tools/syz-declextract: ignore files with non US-ASCII chars

3 files changed, 22 insertions, 9 deletions

diff --git a/pkg/ast/scanner.go b/pkg/ast/scanner.go
index 9a5becf1c..67dd80e52 100644
--- a/pkg/ast/scanner.go
+++ b/pkg/ast/scanner.go
@@ -221,14 +221,11 @@ func (s *scanner) scanStr(pos Pos) string {
 		}
 	}
 	lit := string(s.data[pos.Off+1 : s.off])
-	for i := 0; i < len(lit); i++ {
-		if lit[i] < 0x20 || lit[i] >= 0x80 {
-			pos1 := pos
-			pos1.Col += i + 1
-			pos1.Off += i + 1
-			s.Errorf(pos1, "illegal character %#U in string literal", lit[i])
-			break
-		}
+	if i := IsValidStringLit(lit); i >= 0 {
+		pos1 := pos
+		pos1.Col += i + 1
+		pos1.Off += i + 1
+		s.Errorf(pos1, "illegal character %#U in string literal %q", lit[i], lit)
 	}
 	s.next()
 	if closing != '`' {
@@ -352,3 +349,12 @@ func (s *scanner) pos() Pos {
 		Col:  s.col,
 	}
 }
+
+func IsValidStringLit(lit string) int {
+	for i := 0; i < len(lit); i++ {
+		if lit[i] < 0x20 || lit[i] >= 0x80 {
+			return i
+		}
+	}
+	return -1
+}
diff --git a/pkg/ast/testdata/errors.txt b/pkg/ast/testdata/errors.txt
index a8a4aa197..0d8488134 100644
--- a/pkg/ast/testdata/errors.txt
+++ b/pkg/ast/testdata/errors.txt
@@ -20,7 +20,7 @@ int_flags4 = 1, -2-			### bad integer "-2-"
 
 str_flags0 = "foo", "bar"
 str_flags1 = "non terminated		### string literal is not terminated
-str_flags2 = "bad chars здесь"		### illegal character U+00D0 'Ð' in string literal
+str_flags2 = "bad chars здесь"		### illegal character U+00D0 'Ð' in string literal "bad chars здесь"
 str_flags3 = "string", not a string	### unexpected identifier, expecting '\n'
 str_flags4 = "string", 42		### unexpected int, expecting string, hex string, identifier
 str_flags5 = `x`			### bad hex string literal: encoding/hex: invalid byte: U+0078 'x'
diff --git a/pkg/declextract/fileops.go b/pkg/declextract/fileops.go
index 3a82d13c2..8b2ea629a 100644
--- a/pkg/declextract/fileops.go
+++ b/pkg/declextract/fileops.go
@@ -7,6 +7,8 @@ import (
 	"fmt"
 	"slices"
 	"strings"
+
+	"github.com/google/syzkaller/pkg/ast"
 )
 
 const (
@@ -179,6 +181,11 @@ func (ctx *context) mapFopsToFiles(uniqueFuncs map[*Function]int) map[*FileOps][
 	ctx.FileOps = append(ctx.FileOps, generic)
 	fopsToFiles := make(map[*FileOps][]string)
 	for _, file := range ctx.probe.Files {
+		// There is a single non US-ASCII file in sysfs: "/sys/bus/pci/drivers/CAFÉ NAND".
+		// Ignore it for now as descriptions shouldn't contain non US-ASCII chars.
+		if ast.IsValidStringLit(file.Name) >= 0 {
+			continue
+		}
 		// For each file figure out the potential file_operations that match this file best.
 		best := ctx.mapFileToFops(fileToFuncs[file.Name], funcToFops, uniqueFuncs, generic)
 		for _, fops := range best {