preproc: actually use uleb128 format for encoding macros

The format wasn't actually uleb128 because it was accidentally
bigendian (like UTF-8). That is just begging for confusion in the
future, if and when the uleb128 code gets librarized.

Fix it now.

Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>

asm/preproc.c

@@ -1206,13 +1206,29 @@ static void inject_predefs(void)
     do_predef = false;
 }
 
+static uint64_t get_uleb128(const char **pp)
+{
+    const char *p = *pp;
+    unsigned int shcnt = 0;
+    uint8_t c;
+    uint64_t v = 0;
+
+    do {
+        c = *p++;
+        v += ((uint64_t)(c & 127)) << shcnt;
+        shcnt += 7;
+    } while (c & 128);
+
+    *pp = p;
+    return v;
+}
+
 static char *line_from_stdmac(void)
 {
     static const char *stdmacpos = NULL;
     static char *stdmacbuf = NULL;
     char *line;
     size_t len = 0;
-    uint8_t c;
 
     if (!stdmacpos || !*stdmacpos) {
         macros_t *next = *stdmaclist;
@@ -1234,11 +1250,7 @@ static char *line_from_stdmac(void)
     }
 
     /* Length encoded using uleb128 encoding */
-    while ((c = *stdmacpos++) >= 128) {
-        len += c - 128;
-        len <<= 7;
-    }
-    len += c;
+    len = get_uleb128(&stdmacpos);
 
     line = nasm_malloc(len + 1);
     memcpy(line, stdmacpos, len);

macros/macros.pl

@@ -70,22 +70,26 @@ sub print_data($$) {
 #
 # Prefix a string with its length in uleb128 encoding
 #
+sub uleb128($)
+{
+    my($n) = @_;
+    my $o = '';
+
+    do {
+	my $nn = $n >> 7;
+	$o .= pack('C', ($n & 127) | ($nn ? 128 : 0));
+	$n = $nn;
+    } while ($n);
+
+    return $o;
+}
+
 sub addstringlen($)
 {
     my($s) = @_;
 
     my $l = length($s);
-    return '' if (!$l);		# Drop empty line
-
-    my $lc = '';
-
-    for (my $shcnt = 0; $l >> $shcnt; $shcnt += 7) {
-	my $b = ($l >> $shcnt) & 127;
-	$b += 128 if ($shcnt);
-	$lc = pack('C', $b) . $lc;
-    }
-
-    return $lc.$s;
+    return $l ? uleb128($l).$s : '';
 }
 
 sub init_mac() {