tools/syz-declextract: allow to run on subset of arches

This may be useful for downstream kernels that only build
and are supposed to be used with a subset of arches.
Some esoteric arches may be broken on such kernels.
Allow to ignore them.

Author: dvyukov

pkg/tool/flags.go

@@ -9,9 +9,11 @@ import (
 	"errors"
 	"flag"
 	"fmt"
+	"sort"
 	"strings"
 
 	"github.com/google/syzkaller/pkg/log"
+	"github.com/google/syzkaller/sys/targets"
 )
 
 type Flag struct {
@@ -50,6 +52,30 @@ func ParseFlags(set *flag.FlagSet, args []string) error {
 	return nil
 }
 
+func ParseArchList(OS, archList string) ([]string, error) {
+	allArches := targets.List[OS]
+	if allArches == nil {
+		return nil, fmt.Errorf("bad OS %q", OS)
+	}
+	archMap := make(map[string]bool)
+	if archList != "" {
+		for _, arch := range strings.Split(archList, ",") {
+			if allArches[arch] == nil {
+				return nil, fmt.Errorf("bad arch %q for OS %q in arches flag", arch, OS)
+			}
+			archMap[arch] = true
+		}
+	}
+	var arches []string
+	for arch := range allArches {
+		if len(archMap) == 0 || archMap[arch] {
+			arches = append(arches, arch)
+		}
+	}
+	sort.Strings(arches)
+	return arches, nil
+}
+
 const optionalFlag = "optional"
 
 func serializeFlags(flags []Flag) string {

pkg/tool/flags_test.go

@@ -4,13 +4,15 @@
 package tool
 
 import (
+	"errors"
 	"flag"
 	"fmt"
 	"io"
 	"strings"
 	"testing"
 
 	"github.com/google/go-cmp/cmp"
+	"github.com/stretchr/testify/assert"
 )
 
 func TestParseFlags(t *testing.T) {
@@ -84,3 +86,40 @@ func TestCfgsFlagAlreadySet(t *testing.T) {
 		t.Errorf("cfgs.Set got: nil, want: error")
 	}
 }
+
+func TestParseArchList(t *testing.T) {
+	type Test struct {
+		OS  string
+		In  string
+		Out []string
+		Err error
+	}
+	tests := []Test{
+		{
+			OS:  "foo",
+			Err: errors.New(`bad OS "foo"`),
+		},
+		{
+			OS:  "linux",
+			In:  "amd64,bar",
+			Err: errors.New(`bad arch "bar" for OS "linux" in arches flag`),
+		},
+		{
+			OS:  "linux",
+			In:  "",
+			Out: []string{"386", "amd64", "arm", "arm64", "mips64le", "ppc64le", "riscv64", "s390x"},
+		},
+		{
+			OS:  "linux",
+			In:  "ppc64le,386",
+			Out: []string{"386", "ppc64le"},
+		},
+	}
+	for i, test := range tests {
+		t.Run(fmt.Sprint(i), func(t *testing.T) {
+			got, err := ParseArchList(test.OS, test.In)
+			assert.Equal(t, err, test.Err)
+			assert.Equal(t, got, test.Out)
+		})
+	}
+}

sys/syz-extract/extract.go

@@ -11,7 +11,6 @@ import (
 	"path/filepath"
 	"runtime"
 	"sort"
-	"strings"
 
 	"github.com/google/syzkaller/pkg/ast"
 	"github.com/google/syzkaller/pkg/compiler"
@@ -80,7 +79,11 @@ func main() {
 	if extractor == nil {
 		tool.Failf("unknown os: %v", OS)
 	}
-	arches, nfiles, err := createArches(OS, archList(OS, *flagArch), flag.Args())
+	archList, err := tool.ParseArchList(OS, *flagArch)
+	if err != nil {
+		tool.Failf("failed to parse arch flag: %v", err)
+	}
+	arches, nfiles, err := createArches(OS, archList, flag.Args())
 	if err != nil {
 		tool.Fail(err)
 	}
@@ -242,18 +245,6 @@ func createArches(OS string, archArray, files []string) ([]*Arch, int, error) {
 	return arches, nfiles, nil
 }
 
-func archList(OS, arches string) []string {
-	if arches != "" {
-		return strings.Split(arches, ",")
-	}
-	var archArray []string
-	for arch := range targets.List[OS] {
-		archArray = append(archArray, arch)
-	}
-	sort.Strings(archArray)
-	return archArray
-}
-
 func checkUnsupportedCalls(arches []*Arch) bool {
 	supported := make(map[string]bool)
 	unsupported := make(map[string]string)

tools/syz-declextract/declextract.go

@@ -37,6 +37,7 @@ func main() {
 	var (
 		flagConfig = flag.String("config", "", "manager config file")
 		flagBinary = flag.String("binary", "syz-declextract", "path to syz-declextract binary")
+		flagArches = flag.String("arches", "", "comma-separated list of arches to extract (all if empty)")
 	)
 	defer tool.Init()()
 	cfg, err := mgrconfig.LoadFile(*flagConfig)
@@ -46,7 +47,7 @@ func main() {
 	loadProbeInfo := func() (*ifaceprobe.Info, error) {
 		return probe(cfg, *flagConfig)
 	}
-	if _, err := run(filepath.FromSlash("sys/linux/auto.txt"), loadProbeInfo, &clangtool.Config{
+	if _, err := run(filepath.FromSlash("sys/linux/auto.txt"), loadProbeInfo, *flagArches, &clangtool.Config{
 		ToolBin:    *flagBinary,
 		KernelSrc:  cfg.KernelSrc,
 		KernelObj:  cfg.KernelObj,
@@ -57,9 +58,13 @@ func main() {
 	}
 }
 
-func run(autoFile string, loadProbeInfo func() (*ifaceprobe.Info, error), cfg *clangtool.Config) (
+func run(autoFile string, loadProbeInfo func() (*ifaceprobe.Info, error), archList string, cfg *clangtool.Config) (
 	*declextract.Result, error) {
-	out, probeInfo, syscallRename, err := prepare(loadProbeInfo, cfg)
+	arches, err := tool.ParseArchList(target.OS, archList)
+	if err != nil {
+		return nil, fmt.Errorf("failed to parse arches flag: %v", err)
+	}
+	out, probeInfo, syscallRename, err := prepare(loadProbeInfo, arches, cfg)
 	if err != nil {
 		return nil, err
 	}
@@ -123,7 +128,7 @@ func removeUnused(desc *ast.Description, autoFile string, unusedNodes []ast.Node
 	})
 }
 
-func prepare(loadProbeInfo func() (*ifaceprobe.Info, error), cfg *clangtool.Config) (
+func prepare(loadProbeInfo func() (*ifaceprobe.Info, error), arches []string, cfg *clangtool.Config) (
 	*declextract.Output, *ifaceprobe.Info, map[string][]string, error) {
 	var eg errgroup.Group
 	var out *declextract.Output
@@ -147,7 +152,7 @@ func prepare(loadProbeInfo func() (*ifaceprobe.Info, error), cfg *clangtool.Conf
 	var syscallRename map[string][]string
 	eg.Go(func() error {
 		var err error
-		syscallRename, err = buildSyscallRenameMap(cfg.KernelSrc)
+		syscallRename, err = buildSyscallRenameMap(cfg.KernelSrc, arches)
 		if err != nil {
 			return fmt.Errorf("failed to build syscall rename map: %w", err)
 		}
@@ -234,7 +239,7 @@ func finishInterfaces(interfaces []*declextract.Interface, consts map[string]*co
 	}
 }
 
-func buildSyscallRenameMap(sourceDir string) (map[string][]string, error) {
+func buildSyscallRenameMap(sourceDir string, arches []string) (map[string][]string, error) {
 	// Some syscalls have different names and entry points and thus need to be renamed.
 	// e.g. SYSCALL_DEFINE1(setuid16, old_uid_t, uid) is referred to in the .tbl file with setuid.
 	// Parse *.tbl files that map functions defined with SYSCALL_DEFINE macros to actual syscall names.
@@ -244,7 +249,7 @@ func buildSyscallRenameMap(sourceDir string) (map[string][]string, error) {
 	// and then just order arches by name to have deterministic result.
 	// Note: some syscalls may have no record in the tables for the architectures we support.
 	syscalls := make(map[string][]tblSyscall)
-	tblFiles, err := findTblFiles(sourceDir)
+	tblFiles, err := findTblFiles(sourceDir, arches)
 	if err != nil {
 		return nil, err
 	}
@@ -320,9 +325,10 @@ func parseTblFile(data []byte, arch string, syscalls map[string][]tblSyscall) {
 	}
 }
 
-func findTblFiles(sourceDir string) (map[string][]string, error) {
+func findTblFiles(sourceDir string, arches []string) (map[string][]string, error) {
 	files := make(map[string][]string)
-	for _, arch := range targets.List[target.OS] {
+	for _, name := range arches {
+		arch := targets.List[target.OS][name]
 		err := filepath.WalkDir(filepath.Join(sourceDir, "arch", arch.KernelHeaderArch),
 			func(file string, d fs.DirEntry, err error) error {
 				if err == nil && strings.HasSuffix(file, ".tbl") {

tools/syz-declextract/declextract_test.go

@@ -68,7 +68,7 @@ func TestDeclextract(t *testing.T) {
 			return probeInfo, nil
 		}
 		autoFile := filepath.Join(cfg.KernelObj, filepath.Base(file)+".txt")
-		res, err := run(autoFile, loadProbeInfo, cfg)
+		res, err := run(autoFile, loadProbeInfo, "", cfg)
 		if err != nil {
 			if *flagUpdate {
 				osutil.CopyFile(autoFile, file+".txt")