More EBCDIC stuff:

 - Loose the extra level of function on ASCII.
 - spotted a chr(0) issue in sv.c
 - re-work of UTF-X tr/// ranges to work in Unicode
   space. Still issues with the "0xff is illegal UTF-8" hack.
 - Yet another ad. hoc. utf8 'upgrade' in op.c recoded
   (why do it once when you can do it all over the place :-(
 - Enable HINTS_UTF8 on EBCDIC - then ignore it in toke.c,
   need utf8.pm for swashes.
 - Simplified and commented scan_const() in toke.c
Still something wrong regexp and tr (swashes?).

p4raw-id: //depot/perlio@9267

1 files changed, 3 insertions, 2 deletions

diff --git a/sv.c b/sv.c
index 6d6d39e9f5..315c85b1a0 100644
--- a/sv.c
+++ b/sv.c
@@ -4733,8 +4733,9 @@ Perl_sv_pos_b2u(pTHX_ register SV *sv, I32* offsetp)
     len = 0;
     while (s < send) {
 	STRLEN n;
-        /* We can use low level directly here as we are not looking at the values */
-	if (utf8n_to_uvuni(s, UTF8SKIP(s), &n, 0)) {
+	/* Call utf8n_to_uvchr() to validate the sequence */
+	utf8n_to_uvchr(s, UTF8SKIP(s), &n, 0);
+	if (n > 0) {
 	    s += n;
 	    len++;
 	}