cpus.c 33.0 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
/*
 * QEMU System Emulator
 *
 * Copyright (c) 2003-2008 Fabrice Bellard
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 * THE SOFTWARE.
 */

/* Needed early for CONFIG_BSD etc. */
#include "config-host.h"

#include "monitor.h"
#include "sysemu.h"
#include "gdbstub.h"
#include "dma.h"
#include "kvm.h"
L
Luiz Capitulino 已提交
33
#include "qmp-commands.h"
34

P
Paolo Bonzini 已提交
35
#include "qemu-thread.h"
36
#include "cpus.h"
P
Paolo Bonzini 已提交
37
#include "qtest.h"
P
Paolo Bonzini 已提交
38
#include "main-loop.h"
39
#include "bitmap.h"
J
Jan Kiszka 已提交
40 41

#ifndef _WIN32
M
Marcelo Tosatti 已提交
42
#include "compatfd.h"
J
Jan Kiszka 已提交
43
#endif
44

45 46 47 48
#ifdef CONFIG_LINUX

#include <sys/prctl.h>

M
Marcelo Tosatti 已提交
49 50 51 52
#ifndef PR_MCE_KILL
#define PR_MCE_KILL 33
#endif

53 54 55 56 57 58 59 60 61 62
#ifndef PR_MCE_KILL_SET
#define PR_MCE_KILL_SET 1
#endif

#ifndef PR_MCE_KILL_EARLY
#define PR_MCE_KILL_EARLY 1
#endif

#endif /* CONFIG_LINUX */

63
static CPUArchState *next_cpu;
64

65 66 67 68 69 70 71 72
static bool cpu_thread_is_idle(CPUArchState *env)
{
    if (env->stop || env->queued_work_first) {
        return false;
    }
    if (env->stopped || !runstate_is_running()) {
        return true;
    }
73 74
    if (!env->halted || qemu_cpu_has_work(env) ||
        kvm_async_interrupts_enabled()) {
75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91
        return false;
    }
    return true;
}

static bool all_cpu_threads_idle(void)
{
    CPUArchState *env;

    for (env = first_cpu; env != NULL; env = env->next_cpu) {
        if (!cpu_thread_is_idle(env)) {
            return false;
        }
    }
    return true;
}

P
Paolo Bonzini 已提交
92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120
/***********************************************************/
/* guest cycle counter */

/* Conversion factor from emulated instructions to virtual clock ticks.  */
static int icount_time_shift;
/* Arbitrarily pick 1MIPS as the minimum allowable speed.  */
#define MAX_ICOUNT_SHIFT 10
/* Compensate for varying guest execution speed.  */
static int64_t qemu_icount_bias;
static QEMUTimer *icount_rt_timer;
static QEMUTimer *icount_vm_timer;
static QEMUTimer *icount_warp_timer;
static int64_t vm_clock_warp_start;
static int64_t qemu_icount;

typedef struct TimersState {
    int64_t cpu_ticks_prev;
    int64_t cpu_ticks_offset;
    int64_t cpu_clock_offset;
    int32_t cpu_ticks_enabled;
    int64_t dummy;
} TimersState;

TimersState timers_state;

/* Return the virtual CPU time, based on the instruction counter.  */
int64_t cpu_get_icount(void)
{
    int64_t icount;
121
    CPUArchState *env = cpu_single_env;
P
Paolo Bonzini 已提交
122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269

    icount = qemu_icount;
    if (env) {
        if (!can_do_io(env)) {
            fprintf(stderr, "Bad clock read\n");
        }
        icount -= (env->icount_decr.u16.low + env->icount_extra);
    }
    return qemu_icount_bias + (icount << icount_time_shift);
}

/* return the host CPU cycle counter and handle stop/restart */
int64_t cpu_get_ticks(void)
{
    if (use_icount) {
        return cpu_get_icount();
    }
    if (!timers_state.cpu_ticks_enabled) {
        return timers_state.cpu_ticks_offset;
    } else {
        int64_t ticks;
        ticks = cpu_get_real_ticks();
        if (timers_state.cpu_ticks_prev > ticks) {
            /* Note: non increasing ticks may happen if the host uses
               software suspend */
            timers_state.cpu_ticks_offset += timers_state.cpu_ticks_prev - ticks;
        }
        timers_state.cpu_ticks_prev = ticks;
        return ticks + timers_state.cpu_ticks_offset;
    }
}

/* return the host CPU monotonic timer and handle stop/restart */
int64_t cpu_get_clock(void)
{
    int64_t ti;
    if (!timers_state.cpu_ticks_enabled) {
        return timers_state.cpu_clock_offset;
    } else {
        ti = get_clock();
        return ti + timers_state.cpu_clock_offset;
    }
}

/* enable cpu_get_ticks() */
void cpu_enable_ticks(void)
{
    if (!timers_state.cpu_ticks_enabled) {
        timers_state.cpu_ticks_offset -= cpu_get_real_ticks();
        timers_state.cpu_clock_offset -= get_clock();
        timers_state.cpu_ticks_enabled = 1;
    }
}

/* disable cpu_get_ticks() : the clock is stopped. You must not call
   cpu_get_ticks() after that.  */
void cpu_disable_ticks(void)
{
    if (timers_state.cpu_ticks_enabled) {
        timers_state.cpu_ticks_offset = cpu_get_ticks();
        timers_state.cpu_clock_offset = cpu_get_clock();
        timers_state.cpu_ticks_enabled = 0;
    }
}

/* Correlation between real and virtual time is always going to be
   fairly approximate, so ignore small variation.
   When the guest is idle real and virtual time will be aligned in
   the IO wait loop.  */
#define ICOUNT_WOBBLE (get_ticks_per_sec() / 10)

static void icount_adjust(void)
{
    int64_t cur_time;
    int64_t cur_icount;
    int64_t delta;
    static int64_t last_delta;
    /* If the VM is not running, then do nothing.  */
    if (!runstate_is_running()) {
        return;
    }
    cur_time = cpu_get_clock();
    cur_icount = qemu_get_clock_ns(vm_clock);
    delta = cur_icount - cur_time;
    /* FIXME: This is a very crude algorithm, somewhat prone to oscillation.  */
    if (delta > 0
        && last_delta + ICOUNT_WOBBLE < delta * 2
        && icount_time_shift > 0) {
        /* The guest is getting too far ahead.  Slow time down.  */
        icount_time_shift--;
    }
    if (delta < 0
        && last_delta - ICOUNT_WOBBLE > delta * 2
        && icount_time_shift < MAX_ICOUNT_SHIFT) {
        /* The guest is getting too far behind.  Speed time up.  */
        icount_time_shift++;
    }
    last_delta = delta;
    qemu_icount_bias = cur_icount - (qemu_icount << icount_time_shift);
}

static void icount_adjust_rt(void *opaque)
{
    qemu_mod_timer(icount_rt_timer,
                   qemu_get_clock_ms(rt_clock) + 1000);
    icount_adjust();
}

static void icount_adjust_vm(void *opaque)
{
    qemu_mod_timer(icount_vm_timer,
                   qemu_get_clock_ns(vm_clock) + get_ticks_per_sec() / 10);
    icount_adjust();
}

static int64_t qemu_icount_round(int64_t count)
{
    return (count + (1 << icount_time_shift) - 1) >> icount_time_shift;
}

static void icount_warp_rt(void *opaque)
{
    if (vm_clock_warp_start == -1) {
        return;
    }

    if (runstate_is_running()) {
        int64_t clock = qemu_get_clock_ns(rt_clock);
        int64_t warp_delta = clock - vm_clock_warp_start;
        if (use_icount == 1) {
            qemu_icount_bias += warp_delta;
        } else {
            /*
             * In adaptive mode, do not let the vm_clock run too
             * far ahead of real time.
             */
            int64_t cur_time = cpu_get_clock();
            int64_t cur_icount = qemu_get_clock_ns(vm_clock);
            int64_t delta = cur_time - cur_icount;
            qemu_icount_bias += MIN(warp_delta, delta);
        }
        if (qemu_clock_expired(vm_clock)) {
            qemu_notify_event();
        }
    }
    vm_clock_warp_start = -1;
}

P
Paolo Bonzini 已提交
270 271 272 273 274 275 276 277 278 279 280 281 282 283
void qtest_clock_warp(int64_t dest)
{
    int64_t clock = qemu_get_clock_ns(vm_clock);
    assert(qtest_enabled());
    while (clock < dest) {
        int64_t deadline = qemu_clock_deadline(vm_clock);
        int64_t warp = MIN(dest - clock, deadline);
        qemu_icount_bias += warp;
        qemu_run_timers(vm_clock);
        clock = qemu_get_clock_ns(vm_clock);
    }
    qemu_notify_event();
}

P
Paolo Bonzini 已提交
284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309
void qemu_clock_warp(QEMUClock *clock)
{
    int64_t deadline;

    /*
     * There are too many global variables to make the "warp" behavior
     * applicable to other clocks.  But a clock argument removes the
     * need for if statements all over the place.
     */
    if (clock != vm_clock || !use_icount) {
        return;
    }

    /*
     * If the CPUs have been sleeping, advance the vm_clock timer now.  This
     * ensures that the deadline for the timer is computed correctly below.
     * This also makes sure that the insn counter is synchronized before the
     * CPU starts running, in case the CPU is woken by an event other than
     * the earliest vm_clock timer.
     */
    icount_warp_rt(NULL);
    if (!all_cpu_threads_idle() || !qemu_clock_has_timers(vm_clock)) {
        qemu_del_timer(icount_warp_timer);
        return;
    }

P
Paolo Bonzini 已提交
310 311 312 313 314
    if (qtest_enabled()) {
        /* When testing, qtest commands advance icount.  */
	return;
    }

P
Paolo Bonzini 已提交
315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331
    vm_clock_warp_start = qemu_get_clock_ns(rt_clock);
    deadline = qemu_clock_deadline(vm_clock);
    if (deadline > 0) {
        /*
         * Ensure the vm_clock proceeds even when the virtual CPU goes to
         * sleep.  Otherwise, the CPU might be waiting for a future timer
         * interrupt to wake it up, but the interrupt never comes because
         * the vCPU isn't running any insns and thus doesn't advance the
         * vm_clock.
         *
         * An extreme solution for this problem would be to never let VCPUs
         * sleep in icount mode if there is a pending vm_clock timer; rather
         * time could just advance to the next vm_clock event.  Instead, we
         * do stop VCPUs and only advance vm_clock after some "real" time,
         * (related to the time left until the next event) has passed.  This
         * rt_clock timer will do this.  This avoids that the warps are too
         * visible externally---for example, you will not be sending network
D
Dong Xu Wang 已提交
332
         * packets continuously instead of every 100ms.
P
Paolo Bonzini 已提交
333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385
         */
        qemu_mod_timer(icount_warp_timer, vm_clock_warp_start + deadline);
    } else {
        qemu_notify_event();
    }
}

static const VMStateDescription vmstate_timers = {
    .name = "timer",
    .version_id = 2,
    .minimum_version_id = 1,
    .minimum_version_id_old = 1,
    .fields      = (VMStateField[]) {
        VMSTATE_INT64(cpu_ticks_offset, TimersState),
        VMSTATE_INT64(dummy, TimersState),
        VMSTATE_INT64_V(cpu_clock_offset, TimersState, 2),
        VMSTATE_END_OF_LIST()
    }
};

void configure_icount(const char *option)
{
    vmstate_register(NULL, 0, &vmstate_timers, &timers_state);
    if (!option) {
        return;
    }

    icount_warp_timer = qemu_new_timer_ns(rt_clock, icount_warp_rt, NULL);
    if (strcmp(option, "auto") != 0) {
        icount_time_shift = strtol(option, NULL, 0);
        use_icount = 1;
        return;
    }

    use_icount = 2;

    /* 125MIPS seems a reasonable initial guess at the guest speed.
       It will be corrected fairly quickly anyway.  */
    icount_time_shift = 3;

    /* Have both realtime and virtual time triggers for speed adjustment.
       The realtime trigger catches emulated time passing too slowly,
       the virtual time trigger catches emulated time passing too fast.
       Realtime triggers occur even when idle, so use them less frequently
       than VM triggers.  */
    icount_rt_timer = qemu_new_timer_ms(rt_clock, icount_adjust_rt, NULL);
    qemu_mod_timer(icount_rt_timer,
                   qemu_get_clock_ms(rt_clock) + 1000);
    icount_vm_timer = qemu_new_timer_ns(vm_clock, icount_adjust_vm, NULL);
    qemu_mod_timer(icount_vm_timer,
                   qemu_get_clock_ns(vm_clock) + get_ticks_per_sec() / 10);
}

386 387 388 389
/***********************************************************/
void hw_error(const char *fmt, ...)
{
    va_list ap;
390
    CPUArchState *env;
391 392 393 394 395 396 397

    va_start(ap, fmt);
    fprintf(stderr, "qemu: hardware error: ");
    vfprintf(stderr, fmt, ap);
    fprintf(stderr, "\n");
    for(env = first_cpu; env != NULL; env = env->next_cpu) {
        fprintf(stderr, "CPU #%d:\n", env->cpu_index);
398
        cpu_dump_state(env, stderr, fprintf, CPU_DUMP_FPU);
399 400 401 402 403 404 405
    }
    va_end(ap);
    abort();
}

void cpu_synchronize_all_states(void)
{
406
    CPUArchState *cpu;
407 408 409 410 411 412 413 414

    for (cpu = first_cpu; cpu; cpu = cpu->next_cpu) {
        cpu_synchronize_state(cpu);
    }
}

void cpu_synchronize_all_post_reset(void)
{
415
    CPUArchState *cpu;
416 417 418 419 420 421 422 423

    for (cpu = first_cpu; cpu; cpu = cpu->next_cpu) {
        cpu_synchronize_post_reset(cpu);
    }
}

void cpu_synchronize_all_post_init(void)
{
424
    CPUArchState *cpu;
425 426 427 428 429 430

    for (cpu = first_cpu; cpu; cpu = cpu->next_cpu) {
        cpu_synchronize_post_init(cpu);
    }
}

431
int cpu_is_stopped(CPUArchState *env)
M
Marcelo Tosatti 已提交
432
{
433
    return !runstate_is_running() || env->stopped;
M
Marcelo Tosatti 已提交
434 435
}

436
static void do_vm_stop(RunState state)
437
{
438
    if (runstate_is_running()) {
439 440
        cpu_disable_ticks();
        pause_all_vcpus();
441
        runstate_set(state);
442
        vm_state_notify(0, state);
443
        bdrv_drain_all();
444
        bdrv_flush_all();
445 446 447 448
        monitor_protocol_event(QEVENT_STOP, NULL);
    }
}

449
static int cpu_can_run(CPUArchState *env)
450
{
451
    if (env->stop) {
452
        return 0;
453
    }
454
    if (env->stopped || !runstate_is_running()) {
455
        return 0;
456
    }
457 458 459
    return 1;
}

460
static void cpu_handle_guest_debug(CPUArchState *env)
461
{
462
    gdb_set_stop_cpu(env);
463
    qemu_system_debug_request();
464
    env->stopped = 1;
465 466
}

467 468 469 470 471 472 473 474
static void cpu_signal(int sig)
{
    if (cpu_single_env) {
        cpu_exit(cpu_single_env);
    }
    exit_request = 1;
}

475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513
#ifdef CONFIG_LINUX
static void sigbus_reraise(void)
{
    sigset_t set;
    struct sigaction action;

    memset(&action, 0, sizeof(action));
    action.sa_handler = SIG_DFL;
    if (!sigaction(SIGBUS, &action, NULL)) {
        raise(SIGBUS);
        sigemptyset(&set);
        sigaddset(&set, SIGBUS);
        sigprocmask(SIG_UNBLOCK, &set, NULL);
    }
    perror("Failed to re-raise SIGBUS!\n");
    abort();
}

static void sigbus_handler(int n, struct qemu_signalfd_siginfo *siginfo,
                           void *ctx)
{
    if (kvm_on_sigbus(siginfo->ssi_code,
                      (void *)(intptr_t)siginfo->ssi_addr)) {
        sigbus_reraise();
    }
}

static void qemu_init_sigbus(void)
{
    struct sigaction action;

    memset(&action, 0, sizeof(action));
    action.sa_flags = SA_SIGINFO;
    action.sa_sigaction = (void (*)(int, siginfo_t*, void*))sigbus_handler;
    sigaction(SIGBUS, &action, NULL);

    prctl(PR_MCE_KILL, PR_MCE_KILL_SET, PR_MCE_KILL_EARLY, 0, 0);
}

514
static void qemu_kvm_eat_signals(CPUArchState *env)
515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550
{
    struct timespec ts = { 0, 0 };
    siginfo_t siginfo;
    sigset_t waitset;
    sigset_t chkset;
    int r;

    sigemptyset(&waitset);
    sigaddset(&waitset, SIG_IPI);
    sigaddset(&waitset, SIGBUS);

    do {
        r = sigtimedwait(&waitset, &siginfo, &ts);
        if (r == -1 && !(errno == EAGAIN || errno == EINTR)) {
            perror("sigtimedwait");
            exit(1);
        }

        switch (r) {
        case SIGBUS:
            if (kvm_on_sigbus_vcpu(env, siginfo.si_code, siginfo.si_addr)) {
                sigbus_reraise();
            }
            break;
        default:
            break;
        }

        r = sigpending(&chkset);
        if (r == -1) {
            perror("sigpending");
            exit(1);
        }
    } while (sigismember(&chkset, SIG_IPI) || sigismember(&chkset, SIGBUS));
}

551 552 553 554 555
#else /* !CONFIG_LINUX */

static void qemu_init_sigbus(void)
{
}
556

557
static void qemu_kvm_eat_signals(CPUArchState *env)
558 559
{
}
560 561
#endif /* !CONFIG_LINUX */

562
#ifndef _WIN32
563 564 565 566
static void dummy_signal(int sig)
{
}

567
static void qemu_kvm_init_cpu_signals(CPUArchState *env)
568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600
{
    int r;
    sigset_t set;
    struct sigaction sigact;

    memset(&sigact, 0, sizeof(sigact));
    sigact.sa_handler = dummy_signal;
    sigaction(SIG_IPI, &sigact, NULL);

    pthread_sigmask(SIG_BLOCK, NULL, &set);
    sigdelset(&set, SIG_IPI);
    sigdelset(&set, SIGBUS);
    r = kvm_set_signal_mask(env, &set);
    if (r) {
        fprintf(stderr, "kvm_set_signal_mask: %s\n", strerror(-r));
        exit(1);
    }
}

static void qemu_tcg_init_cpu_signals(void)
{
    sigset_t set;
    struct sigaction sigact;

    memset(&sigact, 0, sizeof(sigact));
    sigact.sa_handler = cpu_signal;
    sigaction(SIG_IPI, &sigact, NULL);

    sigemptyset(&set);
    sigaddset(&set, SIG_IPI);
    pthread_sigmask(SIG_UNBLOCK, &set, NULL);
}

601
#else /* _WIN32 */
602
static void qemu_kvm_init_cpu_signals(CPUArchState *env)
603
{
604 605
    abort();
}
606

607 608
static void qemu_tcg_init_cpu_signals(void)
{
609
}
610
#endif /* _WIN32 */
611

612
static QemuMutex qemu_global_mutex;
613 614
static QemuCond qemu_io_proceeded_cond;
static bool iothread_requesting_mutex;
615 616 617 618 619 620 621 622 623 624

static QemuThread io_thread;

static QemuThread *tcg_cpu_thread;
static QemuCond *tcg_halt_cond;

/* cpu creation */
static QemuCond qemu_cpu_cond;
/* system init */
static QemuCond qemu_pause_cond;
M
Marcelo Tosatti 已提交
625
static QemuCond qemu_work_cond;
626

P
Paolo Bonzini 已提交
627
void qemu_init_cpu_loop(void)
628
{
629
    qemu_init_sigbus();
630 631 632
    qemu_cond_init(&qemu_cpu_cond);
    qemu_cond_init(&qemu_pause_cond);
    qemu_cond_init(&qemu_work_cond);
633
    qemu_cond_init(&qemu_io_proceeded_cond);
634 635
    qemu_mutex_init(&qemu_global_mutex);

J
Jan Kiszka 已提交
636
    qemu_thread_get_self(&io_thread);
637 638
}

639
void run_on_cpu(CPUArchState *env, void (*func)(void *data), void *data)
M
Marcelo Tosatti 已提交
640 641 642
{
    struct qemu_work_item wi;

J
Jan Kiszka 已提交
643
    if (qemu_cpu_is_self(env)) {
M
Marcelo Tosatti 已提交
644 645 646 647 648 649
        func(data);
        return;
    }

    wi.func = func;
    wi.data = data;
650
    if (!env->queued_work_first) {
M
Marcelo Tosatti 已提交
651
        env->queued_work_first = &wi;
652
    } else {
M
Marcelo Tosatti 已提交
653
        env->queued_work_last->next = &wi;
654
    }
M
Marcelo Tosatti 已提交
655 656 657 658 659 660
    env->queued_work_last = &wi;
    wi.next = NULL;
    wi.done = false;

    qemu_cpu_kick(env);
    while (!wi.done) {
661
        CPUArchState *self_env = cpu_single_env;
M
Marcelo Tosatti 已提交
662 663 664 665 666 667

        qemu_cond_wait(&qemu_work_cond, &qemu_global_mutex);
        cpu_single_env = self_env;
    }
}

668
static void flush_queued_work(CPUArchState *env)
M
Marcelo Tosatti 已提交
669 670 671
{
    struct qemu_work_item *wi;

672
    if (!env->queued_work_first) {
M
Marcelo Tosatti 已提交
673
        return;
674
    }
M
Marcelo Tosatti 已提交
675 676 677 678 679 680 681 682 683 684

    while ((wi = env->queued_work_first)) {
        env->queued_work_first = wi->next;
        wi->func(wi->data);
        wi->done = true;
    }
    env->queued_work_last = NULL;
    qemu_cond_broadcast(&qemu_work_cond);
}

685
static void qemu_wait_io_event_common(CPUArchState *env)
686
{
687 688
    CPUState *cpu = ENV_GET_CPU(env);

689 690 691 692 693
    if (env->stop) {
        env->stop = 0;
        env->stopped = 1;
        qemu_cond_signal(&qemu_pause_cond);
    }
M
Marcelo Tosatti 已提交
694
    flush_queued_work(env);
695
    cpu->thread_kicked = false;
696 697
}

698
static void qemu_tcg_wait_io_event(void)
699
{
700
    CPUArchState *env;
701

702
    while (all_cpu_threads_idle()) {
703 704 705
       /* Start accounting real time to the virtual clock if the CPUs
          are idle.  */
        qemu_clock_warp(vm_clock);
706
        qemu_cond_wait(tcg_halt_cond, &qemu_global_mutex);
707
    }
708

709 710 711
    while (iothread_requesting_mutex) {
        qemu_cond_wait(&qemu_io_proceeded_cond, &qemu_global_mutex);
    }
712 713 714 715

    for (env = first_cpu; env != NULL; env = env->next_cpu) {
        qemu_wait_io_event_common(env);
    }
716 717
}

718
static void qemu_kvm_wait_io_event(CPUArchState *env)
719
{
720
    while (cpu_thread_is_idle(env)) {
721
        qemu_cond_wait(env->halt_cond, &qemu_global_mutex);
722
    }
723

J
Jan Kiszka 已提交
724
    qemu_kvm_eat_signals(env);
725 726 727
    qemu_wait_io_event_common(env);
}

728
static void *qemu_kvm_cpu_thread_fn(void *arg)
729
{
730
    CPUArchState *env = arg;
731
    CPUState *cpu = ENV_GET_CPU(env);
J
Jan Kiszka 已提交
732
    int r;
733

734
    qemu_mutex_lock(&qemu_global_mutex);
735
    qemu_thread_get_self(cpu->thread);
J
Jan Kiszka 已提交
736
    env->thread_id = qemu_get_thread_id();
J
Jan Kiszka 已提交
737
    cpu_single_env = env;
738

J
Jan Kiszka 已提交
739 740 741 742 743
    r = kvm_init_vcpu(env);
    if (r < 0) {
        fprintf(stderr, "kvm_init_vcpu failed: %s\n", strerror(-r));
        exit(1);
    }
744

745
    qemu_kvm_init_cpu_signals(env);
746 747 748 749 750 751

    /* signal CPU creation */
    env->created = 1;
    qemu_cond_signal(&qemu_cpu_cond);

    while (1) {
752
        if (cpu_can_run(env)) {
753
            r = kvm_cpu_exec(env);
754
            if (r == EXCP_DEBUG) {
755
                cpu_handle_guest_debug(env);
756
            }
757
        }
758 759 760 761 762 763
        qemu_kvm_wait_io_event(env);
    }

    return NULL;
}

A
Anthony Liguori 已提交
764 765 766 767 768 769 770
static void *qemu_dummy_cpu_thread_fn(void *arg)
{
#ifdef _WIN32
    fprintf(stderr, "qtest is not supported under Windows\n");
    exit(1);
#else
    CPUArchState *env = arg;
771
    CPUState *cpu = ENV_GET_CPU(env);
A
Anthony Liguori 已提交
772 773 774 775
    sigset_t waitset;
    int r;

    qemu_mutex_lock_iothread();
776
    qemu_thread_get_self(cpu->thread);
A
Anthony Liguori 已提交
777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806
    env->thread_id = qemu_get_thread_id();

    sigemptyset(&waitset);
    sigaddset(&waitset, SIG_IPI);

    /* signal CPU creation */
    env->created = 1;
    qemu_cond_signal(&qemu_cpu_cond);

    cpu_single_env = env;
    while (1) {
        cpu_single_env = NULL;
        qemu_mutex_unlock_iothread();
        do {
            int sig;
            r = sigwait(&waitset, &sig);
        } while (r == -1 && (errno == EAGAIN || errno == EINTR));
        if (r == -1) {
            perror("sigwait");
            exit(1);
        }
        qemu_mutex_lock_iothread();
        cpu_single_env = env;
        qemu_wait_io_event_common(env);
    }

    return NULL;
#endif
}

J
Jan Kiszka 已提交
807 808
static void tcg_exec_all(void);

809
static void *qemu_tcg_cpu_thread_fn(void *arg)
810
{
811
    CPUArchState *env = arg;
812
    CPUState *cpu = ENV_GET_CPU(env);
813

814
    qemu_tcg_init_cpu_signals();
815
    qemu_thread_get_self(cpu->thread);
816 817 818

    /* signal CPU creation */
    qemu_mutex_lock(&qemu_global_mutex);
819
    for (env = first_cpu; env != NULL; env = env->next_cpu) {
J
Jan Kiszka 已提交
820
        env->thread_id = qemu_get_thread_id();
821
        env->created = 1;
822
    }
823 824
    qemu_cond_signal(&qemu_cpu_cond);

825 826 827
    /* wait for initial kick-off after machine start */
    while (first_cpu->stopped) {
        qemu_cond_wait(tcg_halt_cond, &qemu_global_mutex);
828 829 830 831 832

        /* process any pending work */
        for (env = first_cpu; env != NULL; env = env->next_cpu) {
            qemu_wait_io_event_common(env);
        }
833
    }
834 835

    while (1) {
J
Jan Kiszka 已提交
836
        tcg_exec_all();
P
Paolo Bonzini 已提交
837
        if (use_icount && qemu_clock_deadline(vm_clock) <= 0) {
838 839
            qemu_notify_event();
        }
840
        qemu_tcg_wait_io_event();
841 842 843 844 845
    }

    return NULL;
}

846
static void qemu_cpu_kick_thread(CPUArchState *env)
P
Paolo Bonzini 已提交
847
{
848
    CPUState *cpu = ENV_GET_CPU(env);
P
Paolo Bonzini 已提交
849 850 851
#ifndef _WIN32
    int err;

852
    err = pthread_kill(cpu->thread->thread, SIG_IPI);
P
Paolo Bonzini 已提交
853 854 855 856 857 858
    if (err) {
        fprintf(stderr, "qemu:%s: %s", __func__, strerror(err));
        exit(1);
    }
#else /* _WIN32 */
    if (!qemu_cpu_is_self(env)) {
859
        SuspendThread(cpu->hThread);
P
Paolo Bonzini 已提交
860
        cpu_signal(0);
861
        ResumeThread(cpu->hThread);
P
Paolo Bonzini 已提交
862 863 864 865
    }
#endif
}

866 867
void qemu_cpu_kick(void *_env)
{
868
    CPUArchState *env = _env;
869
    CPUState *cpu = ENV_GET_CPU(env);
870 871

    qemu_cond_broadcast(env->halt_cond);
872
    if (!tcg_enabled() && !cpu->thread_kicked) {
P
Paolo Bonzini 已提交
873
        qemu_cpu_kick_thread(env);
874
        cpu->thread_kicked = true;
875
    }
876 877
}

878
void qemu_cpu_kick_self(void)
879
{
880
#ifndef _WIN32
881
    assert(cpu_single_env);
882
    CPUState *cpu_single_cpu = ENV_GET_CPU(cpu_single_env);
883

884
    if (!cpu_single_cpu->thread_kicked) {
P
Paolo Bonzini 已提交
885
        qemu_cpu_kick_thread(cpu_single_env);
886
        cpu_single_cpu->thread_kicked = true;
887
    }
888 889 890
#else
    abort();
#endif
891 892
}

J
Jan Kiszka 已提交
893
int qemu_cpu_is_self(void *_env)
894
{
895
    CPUArchState *env = _env;
896
    CPUState *cpu = ENV_GET_CPU(env);
M
Marcelo Tosatti 已提交
897

898
    return qemu_thread_is_self(cpu->thread);
899 900 901 902
}

void qemu_mutex_lock_iothread(void)
{
A
Anthony Liguori 已提交
903
    if (!tcg_enabled()) {
904
        qemu_mutex_lock(&qemu_global_mutex);
905
    } else {
906
        iothread_requesting_mutex = true;
907
        if (qemu_mutex_trylock(&qemu_global_mutex)) {
P
Paolo Bonzini 已提交
908
            qemu_cpu_kick_thread(first_cpu);
909 910
            qemu_mutex_lock(&qemu_global_mutex);
        }
911 912
        iothread_requesting_mutex = false;
        qemu_cond_broadcast(&qemu_io_proceeded_cond);
913
    }
914 915 916 917 918 919 920 921 922
}

void qemu_mutex_unlock_iothread(void)
{
    qemu_mutex_unlock(&qemu_global_mutex);
}

static int all_vcpus_paused(void)
{
923
    CPUArchState *penv = first_cpu;
924 925

    while (penv) {
926
        if (!penv->stopped) {
927
            return 0;
928
        }
929
        penv = penv->next_cpu;
930 931 932 933 934 935 936
    }

    return 1;
}

void pause_all_vcpus(void)
{
937
    CPUArchState *penv = first_cpu;
938

939
    qemu_clock_enable(vm_clock, false);
940 941 942
    while (penv) {
        penv->stop = 1;
        qemu_cpu_kick(penv);
943
        penv = penv->next_cpu;
944 945
    }

946 947 948 949 950 951 952 953 954 955 956 957
    if (!qemu_thread_is_self(&io_thread)) {
        cpu_stop_current();
        if (!kvm_enabled()) {
            while (penv) {
                penv->stop = 0;
                penv->stopped = 1;
                penv = penv->next_cpu;
            }
            return;
        }
    }

958
    while (!all_vcpus_paused()) {
959
        qemu_cond_wait(&qemu_pause_cond, &qemu_global_mutex);
960 961
        penv = first_cpu;
        while (penv) {
962
            qemu_cpu_kick(penv);
963
            penv = penv->next_cpu;
964 965 966 967 968 969
        }
    }
}

void resume_all_vcpus(void)
{
970
    CPUArchState *penv = first_cpu;
971

972
    qemu_clock_enable(vm_clock, true);
973 974 975 976
    while (penv) {
        penv->stop = 0;
        penv->stopped = 0;
        qemu_cpu_kick(penv);
977
        penv = penv->next_cpu;
978 979 980
    }
}

981
static void qemu_tcg_init_vcpu(void *_env)
982
{
983
    CPUArchState *env = _env;
984
    CPUState *cpu = ENV_GET_CPU(env);
985

986 987
    /* share a single thread for all cpus with TCG */
    if (!tcg_cpu_thread) {
988
        cpu->thread = g_malloc0(sizeof(QemuThread));
989
        env->halt_cond = g_malloc0(sizeof(QemuCond));
990
        qemu_cond_init(env->halt_cond);
991
        tcg_halt_cond = env->halt_cond;
992
        qemu_thread_create(cpu->thread, qemu_tcg_cpu_thread_fn, env,
P
Paolo Bonzini 已提交
993 994
                           QEMU_THREAD_JOINABLE);
#ifdef _WIN32
995
        cpu->hThread = qemu_thread_get_handle(cpu->thread);
P
Paolo Bonzini 已提交
996
#endif
997
        while (env->created == 0) {
998
            qemu_cond_wait(&qemu_cpu_cond, &qemu_global_mutex);
999
        }
1000
        tcg_cpu_thread = cpu->thread;
1001
    } else {
1002
        cpu->thread = tcg_cpu_thread;
1003 1004 1005 1006
        env->halt_cond = tcg_halt_cond;
    }
}

1007
static void qemu_kvm_start_vcpu(CPUArchState *env)
1008
{
1009 1010 1011
    CPUState *cpu = ENV_GET_CPU(env);

    cpu->thread = g_malloc0(sizeof(QemuThread));
1012
    env->halt_cond = g_malloc0(sizeof(QemuCond));
1013
    qemu_cond_init(env->halt_cond);
1014
    qemu_thread_create(cpu->thread, qemu_kvm_cpu_thread_fn, env,
P
Paolo Bonzini 已提交
1015
                       QEMU_THREAD_JOINABLE);
1016
    while (env->created == 0) {
1017
        qemu_cond_wait(&qemu_cpu_cond, &qemu_global_mutex);
1018
    }
1019 1020
}

A
Anthony Liguori 已提交
1021 1022
static void qemu_dummy_start_vcpu(CPUArchState *env)
{
1023 1024 1025
    CPUState *cpu = ENV_GET_CPU(env);

    cpu->thread = g_malloc0(sizeof(QemuThread));
A
Anthony Liguori 已提交
1026 1027
    env->halt_cond = g_malloc0(sizeof(QemuCond));
    qemu_cond_init(env->halt_cond);
1028
    qemu_thread_create(cpu->thread, qemu_dummy_cpu_thread_fn, env,
A
Anthony Liguori 已提交
1029 1030 1031 1032 1033 1034
                       QEMU_THREAD_JOINABLE);
    while (env->created == 0) {
        qemu_cond_wait(&qemu_cpu_cond, &qemu_global_mutex);
    }
}

1035 1036
void qemu_init_vcpu(void *_env)
{
1037
    CPUArchState *env = _env;
1038 1039 1040

    env->nr_cores = smp_cores;
    env->nr_threads = smp_threads;
1041
    env->stopped = 1;
1042
    if (kvm_enabled()) {
1043
        qemu_kvm_start_vcpu(env);
A
Anthony Liguori 已提交
1044
    } else if (tcg_enabled()) {
1045
        qemu_tcg_init_vcpu(env);
A
Anthony Liguori 已提交
1046 1047
    } else {
        qemu_dummy_start_vcpu(env);
1048
    }
1049 1050
}

1051
void cpu_stop_current(void)
1052
{
1053
    if (cpu_single_env) {
1054
        cpu_single_env->stop = 0;
1055 1056
        cpu_single_env->stopped = 1;
        cpu_exit(cpu_single_env);
1057
        qemu_cond_signal(&qemu_pause_cond);
1058
    }
1059 1060
}

1061
void vm_stop(RunState state)
1062
{
J
Jan Kiszka 已提交
1063
    if (!qemu_thread_is_self(&io_thread)) {
1064
        qemu_system_vmstop_request(state);
1065 1066 1067 1068
        /*
         * FIXME: should not return to device code in case
         * vm_stop() has been requested.
         */
1069
        cpu_stop_current();
1070 1071
        return;
    }
1072
    do_vm_stop(state);
1073 1074
}

1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085
/* does a state transition even if the VM is already stopped,
   current state is forgotten forever */
void vm_stop_force_state(RunState state)
{
    if (runstate_is_running()) {
        vm_stop(state);
    } else {
        runstate_set(state);
    }
}

1086
static int tcg_cpu_exec(CPUArchState *env)
1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101
{
    int ret;
#ifdef CONFIG_PROFILER
    int64_t ti;
#endif

#ifdef CONFIG_PROFILER
    ti = profile_getclock();
#endif
    if (use_icount) {
        int64_t count;
        int decr;
        qemu_icount -= (env->icount_decr.u16.low + env->icount_extra);
        env->icount_decr.u16.low = 0;
        env->icount_extra = 0;
P
Paolo Bonzini 已提交
1102
        count = qemu_icount_round(qemu_clock_deadline(vm_clock));
1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123
        qemu_icount += count;
        decr = (count > 0xffff) ? 0xffff : count;
        count -= decr;
        env->icount_decr.u16.low = decr;
        env->icount_extra = count;
    }
    ret = cpu_exec(env);
#ifdef CONFIG_PROFILER
    qemu_time += profile_getclock() - ti;
#endif
    if (use_icount) {
        /* Fold pending instructions back into the
           instruction counter, and clear the interrupt flag.  */
        qemu_icount -= (env->icount_decr.u16.low
                        + env->icount_extra);
        env->icount_decr.u32 = 0;
        env->icount_extra = 0;
    }
    return ret;
}

J
Jan Kiszka 已提交
1124
static void tcg_exec_all(void)
1125
{
1126 1127
    int r;

1128 1129 1130
    /* Account partial waits to the vm_clock.  */
    qemu_clock_warp(vm_clock);

1131
    if (next_cpu == NULL) {
1132
        next_cpu = first_cpu;
1133
    }
J
Jan Kiszka 已提交
1134
    for (; next_cpu != NULL && !exit_request; next_cpu = next_cpu->next_cpu) {
1135
        CPUArchState *env = next_cpu;
1136 1137

        qemu_clock_enable(vm_clock,
1138
                          (env->singlestep_enabled & SSTEP_NOTIMER) == 0);
1139

1140
        if (cpu_can_run(env)) {
J
Jan Kiszka 已提交
1141
            r = tcg_cpu_exec(env);
1142
            if (r == EXCP_DEBUG) {
1143
                cpu_handle_guest_debug(env);
1144 1145
                break;
            }
1146
        } else if (env->stop || env->stopped) {
1147 1148 1149
            break;
        }
    }
J
Jan Kiszka 已提交
1150
    exit_request = 0;
1151 1152 1153 1154
}

void set_numa_modes(void)
{
1155
    CPUArchState *env;
1156 1157 1158 1159
    int i;

    for (env = first_cpu; env != NULL; env = env->next_cpu) {
        for (i = 0; i < nb_numa_nodes; i++) {
1160
            if (test_bit(env->cpu_index, node_cpumask[i])) {
1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181
                env->numa_node = i;
            }
        }
    }
}

void set_cpu_log(const char *optarg)
{
    int mask;
    const CPULogItem *item;

    mask = cpu_str_to_log_mask(optarg);
    if (!mask) {
        printf("Log items (comma separated):\n");
        for (item = cpu_log_items; item->mask != 0; item++) {
            printf("%-10s %s\n", item->name, item->help);
        }
        exit(1);
    }
    cpu_set_log(mask);
}
B
Blue Swirl 已提交
1182

1183 1184 1185 1186 1187
void set_cpu_log_filename(const char *optarg)
{
    cpu_set_log_filename(optarg);
}

1188
void list_cpus(FILE *f, fprintf_function cpu_fprintf, const char *optarg)
1189 1190
{
    /* XXX: implement xxx_cpu_list for targets that still miss it */
P
Peter Maydell 已提交
1191 1192
#if defined(cpu_list)
    cpu_list(f, cpu_fprintf);
1193 1194
#endif
}
L
Luiz Capitulino 已提交
1195 1196 1197 1198

CpuInfoList *qmp_query_cpus(Error **errp)
{
    CpuInfoList *head = NULL, *cur_item = NULL;
1199
    CPUArchState *env;
L
Luiz Capitulino 已提交
1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238

    for(env = first_cpu; env != NULL; env = env->next_cpu) {
        CpuInfoList *info;

        cpu_synchronize_state(env);

        info = g_malloc0(sizeof(*info));
        info->value = g_malloc0(sizeof(*info->value));
        info->value->CPU = env->cpu_index;
        info->value->current = (env == first_cpu);
        info->value->halted = env->halted;
        info->value->thread_id = env->thread_id;
#if defined(TARGET_I386)
        info->value->has_pc = true;
        info->value->pc = env->eip + env->segs[R_CS].base;
#elif defined(TARGET_PPC)
        info->value->has_nip = true;
        info->value->nip = env->nip;
#elif defined(TARGET_SPARC)
        info->value->has_pc = true;
        info->value->pc = env->pc;
        info->value->has_npc = true;
        info->value->npc = env->npc;
#elif defined(TARGET_MIPS)
        info->value->has_PC = true;
        info->value->PC = env->active_tc.PC;
#endif

        /* XXX: waiting for the qapi to support GSList */
        if (!cur_item) {
            head = cur_item = info;
        } else {
            cur_item->next = info;
            cur_item = info;
        }
    }

    return head;
}
L
Luiz Capitulino 已提交
1239 1240 1241 1242 1243 1244

void qmp_memsave(int64_t addr, int64_t size, const char *filename,
                 bool has_cpu, int64_t cpu_index, Error **errp)
{
    FILE *f;
    uint32_t l;
1245
    CPUArchState *env;
L
Luiz Capitulino 已提交
1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285
    uint8_t buf[1024];

    if (!has_cpu) {
        cpu_index = 0;
    }

    for (env = first_cpu; env; env = env->next_cpu) {
        if (cpu_index == env->cpu_index) {
            break;
        }
    }

    if (env == NULL) {
        error_set(errp, QERR_INVALID_PARAMETER_VALUE, "cpu-index",
                  "a CPU number");
        return;
    }

    f = fopen(filename, "wb");
    if (!f) {
        error_set(errp, QERR_OPEN_FILE_FAILED, filename);
        return;
    }

    while (size != 0) {
        l = sizeof(buf);
        if (l > size)
            l = size;
        cpu_memory_rw_debug(env, addr, buf, l, 0);
        if (fwrite(buf, 1, l, f) != l) {
            error_set(errp, QERR_IO_ERROR);
            goto exit;
        }
        addr += l;
        size -= l;
    }

exit:
    fclose(f);
}
L
Luiz Capitulino 已提交
1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315

void qmp_pmemsave(int64_t addr, int64_t size, const char *filename,
                  Error **errp)
{
    FILE *f;
    uint32_t l;
    uint8_t buf[1024];

    f = fopen(filename, "wb");
    if (!f) {
        error_set(errp, QERR_OPEN_FILE_FAILED, filename);
        return;
    }

    while (size != 0) {
        l = sizeof(buf);
        if (l > size)
            l = size;
        cpu_physical_memory_rw(addr, buf, l, 0);
        if (fwrite(buf, 1, l, f) != l) {
            error_set(errp, QERR_IO_ERROR);
            goto exit;
        }
        addr += l;
        size -= l;
    }

exit:
    fclose(f);
}
L
Luiz Capitulino 已提交
1316 1317 1318 1319

void qmp_inject_nmi(Error **errp)
{
#if defined(TARGET_I386)
1320
    CPUArchState *env;
L
Luiz Capitulino 已提交
1321 1322

    for (env = first_cpu; env != NULL; env = env->next_cpu) {
1323 1324 1325 1326 1327
        if (!env->apic_state) {
            cpu_interrupt(env, CPU_INTERRUPT_NMI);
        } else {
            apic_deliver_nmi(env->apic_state);
        }
L
Luiz Capitulino 已提交
1328 1329 1330 1331 1332
    }
#else
    error_set(errp, QERR_UNSUPPORTED);
#endif
}