Make Perl_pregcomp() use SvUTF8() of the pattern, rather than the flag

bit in pmflags, to decide whether the pattern is UTF-8.

p4raw-id: //depot/perl@32851

1 files changed, 13 insertions, 1 deletions

diff --git a/regexec.c b/regexec.c
index 95bba2e270..8144f998f7 100644
--- a/regexec.c
+++ b/regexec.c
@@ -3755,7 +3755,19 @@ S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog)
 			U32 pm_flags = 0;
 			const I32 osize = PL_regsize;
 
-			if (DO_UTF8(ret)) pm_flags |= RXf_UTF8;
+			if (DO_UTF8(ret)) {
+			    assert (SvUTF8(ret));
+			} else if (SvUTF8(ret)) {
+			    /* Not doing UTF-8, despite what the SV says. Is
+			       this only if we're trapped in use 'bytes'?  */
+			    /* Make a copy of the octet sequence, but without
+			       the flag on, as the compiler now honours the
+			       SvUTF8 flag on ret.  */
+			    STRLEN len;
+			    const char *const p = SvPV(ret, len);
+			    ret = newSVpvn_flags(p, len, SVs_TEMP);
+			}
+			assert(!(pm_flags & RXf_UTF8));
 			rx = CALLREGCOMP(ret, pm_flags);
 			if (!(SvFLAGS(ret)
 			      & (SVs_TEMP | SVs_PADTMP | SVf_READONLY