s390x/tcg: MVCLU/MVCLE: Process max 4k bytes at a time

Let's stay within single pages.

... and indicate cc=3 in case there is work remaining. Keep unicode
padding simple.

While reworking, properly wrap the addresses.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: David Hildenbrand <david@redhat.com>

target/s390x/mem_helper.c

@@ -768,8 +768,8 @@ static inline uint32_t do_mvcl(CPUS390XState *env,
                                uint64_t *src, uint64_t *srclen,
                                uint16_t pad, int wordsize, uintptr_t ra)
 {
-    uint64_t len = MIN(*srclen, *destlen);
+    int len = MIN(*destlen, -(*dest | TARGET_PAGE_MASK));
-    uint32_t cc;
+    int i, cc;
 
     if (*destlen == *srclen) {
         cc = 0;
@@ -779,32 +779,40 @@ static inline uint32_t do_mvcl(CPUS390XState *env,
         cc = 2;
     }
 
-    /* Copy the src array */
+    if (!*destlen) {
-    fast_memmove(env, *dest, *src, len, ra);
+        return cc;
-    *src += len;
+    }
-    *srclen -= len;
-    *dest += len;
-    *destlen -= len;
 
-    /* Pad the remaining area */
+    /*
-    if (wordsize == 1) {
+     * Only perform one type of type of operation (move/pad) at a time.
-        fast_memset(env, *dest, pad, *destlen, ra);
+     * Stay within single pages.
-        *dest += *destlen;
+     */
-        *destlen = 0;
+    if (*srclen) {
+        /* Copy the src array */
+        len = MIN(MIN(*srclen, -(*src | TARGET_PAGE_MASK)), len);
+        *destlen -= len;
+        *srclen -= len;
+        fast_memmove(env, *dest, *src, len, ra);
+        *src = wrap_address(env, *src + len);
+        *dest = wrap_address(env, *dest + len);
+    } else if (wordsize == 1) {
+        /* Pad the remaining area */
+        *destlen -= len;
+        fast_memset(env, *dest, pad, len, ra);
+        *dest = wrap_address(env, *dest + len);
     } else {
-        /* If remaining length is odd, pad with odd byte first.  */
+        /* The remaining length selects the padding byte. */
-        if (*destlen & 1) {
+        for (i = 0; i < len; (*destlen)--, i++) {
-            cpu_stb_data_ra(env, *dest, pad & 0xff, ra);
+            if (*destlen & 1) {
-            *dest += 1;
+                cpu_stb_data_ra(env, *dest, pad, ra);
-            *destlen -= 1;
+            } else {
-        }
+                cpu_stb_data_ra(env, *dest, pad >> 8, ra);
-        /* The remaining length is even, pad using words.  */
+            }
-        for (; *destlen; *dest += 2, *destlen -= 2) {
+            *dest = wrap_address(env, *dest + 1);
-            cpu_stw_data_ra(env, *dest, pad, ra);
         }
     }
 
-    return cc;
+    return *destlen ? 3 : cc;
 }
 
 /* move long */