kvm-all.c 57.3 KB
Newer Older
1 2 3 4
/*
 * QEMU KVM support
 *
 * Copyright IBM, Corp. 2008
5
 *           Red Hat, Inc. 2008
6 7 8
 *
 * Authors:
 *  Anthony Liguori   <aliguori@us.ibm.com>
9
 *  Glauber Costa     <gcosta@redhat.com>
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>
20 21 22 23

#include <linux/kvm.h>

#include "qemu-common.h"
24 25 26
#include "qemu/atomic.h"
#include "qemu/option.h"
#include "qemu/config-file.h"
27
#include "sysemu/sysemu.h"
28
#include "sysemu/accel.h"
Jan Kiszka's avatar
Jan Kiszka committed
29
#include "hw/hw.h"
30
#include "hw/pci/msi.h"
31
#include "hw/s390x/adapter.h"
32
#include "exec/gdbstub.h"
33
#include "sysemu/kvm.h"
34
#include "qemu/bswap.h"
35
#include "exec/memory.h"
36
#include "exec/ram_addr.h"
37
#include "exec/address-spaces.h"
38
#include "qemu/event_notifier.h"
39
#include "trace.h"
40

41 42
#include "hw/boards.h"

43 44 45 46 47
/* This check must be after config-host.h is included */
#ifdef CONFIG_EVENTFD
#include <sys/eventfd.h>
#endif

48
/* KVM uses PAGE_SIZE in its definition of COALESCED_MMIO_MAX */
aliguori's avatar
aliguori committed
49 50
#define PAGE_SIZE TARGET_PAGE_SIZE

51 52 53
//#define DEBUG_KVM

#ifdef DEBUG_KVM
54
#define DPRINTF(fmt, ...) \
55 56
    do { fprintf(stderr, fmt, ## __VA_ARGS__); } while (0)
#else
57
#define DPRINTF(fmt, ...) \
58 59 60
    do { } while (0)
#endif

61 62
#define KVM_MSI_HASHTAB_SIZE    256

aliguori's avatar
aliguori committed
63 64
typedef struct KVMSlot
{
65
    hwaddr start_addr;
66
    ram_addr_t memory_size;
67
    void *ram;
aliguori's avatar
aliguori committed
68 69 70
    int slot;
    int flags;
} KVMSlot;
71

72 73
typedef struct kvm_dirty_log KVMDirtyLog;

74
struct KVMState
75
{
76 77
    AccelState parent_obj;

78 79
    KVMSlot *slots;
    int nr_slots;
80 81
    int fd;
    int vmfd;
aliguori's avatar
aliguori committed
82
    int coalesced_mmio;
83
    struct kvm_coalesced_mmio_ring *coalesced_mmio_ring;
84
    bool coalesced_flush_in_progress;
85
    int broken_set_mem_region;
86
    int migration_log;
87
    int vcpu_events;
88
    int robust_singlestep;
89
    int debugregs;
90 91 92
#ifdef KVM_CAP_SET_GUEST_DEBUG
    struct kvm_sw_breakpoint_head kvm_sw_breakpoints;
#endif
93
    int pit_state2;
94
    int xsave, xcrs;
95
    int many_ioeventfds;
96
    int intx_set_mask;
97 98 99
    /* The man page (and posix) say ioctl numbers are signed int, but
     * they're not.  Linux, glibc and *BSD all treat ioctl numbers as
     * unsigned, and treating them as signed here can break things */
100
    unsigned irq_set_ioctl;
101
    unsigned int sigmask_len;
102 103 104 105
#ifdef KVM_CAP_IRQ_ROUTING
    struct kvm_irq_routing *irq_routes;
    int nr_allocated_irq_routes;
    uint32_t *used_gsi_bitmap;
106
    unsigned int gsi_count;
107
    QTAILQ_HEAD(msi_hashtab, KVMMSIRoute) msi_hashtab[KVM_MSI_HASHTAB_SIZE];
108
    bool direct_msi;
109
#endif
110
};
111

112 113
#define TYPE_KVM_ACCEL ACCEL_CLASS_NAME("kvm")

114 115 116
#define KVM_STATE(obj) \
    OBJECT_CHECK(KVMState, (obj), TYPE_KVM_ACCEL)

117
KVMState *kvm_state;
118
bool kvm_kernel_irqchip;
119
bool kvm_async_interrupts_allowed;
120
bool kvm_halt_in_kernel_allowed;
121
bool kvm_eventfds_allowed;
122
bool kvm_irqfds_allowed;
123
bool kvm_msi_via_irqfd_allowed;
124
bool kvm_gsi_routing_allowed;
125
bool kvm_gsi_direct_mapping;
126
bool kvm_allowed;
127
bool kvm_readonly_mem_allowed;
128

129 130 131 132 133 134
static const KVMCapabilityInfo kvm_required_capabilites[] = {
    KVM_CAP_INFO(USER_MEMORY),
    KVM_CAP_INFO(DESTROY_MEMORY_REGION_WORKS),
    KVM_CAP_LAST_INFO
};

135
static KVMSlot *kvm_get_free_slot(KVMState *s)
136 137 138
{
    int i;

139
    for (i = 0; i < s->nr_slots; i++) {
140
        if (s->slots[i].memory_size == 0) {
141
            return &s->slots[i];
142
        }
143 144
    }

145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160
    return NULL;
}

bool kvm_has_free_slot(MachineState *ms)
{
    return kvm_get_free_slot(KVM_STATE(ms->accelerator));
}

static KVMSlot *kvm_alloc_slot(KVMState *s)
{
    KVMSlot *slot = kvm_get_free_slot(s);

    if (slot) {
        return slot;
    }

161 162 163 164 165
    fprintf(stderr, "%s: no free slot available\n", __func__);
    abort();
}

static KVMSlot *kvm_lookup_matching_slot(KVMState *s,
166 167
                                         hwaddr start_addr,
                                         hwaddr end_addr)
168 169 170
{
    int i;

171
    for (i = 0; i < s->nr_slots; i++) {
172 173 174 175 176 177 178 179
        KVMSlot *mem = &s->slots[i];

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

180 181 182
    return NULL;
}

183 184 185 186
/*
 * Find overlapping slot with lowest start address
 */
static KVMSlot *kvm_lookup_overlapping_slot(KVMState *s,
187 188
                                            hwaddr start_addr,
                                            hwaddr end_addr)
189
{
190
    KVMSlot *found = NULL;
191 192
    int i;

193
    for (i = 0; i < s->nr_slots; i++) {
194 195
        KVMSlot *mem = &s->slots[i];

196 197 198 199 200 201 202 203 204
        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;
        }
205 206
    }

207
    return found;
208 209
}

210
int kvm_physical_memory_addr_from_host(KVMState *s, void *ram,
211
                                       hwaddr *phys_addr)
212 213 214
{
    int i;

215
    for (i = 0; i < s->nr_slots; i++) {
216 217
        KVMSlot *mem = &s->slots[i];

218 219
        if (ram >= mem->ram && ram < mem->ram + mem->memory_size) {
            *phys_addr = mem->start_addr + (ram - mem->ram);
220 221 222 223 224 225 226
            return 1;
        }
    }

    return 0;
}

227 228 229 230 231 232
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;
233
    mem.userspace_addr = (unsigned long)slot->ram;
234
    mem.flags = slot->flags;
235 236 237
    if (s->migration_log) {
        mem.flags |= KVM_MEM_LOG_DIRTY_PAGES;
    }
238 239

    if (slot->memory_size && mem.flags & KVM_MEM_READONLY) {
240 241 242 243 244 245
        /* Set the slot size to 0 before setting the slot to the desired
         * value. This is needed based on KVM commit 75d61fbc. */
        mem.memory_size = 0;
        kvm_vm_ioctl(s, KVM_SET_USER_MEMORY_REGION, &mem);
    }
    mem.memory_size = slot->memory_size;
246 247 248
    return kvm_vm_ioctl(s, KVM_SET_USER_MEMORY_REGION, &mem);
}

249
int kvm_init_vcpu(CPUState *cpu)
250 251 252 253 254
{
    KVMState *s = kvm_state;
    long mmap_size;
    int ret;

255
    DPRINTF("kvm_init_vcpu\n");
256

257
    ret = kvm_vm_ioctl(s, KVM_CREATE_VCPU, (void *)kvm_arch_vcpu_id(cpu));
258
    if (ret < 0) {
259
        DPRINTF("kvm_create_vcpu failed\n");
260 261 262
        goto err;
    }

263
    cpu->kvm_fd = ret;
264
    cpu->kvm_state = s;
265
    cpu->kvm_vcpu_dirty = true;
266 267 268

    mmap_size = kvm_ioctl(s, KVM_GET_VCPU_MMAP_SIZE, 0);
    if (mmap_size < 0) {
269
        ret = mmap_size;
270
        DPRINTF("KVM_GET_VCPU_MMAP_SIZE failed\n");
271 272 273
        goto err;
    }

274
    cpu->kvm_run = mmap(NULL, mmap_size, PROT_READ | PROT_WRITE, MAP_SHARED,
275
                        cpu->kvm_fd, 0);
276
    if (cpu->kvm_run == MAP_FAILED) {
277
        ret = -errno;
278
        DPRINTF("mmap'ing vcpu state failed\n");
279 280 281
        goto err;
    }

282 283
    if (s->coalesced_mmio && !s->coalesced_mmio_ring) {
        s->coalesced_mmio_ring =
284
            (void *)cpu->kvm_run + s->coalesced_mmio * PAGE_SIZE;
285
    }
286

287
    ret = kvm_arch_init_vcpu(cpu);
288 289 290 291
err:
    return ret;
}

292 293 294
/*
 * dirty pages logging control
 */
295

296
static int kvm_mem_flags(KVMState *s, bool log_dirty, bool readonly)
297
{
298 299 300 301 302 303
    int flags = 0;
    flags = log_dirty ? KVM_MEM_LOG_DIRTY_PAGES : 0;
    if (readonly && kvm_readonly_mem_allowed) {
        flags |= KVM_MEM_READONLY;
    }
    return flags;
304 305 306
}

static int kvm_slot_dirty_pages_log_change(KVMSlot *mem, bool log_dirty)
307 308
{
    KVMState *s = kvm_state;
309
    int flags, mask = KVM_MEM_LOG_DIRTY_PAGES;
310 311 312
    int old_flags;

    old_flags = mem->flags;
313

314
    flags = (mem->flags & ~mask) | kvm_mem_flags(s, log_dirty, false);
315 316
    mem->flags = flags;

317 318 319 320
    /* If nothing changed effectively, no need to issue ioctl */
    if (s->migration_log) {
        flags |= KVM_MEM_LOG_DIRTY_PAGES;
    }
321

322
    if (flags == old_flags) {
323
        return 0;
324 325
    }

326 327 328
    return kvm_set_user_memory_region(s, mem);
}

329
static int kvm_dirty_pages_log_change(hwaddr phys_addr,
330 331 332 333 334 335 336 337
                                      ram_addr_t size, bool log_dirty)
{
    KVMState *s = kvm_state;
    KVMSlot *mem = kvm_lookup_matching_slot(s, phys_addr, phys_addr + size);

    if (mem == NULL)  {
        fprintf(stderr, "BUG: %s: invalid parameters " TARGET_FMT_plx "-"
                TARGET_FMT_plx "\n", __func__, phys_addr,
338
                (hwaddr)(phys_addr + size - 1));
339 340 341 342 343
        return -EINVAL;
    }
    return kvm_slot_dirty_pages_log_change(mem, log_dirty);
}

344 345
static void kvm_log_start(MemoryListener *listener,
                          MemoryRegionSection *section)
346
{
347 348 349
    int r;

    r = kvm_dirty_pages_log_change(section->offset_within_address_space,
350
                                   int128_get64(section->size), true);
351 352 353
    if (r < 0) {
        abort();
    }
354 355
}

356 357
static void kvm_log_stop(MemoryListener *listener,
                          MemoryRegionSection *section)
358
{
359 360 361
    int r;

    r = kvm_dirty_pages_log_change(section->offset_within_address_space,
362
                                   int128_get64(section->size), false);
363 364 365
    if (r < 0) {
        abort();
    }
366 367
}

368
static int kvm_set_migration_log(int enable)
369 370 371 372 373 374 375
{
    KVMState *s = kvm_state;
    KVMSlot *mem;
    int i, err;

    s->migration_log = enable;

376
    for (i = 0; i < s->nr_slots; i++) {
377 378
        mem = &s->slots[i];

379 380 381
        if (!mem->memory_size) {
            continue;
        }
382 383 384 385 386 387 388 389 390 391 392
        if (!!(mem->flags & KVM_MEM_LOG_DIRTY_PAGES) == enable) {
            continue;
        }
        err = kvm_set_user_memory_region(s, mem);
        if (err) {
            return err;
        }
    }
    return 0;
}

393
/* get kvm's dirty pages bitmap and update qemu's */
394 395
static int kvm_get_dirty_pages_log_range(MemoryRegionSection *section,
                                         unsigned long *bitmap)
396
{
397
    ram_addr_t start = section->offset_within_region + section->mr->ram_addr;
398 399 400
    ram_addr_t pages = int128_get64(section->size) / getpagesize();

    cpu_physical_memory_set_dirty_lebitmap(bitmap, start, pages);
401
    return 0;
402 403
}

404 405
#define ALIGN(x, y)  (((x)+(y)-1) & ~((y)-1))

406 407
/**
 * kvm_physical_sync_dirty_bitmap - Grab dirty bitmap from kernel space
408 409 410
 * This function updates qemu's dirty bitmap using
 * memory_region_set_dirty().  This means all bits are set
 * to dirty.
411
 *
412
 * @start_add: start of logged region.
413 414
 * @end_addr: end of logged region.
 */
415
static int kvm_physical_sync_dirty_bitmap(MemoryRegionSection *section)
416 417
{
    KVMState *s = kvm_state;
418 419 420 421
    unsigned long size, allocated_size = 0;
    KVMDirtyLog d;
    KVMSlot *mem;
    int ret = 0;
422
    hwaddr start_addr = section->offset_within_address_space;
423
    hwaddr end_addr = start_addr + int128_get64(section->size);
424

425 426 427 428 429 430
    d.dirty_bitmap = NULL;
    while (start_addr < end_addr) {
        mem = kvm_lookup_overlapping_slot(s, start_addr, end_addr);
        if (mem == NULL) {
            break;
        }
431

432 433 434 435 436 437 438 439 440 441 442 443 444 445
        /* XXX bad kernel interface alert
         * For dirty bitmap, kernel allocates array of size aligned to
         * bits-per-long.  But for case when the kernel is 64bits and
         * the userspace is 32bits, userspace can't align to the same
         * bits-per-long, since sizeof(long) is different between kernel
         * and user space.  This way, userspace will provide buffer which
         * may be 4 bytes less than the kernel will use, resulting in
         * userspace memory corruption (which is not detectable by valgrind
         * too, in most cases).
         * So for now, let's align to 64 instead of HOST_LONG_BITS here, in
         * a hope that sizeof(long) wont become >8 any time soon.
         */
        size = ALIGN(((mem->memory_size) >> TARGET_PAGE_BITS),
                     /*HOST_LONG_BITS*/ 64) / 8;
446
        if (!d.dirty_bitmap) {
447
            d.dirty_bitmap = g_malloc(size);
448
        } else if (size > allocated_size) {
449
            d.dirty_bitmap = g_realloc(d.dirty_bitmap, size);
450 451 452
        }
        allocated_size = size;
        memset(d.dirty_bitmap, 0, allocated_size);
453

454
        d.slot = mem->slot;
455

456
        if (kvm_vm_ioctl(s, KVM_GET_DIRTY_LOG, &d) == -1) {
457
            DPRINTF("ioctl failed %d\n", errno);
458 459 460
            ret = -1;
            break;
        }
461

462
        kvm_get_dirty_pages_log_range(section, d.dirty_bitmap);
463
        start_addr = mem->start_addr + mem->memory_size;
464
    }
465
    g_free(d.dirty_bitmap);
466 467

    return ret;
468 469
}

470 471
static void kvm_coalesce_mmio_region(MemoryListener *listener,
                                     MemoryRegionSection *secion,
472
                                     hwaddr start, hwaddr size)
aliguori's avatar
aliguori committed
473 474 475 476 477 478 479 480
{
    KVMState *s = kvm_state;

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

        zone.addr = start;
        zone.size = size;
481
        zone.pad = 0;
aliguori's avatar
aliguori committed
482

483
        (void)kvm_vm_ioctl(s, KVM_REGISTER_COALESCED_MMIO, &zone);
aliguori's avatar
aliguori committed
484 485 486
    }
}

487 488
static void kvm_uncoalesce_mmio_region(MemoryListener *listener,
                                       MemoryRegionSection *secion,
489
                                       hwaddr start, hwaddr size)
aliguori's avatar
aliguori committed
490 491 492 493 494 495 496 497
{
    KVMState *s = kvm_state;

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

        zone.addr = start;
        zone.size = size;
498
        zone.pad = 0;
aliguori's avatar
aliguori committed
499

500
        (void)kvm_vm_ioctl(s, KVM_UNREGISTER_COALESCED_MMIO, &zone);
aliguori's avatar
aliguori committed
501 502 503
    }
}

504 505 506 507 508 509 510 511 512 513 514 515
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;
}

516 517 518 519 520 521 522 523 524 525 526 527 528
int kvm_vm_check_extension(KVMState *s, unsigned int extension)
{
    int ret;

    ret = kvm_vm_ioctl(s, KVM_CHECK_EXTENSION, extension);
    if (ret < 0) {
        /* VM wide version not implemented, use global one instead */
        ret = kvm_check_extension(s, extension);
    }

    return ret;
}

529
static int kvm_set_ioeventfd_mmio(int fd, hwaddr addr, uint32_t val,
530
                                  bool assign, uint32_t size, bool datamatch)
531 532 533 534
{
    int ret;
    struct kvm_ioeventfd iofd;

535
    iofd.datamatch = datamatch ? val : 0;
536 537
    iofd.addr = addr;
    iofd.len = size;
538
    iofd.flags = 0;
539 540 541 542 543 544
    iofd.fd = fd;

    if (!kvm_enabled()) {
        return -ENOSYS;
    }

545 546 547
    if (datamatch) {
        iofd.flags |= KVM_IOEVENTFD_FLAG_DATAMATCH;
    }
548 549 550 551 552 553 554 555 556 557 558 559 560
    if (!assign) {
        iofd.flags |= KVM_IOEVENTFD_FLAG_DEASSIGN;
    }

    ret = kvm_vm_ioctl(kvm_state, KVM_IOEVENTFD, &iofd);

    if (ret < 0) {
        return -errno;
    }

    return 0;
}

561
static int kvm_set_ioeventfd_pio(int fd, uint16_t addr, uint16_t val,
562
                                 bool assign, uint32_t size, bool datamatch)
563 564
{
    struct kvm_ioeventfd kick = {
565
        .datamatch = datamatch ? val : 0,
566
        .addr = addr,
567
        .flags = KVM_IOEVENTFD_FLAG_PIO,
568
        .len = size,
569 570 571 572 573 574
        .fd = fd,
    };
    int r;
    if (!kvm_enabled()) {
        return -ENOSYS;
    }
575 576 577
    if (datamatch) {
        kick.flags |= KVM_IOEVENTFD_FLAG_DATAMATCH;
    }
578 579 580 581 582 583 584 585 586 587 588
    if (!assign) {
        kick.flags |= KVM_IOEVENTFD_FLAG_DEASSIGN;
    }
    r = kvm_vm_ioctl(kvm_state, KVM_IOEVENTFD, &kick);
    if (r < 0) {
        return r;
    }
    return 0;
}


589 590
static int kvm_check_many_ioeventfds(void)
{
591 592 593 594 595
    /* Userspace can use ioeventfd for io notification.  This requires a host
     * that supports eventfd(2) and an I/O thread; since eventfd does not
     * support SIGIO it cannot interrupt the vcpu.
     *
     * Older kernels have a 6 device limit on the KVM io bus.  Find out so we
596 597
     * can avoid creating too many ioeventfds.
     */
598
#if defined(CONFIG_EVENTFD)
599 600 601 602 603 604 605
    int ioeventfds[7];
    int i, ret = 0;
    for (i = 0; i < ARRAY_SIZE(ioeventfds); i++) {
        ioeventfds[i] = eventfd(0, EFD_CLOEXEC);
        if (ioeventfds[i] < 0) {
            break;
        }
606
        ret = kvm_set_ioeventfd_pio(ioeventfds[i], 0, i, true, 2, true);
607 608 609 610 611 612 613 614 615 616
        if (ret < 0) {
            close(ioeventfds[i]);
            break;
        }
    }

    /* Decide whether many devices are supported or not */
    ret = i == ARRAY_SIZE(ioeventfds);

    while (i-- > 0) {
617
        kvm_set_ioeventfd_pio(ioeventfds[i], 0, i, false, 2, true);
618 619 620 621 622 623 624 625
        close(ioeventfds[i]);
    }
    return ret;
#else
    return 0;
#endif
}

626 627 628 629 630 631 632 633 634 635 636 637
static const KVMCapabilityInfo *
kvm_check_extension_list(KVMState *s, const KVMCapabilityInfo *list)
{
    while (list->name) {
        if (!kvm_check_extension(s, list->value)) {
            return list;
        }
        list++;
    }
    return NULL;
}

638
static void kvm_set_phys_mem(MemoryRegionSection *section, bool add)
639 640 641 642
{
    KVMState *s = kvm_state;
    KVMSlot *mem, old;
    int err;
643 644
    MemoryRegion *mr = section->mr;
    bool log_dirty = memory_region_is_logging(mr);
645 646
    bool writeable = !mr->readonly && !mr->rom_device;
    bool readonly_flag = mr->readonly || memory_region_is_romd(mr);
647
    hwaddr start_addr = section->offset_within_address_space;
648
    ram_addr_t size = int128_get64(section->size);
649
    void *ram = NULL;
Avi Kivity's avatar
Avi Kivity committed
650
    unsigned delta;
651

652
    /* kvm works in page size chunks, but the function may be called
653 654 655 656
       with sub-page size and unaligned start address. Pad the start
       address to next and truncate size to previous page boundary. */
    delta = (TARGET_PAGE_SIZE - (start_addr & ~TARGET_PAGE_MASK));
    delta &= ~TARGET_PAGE_MASK;
Avi Kivity's avatar
Avi Kivity committed
657 658 659 660 661 662 663 664 665
    if (delta > size) {
        return;
    }
    start_addr += delta;
    size -= delta;
    size &= TARGET_PAGE_MASK;
    if (!size || (start_addr & ~TARGET_PAGE_MASK)) {
        return;
    }
666

667
    if (!memory_region_is_ram(mr)) {
668 669 670 671 672 673 674
        if (writeable || !kvm_readonly_mem_allowed) {
            return;
        } else if (!mr->romd_mode) {
            /* If the memory device is not in romd_mode, then we actually want
             * to remove the kvm memory slot so all accesses will trap. */
            add = false;
        }
675 676
    }

Avi Kivity's avatar
Avi Kivity committed
677
    ram = memory_region_get_ram_ptr(mr) + section->offset_within_region + delta;
678

679 680 681 682 683 684
    while (1) {
        mem = kvm_lookup_overlapping_slot(s, start_addr, start_addr + size);
        if (!mem) {
            break;
        }

685
        if (add && start_addr >= mem->start_addr &&
686
            (start_addr + size <= mem->start_addr + mem->memory_size) &&
687
            (ram - start_addr == mem->ram - mem->start_addr)) {
688
            /* The new slot fits into the existing one and comes with
689 690
             * identical parameters - update flags and done. */
            kvm_slot_dirty_pages_log_change(mem, log_dirty);
691 692 693 694 695
            return;
        }

        old = *mem;

696 697 698 699
        if (mem->flags & KVM_MEM_LOG_DIRTY_PAGES) {
            kvm_physical_sync_dirty_bitmap(section);
        }

700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717
        /* 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));
            abort();
        }

        /* 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. */
        if (s->broken_set_mem_region &&
718
            old.start_addr == start_addr && old.memory_size < size && add) {
719 720 721
            mem = kvm_alloc_slot(s);
            mem->memory_size = old.memory_size;
            mem->start_addr = old.start_addr;
722
            mem->ram = old.ram;
723
            mem->flags = kvm_mem_flags(s, log_dirty, readonly_flag);
724 725 726 727 728 729 730 731 732

            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;
733
            ram += old.memory_size;
734 735 736 737 738 739 740 741 742
            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;
743
            mem->ram = old.ram;
744
            mem->flags =  kvm_mem_flags(s, log_dirty, readonly_flag);
745 746 747 748 749

            err = kvm_set_user_memory_region(s, mem);
            if (err) {
                fprintf(stderr, "%s: error registering prefix slot: %s\n",
                        __func__, strerror(-err));
750 751 752 753 754
#ifdef TARGET_PPC
                fprintf(stderr, "%s: This is probably because your kernel's " \
                                "PAGE_SIZE is too big. Please try to use 4k " \
                                "PAGE_SIZE!\n", __func__);
#endif
755 756 757 758 759 760 761 762 763 764 765 766
                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;
767
            mem->ram = old.ram + size_delta;
768
            mem->flags = kvm_mem_flags(s, log_dirty, readonly_flag);
769 770 771 772 773 774 775 776 777 778 779

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

    /* in case the KVM bug workaround already "consumed" the new slot */
780
    if (!size) {
781
        return;
782
    }
783
    if (!add) {
784
        return;
785
    }
786 787 788
    mem = kvm_alloc_slot(s);
    mem->memory_size = size;
    mem->start_addr = start_addr;
789
    mem->ram = ram;
790
    mem->flags = kvm_mem_flags(s, log_dirty, readonly_flag);
791 792 793 794 795 796 797 798 799

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

800 801 802
static void kvm_region_add(MemoryListener *listener,
                           MemoryRegionSection *section)
{
803
    memory_region_ref(section->mr);
804 805 806 807 808 809 810
    kvm_set_phys_mem(section, true);
}

static void kvm_region_del(MemoryListener *listener,
                           MemoryRegionSection *section)
{
    kvm_set_phys_mem(section, false);
811
    memory_region_unref(section->mr);
812 813 814 815
}

static void kvm_log_sync(MemoryListener *listener,
                         MemoryRegionSection *section)
816
{
817 818
    int r;

819
    r = kvm_physical_sync_dirty_bitmap(section);
820 821 822
    if (r < 0) {
        abort();
    }
823 824
}

825
static void kvm_log_global_start(struct MemoryListener *listener)
826
{
827 828 829 830
    int r;

    r = kvm_set_migration_log(1);
    assert(r >= 0);
831 832
}

833
static void kvm_log_global_stop(struct MemoryListener *listener)
834
{
835 836 837 838
    int r;

    r = kvm_set_migration_log(0);
    assert(r >= 0);
839 840
}

841 842 843 844 845 846
static void kvm_mem_ioeventfd_add(MemoryListener *listener,
                                  MemoryRegionSection *section,
                                  bool match_data, uint64_t data,
                                  EventNotifier *e)
{
    int fd = event_notifier_get_fd(e);
847 848
    int r;

849
    r = kvm_set_ioeventfd_mmio(fd, section->offset_within_address_space,
850 851
                               data, true, int128_get64(section->size),
                               match_data);
852
    if (r < 0) {
853 854
        fprintf(stderr, "%s: error adding ioeventfd: %s\n",
                __func__, strerror(-r));
855 856 857 858
        abort();
    }
}

859 860 861 862
static void kvm_mem_ioeventfd_del(MemoryListener *listener,
                                  MemoryRegionSection *section,
                                  bool match_data, uint64_t data,
                                  EventNotifier *e)
863
{
864
    int fd = event_notifier_get_fd(e);
865 866
    int r;

867
    r = kvm_set_ioeventfd_mmio(fd, section->offset_within_address_space,
868 869
                               data, false, int128_get64(section->size),
                               match_data);
870 871 872 873 874
    if (r < 0) {
        abort();
    }
}

875 876 877 878
static void kvm_io_ioeventfd_add(MemoryListener *listener,
                                 MemoryRegionSection *section,
                                 bool match_data, uint64_t data,
                                 EventNotifier *e)
879
{
880
    int fd = event_notifier_get_fd(e);
881 882
    int r;

883
    r = kvm_set_ioeventfd_pio(fd, section->offset_within_address_space,
884 885
                              data, true, int128_get64(section->size),
                              match_data);
886
    if (r < 0) {
887 888
        fprintf(stderr, "%s: error adding ioeventfd: %s\n",
                __func__, strerror(-r));
889 890 891 892
        abort();
    }
}

893 894 895 896
static void kvm_io_ioeventfd_del(MemoryListener *listener,
                                 MemoryRegionSection *section,
                                 bool match_data, uint64_t data,
                                 EventNotifier *e)
897 898

{
899
    int fd = event_notifier_get_fd(e);
900 901
    int r;

902
    r = kvm_set_ioeventfd_pio(fd, section->offset_within_address_space,
903 904
                              data, false, int128_get64(section->size),
                              match_data);
905 906 907 908 909
    if (r < 0) {
        abort();
    }
}

910 911 912
static MemoryListener kvm_memory_listener = {
    .region_add = kvm_region_add,
    .region_del = kvm_region_del,
913 914
    .log_start = kvm_log_start,
    .log_stop = kvm_log_stop,
915 916 917
    .log_sync = kvm_log_sync,
    .log_global_start = kvm_log_global_start,
    .log_global_stop = kvm_log_global_stop,
918 919
    .eventfd_add = kvm_mem_ioeventfd_add,
    .eventfd_del = kvm_mem_ioeventfd_del,
920 921
    .coalesced_mmio_add = kvm_coalesce_mmio_region,
    .coalesced_mmio_del = kvm_uncoalesce_mmio_region,
922 923 924 925 926 927
    .priority = 10,
};

static MemoryListener kvm_io_listener = {
    .eventfd_add = kvm_io_ioeventfd_add,
    .eventfd_del = kvm_io_ioeventfd_del,
928
    .priority = 10,
929 930
};

931
static void kvm_handle_interrupt(CPUState *cpu, int mask)
932
{
933
    cpu->interrupt_request |= mask;
934

935
    if (!qemu_cpu_is_self(cpu)) {
936
        qemu_cpu_kick(cpu);
937 938 939
    }
}

940
int kvm_set_irq(KVMState *s, int irq, int level)
941 942 943 944
{
    struct kvm_irq_level event;
    int ret;

945
    assert(kvm_async_interrupts_enabled());
946 947 948

    event.level = level;
    event.irq = irq;
949
    ret = kvm_vm_ioctl(s, s->irq_set_ioctl, &event);
950
    if (ret < 0) {
951
        perror("kvm_set_irq");
952 953 954
        abort();
    }

955
    return (s->irq_set_ioctl == KVM_IRQ_LINE) ? 1 : event.status;
956 957 958
}

#ifdef KVM_CAP_IRQ_ROUTING
959 960 961 962 963
typedef struct KVMMSIRoute {
    struct kvm_irq_routing_entry kroute;
    QTAILQ_ENTRY(KVMMSIRoute) entry;
} KVMMSIRoute;

964 965 966 967 968
static void set_gsi(KVMState *s, unsigned int gsi)
{
    s->used_gsi_bitmap[gsi / 32] |= 1U << (gsi % 32);
}

969 970 971 972 973
static void clear_gsi(KVMState *s, unsigned int gsi)
{
    s->used_gsi_bitmap[gsi / 32] &= ~(1U << (gsi % 32));
}

974
void kvm_init_irq_routing(KVMState *s)
975
{
976
    int gsi_count, i;
977

978
    gsi_count = kvm_check_extension(s, KVM_CAP_IRQ_ROUTING) - 1;
979 980 981 982
    if (gsi_count > 0) {
        unsigned int gsi_bits, i;

        /* Round up so we can search ints using ffs */
983
        gsi_bits = ALIGN(gsi_count, 32);
984
        s->used_gsi_bitmap = g_malloc0(gsi_bits / 8);
985
        s->gsi_count = gsi_count;
986 987 988 989 990 991 992 993 994 995

        /* Mark any over-allocated bits as already in use */
        for (i = gsi_count; i < gsi_bits; i++) {
            set_gsi(s, i);
        }
    }

    s->irq_routes = g_malloc0(sizeof(*s->irq_routes));
    s->nr_allocated_irq_routes = 0;

996 997 998 999
    if (!s->direct_msi) {
        for (i = 0; i < KVM_MSI_HASHTAB_SIZE; i++) {
            QTAILQ_INIT(&s->msi_hashtab[i]);
        }
1000 1001
    }

1002 1003 1004
    kvm_arch_init_irq_routing(s);
}

1005
void kvm_irqchip_commit_routes(KVMState *s)
1006 1007 1008 1009 1010 1011 1012 1013
{
    int ret;

    s->irq_routes->flags = 0;
    ret = kvm_vm_ioctl(s, KVM_SET_GSI_ROUTING, s->irq_routes);
    assert(ret == 0);
}

1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031
static void kvm_add_routing_entry(KVMState *s,
                                  struct kvm_irq_routing_entry *entry)
{
    struct kvm_irq_routing_entry *new;
    int n, size;

    if (s->irq_routes->nr == s->nr_allocated_irq_routes) {
        n = s->nr_allocated_irq_routes * 2;
        if (n < 64) {
            n = 64;
        }
        size = sizeof(struct kvm_irq_routing);
        size += n * sizeof(*new);
        s->irq_routes = g_realloc(s->irq_routes, size);
        s->nr_allocated_irq_routes = n;
    }
    n = s->irq_routes->nr++;
    new = &s->irq_routes->entries[n];
1032 1033

    *new = *entry;
1034 1035 1036 1037

    set_gsi(s, entry->gsi);
}

1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049
static int kvm_update_routing_entry(KVMState *s,
                                    struct kvm_irq_routing_entry *new_entry)
{
    struct kvm_irq_routing_entry *entry;
    int n;

    for (n = 0; n < s->irq_routes->nr; n++) {
        entry = &s->irq_routes->entries[n];
        if (entry->gsi != new_entry->gsi) {
            continue;
        }

1050 1051 1052 1053
        if(!memcmp(entry, new_entry, sizeof *entry)) {
            return 0;
        }

1054
        *entry = *new_entry;
1055 1056 1057 1058 1059 1060 1061 1062 1063

        kvm_irqchip_commit_routes(s);

        return 0;
    }

    return -ESRCH;
}

1064
void kvm_irqchip_add_irq_route(KVMState *s, int irq, int irqchip, int pin)
1065
{
1066
    struct kvm_irq_routing_entry e = {};
1067

1068 1069
    assert(pin < s->gsi_count);

1070 1071 1072 1073 1074 1075 1076 1077
    e.gsi = irq;
    e.type = KVM_IRQ_ROUTING_IRQCHIP;
    e.flags = 0;
    e.u.irqchip.irqchip = irqchip;
    e.u.irqchip.pin = pin;
    kvm_add_routing_entry(s, &e);
}

1078
void kvm_irqchip_release_virq(KVMState *s, int virq)
1079 1080 1081 1082
{
    struct kvm_irq_routing_entry *e;
    int i;

1083 1084 1085 1086
    if (kvm_gsi_direct_mapping()) {
        return;
    }

1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134
    for (i = 0; i < s->irq_routes->nr; i++) {
        e = &s->irq_routes->entries[i];
        if (e->gsi == virq) {
            s->irq_routes->nr--;
            *e = s->irq_routes->entries[s->irq_routes->nr];
        }
    }
    clear_gsi(s, virq);
}

static unsigned int kvm_hash_msi(uint32_t data)
{
    /* This is optimized for IA32 MSI layout. However, no other arch shall
     * repeat the mistake of not providing a direct MSI injection API. */
    return data & 0xff;
}

static void kvm_flush_dynamic_msi_routes(KVMState *s)
{
    KVMMSIRoute *route, *next;
    unsigned int hash;

    for (hash = 0; hash < KVM_MSI_HASHTAB_SIZE; hash++) {
        QTAILQ_FOREACH_SAFE(route, &s->msi_hashtab[hash], entry, next) {
            kvm_irqchip_release_virq(s, route->kroute.gsi);
            QTAILQ_REMOVE(&s->msi_hashtab[hash], route, entry);
            g_free(route);
        }
    }
}

static int kvm_irqchip_get_virq(KVMState *s)
{
    uint32_t *word = s->used_gsi_bitmap;
    int max_words = ALIGN(s->gsi_count, 32) / 32;
    int i, bit;
    bool retry = true;

again:
    /* Return the lowest unused GSI in the bitmap */
    for (i = 0; i < max_words; i++) {
        bit = ffs(~word[i]);
        if (!bit) {
            continue;
        }

        return bit - 1 + i * 32;
    }
1135
    if (!s->direct_msi && retry) {
1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151
        retry = false;
        kvm_flush_dynamic_msi_routes(s);
        goto again;
    }
    return -ENOSPC;

}

static KVMMSIRoute *kvm_lookup_msi_route(KVMState *s, MSIMessage msg)
{
    unsigned int hash = kvm_hash_msi(msg.data);
    KVMMSIRoute *route;

    QTAILQ_FOREACH(route, &s->msi_hashtab[hash], entry) {
        if (route->kroute.u.msi.address_lo == (uint32_t)msg.address &&
            route->kroute.u.msi.address_hi == (msg.address >> 32) &&
1152
            route->kroute.u.msi.data == le32_to_cpu(msg.data)) {
1153 1154 1155 1156 1157 1158 1159 1160
            return route;
        }
    }
    return NULL;
}

int kvm_irqchip_send_msi(KVMState *s, MSIMessage msg)
{
1161
    struct kvm_msi msi;
1162 1163
    KVMMSIRoute *route;

1164 1165 1166
    if (s->direct_msi) {
        msi.address_lo = (uint32_t)msg.address;
        msi.address_hi = msg.address >> 32;
1167
        msi.data = le32_to_cpu(msg.data);
1168 1169 1170 1171 1172 1173
        msi.flags = 0;
        memset(msi.pad, 0, sizeof(msi.pad));

        return kvm_vm_ioctl(s, KVM_SIGNAL_MSI, &msi);
    }

1174 1175
    route = kvm_lookup_msi_route(s, msg);
    if (!route) {
1176
        int virq;
1177 1178 1179 1180 1181 1182

        virq = kvm_irqchip_get_virq(s);
        if (virq < 0) {
            return virq;
        }

1183
        route = g_malloc0(sizeof(KVMMSIRoute));
1184 1185 1186 1187 1188
        route->kroute.gsi = virq;
        route->kroute.type = KVM_IRQ_ROUTING_MSI;
        route->kroute.flags = 0;
        route->kroute.u.msi.address_lo = (uint32_t)msg.address;
        route->kroute.u.msi.address_hi = msg.address >> 32;
1189
        route->kroute.u.msi.data = le32_to_cpu(msg.data);
1190 1191

        kvm_add_routing_entry(s, &route->kroute);
1192
        kvm_irqchip_commit_routes(s);
1193 1194 1195 1196 1197 1198 1199

        QTAILQ_INSERT_TAIL(&s->msi_hashtab[kvm_hash_msi(msg.data)], route,
                           entry);
    }

    assert(route->kroute.type == KVM_IRQ_ROUTING_MSI);

1200
    return kvm_set_irq(s, route->kroute.gsi, 1);
1201 1202
}

1203 1204
int kvm_irqchip_add_msi_route(KVMState *s, MSIMessage msg)
{
1205
    struct kvm_irq_routing_entry kroute = {};
1206 1207
    int virq;

1208 1209 1210 1211
    if (kvm_gsi_direct_mapping()) {
        return msg.data & 0xffff;
    }

1212
    if (!kvm_gsi_routing_enabled()) {
1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225
        return -ENOSYS;
    }

    virq = kvm_irqchip_get_virq(s);
    if (virq < 0) {
        return virq;
    }

    kroute.gsi = virq;
    kroute.type = KVM_IRQ_ROUTING_MSI;
    kroute.flags = 0;
    kroute.u.msi.address_lo = (uint32_t)msg.address;
    kroute.u.msi.address_hi = msg.address >> 32;
1226
    kroute.u.msi.data = le32_to_cpu(msg.data);
1227 1228

    kvm_add_routing_entry(s, &kroute);
1229
    kvm_irqchip_commit_routes(s);
1230 1231 1232 1233

    return virq;
}

1234 1235
int kvm_irqchip_update_msi_route(KVMState *s, int virq, MSIMessage msg)
{
1236
    struct kvm_irq_routing_entry kroute = {};
1237

1238 1239 1240 1241
    if (kvm_gsi_direct_mapping()) {
        return 0;
    }

1242 1243 1244 1245 1246 1247 1248 1249 1250
    if (!kvm_irqchip_in_kernel()) {
        return -ENOSYS;
    }

    kroute.gsi = virq;
    kroute.type = KVM_IRQ_ROUTING_MSI;
    kroute.flags = 0;
    kroute.u.msi.address_lo = (uint32_t)msg.address;
    kroute.u.msi.address_hi = msg.address >> 32;
1251
    kroute.u.msi.data = le32_to_cpu(msg.data);
1252 1253 1254 1255

    return kvm_update_routing_entry(s, &kroute);
}

1256 1257
static int kvm_irqchip_assign_irqfd(KVMState *s, int fd, int rfd, int virq,
                                    bool assign)
1258 1259 1260 1261 1262 1263 1264
{
    struct kvm_irqfd irqfd = {
        .fd = fd,
        .gsi = virq,
        .flags = assign ? 0 : KVM_IRQFD_FLAG_DEASSIGN,
    };

1265 1266 1267 1268 1269
    if (rfd != -1) {
        irqfd.flags |= KVM_IRQFD_FLAG_RESAMPLE;
        irqfd.resamplefd = rfd;
    }

1270
    if (!kvm_irqfds_enabled()) {
1271 1272 1273 1274 1275 1276
        return -ENOSYS;
    }

    return kvm_vm_ioctl(s, KVM_IRQFD, &irqfd);
}

1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305
int kvm_irqchip_add_adapter_route(KVMState *s, AdapterInfo *adapter)
{
    struct kvm_irq_routing_entry kroute;
    int virq;

    if (!kvm_gsi_routing_enabled()) {
        return -ENOSYS;
    }

    virq = kvm_irqchip_get_virq(s);
    if (virq < 0) {
        return virq;
    }

    kroute.gsi = virq;
    kroute.type = KVM_IRQ_ROUTING_S390_ADAPTER;
    kroute.flags = 0;
    kroute.u.adapter.summary_addr = adapter->summary_addr;
    kroute.u.adapter.ind_addr = adapter->ind_addr;
    kroute.u.adapter.summary_offset = adapter->summary_offset;
    kroute.u.adapter.ind_offset = adapter->ind_offset;
    kroute.u.adapter.adapter_id = adapter->adapter_id;

    kvm_add_routing_entry(s, &kroute);
    kvm_irqchip_commit_routes(s);

    return virq;
}

1306 1307
#else /* !KVM_CAP_IRQ_ROUTING */

1308
void kvm_init_irq_routing(KVMState *s)
1309 1310
{
}
1311

1312 1313 1314 1315
void kvm_irqchip_release_virq(KVMState *s, int virq)
{
}

1316 1317 1318 1319
int kvm_irqchip_send_msi(KVMState *s, MSIMessage msg)
{
    abort();
}
1320 1321 1322

int kvm_irqchip_add_msi_route(KVMState *s, MSIMessage msg)
{
1323
    return -ENOSYS;
1324
}
1325

1326 1327 1328 1329 1330
int kvm_irqchip_add_adapter_route(KVMState *s, AdapterInfo *adapter)
{
    return -ENOSYS;
}

1331 1332 1333 1334
static int kvm_irqchip_assign_irqfd(KVMState *s, int fd, int virq, bool assign)
{
    abort();
}
1335 1336 1337 1338 1339

int kvm_irqchip_update_msi_route(KVMState *s, int virq, MSIMessage msg)
{
    return -ENOSYS;
}
1340 1341
#endif /* !KVM_CAP_IRQ_ROUTING */

1342 1343
int kvm_irqchip_add_irqfd_notifier(KVMState *s, EventNotifier *n,
                                   EventNotifier *rn, int virq)
1344
{
1345 1346
    return kvm_irqchip_assign_irqfd(s, event_notifier_get_fd(n),
           rn ? event_notifier_get_fd(rn) : -1, virq, true);
1347 1348
}

Jan Kiszka's avatar
Jan Kiszka committed
1349
int kvm_irqchip_remove_irqfd_notifier(KVMState *s, EventNotifier *n, int virq)
1350
{
1351 1352
    return kvm_irqchip_assign_irqfd(s, event_notifier_get_fd(n), -1, virq,
           false);
1353 1354
}

1355 1356 1357 1358
static int kvm_irqchip_create(KVMState *s)
{
    int ret;

1359
    if (!qemu_opt_get_bool(qemu_get_machine_opts(), "kernel_irqchip", true) ||
1360 1361
        (!kvm_check_extension(s, KVM_CAP_IRQCHIP) &&
         (kvm_vm_enable_cap(s, KVM_CAP_S390_IRQCHIP, 0) < 0))) {
1362 1363 1364
        return 0;
    }

1365 1366 1367
    /* First probe and see if there's a arch-specific hook to create the
     * in-kernel irqchip for us */
    ret = kvm_arch_irqchip_create(s);
1368 1369
    if (ret < 0) {
        return ret;
1370 1371 1372 1373 1374 1375
    } else if (ret == 0) {
        ret = kvm_vm_ioctl(s, KVM_CREATE_IRQCHIP);
        if (ret < 0) {
            fprintf(stderr, "Create kernel irqchip failed\n");
            return ret;
        }
1376 1377
    }

1378
    kvm_kernel_irqchip = true;