tcg: enable thread-per-vCPU

There are a couple of changes that occur at the same time here:

  - introduce a single vCPU qemu_tcg_cpu_thread_fn

  One of these is spawned per vCPU with its own Thread and Condition
  variables. qemu_tcg_rr_cpu_thread_fn is the new name for the old
  single threaded function.

  - the TLS current_cpu variable is now live for the lifetime of MTTCG
    vCPU threads. This is for future work where async jobs need to know
    the vCPU context they are operating in.

The user to switch on multi-thread behaviour and spawn a thread
per-vCPU. For a simple test kvm-unit-test like:

  ./arm/run ./arm/locking-test.flat -smp 4 -accel tcg,thread=multi

Will now use 4 vCPU threads and have an expected FAIL (instead of the
unexpected PASS) as the default mode of the test has no protection when
incrementing a shared variable.

We enable the parallel_cpus flag to ensure we generate correct barrier
and atomic code if supported by the front and backends. This doesn't
automatically enable MTTCG until default_mttcg_enabled() is updated to
check the configuration is supported.
Signed-off-by: NKONRAD Frederic <fred.konrad@greensocs.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>
[AJB: Some fixes, conditionally, commit rewording]
Signed-off-by: NAlex Bennée <alex.bennee@linaro.org>
Reviewed-by: NRichard Henderson <rth@twiddle.net>

cpu-exec.c

@@ -396,7 +396,6 @@ static inline bool cpu_handle_halt(CPUState *cpu)
         }
 #endif
         if (!cpu_has_work(cpu)) {
-            current_cpu = NULL;
             return true;
         }
 
@@ -675,8 +674,5 @@ int cpu_exec(CPUState *cpu)
     cc->cpu_exec_exit(cpu);
     rcu_read_unlock();
 
-    /* fail safe : never use current_cpu outside cpu_exec() */
-    current_cpu = NULL;
-
     return ret;
 }

cpus.c

@@ -809,7 +809,7 @@ static void kick_tcg_thread(void *opaque)
 
 static void start_tcg_kick_timer(void)
 {
-    if (!tcg_kick_vcpu_timer && CPU_NEXT(first_cpu)) {
+    if (!mttcg_enabled && !tcg_kick_vcpu_timer && CPU_NEXT(first_cpu)) {
         tcg_kick_vcpu_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL,
                                            kick_tcg_thread, NULL);
         timer_mod(tcg_kick_vcpu_timer, qemu_tcg_next_kick());
@@ -1063,27 +1063,34 @@ static void qemu_tcg_destroy_vcpu(CPUState *cpu)
 
 static void qemu_wait_io_event_common(CPUState *cpu)
 {
+    atomic_mb_set(&cpu->thread_kicked, false);
     if (cpu->stop) {
         cpu->stop = false;
         cpu->stopped = true;
         qemu_cond_broadcast(&qemu_pause_cond);
     }
     process_queued_cpu_work(cpu);
-    cpu->thread_kicked = false;
+}
+
+static bool qemu_tcg_should_sleep(CPUState *cpu)
+{
+    if (mttcg_enabled) {
+        return cpu_thread_is_idle(cpu);
+    } else {
+        return all_cpu_threads_idle();
+    }
 }
 
 static void qemu_tcg_wait_io_event(CPUState *cpu)
 {
-    while (all_cpu_threads_idle()) {
+    while (qemu_tcg_should_sleep(cpu)) {
         stop_tcg_kick_timer();
         qemu_cond_wait(cpu->halt_cond, &qemu_global_mutex);
     }
 
     start_tcg_kick_timer();
 
-    CPU_FOREACH(cpu) {
-        qemu_wait_io_event_common(cpu);
-    }
+    qemu_wait_io_event_common(cpu);
 }
 
 static void qemu_kvm_wait_io_event(CPUState *cpu)
@@ -1154,6 +1161,7 @@ static void *qemu_dummy_cpu_thread_fn(void *arg)
     qemu_thread_get_self(cpu->thread);
     cpu->thread_id = qemu_get_thread_id();
     cpu->can_do_io = 1;
+    current_cpu = cpu;
 
     sigemptyset(&waitset);
     sigaddset(&waitset, SIG_IPI);
@@ -1162,9 +1170,7 @@ static void *qemu_dummy_cpu_thread_fn(void *arg)
     cpu->created = true;
     qemu_cond_signal(&qemu_cpu_cond);
 
-    current_cpu = cpu;
     while (1) {
-        current_cpu = NULL;
         qemu_mutex_unlock_iothread();
         do {
             int sig;
@@ -1175,7 +1181,6 @@ static void *qemu_dummy_cpu_thread_fn(void *arg)
             exit(1);
         }
         qemu_mutex_lock_iothread();
-        current_cpu = cpu;
         qemu_wait_io_event_common(cpu);
     }
 
@@ -1287,7 +1292,7 @@ static void deal_with_unplugged_cpus(void)
  * elsewhere.
  */
 
-static void *qemu_tcg_cpu_thread_fn(void *arg)
+static void *qemu_tcg_rr_cpu_thread_fn(void *arg)
 {
     CPUState *cpu = arg;
 
@@ -1309,6 +1314,7 @@ static void *qemu_tcg_cpu_thread_fn(void *arg)
 
         /* process any pending work */
         CPU_FOREACH(cpu) {
+            current_cpu = cpu;
             qemu_wait_io_event_common(cpu);
         }
     }
@@ -1331,6 +1337,7 @@ static void *qemu_tcg_cpu_thread_fn(void *arg)
         while (cpu && !cpu->queued_work_first && !cpu->exit_request) {
 
             atomic_mb_set(&tcg_current_rr_cpu, cpu);
+            current_cpu = cpu;
 
             qemu_clock_enable(QEMU_CLOCK_VIRTUAL,
                               (cpu->singlestep_enabled & SSTEP_NOTIMER) == 0);
@@ -1342,7 +1349,7 @@ static void *qemu_tcg_cpu_thread_fn(void *arg)
                     cpu_handle_guest_debug(cpu);
                     break;
                 }
-            } else if (cpu->stop || cpu->stopped) {
+            } else if (cpu->stop) {
                 if (cpu->unplug) {
                     cpu = CPU_NEXT(cpu);
                 }
@@ -1361,7 +1368,7 @@ static void *qemu_tcg_cpu_thread_fn(void *arg)
 
         handle_icount_deadline();
 
-        qemu_tcg_wait_io_event(QTAILQ_FIRST(&cpus));
+        qemu_tcg_wait_io_event(cpu ? cpu : QTAILQ_FIRST(&cpus));
         deal_with_unplugged_cpus();
     }
 
@@ -1408,6 +1415,64 @@ static void CALLBACK dummy_apc_func(ULONG_PTR unused)
 }
 #endif
 
+/* Multi-threaded TCG
+ *
+ * In the multi-threaded case each vCPU has its own thread. The TLS
+ * variable current_cpu can be used deep in the code to find the
+ * current CPUState for a given thread.
+ */
+
+static void *qemu_tcg_cpu_thread_fn(void *arg)
+{
+    CPUState *cpu = arg;
+
+    rcu_register_thread();
+
+    qemu_mutex_lock_iothread();
+    qemu_thread_get_self(cpu->thread);
+
+    cpu->thread_id = qemu_get_thread_id();
+    cpu->created = true;
+    cpu->can_do_io = 1;
+    current_cpu = cpu;
+    qemu_cond_signal(&qemu_cpu_cond);
+
+    /* process any pending work */
+    cpu->exit_request = 1;
+
+    while (1) {
+        if (cpu_can_run(cpu)) {
+            int r;
+            r = tcg_cpu_exec(cpu);
+            switch (r) {
+            case EXCP_DEBUG:
+                cpu_handle_guest_debug(cpu);
+                break;
+            case EXCP_HALTED:
+                /* during start-up the vCPU is reset and the thread is
+                 * kicked several times. If we don't ensure we go back
+                 * to sleep in the halted state we won't cleanly
+                 * start-up when the vCPU is enabled.
+                 *
+                 * cpu->halted should ensure we sleep in wait_io_event
+                 */
+                g_assert(cpu->halted);
+                break;
+            default:
+                /* Ignore everything else? */
+                break;
+            }
+        }
+
+        handle_icount_deadline();
+
+        atomic_mb_set(&cpu->exit_request, 0);
+        qemu_tcg_wait_io_event(cpu);
+    }
+
+    return NULL;
+}
+
 static void qemu_cpu_kick_thread(CPUState *cpu)
 {
 #ifndef _WIN32
@@ -1438,7 +1503,7 @@ void qemu_cpu_kick(CPUState *cpu)
     qemu_cond_broadcast(cpu->halt_cond);
     if (tcg_enabled()) {
         cpu_exit(cpu);
-        /* Also ensure current RR cpu is kicked */
+        /* NOP unless doing single-thread RR */
         qemu_cpu_kick_rr_cpu();
     } else {
         if (hax_enabled()) {
@@ -1514,13 +1579,6 @@ void pause_all_vcpus(void)
 
     if (qemu_in_vcpu_thread()) {
         cpu_stop_current();
-        if (!kvm_enabled()) {
-            CPU_FOREACH(cpu) {
-                cpu->stop = false;
-                cpu->stopped = true;
-            }
-            return;
-        }
     }
 
     while (!all_vcpus_paused()) {
@@ -1569,29 +1627,43 @@ void cpu_remove_sync(CPUState *cpu)
 static void qemu_tcg_init_vcpu(CPUState *cpu)
 {
     char thread_name[VCPU_THREAD_NAME_SIZE];
-    static QemuCond *tcg_halt_cond;
-    static QemuThread *tcg_cpu_thread;
+    static QemuCond *single_tcg_halt_cond;
+    static QemuThread *single_tcg_cpu_thread;
 
-    /* share a single thread for all cpus with TCG */
-    if (!tcg_cpu_thread) {
+    if (qemu_tcg_mttcg_enabled() || !single_tcg_cpu_thread) {
         cpu->thread = g_malloc0(sizeof(QemuThread));
         cpu->halt_cond = g_malloc0(sizeof(QemuCond));
         qemu_cond_init(cpu->halt_cond);
-        tcg_halt_cond = cpu->halt_cond;
-        snprintf(thread_name, VCPU_THREAD_NAME_SIZE, "CPU %d/TCG",
+
+        if (qemu_tcg_mttcg_enabled()) {
+            /* create a thread per vCPU with TCG (MTTCG) */
+            parallel_cpus = true;
+            snprintf(thread_name, VCPU_THREAD_NAME_SIZE, "CPU %d/TCG",
                  cpu->cpu_index);
-        qemu_thread_create(cpu->thread, thread_name, qemu_tcg_cpu_thread_fn,
-                           cpu, QEMU_THREAD_JOINABLE);
+
+            qemu_thread_create(cpu->thread, thread_name, qemu_tcg_cpu_thread_fn,
+                               cpu, QEMU_THREAD_JOINABLE);
+
+        } else {
+            /* share a single thread for all cpus with TCG */
+            snprintf(thread_name, VCPU_THREAD_NAME_SIZE, "ALL CPUs/TCG");
+            qemu_thread_create(cpu->thread, thread_name,
+                               qemu_tcg_rr_cpu_thread_fn,
+                               cpu, QEMU_THREAD_JOINABLE);
+
+            single_tcg_halt_cond = cpu->halt_cond;
+            single_tcg_cpu_thread = cpu->thread;
+        }
 #ifdef _WIN32
         cpu->hThread = qemu_thread_get_handle(cpu->thread);
 #endif
         while (!cpu->created) {
             qemu_cond_wait(&qemu_cpu_cond, &qemu_global_mutex);
         }
-        tcg_cpu_thread = cpu->thread;
     } else {
-        cpu->thread = tcg_cpu_thread;
-        cpu->halt_cond = tcg_halt_cond;
+        /* For non-MTTCG cases we share the thread */
+        cpu->thread = single_tcg_cpu_thread;
+        cpu->halt_cond = single_tcg_halt_cond;
     }
 }