s390x/tcg: MVST: Fix storing back the addresses to registers

24 and 31-bit address space handling is wrong when it comes to storing
back the addresses to the register.

While at it, read gprs 0 implicitly.
Reviewed-by: NRichard Henderson <richard.henderson@linaro.org>
Signed-off-by: NDavid Hildenbrand <david@redhat.com>

target/s390x/helper.h

@@ -20,7 +20,7 @@ DEF_HELPER_FLAGS_4(mvn, TCG_CALL_NO_WG, void, env, i32, i64, i64)
 DEF_HELPER_FLAGS_4(mvo, TCG_CALL_NO_WG, void, env, i32, i64, i64)
 DEF_HELPER_FLAGS_4(mvpg, TCG_CALL_NO_WG, i32, env, i64, i64, i64)
 DEF_HELPER_FLAGS_4(mvz, TCG_CALL_NO_WG, void, env, i32, i64, i64)
-DEF_HELPER_4(mvst, i64, env, i64, i64, i64)
+DEF_HELPER_3(mvst, i32, env, i32, i32)
 DEF_HELPER_4(ex, void, env, i32, i64, i64)
 DEF_HELPER_FLAGS_4(stam, TCG_CALL_NO_WG, void, env, i32, i64, i32)
 DEF_HELPER_FLAGS_4(lam, TCG_CALL_NO_WG, void, env, i32, i64, i32)

target/s390x/insn-data.def

@@ -637,7 +637,7 @@
 /* MOVE PAGE */
     C(0xb254, MVPG,    RRE,   Z,   r1_o, r2_o, 0, 0, mvpg, 0)
 /* MOVE STRING */
-    C(0xb255, MVST,    RRE,   Z,   r1_o, r2_o, 0, 0, mvst, 0)
+    C(0xb255, MVST,    RRE,   Z,   0, 0, 0, 0, mvst, 0)
 /* MOVE WITH OPTIONAL SPECIFICATION */
     C(0xc800, MVCOS,   SSF,   MVCOS, la1, a2, 0, 0, mvcos, 0)
 /* MOVE WITH OFFSET */

target/s390x/mem_helper.c

@@ -700,18 +700,18 @@ uint32_t HELPER(mvpg)(CPUS390XState *env, uint64_t r0, uint64_t r1, uint64_t r2)
     return 0; /* data moved */
 }
 
-/* string copy (c is string terminator) */
-uint64_t HELPER(mvst)(CPUS390XState *env, uint64_t c, uint64_t d, uint64_t s)
+/* string copy */
+uint32_t HELPER(mvst)(CPUS390XState *env, uint32_t r1, uint32_t r2)
 {
+    const uint64_t d = get_address(env, r1);
+    const uint64_t s = get_address(env, r2);
+    const uint8_t c = env->regs[0];
     uintptr_t ra = GETPC();
     uint32_t len;
 
-    if (c & 0xffffff00ull) {
+    if (env->regs[0] & 0xffffff00ull) {
         s390_program_interrupt(env, PGM_SPECIFICATION, ILEN_AUTO, ra);
     }
-    c = c & 0xff;
-    d = wrap_address(env, d);
-    s = wrap_address(env, s);
 
     /* Lest we fail to service interrupts in a timely manner, limit the
        amount of work we're willing to do.  For now, let's cap at 8k.  */
@@ -719,17 +719,13 @@ uint64_t HELPER(mvst)(CPUS390XState *env, uint64_t c, uint64_t d, uint64_t s)
         uint8_t v = cpu_ldub_data_ra(env, s + len, ra);
         cpu_stb_data_ra(env, d + len, v, ra);
         if (v == c) {
-            /* Complete.  Set CC=1 and advance R1.  */
-            env->cc_op = 1;
-            env->retxl = s;
-            return d + len;
+            set_address_zero(env, r1, d + len);
+            return 1;
         }
     }
-
-    /* Incomplete.  Set CC=3 and signal to advance R1 and R2.  */
-    env->cc_op = 3;
-    env->retxl = s + len;
-    return d + len;
+    set_address_zero(env, r1, d + len);
+    set_address_zero(env, r2, s + len);
+    return 3;
 }
 
 /* load access registers r1 to r3 from memory at a2 */

target/s390x/translate.c

@@ -3488,9 +3488,13 @@ static DisasJumpType op_mvpg(DisasContext *s, DisasOps *o)
 
 static DisasJumpType op_mvst(DisasContext *s, DisasOps *o)
 {
-    gen_helper_mvst(o->in1, cpu_env, regs[0], o->in1, o->in2);
+    TCGv_i32 t1 = tcg_const_i32(get_field(s->fields, r1));
+    TCGv_i32 t2 = tcg_const_i32(get_field(s->fields, r2));
+
+    gen_helper_mvst(cc_op, cpu_env, t1, t2);
+    tcg_temp_free_i32(t1);
+    tcg_temp_free_i32(t2);
     set_cc_static(s);
-    return_low128(o->in2);
     return DISAS_NEXT;
 }