append(emails, user.Address.Address)
+		}
+	}
+	sort.Strings(emails)
+	return emails
+}
+
+func NewRecipients(emails []string, t RecipientType) Recipients {
+	r := Recipients{}
+	for _, e := range emails {
+		r = append(r, RecipientInfo{mail.Address{Address: e}, t})
+	}
+	sort.Sort(r)
+	return r
+}
+
+func (r Recipients) Len() int           { return len(r) }
+func (r Recipients) Less(i, j int) bool { return r[i].Address.Address < r[j].Address.Address }
+func (r Recipients) Swap(i, j int)      { r[i], r[j] = r[j], r[i] }
+
+func (r Recipients) ToDash() dashapi.Recipients {
+	d := dashapi.Recipients{}
+	for _, user := range r {
+		d = append(d, dashapi.RecipientInfo{Address: user.Address, Type: dashapi.RecipientType(user.Type)})
+	}
+	return d
+}
+
 type Repo interface {
 	// Poll checkouts the specified repository/branch.
 	// This involves fetching/resetting/cloning as necessary to recover from all possible problems.
@@ -78,7 +131,7 @@ type Commit struct {
 	Title      string
 	Author     string
 	AuthorName string
-	CC         []string
+	Recipients Recipients
 	Tags       []string
 	Parents    []string
 	Date       time.Time

pkg/vcs/vcs_test.go

@@ -4,7 +4,10 @@
 package vcs
 
 import (
+	"net/mail"
 	"testing"
+
+	"github.com/google/go-cmp/cmp"
 )
 
 func TestCanonicalizeCommit(t *testing.T) {
@@ -155,3 +158,37 @@ func TestCommitLink(t *testing.T) {
 		}
 	}
 }
+
+func TestParse(t *testing.T) {
+	// nolint: lll
+	test1 := []byte(`Foo (Maintainer) Bar <a@email.com> (maintainer:KERNEL)
+	Foo Bar(Reviewer) <b@email.com> (reviewer:KERNEL)
+	<somelist@list.com> (open list:FOO)
+	"Supporter Foo" <c@email.com> (supporter:KERNEL)
+	linux-kernel@vger.kernel.org (open list)`)
+	// nolint: lll
+	test2 := []byte(`Foo (Maintainer) Bar <a@email.com> (maintainer:KERNEL)
+	Foo Bar(Reviewer) <b@email.com> (reviewer:KERNEL)
+	"Supporter Foo" <c@email.com> (supporter:KERNEL)
+	linux-kernel@vger.kernel.org (open list)`)
+
+	maintainers1 := Recipients{{mail.Address{Name: "Foo (Maintainer) Bar", Address: "a@email.com"}, To},
+		{mail.Address{Name: "Foo Bar(Reviewer)", Address: "b@email.com"}, Cc},
+		{mail.Address{Name: "Supporter Foo", Address: "c@email.com"}, To},
+		{mail.Address{Name: "", Address: "linux-kernel@vger.kernel.org"}, Cc},
+		{mail.Address{Name: "", Address: "somelist@list.com"}, To}}
+	maintainers2 := Recipients{{mail.Address{Name: "Foo (Maintainer) Bar", Address: "a@email.com"}, To},
+		{mail.Address{Name: "Foo Bar(Reviewer)", Address: "b@email.com"}, Cc},
+		{mail.Address{Name: "Supporter Foo", Address: "c@email.com"}, To},
+		{mail.Address{Name: "", Address: "linux-kernel@vger.kernel.org"}, To}}
+
+	if diff := cmp.Diff(ParseMaintainersLinux(test1), maintainers1); diff != "" {
+		t.Fatal(diff)
+	}
+	if diff := cmp.Diff(ParseMaintainersLinux(test2), maintainers2); diff != "" {
+		t.Fatal(diff)
+	}
+	if diff := cmp.Diff(ParseMaintainersLinux([]byte("")), Recipients(nil)); diff != "" {
+		t.Fatal(diff)
+	}
+}

syz-ci/jobs.go

@@ -438,7 +438,7 @@ func (jp *JobProcessor) bisect(job *Job, mgrcfg *mgrconfig.Config) error {
 			Title:      com.Title,
 			Author:     com.Author,
 			AuthorName: com.AuthorName,
-			CC:         com.CC,
+			Recipients: com.Recipients.ToDash(),
 			Date:       com.Date,
 		})
 	}
@@ -469,7 +469,7 @@ func (jp *JobProcessor) bisect(job *Job, mgrcfg *mgrconfig.Config) error {
 		resp.CrashReport = res.Report.Report
 		resp.CrashLog = res.Report.Output
 		if len(resp.Commits) != 0 {
-			resp.Commits[0].CC = append(resp.Commits[0].CC, res.Report.Maintainers...)
+			resp.Commits[0].Recipients = append(resp.Commits[0].Recipients, res.Report.Recipients.ToDash()...)
 		} else {
 			// If there is a report and there is no commit, it means a crash
 			// occurred on HEAD(for BisectFix) and oldest tested release(for BisectCause).

syz-ci/manager.go

@@ -313,7 +313,7 @@ func (mgr *Manager) build(kernelCommit *vcs.Commit) error {
 		case *build.KernelError:
 			rep.Report = err1.Report
 			rep.Output = err1.Output
-			rep.Maintainers = err1.Maintainers
+			rep.Recipients = err1.Recipients
 		case *osutil.VerboseError:
 			rep.Report = []byte(err1.Title)
 			rep.Output = err1.Output
@@ -440,11 +440,11 @@ func (mgr *Manager) reportBuildError(rep *report.Report, info *BuildInfo, imageD
 	req := &dashapi.BuildErrorReq{
 		Build: *build,
 		Crash: dashapi.Crash{
-			Title:       rep.Title,
-			Corrupted:   false, // Otherwise they get merged with other corrupted reports.
-			Maintainers: rep.Maintainers,
-			Log:         rep.Output,
-			Report:      rep.Report,
+			Title:      rep.Title,
+			Corrupted:  false, // Otherwise they get merged with other corrupted reports.
+			Recipients: rep.Recipients.ToDash(),
+			Log:        rep.Output,
+			Report:     rep.Report,
 		},
 	}
 	return mgr.dash.ReportBuildError(req)

syz-manager/manager.go

@@ -645,12 +645,12 @@ func (mgr *Manager) saveCrash(crash *Crash) bool {
 			return true
 		}
 		dc := &dashapi.Crash{
-			BuildID:     mgr.cfg.Tag,
-			Title:       crash.Title,
-			Corrupted:   crash.Corrupted,
-			Maintainers: crash.Maintainers,
-			Log:         crash.Output,
-			Report:      crash.Report.Report,
+			BuildID:    mgr.cfg.Tag,
+			Title:      crash.Title,
+			Corrupted:  crash.Corrupted,
+			Recipients: crash.Recipients.ToDash(),
+			Log:        crash.Output,
+			Report:     crash.Report.Report,
 		}
 		resp, err := mgr.dash.ReportCrash(dc)
 		if err != nil {
@@ -811,14 +811,14 @@ func (mgr *Manager) saveRepro(res *repro.Result, stats *repro.Stats, hub bool) {
 		//    so maybe corrupted report detection is broken.
 		// 3. Reproduction is expensive so it's good to persist the result.
 		dc := &dashapi.Crash{
-			BuildID:     mgr.cfg.Tag,
-			Title:       res.Report.Title,
-			Maintainers: res.Report.Maintainers,
-			Log:         res.Report.Output,
-			Report:      res.Report.Report,
-			ReproOpts:   res.Opts.Serialize(),
-			ReproSyz:    res.Prog.Serialize(),
-			ReproC:      cprogText,
+			BuildID:    mgr.cfg.Tag,
+			Title:      res.Report.Title,
+			Recipients: res.Report.Recipients.ToDash(),
+			Log:        res.Report.Output,
+			Report:     res.Report.Report,
+			ReproOpts:  res.Opts.Serialize(),
+			ReproSyz:   res.Prog.Serialize(),
+			ReproC:     cprogText,
 		}
 		if _, err := mgr.dash.ReportCrash(dc); err != nil {
 			log.Logf(0, "failed to report repro to dashboard: %v", err)

tools/syz-symbolize/symbolize.go

@@ -15,6 +15,7 @@ import (
 	"github.com/google/syzkaller/pkg/mgrconfig"
 	"github.com/google/syzkaller/pkg/osutil"
 	"github.com/google/syzkaller/pkg/report"
+	"github.com/google/syzkaller/pkg/vcs"
 )
 
 var (
@@ -69,7 +70,8 @@ func main() {
 		}
 		fmt.Printf("TITLE: %v\n", rep.Title)
 		fmt.Printf("CORRUPTED: %v (%v)\n", rep.Corrupted, rep.CorruptedReason)
-		fmt.Printf("MAINTAINERS: %v\n", rep.Maintainers)
+		fmt.Printf("MAINTAINERS (TO): %v\n", rep.Recipients.GetEmails(vcs.To))
+		fmt.Printf("MAINTAINERS (CC): %v\n", rep.Recipients.GetEmails(vcs.Cc))
 		fmt.Printf("\n")
 		os.Stdout.Write(rep.Report)
 		fmt.Printf("\n\n")