target/s390x: Use atomic operations for COMPARE SWAP PURGE

Also provide the cross-cpu tlb flushing required by the PoO.

Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Richard Henderson <rth@twiddle.net>

target/s390x/helper.h

@@ -107,13 +107,13 @@ DEF_HELPER_FLAGS_2(tprot, TCG_CALL_NO_RWG, i32, i64, i64)
 DEF_HELPER_FLAGS_2(iske, TCG_CALL_NO_RWG_SE, i64, env, i64)
 DEF_HELPER_FLAGS_3(sske, TCG_CALL_NO_RWG, void, env, i64, i64)
 DEF_HELPER_FLAGS_2(rrbe, TCG_CALL_NO_RWG, i32, env, i64)
-DEF_HELPER_3(csp, i32, env, i32, i64)
 DEF_HELPER_4(mvcs, i32, env, i64, i64, i64)
 DEF_HELPER_4(mvcp, i32, env, i64, i64, i64)
 DEF_HELPER_4(sigp, i32, env, i64, i32, i64)
 DEF_HELPER_FLAGS_2(sacf, TCG_CALL_NO_WG, void, env, i64)
 DEF_HELPER_FLAGS_3(ipte, TCG_CALL_NO_RWG, void, env, i64, i64)
 DEF_HELPER_FLAGS_1(ptlb, TCG_CALL_NO_RWG, void, env)
+DEF_HELPER_FLAGS_1(purge, TCG_CALL_NO_RWG, void, env)
 DEF_HELPER_2(lra, i64, env, i64)
 DEF_HELPER_FLAGS_2(lura, TCG_CALL_NO_WG, i64, env, i64)
 DEF_HELPER_FLAGS_2(lurag, TCG_CALL_NO_WG, i64, env, i64)

target/s390x/insn-data.def

@@ -837,7 +837,7 @@
 
 #ifndef CONFIG_USER_ONLY
 /* COMPARE AND SWAP AND PURGE */
-    C(0xb250, CSP,     RRE,   Z,   0, ra2, 0, 0, csp, 0)
+    D(0xb250, CSP,     RRE,   Z,   r1_32u, ra2, r1_P, 0, csp, 0, MO_TEUL)
 /* DIAGNOSE (KVM hypercall) */
     C(0x8300, DIAG,    RSI,   Z,   0, 0, 0, 0, diag, 0)
 /* INSERT STORAGE KEY EXTENDED */

target/s390x/mem_helper.c

@@ -1056,30 +1056,6 @@ uint32_t HELPER(rrbe)(CPUS390XState *env, uint64_t r2)
     return re >> 1;
 }
 
-/* compare and swap and purge */
-uint32_t HELPER(csp)(CPUS390XState *env, uint32_t r1, uint64_t r2)
-{
-    S390CPU *cpu = s390_env_get_cpu(env);
-    uint32_t cc;
-    uint32_t o1 = env->regs[r1];
-    uint64_t a2 = r2 & ~3ULL;
-    uint32_t o2 = cpu_ldl_data(env, a2);
-
-    if (o1 == o2) {
-        cpu_stl_data(env, a2, env->regs[(r1 + 1) & 15]);
-        if (r2 & 0x3) {
-            /* flush TLB / ALB */
-            tlb_flush(CPU(cpu));
-        }
-        cc = 0;
-    } else {
-        env->regs[r1] = (env->regs[r1] & 0xffffffff00000000ULL) | o2;
-        cc = 1;
-    }
-
-    return cc;
-}
-
 uint32_t HELPER(mvcs)(CPUS390XState *env, uint64_t l, uint64_t a1, uint64_t a2)
 {
     uintptr_t ra = GETPC();
@@ -1161,6 +1137,14 @@ void HELPER(ptlb)(CPUS390XState *env)
     tlb_flush(CPU(cpu));
 }
 
+/* flush global tlb */
+void HELPER(purge)(CPUS390XState *env)
+{
+    S390CPU *cpu = s390_env_get_cpu(env);
+
+    tlb_flush_all_cpus_synced(CPU(cpu));
+}
+
 /* load using real address */
 uint64_t HELPER(lura)(CPUS390XState *env, uint64_t addr)
 {

target/s390x/translate.c

@@ -2006,11 +2006,45 @@ static ExitStatus op_cdsg(DisasContext *s, DisasOps *o)
 #ifndef CONFIG_USER_ONLY
 static ExitStatus op_csp(DisasContext *s, DisasOps *o)
 {
-    TCGv_i32 r1 = tcg_const_i32(get_field(s->fields, r1));
+    TCGMemOp mop = s->insn->data;
+    TCGv_i64 addr, old, cc;
+    TCGLabel *lab = gen_new_label();
+
+    /* Note that in1 = R1 (zero-extended expected value),
+       out = R1 (original reg), out2 = R1+1 (new value).  */
+
     check_privileged(s);
-    gen_helper_csp(cc_op, cpu_env, r1, o->in2);
+    addr = tcg_temp_new_i64();
-    tcg_temp_free_i32(r1);
+    old = tcg_temp_new_i64();
-    set_cc_static(s);
+    tcg_gen_andi_i64(addr, o->in2, -1ULL << (mop & MO_SIZE));
+    tcg_gen_atomic_cmpxchg_i64(old, addr, o->in1, o->out2,
+                               get_mem_index(s), mop | MO_ALIGN);
+    tcg_temp_free_i64(addr);
+
+    /* Are the memory and expected values (un)equal?  */
+    cc = tcg_temp_new_i64();
+    tcg_gen_setcond_i64(TCG_COND_NE, cc, o->in1, old);
+    tcg_gen_extrl_i64_i32(cc_op, cc);
+
+    /* Write back the output now, so that it happens before the
+       following branch, so that we don't need local temps.  */
+    if ((mop & MO_SIZE) == MO_32) {
+        tcg_gen_deposit_i64(o->out, o->out, old, 0, 32);
+    } else {
+        tcg_gen_mov_i64(o->out, old);
+    }
+    tcg_temp_free_i64(old);
+
+    /* If the comparison was equal, and the LSB of R2 was set,
+       then we need to flush the TLB (for all cpus).  */
+    tcg_gen_xori_i64(cc, cc, 1);
+    tcg_gen_and_i64(cc, cc, o->in2);
+    tcg_gen_brcondi_i64(TCG_COND_EQ, cc, 0, lab);
+    tcg_temp_free_i64(cc);
+
+    gen_helper_purge(cpu_env);
+    gen_set_label(lab);
+
     return NO_EXIT;
 }
 #endif