Re-order the flags values for struct refcounted_he to allow the

possibility of storing 2 futher types. Add a void * to the union
as it won't increase the size, but may become useful.

p4raw-id: //depot/perl@30683

2 files changed, 31 insertions, 23 deletions

diff --git a/hv.c b/hv.c
index 4831ccd563..4266e8be9b 100644
--- a/hv.c
+++ b/hv.c
@@ -2581,11 +2581,13 @@ S_refcounted_he_value(pTHX_ const struct refcounted_he *he)
 	value = &PL_sv_placeholder;
 	break;
     case HVrhek_IV:
-	value = (he->refcounted_he_data[0] & HVrhek_UV)
-	    ? newSVuv(he->refcounted_he_val.refcounted_he_u_iv)
-	    : newSViv(he->refcounted_he_val.refcounted_he_u_uv);
+	value = newSViv(he->refcounted_he_val.refcounted_he_u_iv);
+	break;
+    case HVrhek_UV:
+	value = newSVuv(he->refcounted_he_val.refcounted_he_u_uv);
 	break;
     case HVrhek_PV:
+    case HVrhek_PV_UTF8:
 	/* Create a string SV that directly points to the bytes in our
 	   structure.  */
 	value = newSV_type(SVt_PV);
@@ -2595,7 +2597,7 @@ S_refcounted_he_value(pTHX_ const struct refcounted_he *he)
 	SvLEN_set(value, 0);
 	SvPOK_on(value);
 	SvREADONLY_on(value);
-	if (he->refcounted_he_data[0] & HVrhek_UTF8)
+	if ((he->refcounted_he_data[0] & HVrhek_typemask) == HVrhek_PV_UTF8)
 	    SvUTF8_on(value);
 	break;
     default:
@@ -2605,14 +2607,6 @@ S_refcounted_he_value(pTHX_ const struct refcounted_he *he)
     return value;
 }
 
-#ifdef USE_ITHREADS
-/* A big expression to find the key offset */
-#define REF_HE_KEY(chain) \
-	((((chain->refcounted_he_data[0] & HVrhek_typemask) == HVrhek_PV) \
-	    ? chain->refcounted_he_val.refcounted_he_u_len + 1 : 0)	  \
-	 + 1 + chain->refcounted_he_data)
-#endif
-
 /*
 =for apidoc refcounted_he_chain_2hv
 
@@ -2821,7 +2815,6 @@ Perl_refcounted_he_new(pTHX_ struct refcounted_he *const parent,
 	value_len = 0;
 	key_offset = 1;
     }
-    flags = value_type;
 
 #ifdef USE_ITHREADS
     he = (struct refcounted_he*)
@@ -2840,17 +2833,19 @@ Perl_refcounted_he_new(pTHX_ struct refcounted_he *const parent,
     if (value_type == HVrhek_PV) {
 	Copy(value_p, he->refcounted_he_data + 1, value_len + 1, char);
 	he->refcounted_he_val.refcounted_he_u_len = value_len;
-	if (SvUTF8(value)) {
-	    flags |= HVrhek_UTF8;
-	}
+	/* Do it this way so that the SvUTF8() test is after the SvPV, in case
+	   the value is overloaded, and doesn't yet have the UTF-8flag set.  */
+	if (SvUTF8(value))
+	    value_type = HVrhek_PV_UTF8;
     } else if (value_type == HVrhek_IV) {
 	if (SvUOK(value)) {
 	    he->refcounted_he_val.refcounted_he_u_uv = SvUVX(value);
-	    flags |= HVrhek_UV;
+	    value_type = HVrhek_UV;
 	} else {
 	    he->refcounted_he_val.refcounted_he_u_iv = SvIVX(value);
 	}
     }
+    flags = value_type;
 
     if (is_utf8) {
 	/* Hash keys are always stored normalised to (yes) ISO-8859-1.
diff --git a/hv.h b/hv.h
index 6e19c84213..d6a036109e 100644
--- a/hv.h
+++ b/hv.h
@@ -390,6 +390,8 @@ C<SV*>.
    between threads (because it hangs from OPs, which are shared), hence the
    alternate definition and mutex.  */
 
+struct refcounted_he;
+
 #ifdef PERL_CORE
 
 /* Gosh. This really isn't a good name any longer.  */
@@ -406,6 +408,7 @@ struct refcounted_he {
 	IV                refcounted_he_u_iv;
 	UV                refcounted_he_u_uv;
 	STRLEN            refcounted_he_u_len;
+	void		 *refcounted_he_u_ptr;	/* Might be useful in future */
     } refcounted_he_val;
     /* First byte is flags. Then NUL-terminated value. Then for ithreads,
        non-NUL terminated key.  */
@@ -414,12 +417,22 @@ struct refcounted_he {
 
 /* Flag bits are HVhek_UTF8, HVhek_WASUTF8, then */
 #define HVrhek_undef	0x00 /* Value is undef. */
-#define HVrhek_PV	0x10 /* Value is a string. */
-#define HVrhek_IV	0x20 /* Value is IV/UV. */
-#define HVrhek_delete	0x30 /* Value is placeholder - signifies delete. */
-#define HVrhek_typemask	0x30
-#define HVrhek_UTF8	0x40 /* string value is utf8. */
-#define HVrhek_UV	0x40 /* integer value is UV. */
+#define HVrhek_delete	0x10 /* Value is placeholder - signifies delete. */
+#define HVrhek_IV	0x20 /* Value is IV. */
+#define HVrhek_UV	0x30 /* Value is UV. */
+#define HVrhek_PV	0x40 /* Value is a (byte) string. */
+#define HVrhek_PV_UTF8	0x50 /* Value is a (utf8) string. */
+/* Two spare. As these have to live in the optree, you can't store anything
+   interpreter specific, such as SVs. :-( */
+#define HVrhek_typemask 0x70
+
+#ifdef USE_ITHREADS
+/* A big expression to find the key offset */
+#define REF_HE_KEY(chain)						\
+	((((chain->refcounted_he_data[0] & 0x60) == 0x40)		\
+	    ? chain->refcounted_he_val.refcounted_he_u_len + 1 : 0)	\
+	 + 1 + chain->refcounted_he_data)
+#endif
 
 #  ifdef USE_ITHREADS
 #    define HINTS_REFCNT_LOCK		MUTEX_LOCK(&PL_hints_mutex)