The problem is: currently our single byte ctype(3) functions are broken

for wide characters locales in the argument range >= 0x80 - they may
return false positives.

Example 1: for UTF-8 locale we currently have:
iswspace(0xA0)==1 and isspace(0xA0)==1
(because iswspace() and isspace() are the same code)
but must have
iswspace(0xA0)==1 and isspace(0xA0)==0
(because there is no such character and all others in the range
0x80..0xff for the UTF-8 locale, it keeps ASCII only in the single byte
range because our internal wchar_t representation for UTF-8 is UCS-4).

Example 2: for all wide character locales isalpha(arg) when arg > 0xFF may
return false positives (must be 0).
(because iswalpha() and isalpha() are the same code)

This change address this issue separating single byte and wide ctype
and also fix iswascii() (currently iswascii() is broken for
arguments > 0xFF).
This change is 100% binary compatible with old binaries.

Reviewied by: i18n@

include/_ctype.h

@@ -87,6 +87,8 @@ __END_DECLS
 #define	__inline
 #endif
 
+extern int __mb_sb_limit;
+
 /*
  * Use inline functions if we are allowed to and the compiler supports them.
  */
@@ -102,16 +104,29 @@ __maskrune(__ct_rune_t _c, unsigned long _f)
 		_CurrentRuneLocale->__runetype[_c]) & _f;
 }
 
+static __inline int
+__sbmaskrune(__ct_rune_t _c, unsigned long _f)
+{
+	return (_c < 0 || _c >= __mb_sb_limit) ? 0 :
+	       _CurrentRuneLocale->__runetype[_c] & _f;
+}
+
 static __inline int
 __istype(__ct_rune_t _c, unsigned long _f)
 {
 	return (!!__maskrune(_c, _f));
 }
 
+static __inline int
+__sbistype(__ct_rune_t _c, unsigned long _f)
+{
+	return (!!__sbmaskrune(_c, _f));
+}
+
 static __inline int
 __isctype(__ct_rune_t _c, unsigned long _f)
 {
-	return (_c < 0 || _c >= _CACHED_RUNES) ? 0 :
+	return (_c < 0 || _c >= __mb_sb_limit) ? 0 :
 	       !!(_DefaultRuneLocale.__runetype[_c] & _f);
 }
 
@@ -122,6 +137,13 @@ __toupper(__ct_rune_t _c)
 	       _CurrentRuneLocale->__mapupper[_c];
 }
 
+static __inline __ct_rune_t
+__sbtoupper(__ct_rune_t _c)
+{
+	return (_c < 0 || _c >= __mb_sb_limit) ? _c :
+	       _CurrentRuneLocale->__mapupper[_c];
+}
+
 static __inline __ct_rune_t
 __tolower(__ct_rune_t _c)
 {
@@ -129,6 +151,13 @@ __tolower(__ct_rune_t _c)
 	       _CurrentRuneLocale->__maplower[_c];
 }
 
+static __inline __ct_rune_t
+__sbtolower(__ct_rune_t _c)
+{
+	return (_c < 0 || _c >= __mb_sb_limit) ? _c :
+	       _CurrentRuneLocale->__maplower[_c];
+}
+
 static __inline int
 __wcwidth(__ct_rune_t _c)
 {
@@ -146,10 +175,14 @@ __wcwidth(__ct_rune_t _c)
 
 __BEGIN_DECLS
 int		__maskrune(__ct_rune_t, unsigned long);
+int		__sbmaskrune(__ct_rune_t, unsigned long);
 int		__istype(__ct_rune_t, unsigned long);
+int		__sbistype(__ct_rune_t, unsigned long);
 int		__isctype(__ct_rune_t, unsigned long);
 __ct_rune_t	__toupper(__ct_rune_t);
+__ct_rune_t	__sbtoupper(__ct_rune_t);
 __ct_rune_t	__tolower(__ct_rune_t);
+__ct_rune_t	__sbtolower(__ct_rune_t);
 int		__wcwidth(__ct_rune_t);
 __END_DECLS
 #endif /* using inlines */

include/ctype.h

@@ -86,19 +86,19 @@ int	isspecial(int);
 #endif
 __END_DECLS
 
-#define	isalnum(c)	__istype((c), _CTYPE_A|_CTYPE_D)
+#define	isalnum(c)	__sbistype((c), _CTYPE_A|_CTYPE_D)
-#define	isalpha(c)	__istype((c), _CTYPE_A)
+#define	isalpha(c)	__sbistype((c), _CTYPE_A)
-#define	iscntrl(c)	__istype((c), _CTYPE_C)
+#define	iscntrl(c)	__sbistype((c), _CTYPE_C)
 #define	isdigit(c)	__isctype((c), _CTYPE_D) /* ANSI -- locale independent */
-#define	isgraph(c)	__istype((c), _CTYPE_G)
+#define	isgraph(c)	__sbistype((c), _CTYPE_G)
-#define	islower(c)	__istype((c), _CTYPE_L)
+#define	islower(c)	__sbistype((c), _CTYPE_L)
-#define	isprint(c)	__istype((c), _CTYPE_R)
+#define	isprint(c)	__sbistype((c), _CTYPE_R)
-#define	ispunct(c)	__istype((c), _CTYPE_P)
+#define	ispunct(c)	__sbistype((c), _CTYPE_P)
-#define	isspace(c)	__istype((c), _CTYPE_S)
+#define	isspace(c)	__sbistype((c), _CTYPE_S)
-#define	isupper(c)	__istype((c), _CTYPE_U)
+#define	isupper(c)	__sbistype((c), _CTYPE_U)
 #define	isxdigit(c)	__isctype((c), _CTYPE_X) /* ANSI -- locale independent */
-#define	tolower(c)	__tolower(c)
+#define	tolower(c)	__sbtolower(c)
-#define	toupper(c)	__toupper(c)
+#define	toupper(c)	__sbtoupper(c)
 
 #if __XSI_VISIBLE
 /*
@@ -112,24 +112,24 @@ __END_DECLS
  *
  * XXX isascii() and toascii() should similarly be undocumented.
  */
-#define	_tolower(c)	__tolower(c)
+#define	_tolower(c)	__sbtolower(c)
-#define	_toupper(c)	__toupper(c)
+#define	_toupper(c)	__sbtoupper(c)
 #define	isascii(c)	(((c) & ~0x7F) == 0)
 #define	toascii(c)	((c) & 0x7F)
 #endif
 
 #if __ISO_C_VISIBLE >= 1999
-#define	isblank(c)	__istype((c), _CTYPE_B)
+#define	isblank(c)	__sbistype((c), _CTYPE_B)
 #endif
 
 #if __BSD_VISIBLE
-#define	digittoint(c)	__maskrune((c), 0xFF)
+#define	digittoint(c)	__sbmaskrune((c), 0xFF)
-#define	ishexnumber(c)	__istype((c), _CTYPE_X)
+#define	ishexnumber(c)	__sbistype((c), _CTYPE_X)
-#define	isideogram(c)	__istype((c), _CTYPE_I)
+#define	isideogram(c)	__sbistype((c), _CTYPE_I)
-#define	isnumber(c)	__istype((c), _CTYPE_D)
+#define	isnumber(c)	__sbistype((c), _CTYPE_D)
-#define	isphonogram(c)	__istype((c), _CTYPE_Q)
+#define	isphonogram(c)	__sbistype((c), _CTYPE_Q)
-#define	isrune(c)	__istype((c), 0xFFFFFF00L)
+#define	isrune(c)	__sbistype((c), 0xFFFFFF00L)
-#define	isspecial(c)	__istype((c), _CTYPE_T)
+#define	isspecial(c)	__sbistype((c), _CTYPE_T)
 #endif
 
 #endif /* !_CTYPE_H_ */

include/wctype.h

@@ -106,7 +106,7 @@ __END_DECLS
 #define	towupper(wc)		__toupper(wc)
 
 #if __BSD_VISIBLE
-#define	iswascii(wc)		(((wc) & ~0x7F) == 0)
+#define	iswascii(wc)		((wc) < 0x80)
 #define	iswhexnumber(wc)	__istype((wc), _CTYPE_X)
 #define	iswideogram(wc)		__istype((wc), _CTYPE_I)
 #define	iswnumber(wc)		__istype((wc), _CTYPE_D)

lib/libc/locale/Symbol.map

@@ -60,12 +60,17 @@ FBSD_1.0 {
 	nextwctype;
 	nl_langinfo;
 	__maskrune;
+	__sbmaskrune;
 	__istype;
+	__sbistype;
 	__isctype;
 	__toupper;
+	__sbtoupper;
 	__tolower;
+	__sbtolower;
 	__wcwidth;
 	__mb_cur_max;
+	__mb_sb_limit;
 	rpmatch;
 	___runetype;
 	setlocale;

lib/libc/locale/big5.c

@@ -49,6 +49,8 @@ __FBSDID("$FreeBSD$");
 #include <wchar.h>
 #include "mblocal.h"
 
+extern int __mb_sb_limit;
+
 static size_t	_BIG5_mbrtowc(wchar_t * __restrict, const char * __restrict,
 		    size_t, mbstate_t * __restrict);
 static int	_BIG5_mbsinit(const mbstate_t *);
@@ -68,6 +70,7 @@ _BIG5_init(_RuneLocale *rl)
 	__mbsinit = _BIG5_mbsinit;
 	_CurrentRuneLocale = rl;
 	__mb_cur_max = 2;
+	__mb_sb_limit = 128;
 	return (0);
 }
 

lib/libc/locale/euc.c

@@ -49,6 +49,8 @@ __FBSDID("$FreeBSD$");
 #include <wchar.h>
 #include "mblocal.h"
 
+extern int __mb_sb_limit;
+
 static size_t	_EUC_mbrtowc(wchar_t * __restrict, const char * __restrict,
 		    size_t, mbstate_t * __restrict);
 static int	_EUC_mbsinit(const mbstate_t *);
@@ -116,6 +118,7 @@ _EUC_init(_RuneLocale *rl)
 	__mbrtowc = _EUC_mbrtowc;
 	__wcrtomb = _EUC_wcrtomb;
 	__mbsinit = _EUC_mbsinit;
+	__mb_sb_limit = 256;
 	return (0);
 }
 

lib/libc/locale/gb18030.c

@@ -39,6 +39,8 @@ __FBSDID("$FreeBSD$");
 #include <wchar.h>
 #include "mblocal.h"
 
+extern int __mb_sb_limit;
+
 static size_t	_GB18030_mbrtowc(wchar_t * __restrict, const char * __restrict,
 		    size_t, mbstate_t * __restrict);
 static int	_GB18030_mbsinit(const mbstate_t *);
@@ -59,6 +61,7 @@ _GB18030_init(_RuneLocale *rl)
 	__mbsinit = _GB18030_mbsinit;
 	_CurrentRuneLocale = rl;
 	__mb_cur_max = 4;
+	__mb_sb_limit = 128;
 
 	return (0);
 }

lib/libc/locale/gb2312.c

@@ -35,6 +35,8 @@ __FBSDID("$FreeBSD$");
 #include <wchar.h>
 #include "mblocal.h"
 
+extern int __mb_sb_limit;
+
 static size_t	_GB2312_mbrtowc(wchar_t * __restrict, const char * __restrict,
 		    size_t, mbstate_t * __restrict);
 static int	_GB2312_mbsinit(const mbstate_t *);
@@ -55,6 +57,7 @@ _GB2312_init(_RuneLocale *rl)
 	__wcrtomb = _GB2312_wcrtomb;
 	__mbsinit = _GB2312_mbsinit;
 	__mb_cur_max = 2;
+	__mb_sb_limit = 128;
 	return (0);
 }
 

lib/libc/locale/gbk.c

@@ -42,6 +42,8 @@ __FBSDID("$FreeBSD$");
 #include <wchar.h>
 #include "mblocal.h"
 
+extern int __mb_sb_limit;
+
 static size_t	_GBK_mbrtowc(wchar_t * __restrict, const char * __restrict,
 		    size_t, mbstate_t * __restrict);
 static int	_GBK_mbsinit(const mbstate_t *);
@@ -61,6 +63,7 @@ _GBK_init(_RuneLocale *rl)
 	__mbsinit = _GBK_mbsinit;
 	_CurrentRuneLocale = rl;
 	__mb_cur_max = 2;
+	__mb_sb_limit = 128;
 	return (0);
 }
 

lib/libc/locale/isctype.c

@@ -48,7 +48,7 @@ int
 digittoint(c)
 	int c;
 {
-	return (__maskrune(c, 0xFF));
+	return (__sbmaskrune(c, 0xFF));
 }
 
 #undef isalnum
@@ -56,7 +56,7 @@ int
 isalnum(c)
 	int c;
 {
-	return (__istype(c, _CTYPE_A|_CTYPE_D));
+	return (__sbistype(c, _CTYPE_A|_CTYPE_D));
 }
 
 #undef isalpha
@@ -64,7 +64,7 @@ int
 isalpha(c)
 	int c;
 {
-	return (__istype(c, _CTYPE_A));
+	return (__sbistype(c, _CTYPE_A));
 }
 
 #undef isascii
@@ -80,7 +80,7 @@ int
 isblank(c)
 	int c;
 {
-	return (__istype(c, _CTYPE_B));
+	return (__sbistype(c, _CTYPE_B));
 }
 
 #undef iscntrl
@@ -88,7 +88,7 @@ int
 iscntrl(c)
 	int c;
 {
-	return (__istype(c, _CTYPE_C));
+	return (__sbistype(c, _CTYPE_C));
 }
 
 #undef isdigit
@@ -104,7 +104,7 @@ int
 isgraph(c)
 	int c;
 {
-	return (__istype(c, _CTYPE_G));
+	return (__sbistype(c, _CTYPE_G));
 }
 
 #undef ishexnumber 
@@ -112,7 +112,7 @@ int
 ishexnumber(c)
 	int c;
 {
-	return (__istype(c, _CTYPE_X));
+	return (__sbistype(c, _CTYPE_X));
 }
 
 #undef isideogram
@@ -120,7 +120,7 @@ int
 isideogram(c)
 	int c;
 {
-	return (__istype(c, _CTYPE_I));
+	return (__sbistype(c, _CTYPE_I));
 }
 
 #undef islower
@@ -128,7 +128,7 @@ int
 islower(c)
 	int c;
 {
-	return (__istype(c, _CTYPE_L));
+	return (__sbistype(c, _CTYPE_L));
 }
 
 #undef isnumber
@@ -136,7 +136,7 @@ int
 isnumber(c)
 	int c;
 {
-	return (__istype(c, _CTYPE_D));
+	return (__sbistype(c, _CTYPE_D));
 }
 
 #undef isphonogram	
@@ -144,7 +144,7 @@ int
 isphonogram(c)
 	int c;
 {
-	return (__istype(c, _CTYPE_Q));
+	return (__sbistype(c, _CTYPE_Q));
 }
 
 #undef isprint
@@ -152,7 +152,7 @@ int
 isprint(c)
 	int c;
 {
-	return (__istype(c, _CTYPE_R));
+	return (__sbistype(c, _CTYPE_R));
 }
 
 #undef ispunct
@@ -160,7 +160,7 @@ int
 ispunct(c)
 	int c;
 {
-	return (__istype(c, _CTYPE_P));
+	return (__sbistype(c, _CTYPE_P));
 }
 
 #undef isrune
@@ -168,7 +168,7 @@ int
 isrune(c)
 	int c;
 {
-	return (__istype(c, 0xFFFFFF00L));
+	return (__sbistype(c, 0xFFFFFF00L));
 }
 
 #undef isspace
@@ -176,7 +176,7 @@ int
 isspace(c)
 	int c;
 {
-	return (__istype(c, _CTYPE_S));
+	return (__sbistype(c, _CTYPE_S));
 }
 
 #undef isspecial
@@ -184,7 +184,7 @@ int
 isspecial(c)
 	int c;
 {
-	return (__istype(c, _CTYPE_T));
+	return (__sbistype(c, _CTYPE_T));
 }
 
 #undef isupper
@@ -192,7 +192,7 @@ int
 isupper(c)
 	int c;
 {
-	return (__istype(c, _CTYPE_U));
+	return (__sbistype(c, _CTYPE_U));
 }
 
 #undef isxdigit
@@ -216,7 +216,7 @@ int
 tolower(c)
 	int c;
 {
-        return (__tolower(c));
+	return (__sbtolower(c));
 }
 
 #undef toupper
@@ -224,6 +224,6 @@ int
 toupper(c)
 	int c;
 {
-        return (__toupper(c));
+	return (__sbtoupper(c));
 }
 

lib/libc/locale/iswctype.c

@@ -61,7 +61,7 @@ int
 iswascii(wc)
 	wint_t wc;
 {
-	return ((wc & ~0x7F) == 0);
+	return (wc < 0x80);
 }
 
 #undef iswblank

lib/libc/locale/mskanji.c

@@ -47,6 +47,8 @@ __FBSDID("$FreeBSD$");
 #include <wchar.h>
 #include "mblocal.h"
 
+extern int __mb_sb_limit;
+
 static size_t	_MSKanji_mbrtowc(wchar_t * __restrict, const char * __restrict,
 		    size_t, mbstate_t * __restrict);
 static int	_MSKanji_mbsinit(const mbstate_t *);
@@ -66,6 +68,7 @@ _MSKanji_init(_RuneLocale *rl)
 	__mbsinit = _MSKanji_mbsinit;
 	_CurrentRuneLocale = rl;
 	__mb_cur_max = 2;
+	__mb_sb_limit = 256;
 	return (0);
 }
 

lib/libc/locale/none.c

@@ -58,6 +58,11 @@ static size_t	_none_wcrtomb(char * __restrict, wchar_t,
 static size_t	_none_wcsnrtombs(char * __restrict, const wchar_t ** __restrict,
 		    size_t, size_t, mbstate_t * __restrict);
 
+/* setup defaults */
+
+int __mb_cur_max = 1;
+int __mb_sb_limit = 256; /* Expected to be <= _CACHED_RUNES */
+
 int
 _none_init(_RuneLocale *rl)
 {
@@ -69,6 +74,7 @@ _none_init(_RuneLocale *rl)
 	__wcsnrtombs = _none_wcsnrtombs;
 	_CurrentRuneLocale = rl;
 	__mb_cur_max = 1;
+	__mb_sb_limit = 256;
 	return(0);
 }
 
@@ -176,7 +182,6 @@ _none_wcsnrtombs(char * __restrict dst, const wchar_t ** __restrict src,
 
 /* setup defaults */
 
-int __mb_cur_max = 1;
 size_t (*__mbrtowc)(wchar_t * __restrict, const char * __restrict, size_t,
     mbstate_t * __restrict) = _none_mbrtowc;
 int (*__mbsinit)(const mbstate_t *) = _none_mbsinit;

lib/libc/locale/setrunelocale.c

@@ -45,6 +45,8 @@ __FBSDID("$FreeBSD$");
 #include "mblocal.h"
 #include "setlocale.h"
 
+extern int __mb_sb_limit;
+
 extern _RuneLocale	*_Read_RuneMagi(FILE *);
 
 static int		__setrunelocale(const char *);
@@ -59,6 +61,7 @@ __setrunelocale(const char *encoding)
 	static char ctype_encoding[ENCODING_LEN + 1];
 	static _RuneLocale *CachedRuneLocale;
 	static int Cached__mb_cur_max;
+	static int Cached__mb_sb_limit;
 	static size_t (*Cached__mbrtowc)(wchar_t * __restrict,
 	    const char * __restrict, size_t, mbstate_t * __restrict);
 	static size_t (*Cached__wcrtomb)(char * __restrict, wchar_t,
@@ -85,6 +88,7 @@ __setrunelocale(const char *encoding)
 	    strcmp(encoding, ctype_encoding) == 0) {
 		_CurrentRuneLocale = CachedRuneLocale;
 		__mb_cur_max = Cached__mb_cur_max;
+		__mb_sb_limit = Cached__mb_sb_limit;
 		__mbrtowc = Cached__mbrtowc;
 		__mbsinit = Cached__mbsinit;
 		__mbsnrtowcs = Cached__mbsnrtowcs;
@@ -147,6 +151,7 @@ __setrunelocale(const char *encoding)
 		}
 		CachedRuneLocale = _CurrentRuneLocale;
 		Cached__mb_cur_max = __mb_cur_max;
+		Cached__mb_sb_limit = __mb_sb_limit;
 		Cached__mbrtowc = __mbrtowc;
 		Cached__mbsinit = __mbsinit;
 		Cached__mbsnrtowcs = __mbsnrtowcs;

lib/libc/locale/utf8.c

@@ -35,6 +35,8 @@ __FBSDID("$FreeBSD$");
 #include <wchar.h>
 #include "mblocal.h"
 
+extern int __mb_sb_limit;
+
 static size_t	_UTF8_mbrtowc(wchar_t * __restrict, const char * __restrict,
 		    size_t, mbstate_t * __restrict);
 static int	_UTF8_mbsinit(const mbstate_t *);
@@ -63,6 +65,7 @@ _UTF8_init(_RuneLocale *rl)
 	__wcsnrtombs = _UTF8_wcsnrtombs;
 	_CurrentRuneLocale = rl;
 	__mb_cur_max = 6;
+	__mb_sb_limit = 128;
 
 	return (0);
 }