kvm-all.c 27.3 KB
Newer Older
A
aliguori 已提交
1 2 3 4
/*
 * QEMU KVM support
 *
 * Copyright IBM, Corp. 2008
5
 *           Red Hat, Inc. 2008
A
aliguori 已提交
6 7 8
 *
 * Authors:
 *  Anthony Liguori   <aliguori@us.ibm.com>
9
 *  Glauber Costa     <gcosta@redhat.com>
A
aliguori 已提交
10 11 12 13 14 15 16 17 18
 *
 * This work is licensed under the terms of the GNU GPL, version 2 or later.
 * See the COPYING file in the top-level directory.
 *
 */

#include <sys/types.h>
#include <sys/ioctl.h>
#include <sys/mman.h>
19
#include <stdarg.h>
A
aliguori 已提交
20 21 22 23 24

#include <linux/kvm.h>

#include "qemu-common.h"
#include "sysemu.h"
J
Jan Kiszka 已提交
25
#include "hw/hw.h"
26
#include "gdbstub.h"
A
aliguori 已提交
27 28
#include "kvm.h"

A
aliguori 已提交
29 30 31
/* KVM uses PAGE_SIZE in it's definition of COALESCED_MMIO_MAX */
#define PAGE_SIZE TARGET_PAGE_SIZE

A
aliguori 已提交
32 33 34 35 36 37 38 39 40 41
//#define DEBUG_KVM

#ifdef DEBUG_KVM
#define dprintf(fmt, ...) \
    do { fprintf(stderr, fmt, ## __VA_ARGS__); } while (0)
#else
#define dprintf(fmt, ...) \
    do { } while (0)
#endif

A
aliguori 已提交
42 43 44 45 46 47 48 49
typedef struct KVMSlot
{
    target_phys_addr_t start_addr;
    ram_addr_t memory_size;
    ram_addr_t phys_offset;
    int slot;
    int flags;
} KVMSlot;
A
aliguori 已提交
50

51 52
typedef struct kvm_dirty_log KVMDirtyLog;

A
aliguori 已提交
53 54 55 56 57 58 59
int kvm_allowed = 0;

struct KVMState
{
    KVMSlot slots[32];
    int fd;
    int vmfd;
A
aliguori 已提交
60
    int coalesced_mmio;
61
    int broken_set_mem_region;
62
    int migration_log;
63 64 65
#ifdef KVM_CAP_SET_GUEST_DEBUG
    struct kvm_sw_breakpoint_head kvm_sw_breakpoints;
#endif
66 67
    int irqchip_in_kernel;
    int pit_in_kernel;
A
aliguori 已提交
68 69 70 71 72 73 74 75 76
};

static KVMState *kvm_state;

static KVMSlot *kvm_alloc_slot(KVMState *s)
{
    int i;

    for (i = 0; i < ARRAY_SIZE(s->slots); i++) {
A
aliguori 已提交
77 78 79
        /* KVM private memory slots */
        if (i >= 8 && i < 12)
            continue;
A
aliguori 已提交
80 81 82 83
        if (s->slots[i].memory_size == 0)
            return &s->slots[i];
    }

84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102
    fprintf(stderr, "%s: no free slot available\n", __func__);
    abort();
}

static KVMSlot *kvm_lookup_matching_slot(KVMState *s,
                                         target_phys_addr_t start_addr,
                                         target_phys_addr_t end_addr)
{
    int i;

    for (i = 0; i < ARRAY_SIZE(s->slots); i++) {
        KVMSlot *mem = &s->slots[i];

        if (start_addr == mem->start_addr &&
            end_addr == mem->start_addr + mem->memory_size) {
            return mem;
        }
    }

A
aliguori 已提交
103 104 105
    return NULL;
}

106 107 108 109 110 111
/*
 * Find overlapping slot with lowest start address
 */
static KVMSlot *kvm_lookup_overlapping_slot(KVMState *s,
                                            target_phys_addr_t start_addr,
                                            target_phys_addr_t end_addr)
A
aliguori 已提交
112
{
113
    KVMSlot *found = NULL;
A
aliguori 已提交
114 115 116 117 118
    int i;

    for (i = 0; i < ARRAY_SIZE(s->slots); i++) {
        KVMSlot *mem = &s->slots[i];

119 120 121 122 123 124 125 126 127
        if (mem->memory_size == 0 ||
            (found && found->start_addr < mem->start_addr)) {
            continue;
        }

        if (end_addr > mem->start_addr &&
            start_addr < mem->start_addr + mem->memory_size) {
            found = mem;
        }
A
aliguori 已提交
128 129
    }

130
    return found;
A
aliguori 已提交
131 132
}

133 134 135 136 137 138 139
static int kvm_set_user_memory_region(KVMState *s, KVMSlot *slot)
{
    struct kvm_userspace_memory_region mem;

    mem.slot = slot->slot;
    mem.guest_phys_addr = slot->start_addr;
    mem.memory_size = slot->memory_size;
P
pbrook 已提交
140
    mem.userspace_addr = (unsigned long)qemu_get_ram_ptr(slot->phys_offset);
141
    mem.flags = slot->flags;
142 143 144
    if (s->migration_log) {
        mem.flags |= KVM_MEM_LOG_DIRTY_PAGES;
    }
145 146 147
    return kvm_vm_ioctl(s, KVM_SET_USER_MEMORY_REGION, &mem);
}

J
Jan Kiszka 已提交
148 149 150 151 152 153 154 155 156
static void kvm_reset_vcpu(void *opaque)
{
    CPUState *env = opaque;

    if (kvm_arch_put_registers(env)) {
        fprintf(stderr, "Fatal: kvm vcpu reset failed\n");
        abort();
    }
}
157

G
Glauber Costa 已提交
158 159 160 161 162 163 164 165 166
static void on_vcpu(CPUState *env, void (*func)(void *data), void *data)
{
    if (env == cpu_single_env) {
        func(data);
        return;
    }
    abort();
}

167 168 169 170 171 172 173 174 175 176 177
int kvm_irqchip_in_kernel(void)
{
    return kvm_state->irqchip_in_kernel;
}

int kvm_pit_in_kernel(void)
{
    return kvm_state->pit_in_kernel;
}


A
aliguori 已提交
178 179 180 181 182 183 184 185
int kvm_init_vcpu(CPUState *env)
{
    KVMState *s = kvm_state;
    long mmap_size;
    int ret;

    dprintf("kvm_init_vcpu\n");

186
    ret = kvm_vm_ioctl(s, KVM_CREATE_VCPU, env->cpu_index);
A
aliguori 已提交
187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209
    if (ret < 0) {
        dprintf("kvm_create_vcpu failed\n");
        goto err;
    }

    env->kvm_fd = ret;
    env->kvm_state = s;

    mmap_size = kvm_ioctl(s, KVM_GET_VCPU_MMAP_SIZE, 0);
    if (mmap_size < 0) {
        dprintf("KVM_GET_VCPU_MMAP_SIZE failed\n");
        goto err;
    }

    env->kvm_run = mmap(NULL, mmap_size, PROT_READ | PROT_WRITE, MAP_SHARED,
                        env->kvm_fd, 0);
    if (env->kvm_run == MAP_FAILED) {
        ret = -errno;
        dprintf("mmap'ing vcpu state failed\n");
        goto err;
    }

    ret = kvm_arch_init_vcpu(env);
J
Jan Kiszka 已提交
210
    if (ret == 0) {
211
        qemu_register_reset(kvm_reset_vcpu, env);
J
Jan Kiszka 已提交
212 213
        ret = kvm_arch_put_registers(env);
    }
A
aliguori 已提交
214 215 216 217
err:
    return ret;
}

218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237
int kvm_put_mp_state(CPUState *env)
{
    struct kvm_mp_state mp_state = { .mp_state = env->mp_state };

    return kvm_vcpu_ioctl(env, KVM_SET_MP_STATE, &mp_state);
}

int kvm_get_mp_state(CPUState *env)
{
    struct kvm_mp_state mp_state;
    int ret;

    ret = kvm_vcpu_ioctl(env, KVM_GET_MP_STATE, &mp_state);
    if (ret < 0) {
        return ret;
    }
    env->mp_state = mp_state.mp_state;
    return 0;
}

238 239 240
/*
 * dirty pages logging control
 */
241
static int kvm_dirty_pages_log_change(target_phys_addr_t phys_addr,
242
                                      ram_addr_t size, int flags, int mask)
243 244
{
    KVMState *s = kvm_state;
245
    KVMSlot *mem = kvm_lookup_matching_slot(s, phys_addr, phys_addr + size);
246 247
    int old_flags;

248
    if (mem == NULL)  {
249 250
            fprintf(stderr, "BUG: %s: invalid parameters " TARGET_FMT_plx "-"
                    TARGET_FMT_plx "\n", __func__, phys_addr,
A
Alexander Graf 已提交
251
                    (target_phys_addr_t)(phys_addr + size - 1));
252 253 254
            return -EINVAL;
    }

255
    old_flags = mem->flags;
256

257
    flags = (mem->flags & ~mask) | flags;
258 259
    mem->flags = flags;

260 261 262 263 264 265 266 267
    /* If nothing changed effectively, no need to issue ioctl */
    if (s->migration_log) {
        flags |= KVM_MEM_LOG_DIRTY_PAGES;
    }
    if (flags == old_flags) {
            return 0;
    }

268 269 270
    return kvm_set_user_memory_region(s, mem);
}

271
int kvm_log_start(target_phys_addr_t phys_addr, ram_addr_t size)
272
{
273
        return kvm_dirty_pages_log_change(phys_addr, size,
274 275 276 277
                                          KVM_MEM_LOG_DIRTY_PAGES,
                                          KVM_MEM_LOG_DIRTY_PAGES);
}

278
int kvm_log_stop(target_phys_addr_t phys_addr, ram_addr_t size)
279
{
280
        return kvm_dirty_pages_log_change(phys_addr, size,
281 282 283 284
                                          0,
                                          KVM_MEM_LOG_DIRTY_PAGES);
}

285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306
int kvm_set_migration_log(int enable)
{
    KVMState *s = kvm_state;
    KVMSlot *mem;
    int i, err;

    s->migration_log = enable;

    for (i = 0; i < ARRAY_SIZE(s->slots); i++) {
        mem = &s->slots[i];

        if (!!(mem->flags & KVM_MEM_LOG_DIRTY_PAGES) == enable) {
            continue;
        }
        err = kvm_set_user_memory_region(s, mem);
        if (err) {
            return err;
        }
    }
    return 0;
}

A
Alexander Graf 已提交
307 308 309 310 311
static int test_le_bit(unsigned long nr, unsigned char *addr)
{
    return (addr[nr >> 3] >> (nr & 7)) & 1;
}

312 313 314 315 316
/**
 * kvm_physical_sync_dirty_bitmap - Grab dirty bitmap from kernel space
 * This function updates qemu's dirty bitmap using cpu_physical_memory_set_dirty().
 * This means all bits are set to dirty.
 *
317
 * @start_add: start of logged region.
318 319
 * @end_addr: end of logged region.
 */
320 321
int kvm_physical_sync_dirty_bitmap(target_phys_addr_t start_addr,
                                   target_phys_addr_t end_addr)
322 323
{
    KVMState *s = kvm_state;
324 325
    unsigned long size, allocated_size = 0;
    target_phys_addr_t phys_addr;
326
    ram_addr_t addr;
327 328 329
    KVMDirtyLog d;
    KVMSlot *mem;
    int ret = 0;
330
    int r;
331

332 333 334 335 336 337
    d.dirty_bitmap = NULL;
    while (start_addr < end_addr) {
        mem = kvm_lookup_overlapping_slot(s, start_addr, end_addr);
        if (mem == NULL) {
            break;
        }
338

339 340 341 342 343
        /* We didn't activate dirty logging? Don't care then. */
        if(!(mem->flags & KVM_MEM_LOG_DIRTY_PAGES)) {
            continue;
        }

344 345 346 347 348 349 350 351
        size = ((mem->memory_size >> TARGET_PAGE_BITS) + 7) / 8;
        if (!d.dirty_bitmap) {
            d.dirty_bitmap = qemu_malloc(size);
        } else if (size > allocated_size) {
            d.dirty_bitmap = qemu_realloc(d.dirty_bitmap, size);
        }
        allocated_size = size;
        memset(d.dirty_bitmap, 0, allocated_size);
352

353
        d.slot = mem->slot;
354

355 356
        r = kvm_vm_ioctl(s, KVM_GET_DIRTY_LOG, &d);
        if (r == -EINVAL) {
357 358 359 360
            dprintf("ioctl failed %d\n", errno);
            ret = -1;
            break;
        }
361

362 363 364
        for (phys_addr = mem->start_addr, addr = mem->phys_offset;
             phys_addr < mem->start_addr + mem->memory_size;
             phys_addr += TARGET_PAGE_SIZE, addr += TARGET_PAGE_SIZE) {
A
Alexander Graf 已提交
365
            unsigned char *bitmap = (unsigned char *)d.dirty_bitmap;
366 367
            unsigned nr = (phys_addr - mem->start_addr) >> TARGET_PAGE_BITS;

A
Alexander Graf 已提交
368
            if (test_le_bit(nr, bitmap)) {
369
                cpu_physical_memory_set_dirty(addr);
370 371 372 373
            } else if (r < 0) {
                /* When our KVM implementation doesn't know about dirty logging
                 * we can just assume it's always dirty and be fine. */
                cpu_physical_memory_set_dirty(addr);
374 375 376
            }
        }
        start_addr = phys_addr;
377 378
    }
    qemu_free(d.dirty_bitmap);
379 380

    return ret;
381 382
}

A
aliguori 已提交
383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420
int kvm_coalesce_mmio_region(target_phys_addr_t start, ram_addr_t size)
{
    int ret = -ENOSYS;
#ifdef KVM_CAP_COALESCED_MMIO
    KVMState *s = kvm_state;

    if (s->coalesced_mmio) {
        struct kvm_coalesced_mmio_zone zone;

        zone.addr = start;
        zone.size = size;

        ret = kvm_vm_ioctl(s, KVM_REGISTER_COALESCED_MMIO, &zone);
    }
#endif

    return ret;
}

int kvm_uncoalesce_mmio_region(target_phys_addr_t start, ram_addr_t size)
{
    int ret = -ENOSYS;
#ifdef KVM_CAP_COALESCED_MMIO
    KVMState *s = kvm_state;

    if (s->coalesced_mmio) {
        struct kvm_coalesced_mmio_zone zone;

        zone.addr = start;
        zone.size = size;

        ret = kvm_vm_ioctl(s, KVM_UNREGISTER_COALESCED_MMIO, &zone);
    }
#endif

    return ret;
}

421 422 423 424 425 426 427 428 429 430 431 432
int kvm_check_extension(KVMState *s, unsigned int extension)
{
    int ret;

    ret = kvm_ioctl(s, KVM_CHECK_EXTENSION, extension);
    if (ret < 0) {
        ret = 0;
    }

    return ret;
}

A
aliguori 已提交
433 434
int kvm_init(int smp_cpus)
{
435 436 437
    static const char upgrade_note[] =
        "Please upgrade to at least kernel 2.6.29 or recent kvm-kmod\n"
        "(see http://sourceforge.net/projects/kvm).\n";
A
aliguori 已提交
438 439 440 441
    KVMState *s;
    int ret;
    int i;

442 443
    if (smp_cpus > 1) {
        fprintf(stderr, "No SMP KVM support, use '-smp 1'\n");
A
aliguori 已提交
444
        return -EINVAL;
445
    }
A
aliguori 已提交
446 447 448

    s = qemu_mallocz(sizeof(KVMState));

449 450 451
#ifdef KVM_CAP_SET_GUEST_DEBUG
    TAILQ_INIT(&s->kvm_sw_breakpoints);
#endif
A
aliguori 已提交
452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482
    for (i = 0; i < ARRAY_SIZE(s->slots); i++)
        s->slots[i].slot = i;

    s->vmfd = -1;
    s->fd = open("/dev/kvm", O_RDWR);
    if (s->fd == -1) {
        fprintf(stderr, "Could not access KVM kernel module: %m\n");
        ret = -errno;
        goto err;
    }

    ret = kvm_ioctl(s, KVM_GET_API_VERSION, 0);
    if (ret < KVM_API_VERSION) {
        if (ret > 0)
            ret = -EINVAL;
        fprintf(stderr, "kvm version too old\n");
        goto err;
    }

    if (ret > KVM_API_VERSION) {
        ret = -EINVAL;
        fprintf(stderr, "kvm version not supported\n");
        goto err;
    }

    s->vmfd = kvm_ioctl(s, KVM_CREATE_VM, 0);
    if (s->vmfd < 0)
        goto err;

    /* initially, KVM allocated its own memory and we had to jump through
     * hooks to make phys_ram_base point to this.  Modern versions of KVM
P
pbrook 已提交
483
     * just use a user allocated buffer so we can use regular pages
A
aliguori 已提交
484 485
     * unmodified.  Make sure we have a sufficiently modern version of KVM.
     */
486 487
    if (!kvm_check_extension(s, KVM_CAP_USER_MEMORY)) {
        ret = -EINVAL;
488 489
        fprintf(stderr, "kvm does not support KVM_CAP_USER_MEMORY\n%s",
                upgrade_note);
A
aliguori 已提交
490 491 492
        goto err;
    }

493 494 495
    /* There was a nasty bug in < kvm-80 that prevents memory slots from being
     * destroyed properly.  Since we rely on this capability, refuse to work
     * with any kernel without this capability. */
496 497
    if (!kvm_check_extension(s, KVM_CAP_DESTROY_MEMORY_REGION_WORKS)) {
        ret = -EINVAL;
498 499

        fprintf(stderr,
500 501
                "KVM kernel module broken (DESTROY_MEMORY_REGION).\n%s",
                upgrade_note);
502 503 504
        goto err;
    }

A
aliguori 已提交
505
#ifdef KVM_CAP_COALESCED_MMIO
506 507 508
    s->coalesced_mmio = kvm_check_extension(s, KVM_CAP_COALESCED_MMIO);
#else
    s->coalesced_mmio = 0;
A
aliguori 已提交
509 510
#endif

511 512 513 514 515 516 517 518
    s->broken_set_mem_region = 1;
#ifdef KVM_CAP_JOIN_MEMORY_REGIONS_WORKS
    ret = kvm_ioctl(s, KVM_CHECK_EXTENSION, KVM_CAP_JOIN_MEMORY_REGIONS_WORKS);
    if (ret > 0) {
        s->broken_set_mem_region = 0;
    }
#endif

A
aliguori 已提交
519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577
    ret = kvm_arch_init(s, smp_cpus);
    if (ret < 0)
        goto err;

    kvm_state = s;

    return 0;

err:
    if (s) {
        if (s->vmfd != -1)
            close(s->vmfd);
        if (s->fd != -1)
            close(s->fd);
    }
    qemu_free(s);

    return ret;
}

static int kvm_handle_io(CPUState *env, uint16_t port, void *data,
                         int direction, int size, uint32_t count)
{
    int i;
    uint8_t *ptr = data;

    for (i = 0; i < count; i++) {
        if (direction == KVM_EXIT_IO_IN) {
            switch (size) {
            case 1:
                stb_p(ptr, cpu_inb(env, port));
                break;
            case 2:
                stw_p(ptr, cpu_inw(env, port));
                break;
            case 4:
                stl_p(ptr, cpu_inl(env, port));
                break;
            }
        } else {
            switch (size) {
            case 1:
                cpu_outb(env, port, ldub_p(ptr));
                break;
            case 2:
                cpu_outw(env, port, lduw_p(ptr));
                break;
            case 4:
                cpu_outl(env, port, ldl_p(ptr));
                break;
            }
        }

        ptr += size;
    }

    return 1;
}

A
aliguori 已提交
578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598
static void kvm_run_coalesced_mmio(CPUState *env, struct kvm_run *run)
{
#ifdef KVM_CAP_COALESCED_MMIO
    KVMState *s = kvm_state;
    if (s->coalesced_mmio) {
        struct kvm_coalesced_mmio_ring *ring;

        ring = (void *)run + (s->coalesced_mmio * TARGET_PAGE_SIZE);
        while (ring->first != ring->last) {
            struct kvm_coalesced_mmio *ent;

            ent = &ring->coalesced_mmio[ring->first];

            cpu_physical_memory_write(ent->phys_addr, ent->data, ent->len);
            /* FIXME smp_wmb() */
            ring->first = (ring->first + 1) % KVM_COALESCED_MMIO_MAX;
        }
    }
#endif
}

A
aliguori 已提交
599 600 601 602 603 604 605 606
int kvm_cpu_exec(CPUState *env)
{
    struct kvm_run *run = env->kvm_run;
    int ret;

    dprintf("kvm_cpu_exec()\n");

    do {
607
        if (env->exit_request) {
A
aliguori 已提交
608 609 610 611 612
            dprintf("interrupt exit requested\n");
            ret = 0;
            break;
        }

613
        kvm_arch_pre_run(env, run);
A
aliguori 已提交
614 615 616 617 618 619 620 621 622 623 624 625 626 627
        ret = kvm_vcpu_ioctl(env, KVM_RUN, 0);
        kvm_arch_post_run(env, run);

        if (ret == -EINTR || ret == -EAGAIN) {
            dprintf("io window exit\n");
            ret = 0;
            break;
        }

        if (ret < 0) {
            dprintf("kvm run failed %s\n", strerror(-ret));
            abort();
        }

A
aliguori 已提交
628 629
        kvm_run_coalesced_mmio(env, run);

A
aliguori 已提交
630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666
        ret = 0; /* exit loop */
        switch (run->exit_reason) {
        case KVM_EXIT_IO:
            dprintf("handle_io\n");
            ret = kvm_handle_io(env, run->io.port,
                                (uint8_t *)run + run->io.data_offset,
                                run->io.direction,
                                run->io.size,
                                run->io.count);
            break;
        case KVM_EXIT_MMIO:
            dprintf("handle_mmio\n");
            cpu_physical_memory_rw(run->mmio.phys_addr,
                                   run->mmio.data,
                                   run->mmio.len,
                                   run->mmio.is_write);
            ret = 1;
            break;
        case KVM_EXIT_IRQ_WINDOW_OPEN:
            dprintf("irq_window_open\n");
            break;
        case KVM_EXIT_SHUTDOWN:
            dprintf("shutdown\n");
            qemu_system_reset_request();
            ret = 1;
            break;
        case KVM_EXIT_UNKNOWN:
            dprintf("kvm_exit_unknown\n");
            break;
        case KVM_EXIT_FAIL_ENTRY:
            dprintf("kvm_exit_fail_entry\n");
            break;
        case KVM_EXIT_EXCEPTION:
            dprintf("kvm_exit_exception\n");
            break;
        case KVM_EXIT_DEBUG:
            dprintf("kvm_exit_debug\n");
667 668 669 670 671 672 673 674 675 676
#ifdef KVM_CAP_SET_GUEST_DEBUG
            if (kvm_arch_debug(&run->debug.arch)) {
                gdb_set_stop_cpu(env);
                vm_stop(EXCP_DEBUG);
                env->exception_index = EXCP_DEBUG;
                return 0;
            }
            /* re-enter, this exception was guest-internal */
            ret = 1;
#endif /* KVM_CAP_SET_GUEST_DEBUG */
A
aliguori 已提交
677 678 679 680 681 682 683 684
            break;
        default:
            dprintf("kvm_arch_handle_exit\n");
            ret = kvm_arch_handle_exit(env, run);
            break;
        }
    } while (ret > 0);

685 686
    if (env->exit_request) {
        env->exit_request = 0;
A
aliguori 已提交
687 688 689
        env->exception_index = EXCP_INTERRUPT;
    }

A
aliguori 已提交
690 691 692 693 694 695 696 697 698
    return ret;
}

void kvm_set_phys_mem(target_phys_addr_t start_addr,
                      ram_addr_t size,
                      ram_addr_t phys_offset)
{
    KVMState *s = kvm_state;
    ram_addr_t flags = phys_offset & ~TARGET_PAGE_MASK;
699 700
    KVMSlot *mem, old;
    int err;
A
aliguori 已提交
701

702
    if (start_addr & ~TARGET_PAGE_MASK) {
703 704 705 706 707 708 709 710 711
        if (flags >= IO_MEM_UNASSIGNED) {
            if (!kvm_lookup_overlapping_slot(s, start_addr,
                                             start_addr + size)) {
                return;
            }
            fprintf(stderr, "Unaligned split of a KVM memory slot\n");
        } else {
            fprintf(stderr, "Only page-aligned memory slots supported\n");
        }
712 713 714
        abort();
    }

A
aliguori 已提交
715 716 717
    /* KVM does not support read-only slots */
    phys_offset &= ~IO_MEM_ROM;

718 719 720 721 722
    while (1) {
        mem = kvm_lookup_overlapping_slot(s, start_addr, start_addr + size);
        if (!mem) {
            break;
        }
A
aliguori 已提交
723

724 725 726 727 728
        if (flags < IO_MEM_UNASSIGNED && start_addr >= mem->start_addr &&
            (start_addr + size <= mem->start_addr + mem->memory_size) &&
            (phys_offset - start_addr == mem->phys_offset - mem->start_addr)) {
            /* The new slot fits into the existing one and comes with
             * identical parameters - nothing to be done. */
A
aliguori 已提交
729
            return;
730 731 732 733 734 735 736 737 738 739
        }

        old = *mem;

        /* unregister the overlapping slot */
        mem->memory_size = 0;
        err = kvm_set_user_memory_region(s, mem);
        if (err) {
            fprintf(stderr, "%s: error unregistering overlapping slot: %s\n",
                    __func__, strerror(-err));
A
aliguori 已提交
740 741
            abort();
        }
742 743 744 745 746 747 748 749 750

        /* Workaround for older KVM versions: we can't join slots, even not by
         * unregistering the previous ones and then registering the larger
         * slot. We have to maintain the existing fragmentation. Sigh.
         *
         * This workaround assumes that the new slot starts at the same
         * address as the first existing one. If not or if some overlapping
         * slot comes around later, we will fail (not seen in practice so far)
         * - and actually require a recent KVM version. */
751 752
        if (s->broken_set_mem_region &&
            old.start_addr == start_addr && old.memory_size < size &&
753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806
            flags < IO_MEM_UNASSIGNED) {
            mem = kvm_alloc_slot(s);
            mem->memory_size = old.memory_size;
            mem->start_addr = old.start_addr;
            mem->phys_offset = old.phys_offset;
            mem->flags = 0;

            err = kvm_set_user_memory_region(s, mem);
            if (err) {
                fprintf(stderr, "%s: error updating slot: %s\n", __func__,
                        strerror(-err));
                abort();
            }

            start_addr += old.memory_size;
            phys_offset += old.memory_size;
            size -= old.memory_size;
            continue;
        }

        /* register prefix slot */
        if (old.start_addr < start_addr) {
            mem = kvm_alloc_slot(s);
            mem->memory_size = start_addr - old.start_addr;
            mem->start_addr = old.start_addr;
            mem->phys_offset = old.phys_offset;
            mem->flags = 0;

            err = kvm_set_user_memory_region(s, mem);
            if (err) {
                fprintf(stderr, "%s: error registering prefix slot: %s\n",
                        __func__, strerror(-err));
                abort();
            }
        }

        /* register suffix slot */
        if (old.start_addr + old.memory_size > start_addr + size) {
            ram_addr_t size_delta;

            mem = kvm_alloc_slot(s);
            mem->start_addr = start_addr + size;
            size_delta = mem->start_addr - old.start_addr;
            mem->memory_size = old.memory_size - size_delta;
            mem->phys_offset = old.phys_offset + size_delta;
            mem->flags = 0;

            err = kvm_set_user_memory_region(s, mem);
            if (err) {
                fprintf(stderr, "%s: error registering suffix slot: %s\n",
                        __func__, strerror(-err));
                abort();
            }
        }
A
aliguori 已提交
807
    }
808 809 810 811 812

    /* in case the KVM bug workaround already "consumed" the new slot */
    if (!size)
        return;

A
aliguori 已提交
813 814 815 816 817 818
    /* KVM does not need to know about this memory */
    if (flags >= IO_MEM_UNASSIGNED)
        return;

    mem = kvm_alloc_slot(s);
    mem->memory_size = size;
A
aliguori 已提交
819 820
    mem->start_addr = start_addr;
    mem->phys_offset = phys_offset;
A
aliguori 已提交
821 822
    mem->flags = 0;

823 824 825 826 827 828
    err = kvm_set_user_memory_region(s, mem);
    if (err) {
        fprintf(stderr, "%s: error registering slot: %s\n", __func__,
                strerror(-err));
        abort();
    }
A
aliguori 已提交
829 830
}

831
int kvm_ioctl(KVMState *s, int type, ...)
A
aliguori 已提交
832 833
{
    int ret;
834 835
    void *arg;
    va_list ap;
A
aliguori 已提交
836

837 838 839 840 841
    va_start(ap, type);
    arg = va_arg(ap, void *);
    va_end(ap);

    ret = ioctl(s->fd, type, arg);
A
aliguori 已提交
842 843 844 845 846 847
    if (ret == -1)
        ret = -errno;

    return ret;
}

848
int kvm_vm_ioctl(KVMState *s, int type, ...)
A
aliguori 已提交
849 850
{
    int ret;
851 852 853 854 855 856
    void *arg;
    va_list ap;

    va_start(ap, type);
    arg = va_arg(ap, void *);
    va_end(ap);
A
aliguori 已提交
857

858
    ret = ioctl(s->vmfd, type, arg);
A
aliguori 已提交
859 860 861 862 863 864
    if (ret == -1)
        ret = -errno;

    return ret;
}

865
int kvm_vcpu_ioctl(CPUState *env, int type, ...)
A
aliguori 已提交
866 867
{
    int ret;
868 869 870 871 872 873
    void *arg;
    va_list ap;

    va_start(ap, type);
    arg = va_arg(ap, void *);
    va_end(ap);
A
aliguori 已提交
874

875
    ret = ioctl(env->kvm_fd, type, arg);
A
aliguori 已提交
876 877 878 879 880
    if (ret == -1)
        ret = -errno;

    return ret;
}
A
aliguori 已提交
881 882 883

int kvm_has_sync_mmu(void)
{
A
aurel32 已提交
884
#ifdef KVM_CAP_SYNC_MMU
A
aliguori 已提交
885 886
    KVMState *s = kvm_state;

887 888
    return kvm_check_extension(s, KVM_CAP_SYNC_MMU);
#else
A
aliguori 已提交
889
    return 0;
890
#endif
A
aliguori 已提交
891
}
892

893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910
void kvm_setup_guest_memory(void *start, size_t size)
{
    if (!kvm_has_sync_mmu()) {
#ifdef MADV_DONTFORK
        int ret = madvise(start, size, MADV_DONTFORK);

        if (ret) {
            perror("madvice");
            exit(1);
        }
#else
        fprintf(stderr,
                "Need MADV_DONTFORK in absence of synchronous KVM MMU\n");
        exit(1);
#endif
    }
}

911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928
#ifdef KVM_CAP_SET_GUEST_DEBUG
struct kvm_sw_breakpoint *kvm_find_sw_breakpoint(CPUState *env,
                                                 target_ulong pc)
{
    struct kvm_sw_breakpoint *bp;

    TAILQ_FOREACH(bp, &env->kvm_state->kvm_sw_breakpoints, entry) {
        if (bp->pc == pc)
            return bp;
    }
    return NULL;
}

int kvm_sw_breakpoints_active(CPUState *env)
{
    return !TAILQ_EMPTY(&env->kvm_state->kvm_sw_breakpoints);
}

G
Glauber Costa 已提交
929 930 931 932 933 934 935 936 937 938 939 940
struct kvm_set_guest_debug_data {
    struct kvm_guest_debug dbg;
    CPUState *env;
    int err;
};

static void kvm_invoke_set_guest_debug(void *data)
{
    struct kvm_set_guest_debug_data *dbg_data = data;
    dbg_data->err = kvm_vcpu_ioctl(dbg_data->env, KVM_SET_GUEST_DEBUG, &dbg_data->dbg);
}

941 942
int kvm_update_guest_debug(CPUState *env, unsigned long reinject_trap)
{
G
Glauber Costa 已提交
943
    struct kvm_set_guest_debug_data data;
944

G
Glauber Costa 已提交
945
    data.dbg.control = 0;
946
    if (env->singlestep_enabled)
G
Glauber Costa 已提交
947
        data.dbg.control = KVM_GUESTDBG_ENABLE | KVM_GUESTDBG_SINGLESTEP;
948

G
Glauber Costa 已提交
949 950 951
    kvm_arch_update_guest_debug(env, &data.dbg);
    data.dbg.control |= reinject_trap;
    data.env = env;
952

G
Glauber Costa 已提交
953 954
    on_vcpu(env, kvm_invoke_set_guest_debug, &data);
    return data.err;
955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079
}

int kvm_insert_breakpoint(CPUState *current_env, target_ulong addr,
                          target_ulong len, int type)
{
    struct kvm_sw_breakpoint *bp;
    CPUState *env;
    int err;

    if (type == GDB_BREAKPOINT_SW) {
        bp = kvm_find_sw_breakpoint(current_env, addr);
        if (bp) {
            bp->use_count++;
            return 0;
        }

        bp = qemu_malloc(sizeof(struct kvm_sw_breakpoint));
        if (!bp)
            return -ENOMEM;

        bp->pc = addr;
        bp->use_count = 1;
        err = kvm_arch_insert_sw_breakpoint(current_env, bp);
        if (err) {
            free(bp);
            return err;
        }

        TAILQ_INSERT_HEAD(&current_env->kvm_state->kvm_sw_breakpoints,
                          bp, entry);
    } else {
        err = kvm_arch_insert_hw_breakpoint(addr, len, type);
        if (err)
            return err;
    }

    for (env = first_cpu; env != NULL; env = env->next_cpu) {
        err = kvm_update_guest_debug(env, 0);
        if (err)
            return err;
    }
    return 0;
}

int kvm_remove_breakpoint(CPUState *current_env, target_ulong addr,
                          target_ulong len, int type)
{
    struct kvm_sw_breakpoint *bp;
    CPUState *env;
    int err;

    if (type == GDB_BREAKPOINT_SW) {
        bp = kvm_find_sw_breakpoint(current_env, addr);
        if (!bp)
            return -ENOENT;

        if (bp->use_count > 1) {
            bp->use_count--;
            return 0;
        }

        err = kvm_arch_remove_sw_breakpoint(current_env, bp);
        if (err)
            return err;

        TAILQ_REMOVE(&current_env->kvm_state->kvm_sw_breakpoints, bp, entry);
        qemu_free(bp);
    } else {
        err = kvm_arch_remove_hw_breakpoint(addr, len, type);
        if (err)
            return err;
    }

    for (env = first_cpu; env != NULL; env = env->next_cpu) {
        err = kvm_update_guest_debug(env, 0);
        if (err)
            return err;
    }
    return 0;
}

void kvm_remove_all_breakpoints(CPUState *current_env)
{
    struct kvm_sw_breakpoint *bp, *next;
    KVMState *s = current_env->kvm_state;
    CPUState *env;

    TAILQ_FOREACH_SAFE(bp, &s->kvm_sw_breakpoints, entry, next) {
        if (kvm_arch_remove_sw_breakpoint(current_env, bp) != 0) {
            /* Try harder to find a CPU that currently sees the breakpoint. */
            for (env = first_cpu; env != NULL; env = env->next_cpu) {
                if (kvm_arch_remove_sw_breakpoint(env, bp) == 0)
                    break;
            }
        }
    }
    kvm_arch_remove_all_hw_breakpoints();

    for (env = first_cpu; env != NULL; env = env->next_cpu)
        kvm_update_guest_debug(env, 0);
}

#else /* !KVM_CAP_SET_GUEST_DEBUG */

int kvm_update_guest_debug(CPUState *env, unsigned long reinject_trap)
{
    return -EINVAL;
}

int kvm_insert_breakpoint(CPUState *current_env, target_ulong addr,
                          target_ulong len, int type)
{
    return -EINVAL;
}

int kvm_remove_breakpoint(CPUState *current_env, target_ulong addr,
                          target_ulong len, int type)
{
    return -EINVAL;
}

void kvm_remove_all_breakpoints(CPUState *current_env)
{
}
#endif /* !KVM_CAP_SET_GUEST_DEBUG */