12358 Need mbrtowc variant that indicates consumed zero bytes

Reviewed by: John Levon <john.levon@joyent.com>
Reviewed by: Yuri Pankov <ypankov@fastmail.com>
Approved by: Dan McDonald <danmcd@joyent.com>

14 files changed, 123 insertions, 50 deletions

diff --git a/usr/src/lib/libc/inc/libc.h b/usr/src/lib/libc/inc/libc.h
index 448523df01..90a2859b33 100644
--- a/usr/src/lib/libc/inc/libc.h
+++ b/usr/src/lib/libc/inc/libc.h
@@ -333,6 +333,14 @@ extern void __throw_constraint_handler_s(const char *_RESTRICT_KYWD, int);
  */
 extern void common_panic(const char *, const char *);
 
+/*
+ * defined in mbrtowc.c.
+ */
+extern size_t mbrtowc_nz_l(wchar_t *_RESTRICT_KYWD, const char *_RESTRICT_KYWD,
+    size_t, mbstate_t *_RESTRICT_KYWD, locale_t);
+extern size_t mbrtowc_nz(wchar_t *_RESTRICT_KYWD, const char *_RESTRICT_KYWD,
+    size_t, mbstate_t *_RESTRICT_KYWD);
+
 #ifdef	__cplusplus
 }
 #endif
diff --git a/usr/src/lib/libc/port/locale/big5.c b/usr/src/lib/libc/port/locale/big5.c
index 2729ab0289..889b182de5 100644
--- a/usr/src/lib/libc/port/locale/big5.c
+++ b/usr/src/lib/libc/port/locale/big5.c
@@ -44,7 +44,7 @@
 
 static size_t	_BIG5_mbrtowc(wchar_t *_RESTRICT_KYWD,
 		    const char *_RESTRICT_KYWD,
-		    size_t, mbstate_t *RESTRICT_KYWD);
+		    size_t, mbstate_t *RESTRICT_KYWD, boolean_t);
 static int	_BIG5_mbsinit(const mbstate_t *);
 static size_t	_BIG5_wcrtomb(char *_RESTRICT_KYWD, wchar_t,
 		    mbstate_t *_RESTRICT_KYWD);
@@ -88,7 +88,7 @@ _big5_check(uint_t c)
 
 static size_t
 _BIG5_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
-    size_t n, mbstate_t *_RESTRICT_KYWD ps)
+    size_t n, mbstate_t *_RESTRICT_KYWD ps, boolean_t zero)
 {
 	_BIG5State *bs;
 	wchar_t wc;
@@ -143,7 +143,11 @@ _BIG5_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
 	} else {
 		if (pwc != NULL)
 			*pwc = wc;
-		return (wc == L'\0' ? 0 : 1);
+		if (zero || wc != L'\0') {
+			return (1);
+		} else {
+			return (0);
+		}
 	}
 }
 
diff --git a/usr/src/lib/libc/port/locale/euc.c b/usr/src/lib/libc/port/locale/euc.c
index a99cdf8689..1d1d25b17b 100644
--- a/usr/src/lib/libc/port/locale/euc.c
+++ b/usr/src/lib/libc/port/locale/euc.c
@@ -46,22 +46,23 @@
 
 static size_t	_EUC_mbrtowc_impl(wchar_t *_RESTRICT_KYWD,
     const char *_RESTRICT_KYWD,
-    size_t, mbstate_t *_RESTRICT_KYWD, uint8_t, uint8_t, uint8_t, uint8_t);
+    size_t, mbstate_t *_RESTRICT_KYWD, uint8_t, uint8_t, uint8_t, uint8_t,
+    boolean_t);
 static size_t	_EUC_wcrtomb_impl(char *_RESTRICT_KYWD, wchar_t,
     mbstate_t *_RESTRICT_KYWD, uint8_t, uint8_t, uint8_t, uint8_t);
 
 static size_t	_EUC_CN_mbrtowc(wchar_t *_RESTRICT_KYWD,
 		    const char *_RESTRICT_KYWD,
-		    size_t, mbstate_t *_RESTRICT_KYWD);
+		    size_t, mbstate_t *_RESTRICT_KYWD, boolean_t);
 static size_t	_EUC_JP_mbrtowc(wchar_t *_RESTRICT_KYWD,
 		    const char *_RESTRICT_KYWD,
-		    size_t, mbstate_t *_RESTRICT_KYWD);
+		    size_t, mbstate_t *_RESTRICT_KYWD, boolean_t);
 static size_t	_EUC_KR_mbrtowc(wchar_t *_RESTRICT_KYWD,
 		    const char *_RESTRICT_KYWD,
-		    size_t, mbstate_t *_RESTRICT_KYWD);
+		    size_t, mbstate_t *_RESTRICT_KYWD, boolean_t);
 static size_t	_EUC_TW_mbrtowc(wchar_t *_RESTRICT_KYWD,
 		    const char *_RESTRICT_KYWD,
-		    size_t, mbstate_t *_RESTRICT_KYWD);
+		    size_t, mbstate_t *_RESTRICT_KYWD, boolean_t);
 
 static size_t	_EUC_CN_wcrtomb(char *_RESTRICT_KYWD, wchar_t,
 		    mbstate_t *_RESTRICT_KYWD);
@@ -131,9 +132,9 @@ _EUC_CN_init(struct lc_ctype *lct)
 
 static size_t
 _EUC_CN_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
-    size_t n, mbstate_t *_RESTRICT_KYWD ps)
+    size_t n, mbstate_t *_RESTRICT_KYWD ps, boolean_t zero)
 {
-	return (_EUC_mbrtowc_impl(pwc, s, n, ps, SS2, 4, 0, 0));
+	return (_EUC_mbrtowc_impl(pwc, s, n, ps, SS2, 4, 0, 0, zero));
 }
 
 static size_t
@@ -176,9 +177,9 @@ _EUC_KR_init(struct lc_ctype *lct)
 
 static size_t
 _EUC_KR_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
-    size_t n, mbstate_t *_RESTRICT_KYWD ps)
+    size_t n, mbstate_t *_RESTRICT_KYWD ps, boolean_t zero)
 {
-	return (_EUC_mbrtowc_impl(pwc, s, n, ps, 0, 0, 0, 0));
+	return (_EUC_mbrtowc_impl(pwc, s, n, ps, 0, 0, 0, 0, zero));
 }
 
 static size_t
@@ -221,9 +222,9 @@ _EUC_JP_init(struct lc_ctype *lct)
 
 static size_t
 _EUC_JP_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
-    size_t n, mbstate_t *_RESTRICT_KYWD ps)
+    size_t n, mbstate_t *_RESTRICT_KYWD ps, boolean_t zero)
 {
-	return (_EUC_mbrtowc_impl(pwc, s, n, ps, SS2, 2, SS3, 3));
+	return (_EUC_mbrtowc_impl(pwc, s, n, ps, SS2, 2, SS3, 3, zero));
 }
 
 static size_t
@@ -266,9 +267,9 @@ _EUC_TW_init(struct lc_ctype *lct)
 
 static size_t
 _EUC_TW_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
-    size_t n, mbstate_t *_RESTRICT_KYWD ps)
+    size_t n, mbstate_t *_RESTRICT_KYWD ps, boolean_t zero)
 {
-	return (_EUC_mbrtowc_impl(pwc, s, n, ps, SS2, 4, 0, 0));
+	return (_EUC_mbrtowc_impl(pwc, s, n, ps, SS2, 4, 0, 0, zero));
 }
 
 static size_t
@@ -300,7 +301,8 @@ _EUC_TW_wcsnrtombs(char *_RESTRICT_KYWD dst, const wchar_t **_RESTRICT_KYWD src,
 static size_t
 _EUC_mbrtowc_impl(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
     size_t n, mbstate_t *_RESTRICT_KYWD ps,
-    uint8_t cs2, uint8_t cs2width, uint8_t cs3, uint8_t cs3width)
+    uint8_t cs2, uint8_t cs2width, uint8_t cs3, uint8_t cs3width,
+    boolean_t zero)
 {
 	_EucState *es;
 	int i, want;
@@ -329,7 +331,11 @@ _EUC_mbrtowc_impl(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
 		if (((ch = (unsigned char)*s) & 0x80) == 0) {
 			if (pwc != NULL)
 				*pwc = ch;
-			return (ch != '\0' ? 1 : 0);
+			if (zero || ch != '\0') {
+				return (1);
+			} else {
+				return (0);
+			}
 		}
 
 		if (ch >= 0xa1) {
@@ -367,7 +373,11 @@ _EUC_mbrtowc_impl(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
 	if (pwc != NULL)
 		*pwc = wc;
 	es->want = 0;
-	return (wc == L'\0' ? 0 : want);
+	if (zero || wc != L'\0') {
+		return (want);
+	} else {
+		return (0);
+	}
 }
 
 static size_t
diff --git a/usr/src/lib/libc/port/locale/fgetwc.c b/usr/src/lib/libc/port/locale/fgetwc.c
index 282b65cebc..254a785891 100644
--- a/usr/src/lib/libc/port/locale/fgetwc.c
+++ b/usr/src/lib/libc/port/locale/fgetwc.c
@@ -66,7 +66,7 @@ _fgetwc_unlocked_l(FILE *fp, locale_t loc)
 	}
 	do {
 		char	x = (char)c;
-		nconv = lct->lc_mbrtowc(&wc, &x, 1, statep);
+		nconv = lct->lc_mbrtowc(&wc, &x, 1, statep, B_FALSE);
 		if (nconv == (size_t)-1) {
 			break;
 		} else if (nconv == (size_t)-2) {
diff --git a/usr/src/lib/libc/port/locale/gb18030.c b/usr/src/lib/libc/port/locale/gb18030.c
index 232daade50..36c48c5cc5 100644
--- a/usr/src/lib/libc/port/locale/gb18030.c
+++ b/usr/src/lib/libc/port/locale/gb18030.c
@@ -44,7 +44,7 @@
 
 static size_t	_GB18030_mbrtowc(wchar_t *_RESTRICT_KYWD,
 		    const char *_RESTRICT_KYWD,
-		    size_t, mbstate_t *_RESTRICT_KYWD);
+		    size_t, mbstate_t *_RESTRICT_KYWD, boolean_t);
 static int	_GB18030_mbsinit(const mbstate_t *);
 static size_t	_GB18030_wcrtomb(char *_RESTRICT_KYWD, wchar_t,
 		    mbstate_t *_RESTRICT_KYWD);
@@ -83,7 +83,7 @@ _GB18030_mbsinit(const mbstate_t *ps)
 
 static size_t
 _GB18030_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
-    size_t n, mbstate_t *_RESTRICT_KYWD ps)
+    size_t n, mbstate_t *_RESTRICT_KYWD ps, boolean_t zero)
 {
 	_GB18030State *gs;
 	wchar_t wch;
@@ -159,7 +159,11 @@ _GB18030_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
 	if (pwc != NULL)
 		*pwc = wch;
 	gs->count = 0;
-	return (wch == L'\0' ? 0 : len - ocount);
+	if (zero || wch != L'\0') {
+		return (len - ocount);
+	} else {
+		return (0);
+	}
 ilseq:
 	errno = EILSEQ;
 	return ((size_t)-1);
diff --git a/usr/src/lib/libc/port/locale/gb2312.c b/usr/src/lib/libc/port/locale/gb2312.c
index a25af781b4..bfb6c0177b 100644
--- a/usr/src/lib/libc/port/locale/gb2312.c
+++ b/usr/src/lib/libc/port/locale/gb2312.c
@@ -39,7 +39,7 @@
 
 static size_t	_GB2312_mbrtowc(wchar_t *_RESTRICT_KYWD,
 		    const char *_RESTRICT_KYWD,
-		    size_t, mbstate_t *_RESTRICT_KYWD);
+		    size_t, mbstate_t *_RESTRICT_KYWD, boolean_t);
 static int	_GB2312_mbsinit(const mbstate_t *);
 static size_t	_GB2312_wcrtomb(char *_RESTRICT_KYWD, wchar_t,
 		    mbstate_t *_RESTRICT_KYWD);
@@ -101,7 +101,7 @@ _GB2312_check(const char *str, size_t n)
 
 static size_t
 _GB2312_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
-    size_t n, mbstate_t *_RESTRICT_KYWD ps)
+    size_t n, mbstate_t *_RESTRICT_KYWD ps, boolean_t zero)
 {
 	_GB2312State *gs;
 	wchar_t wc;
@@ -137,7 +137,11 @@ _GB2312_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
 	if (pwc != NULL)
 		*pwc = wc;
 	gs->count = 0;
-	return (wc == L'\0' ? 0 : len - ocount);
+	if (zero || wc != L'\0') {
+		return (len - ocount);
+	} else {
+		return (0);
+	}
 }
 
 static size_t
diff --git a/usr/src/lib/libc/port/locale/gbk.c b/usr/src/lib/libc/port/locale/gbk.c
index 5c94b9954b..f422ce8fb5 100644
--- a/usr/src/lib/libc/port/locale/gbk.c
+++ b/usr/src/lib/libc/port/locale/gbk.c
@@ -44,7 +44,7 @@
 
 static size_t	_GBK_mbrtowc(wchar_t *_RESTRICT_KYWD,
 		    const char *_RESTRICT_KYWD,
-		    size_t, mbstate_t *_RESTRICT_KYWD);
+		    size_t, mbstate_t *_RESTRICT_KYWD, boolean_t);
 static int	_GBK_mbsinit(const mbstate_t *);
 static size_t	_GBK_wcrtomb(char *_RESTRICT_KYWD, wchar_t,
 		    mbstate_t *_RESTRICT_KYWD);
@@ -88,7 +88,7 @@ _gbk_check(uint_t c)
 
 static size_t
 _GBK_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
-    size_t n, mbstate_t *_RESTRICT_KYWD ps)
+    size_t n, mbstate_t *_RESTRICT_KYWD ps, boolean_t zero)
 {
 	_GBKState *gs;
 	wchar_t wc;
@@ -143,7 +143,11 @@ _GBK_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
 	} else {
 		if (pwc != NULL)
 			*pwc = wc;
-		return (wc == L'\0' ? 0 : 1);
+		if (zero || wc != L'\0') {
+			return (1);
+		} else {
+			return (0);
+		}
 	}
 }
 
diff --git a/usr/src/lib/libc/port/locale/lctype.h b/usr/src/lib/libc/port/locale/lctype.h
index 25652da1d4..0a9e1335bf 100644
--- a/usr/src/lib/libc/port/locale/lctype.h
+++ b/usr/src/lib/libc/port/locale/lctype.h
@@ -17,6 +17,7 @@
 #define	_LCTYPE_H_
 
 #include <wchar.h>
+#include <sys/types.h>
 
 /* private LC_CTYPE related structures */
 
@@ -24,7 +25,8 @@
 struct lc_ctype {
 
 	size_t (*lc_mbrtowc)(wchar_t *_RESTRICT_KYWD,
-	    const char *_RESTRICT_KYWD, size_t, mbstate_t *_RESTRICT_KYWD);
+	    const char *_RESTRICT_KYWD, size_t, mbstate_t *_RESTRICT_KYWD,
+	    boolean_t);
 
 	int (*lc_mbsinit)(const mbstate_t *);
 
@@ -51,7 +53,7 @@ struct lc_ctype {
  * Default implementation (C locale, i.e. ASCII).
  */
 size_t	__mbrtowc_ascii(wchar_t *_RESTRICT_KYWD,
-    const char *_RESTRICT_KYWD, size_t, mbstate_t *_RESTRICT_KYWD);
+    const char *_RESTRICT_KYWD, size_t, mbstate_t *_RESTRICT_KYWD, boolean_t);
 int	__mbsinit_ascii(const mbstate_t *);
 size_t	__mbsnrtowcs_ascii(wchar_t *_RESTRICT_KYWD dst,
     const char **_RESTRICT_KYWD src, size_t nms, size_t len,
diff --git a/usr/src/lib/libc/port/locale/mblocal.h b/usr/src/lib/libc/port/locale/mblocal.h
index 4412bec7cb..3d958e364e 100644
--- a/usr/src/lib/libc/port/locale/mblocal.h
+++ b/usr/src/lib/libc/port/locale/mblocal.h
@@ -48,7 +48,7 @@ void	_BIG5_init(struct lc_ctype *);
 void	_MSKanji_init(struct lc_ctype *);
 
 typedef size_t (*mbrtowc_pfn_t)(wchar_t *_RESTRICT_KYWD,
-    const char *_RESTRICT_KYWD, size_t, mbstate_t *_RESTRICT_KYWD);
+    const char *_RESTRICT_KYWD, size_t, mbstate_t *_RESTRICT_KYWD, boolean_t);
 typedef size_t (*wcrtomb_pfn_t)(char *_RESTRICT_KYWD, wchar_t,
     mbstate_t *_RESTRICT_KYWD);
 size_t __mbsnrtowcs_std(wchar_t *_RESTRICT_KYWD, const char **_RESTRICT_KYWD,
diff --git a/usr/src/lib/libc/port/locale/mbrtowc.c b/usr/src/lib/libc/port/locale/mbrtowc.c
index cd1e7b064f..c0934c88d1 100644
--- a/usr/src/lib/libc/port/locale/mbrtowc.c
+++ b/usr/src/lib/libc/port/locale/mbrtowc.c
@@ -40,7 +40,7 @@ mbrtowc_l(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
 
 	if (ps == NULL)
 		ps = &mbs;
-	return (loc->ctype->lc_mbrtowc(pwc, s, n, ps));
+	return (loc->ctype->lc_mbrtowc(pwc, s, n, ps, B_FALSE));
 }
 
 size_t
@@ -49,3 +49,21 @@ mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
 {
 	return (mbrtowc_l(pwc, s, n, ps, uselocale(NULL)));
 }
+
+size_t
+mbrtowc_nz_l(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
+    size_t n, mbstate_t *_RESTRICT_KYWD ps, locale_t loc)
+{
+	static mbstate_t mbs;
+
+	if (ps == NULL)
+		ps = &mbs;
+	return (loc->ctype->lc_mbrtowc(pwc, s, n, ps, B_TRUE));
+}
+
+size_t
+mbrtowc_nz(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
+    size_t n, mbstate_t *_RESTRICT_KYWD ps)
+{
+	return (mbrtowc_nz_l(pwc, s, n, ps, uselocale(NULL)));
+}
diff --git a/usr/src/lib/libc/port/locale/mbsnrtowcs.c b/usr/src/lib/libc/port/locale/mbsnrtowcs.c
index 533b7776e0..6411cfcb74 100644
--- a/usr/src/lib/libc/port/locale/mbsnrtowcs.c
+++ b/usr/src/lib/libc/port/locale/mbsnrtowcs.c
@@ -68,11 +68,13 @@ __mbsnrtowcs_std(wchar_t *_RESTRICT_KYWD dst, const char **_RESTRICT_KYWD src,
 
 	if (dst == NULL) {
 		for (;;) {
-			if ((nb = pmbrtowc(&wc, s, nms, ps)) == (size_t)-1)
+			if ((nb = pmbrtowc(&wc, s, nms, ps, B_FALSE)) ==
+			    (size_t)-1) {
 				/* Invalid sequence - mbrtowc() sets errno. */
 				return ((size_t)-1);
-			else if (nb == 0 || nb == (size_t)-2)
+			} else if (nb == 0 || nb == (size_t)-2) {
 				return (nchr);
+			}
 			s += nb;
 			nms -= nb;
 			nchr++;
@@ -81,7 +83,7 @@ __mbsnrtowcs_std(wchar_t *_RESTRICT_KYWD dst, const char **_RESTRICT_KYWD src,
 	}
 
 	while (len-- > 0) {
-		if ((nb = pmbrtowc(dst, s, nms, ps)) == (size_t)-1) {
+		if ((nb = pmbrtowc(dst, s, nms, ps, B_FALSE)) == (size_t)-1) {
 			*src = s;
 			return ((size_t)-1);
 		} else if (nb == (size_t)-2) {
diff --git a/usr/src/lib/libc/port/locale/mskanji.c b/usr/src/lib/libc/port/locale/mskanji.c
index 5d9b899aae..69955e5afa 100644
--- a/usr/src/lib/libc/port/locale/mskanji.c
+++ b/usr/src/lib/libc/port/locale/mskanji.c
@@ -46,7 +46,7 @@
 
 static size_t	_MSKanji_mbrtowc(wchar_t *_RESTRICT_KYWD,
 		    const char *_RESTRICT_KYWD,
-		    size_t, mbstate_t *_RESTRICT_KYWD);
+		    size_t, mbstate_t *_RESTRICT_KYWD, boolean_t);
 static int	_MSKanji_mbsinit(const mbstate_t *);
 static size_t	_MSKanji_wcrtomb(char *_RESTRICT_KYWD, wchar_t,
 		    mbstate_t *_RESTRICT_KYWD);
@@ -83,7 +83,7 @@ _MSKanji_mbsinit(const mbstate_t *ps)
 
 static size_t
 _MSKanji_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
-    size_t n, mbstate_t *_RESTRICT_KYWD ps)
+    size_t n, mbstate_t *_RESTRICT_KYWD ps, boolean_t zero)
 {
 	_MSKanjiState *ms;
 	wchar_t wc;
@@ -135,7 +135,11 @@ _MSKanji_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
 	} else {
 		if (pwc != NULL)
 			*pwc = wc;
-		return (wc == L'\0' ? 0 : 1);
+		if (zero || wc != L'\0') {
+			return (1);
+		} else {
+			return (0);
+		}
 	}
 }
 
diff --git a/usr/src/lib/libc/port/locale/none.c b/usr/src/lib/libc/port/locale/none.c
index 0511563cb1..5eefc746aa 100644
--- a/usr/src/lib/libc/port/locale/none.c
+++ b/usr/src/lib/libc/port/locale/none.c
@@ -73,7 +73,7 @@ __mbsinit_ascii(const mbstate_t *unused)
 
 size_t
 __mbrtowc_ascii(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
-    size_t n, mbstate_t *_RESTRICT_KYWD unused)
+    size_t n, mbstate_t *_RESTRICT_KYWD unused, boolean_t zero)
 {
 	_NOTE(ARGUNUSED(unused));
 
@@ -85,7 +85,11 @@ __mbrtowc_ascii(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
 		return ((size_t)-2);
 	if (pwc != NULL)
 		*pwc = (unsigned char)*s;
-	return (*s == '\0' ? 0 : 1);
+	if (zero || *s != '\0') {
+		return (1);
+	} else {
+		return (0);
+	}
 }
 
 size_t
diff --git a/usr/src/lib/libc/port/locale/utf8.c b/usr/src/lib/libc/port/locale/utf8.c
index e919b9dd4a..a6e037d94e 100644
--- a/usr/src/lib/libc/port/locale/utf8.c
+++ b/usr/src/lib/libc/port/locale/utf8.c
@@ -37,7 +37,7 @@
 
 static size_t	_UTF8_mbrtowc(wchar_t *_RESTRICT_KYWD,
 		    const char *_RESTRICT_KYWD,
-		    size_t, mbstate_t *_RESTRICT_KYWD);
+		    size_t, mbstate_t *_RESTRICT_KYWD, boolean_t);
 static int	_UTF8_mbsinit(const mbstate_t *);
 static size_t	_UTF8_mbsnrtowcs(wchar_t *_RESTRICT_KYWD,
 		    const char **_RESTRICT_KYWD, size_t, size_t,
@@ -75,7 +75,7 @@ _UTF8_mbsinit(const mbstate_t *ps)
 
 static size_t
 _UTF8_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
-    size_t n, mbstate_t *_RESTRICT_KYWD ps)
+    size_t n, mbstate_t *_RESTRICT_KYWD ps, boolean_t zero)
 {
 	_UTF8State *us;
 	int ch, i, mask, want;
@@ -116,7 +116,11 @@ _UTF8_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
 			/* Fast path for plain ASCII characters. */
 			if (pwc != NULL)
 				*pwc = ch;
-			return (ch != '\0' ? 1 : 0);
+			if (zero || ch != '\0') {
+				return (1);
+			} else {
+				return (0);
+			}
 		}
 		if ((ch & 0xe0) == 0xc0) {
 			mask = 0x1f;
@@ -192,7 +196,11 @@ _UTF8_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
 	if (pwc != NULL)
 		*pwc = wch;
 	us->want = 0;
-	return (wch == L'\0' ? 0 : want);
+	if (zero || wch != L'\0') {
+		return (want);
+	} else {
+		return (0);
+	}
 }
 
 static size_t
@@ -221,18 +229,19 @@ _UTF8_mbsnrtowcs(wchar_t *_RESTRICT_KYWD dst, const char **_RESTRICT_KYWD src,
 			return ((size_t)-1);
 		}
 		for (;;) {
-			if (nms > 0 && (signed char)*s > 0)
+			if (nms > 0 && (signed char)*s > 0) {
 				/*
 				 * Fast path for plain ASCII characters
 				 * excluding NUL.
 				 */
 				nb = 1;
-			else if ((nb = _UTF8_mbrtowc(&wc, s, nms, ps)) ==
-			    (size_t)-1)
+			} else if ((nb = _UTF8_mbrtowc(&wc, s, nms, ps,
+			    B_FALSE)) == (size_t)-1) {
 				/* Invalid sequence - mbrtowc() sets errno. */
 				return ((size_t)-1);
-			else if (nb == 0 || nb == (size_t)-2)
+			} else if (nb == 0 || nb == (size_t)-2) {
 				return (nchr);
+			}
 			s += nb;
 			nms -= nb;
 			nchr++;
@@ -257,7 +266,7 @@ _UTF8_mbsnrtowcs(wchar_t *_RESTRICT_KYWD dst, const char **_RESTRICT_KYWD src,
 			 */
 			*dst = (wchar_t)*s;
 			nb = 1;
-		} else if ((nb = _UTF8_mbrtowc(dst, s, nms, ps)) ==
+		} else if ((nb = _UTF8_mbrtowc(dst, s, nms, ps, B_FALSE)) ==
 		    (size_t)-1) {
 			*src = s;
 			return ((size_t)-1);