fix(libc): resolve gettext catalogue fallbacks

--- a/usr/src/lib/libc/port/i18n/gettext_real.c
+++ b/usr/src/lib/libc/port/i18n/gettext_real.c
@@ -263,24 +263,98 @@
 		mp->status = 0;
 	}

-	/*
-	 * Finally, handle a single binding
-	 */
-#ifdef GETTEXT_DEBUG
-	*mp->msgfile = '\0';
-#endif
-	if (mk_msgfile(mp) == NULL) {
-		DFLTMSG(result, msgid1, msgid2, n, plural);
-		return (result);
-	}
-
-	result = handle_mo(mp);
+	result = handle_locale(mp);
 	if (result) {
 		return (result);
 	}
 	DFLTMSG(result, msgid1, msgid2, n, plural);
 	return (result);
 } /* _real_gettext_u */
+
+/* Search catalogue names from most to least specific. */
+char *
+handle_locale(struct msg_pack *mp)
+{
+	const char *saved = mp->locale;
+	char name[MAXPATHLEN], normalized[MAXPATHLEN], path[MAXPATHLEN];
+	char *territory, *codeset, *modifier, *p, *out, *result;
+	unsigned int parts = 0;
+	int mask, length;
+	int digits = 1;
+
+	if (mk_msgfile(mp) == NULL)
+		return (NULL);
+	result = handle_mo(mp);
+	if ((mp->status & (ST_GNU_MSG_FOUND | ST_SUN_MO_FOUND)) != 0)
+		return (result);
+	if (strlen(saved) >= sizeof (name) || strchr(saved, '/') != NULL)
+		return (result);
+	(void) strcpy(name, saved);
+
+	modifier = strchr(name, '@');
+	if (modifier != NULL) {
+		*modifier++ = '\0';
+		parts |= 8;
+	}
+	codeset = strchr(name, '.');
+	if (codeset != NULL) {
+		*codeset++ = '\0';
+		parts |= 2;
+	}
+	territory = strchr(name, '_');
+	if (territory != NULL) {
+		*territory++ = '\0';
+		parts |= 4;
+	}
+	if (strcmp(name, "C") == 0 || strcmp(name, "POSIX") == 0)
+		return (result);
+
+	/* Normalize codesets without depending on the current locale. */
+	out = normalized;
+	if (codeset != NULL) {
+		for (p = codeset; *p != '\0'; p++) {
+			if (*p >= '0' && *p <= '9') {
+				*out++ = *p;
+			} else if ((*p >= 'A' && *p <= 'Z') ||
+			    (*p >= 'a' && *p <= 'z')) {
+				*out++ = (*p >= 'A' && *p <= 'Z') ?
+				    *p + ('a' - 'A') : *p;
+				digits = 0;
+			}
+		}
+	}
+	*out = '\0';
+	if (out != normalized) {
+		if (digits && strlen(normalized) + 4 <= sizeof (normalized)) {
+			(void) memmove(normalized + 3, normalized,
+			    strlen(normalized) + 1);
+			(void) memcpy(normalized, "iso", 3);
+		}
+		if (strcmp(normalized, codeset) != 0)
+			parts |= 1;
+	}
+
+	for (mask = parts; mask >= 0; mask--) {
+		if ((mask & ~parts) != 0 || (mask & 3) == 3)
+			continue;
+		length = snprintf(path, sizeof (path), "%s%s%s%s%s%s%s",
+		    name, (mask & 4) ? "_" : "", (mask & 4) ? territory : "",
+		    (mask & 3) ? "." : "",
+		    (mask & 2) ? codeset : ((mask & 1) ? normalized : ""),
+		    (mask & 8) ? "@" : "", (mask & 8) ? modifier : "");
+		if (length < 0 || (size_t)length >= sizeof (path) ||
+		    strcmp(path, saved) == 0)
+			continue;
+		mp->locale = path;
+		if (mk_msgfile(mp) == NULL)
+			continue;
+		result = handle_mo(mp);
+		if ((mp->status & (ST_GNU_MSG_FOUND | ST_SUN_MO_FOUND)) != 0)
+			break;
+	}
+	mp->locale = saved;
+	return (result);
+}

 #define	ALLFREE	\
 	free_all(nlstmp, nnp, pathname, ppaths, lang)
--- a/usr/src/lib/libc/port/i18n/gettext_gnu.c
+++ b/usr/src/lib/libc/port/i18n/gettext_gnu.c
@@ -319,12 +319,7 @@
 #ifdef GETTEXT_DEBUG
 		*mp->msgfile = '\0';
 #endif
-		if (mk_msgfile(mp) == NULL) {
-			/* illegal locale name */
-			continue;
-		}
-
-		result = handle_mo(mp);
+		result = handle_locale(mp);
 		if (mp->status & ST_GNU_MSG_FOUND)
 			return (result);

--- a/usr/src/lib/libc/port/i18n/gettext.h
+++ b/usr/src/lib/libc/port/i18n/gettext.h
@@ -191,6 +191,7 @@
 extern char	*_real_bindtextdomain_u(const char *, const char *, int);
 extern char	*_real_gettext_u(const char *, const char *,
     const char *, unsigned long int, int, int, locale_t);
+extern char	*handle_locale(struct msg_pack *);
 extern char	*handle_mo(struct msg_pack *);

 extern int	gnu_setmsg(Msg_node *, char *, size_t);
--- a/usr/src/man/man3c/gettext.3c
+++ b/usr/src/man/man3c/gettext.3c
@@ -128,6 +128,11 @@
 \fBdcngettext()\fR.
 .sp
 .LP
+Catalogue lookup also tries less specific locale names by removing the codeset,
+territory, and modifier, in that order. Normalized codeset names are tried before
+removing the codeset. Each message domain is searched independently.
+.sp
+.LP
 For \fBgettext()\fR and \fBngettext()\fR, the domain used is set by the last
 valid call to \fBtextdomain()\fR. If a valid call to \fBtextdomain()\fR has not
 been made, the default domain  (called \fBmessages\fR) is used.