1 files changed, 108 insertions, 22 deletions

diff --git a/usr/src/lib/libdemangle/common/demangle.c b/usr/src/lib/libdemangle/common/demangle.c
index e827fd8cec..4f8e9ad678 100644
--- a/usr/src/lib/libdemangle/common/demangle.c
+++ b/usr/src/lib/libdemangle/common/demangle.c
@@ -11,13 +11,17 @@
 
 /*
  * Copyright 2018 Jason King
+ * Copyright 2019, Joyent, Inc.
  */
 
 #include <stdlib.h>
+#include <stdio.h>
 #include <string.h>
 #include <errno.h>
 #include <pthread.h>
+#include <sys/ctype.h>
 #include <sys/debug.h>
+#include <stdarg.h>
 #include "demangle-sys.h"
 #include "demangle_int.h"
 
@@ -25,31 +29,63 @@
 
 static pthread_once_t debug_once = PTHREAD_ONCE_INIT;
 volatile boolean_t demangle_debug;
+FILE *debugf = stderr;
+
+static const char *
+langstr(sysdem_lang_t lang)
+{
+	switch (lang) {
+	case SYSDEM_LANG_AUTO:
+		return ("auto");
+	case SYSDEM_LANG_CPP:
+		return ("c++");
+	case SYSDEM_LANG_RUST:
+		return ("rust");
+	default:
+		return ("invalid");
+	}
+}
 
 static sysdem_lang_t
-detect_lang(const char *str)
+detect_lang(const char *str, size_t n)
 {
-	size_t n = strlen(str);
+	const char *p = str;
+	size_t len;
 
 	if (n < 3 || str[0] != '_')
 		return (SYSDEM_LANG_AUTO);
 
-	switch (str[1]) {
-	case 'Z':
+	/*
+	 * Check for ^_Z or ^__Z
+	 */
+	p = str + 1;
+	if (*p == '_') {
+		p++;
+	}
+
+	if (*p != 'Z')
+		return (SYSDEM_LANG_AUTO);
+
+	/*
+	 * Sadly, rust currently uses the same prefix as C++, however
+	 * demangling rust as a C++ mangled name yields less than desirable
+	 * results.  However rust names end with a hash.  We use that to
+	 * attempt to disambiguate
+	 */
+
+	/* Find 'h'<hexdigit>+E$ */
+	if ((p = strrchr(p, 'h')) == NULL)
 		return (SYSDEM_LANG_CPP);
 
-	case '_':
-		break;
+	if ((len = strspn(p + 1, "0123456789abcdef")) == 0)
+		return (SYSDEM_LANG_CPP);
 
-	default:
-		return (SYSDEM_LANG_AUTO);
-	}
+	p += len + 1;
 
-	/* why they use ___Z sometimes is puzzling... *sigh* */
-	if (str[2] == '_' && str[3] == 'Z')
+	if (p[0] != 'E' || p[1] != '\0')
 		return (SYSDEM_LANG_CPP);
 
-	return (SYSDEM_LANG_AUTO);
+	return (SYSDEM_LANG_RUST);
 }
 
 static void
@@ -62,26 +98,76 @@ check_debug(void)
 char *
 sysdemangle(const char *str, sysdem_lang_t lang, sysdem_ops_t *ops)
 {
+	/*
+	 * While the language specific demangler code can handle non-NUL
+	 * terminated strings, we currently don't expose this to consumers.
+	 * Consumers should still pass in a NUL-terminated string.
+	 */
+	size_t slen;
+
 	VERIFY0(pthread_once(&debug_once, check_debug));
 
+	DEMDEBUG("name = '%s'", (str == NULL) ? "(NULL)" : str);
+	DEMDEBUG("lang = %s (%d)", langstr(lang), lang);
+
+	if (str == NULL) {
+		errno = EINVAL;
+		return (NULL);
+	}
+
+	slen = strlen(str);
+
+	switch (lang) {
+		case SYSDEM_LANG_AUTO:
+		case SYSDEM_LANG_CPP:
+		case SYSDEM_LANG_RUST:
+			break;
+		default:
+			errno = EINVAL;
+			return (NULL);
+	}
+
 	if (ops == NULL)
 		ops = sysdem_ops_default;
 
 	if (lang == SYSDEM_LANG_AUTO) {
-		lang = detect_lang(str);
-		if (lang == SYSDEM_LANG_AUTO) {
-			errno = ENOTSUP;
-			return (NULL);
-		}
+		lang = detect_lang(str, slen);
+		if (lang != SYSDEM_LANG_AUTO)
+			DEMDEBUG("detected language is %s", langstr(lang));
 	}
 
 	switch (lang) {
-	case SYSDEM_LANG_AUTO:
-		break;
 	case SYSDEM_LANG_CPP:
-		return (cpp_demangle(str, ops));
+		return (cpp_demangle(str, slen, ops));
+	case SYSDEM_LANG_RUST:
+		return (rust_demangle(str, slen, ops));
+	case SYSDEM_LANG_AUTO:
+		DEMDEBUG("could not detect language");
+		errno = ENOTSUP;
+		return (NULL);
+	default:
+		/*
+		 * This can't happen unless there's a bug with detect_lang,
+		 * but gcc doesn't know that.
+		 */
+		errno = EINVAL;
+		return (NULL);
 	}
+}
 
-	errno = ENOTSUP;
-	return (NULL);
+int
+demdebug(const char *fmt, ...)
+{
+	va_list ap;
+
+	flockfile(debugf);
+	(void) fprintf(debugf, "LIBDEMANGLE: ");
+	va_start(ap, fmt);
+	(void) vfprintf(debugf, fmt, ap);
+	(void) fputc('\n', debugf);
+	(void) fflush(debugf);
+	va_end(ap);
+	funlockfile(debugf);
+
+	return (0);
 }