qemu_capabilities.c 165.9 KB
Newer Older
1 2 3
/*
 * qemu_capabilities.c: QEMU capabilities generation
 *
4
 * Copyright (C) 2006-2016 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24
 */

#include <config.h>

#include "qemu_capabilities.h"
25
#include "viralloc.h"
26
#include "vircrypto.h"
27
#include "virlog.h"
28
#include "virerror.h"
E
Eric Blake 已提交
29
#include "virfile.h"
30
#include "virfilecache.h"
31 32
#include "virpidfile.h"
#include "virprocess.h"
33
#include "cpu/cpu.h"
34
#include "cpu/cpu_x86.h"
35
#include "domain_conf.h"
36
#include "vircommand.h"
37
#include "virbitmap.h"
38
#include "virnodesuspend.h"
39
#include "virnuma.h"
40
#include "virhostcpu.h"
41
#include "qemu_monitor.h"
42
#include "virstring.h"
43
#include "qemu_hostdev.h"
44
#include "qemu_domain.h"
45
#define LIBVIRT_QEMU_CAPSPRIV_H_ALLOW
46
#include "qemu_capspriv.h"
47
#include "qemu_qapi.h"
48

49
#include <fcntl.h>
50 51 52
#include <sys/stat.h>
#include <unistd.h>
#include <sys/wait.h>
53
#include <stdarg.h>
54
#include <sys/utsname.h>
55

56 57 58 59
#if WITH_CAPNG
# include <cap-ng.h>
#endif

60 61
#define VIR_FROM_THIS VIR_FROM_QEMU

62 63
VIR_LOG_INIT("qemu.qemu_capabilities");

64 65 66 67
/* While not public, these strings must not change. They
 * are used in domain status files which are read on
 * daemon restarts
 */
68
VIR_ENUM_IMPL(virQEMUCaps, QEMU_CAPS_LAST,
69 70
              /* 0 */
              "kqemu",
71 72 73 74 75
              "vnc-colon",
              "no-reboot",
              "drive",
              "drive-boot",

76 77
              /* 5 */
              "name",
78 79 80 81 82
              "uuid",
              "domid",
              "vnet-hdr",
              "migrate-kvm-stdio",

83 84
              /* 10 */
              "migrate-qemu-tcp",
85 86 87 88 89
              "migrate-qemu-exec",
              "drive-cache-v2",
              "kvm",
              "drive-format",

90 91
              /* 15 */
              "vga",
92 93 94 95 96
              "0.10",
              "pci-device",
              "mem-path",
              "drive-serial",

97 98
              /* 20 */
              "xen-domid",
99 100 101 102 103
              "migrate-qemu-unix",
              "chardev",
              "enable-kvm",
              "monitor-json",

104 105
              /* 25 */
              "balloon",
106 107 108 109 110
              "device",
              "sdl",
              "smp-topology",
              "netdev",

111 112
              /* 30 */
              "rtc",
113
              "vhost-net",
114 115 116 117
              "rtc-td-hack",
              "no-hpet",
              "no-kvm-pit",

118 119
              /* 35 */
              "tdf",
120 121 122 123 124
              "pci-configfd",
              "nodefconfig",
              "boot-menu",
              "enable-kqemu",

125 126
              /* 40 */
              "fsdev",
127 128 129 130 131
              "nesting",
              "name-process",
              "drive-readonly",
              "smbios-type",

132 133
              /* 45 */
              "vga-qxl",
134 135 136 137 138
              "spice",
              "vga-none",
              "migrate-qemu-fd",
              "boot-index",

139 140
              /* 50 */
              "hda-duplex",
141 142 143 144 145
              "drive-aio",
              "pci-multibus",
              "pci-bootindex",
              "ccid-emulated",

146 147
              /* 55 */
              "ccid-passthru",
148 149 150 151
              "chardev-spicevmc",
              "device-spicevmc",
              "virtio-tx-alg",
              "device-qxl-vga",
152

153 154
              /* 60 */
              "pci-multifunction",
155
              "virtio-blk-pci.ioeventfd",
M
Michal Privoznik 已提交
156
              "sga",
157 158
              "virtio-blk-pci.event_idx",
              "virtio-net-pci.event_idx",
159

160 161
              /* 65 */
              "cache-directsync",
162 163 164 165 166
              "piix3-usb-uhci",
              "piix4-usb-uhci",
              "usb-ehci",
              "ich9-usb-ehci1",

167 168
              /* 70 */
              "vt82c686b-usb-uhci",
169 170
              "pci-ohci",
              "usb-redir",
M
Marc-André Lureau 已提交
171
              "usb-hub",
172
              "no-shutdown",
173

174 175
              /* 75 */
              "cache-unsafe",
176
              "rombar",
J
Jim Fehlig 已提交
177
              "ich9-ahci",
178
              "no-acpi",
179
              "fsdev-readonly",
180

181 182
              /* 80 */
              "virtio-blk-pci.scsi",
183
              "blk-sg-io",
O
Osier Yang 已提交
184
              "drive-copy-on-read",
185
              "cpu-host",
186
              "fsdev-writeout",
187

188 189
              /* 85 */
              "drive-iotune",
190
              "system_wakeup",
191
              "scsi-disk.channel",
192
              "scsi-block",
193
              "transaction",
194

195 196
              /* 90 */
              "block-job-sync",
197
              "block-job-async",
198
              "scsi-cd",
199
              "ide-cd",
200
              "no-user-config",
M
Marc-André Lureau 已提交
201

202 203
              /* 95 */
              "hda-micro",
204
              "dump-guest-memory",
G
Gerd Hoffmann 已提交
205
              "nec-usb-xhci",
206
              "virtio-s390",
207
              "balloon-event",
M
Marc-André Lureau 已提交
208

209 210
              /* 100 */
              "bridge",
211 212
              "lsi",
              "virtio-scsi-pci",
V
Viktor Mihajlovski 已提交
213
              "blockio",
214
              "disable-s3",
R
Richa Marwaha 已提交
215

216 217
              /* 105 */
              "disable-s4",
218
              "usb-redir.filter",
219 220
              "ide-drive.wwn",
              "scsi-disk.wwn",
221
              "seccomp-sandbox",
222

223 224
              /* 110 */
              "reboot-timeout",
225
              "dump-guest-core",
226
              "seamless-migration",
227
              "block-commit",
228
              "vnc",
229

230 231
              /* 115 */
              "drive-mirror",
232 233
              "usb-redir.bootindex",
              "usb-host.bootindex",
234
              "blockdev-snapshot-sync",
235 236
              "qxl",

237 238
              /* 120 */
              "VGA",
239 240 241
              "cirrus-vga",
              "vmware-svga",
              "device-video-primary",
242
              "s390-sclp",
243

244 245
              /* 125 */
              "usb-serial",
G
Guannan Ren 已提交
246
              "usb-net",
247
              "add-fd",
248
              "nbd-server",
249 250
              "virtio-rng",

251 252
              /* 130 */
              "rng-random",
253
              "rng-egd",
O
Olivia Yin 已提交
254 255
              "virtio-ccw",
              "dtb",
256
              "megasas",
J
Ján Tomko 已提交
257

258 259
              /* 135 */
              "ipv6-migration",
260
              "machine-opt",
L
Li Zhang 已提交
261
              "machine-usb-opt",
S
Stefan Berger 已提交
262 263
              "tpm-passthrough",
              "tpm-tis",
264

265 266
              /* 140 */
              "nvram",
H
Han Cheng 已提交
267 268 269 270 271
              "pci-bridge",
              "vfio-pci",
              "vfio-pci.bootindex",
              "scsi-generic",

272 273
              /* 145 */
              "scsi-generic.bootindex",
274
              "mem-merge",
275
              "vnc-websocket",
O
Osier Yang 已提交
276
              "drive-discard",
277
              "mlock",
278

279 280
              /* 150 */
              "vnc-share-policy",
281
              "device-del-event",
282
              "dmi-to-pci-bridge",
283 284
              "i440fx-pci-hole64-size",
              "q35-pci-hole64-size",
285

286 287
              /* 155 */
              "usb-storage",
288
              "usb-storage.removable",
289
              "virtio-mmio",
290
              "ich9-intel-hda",
291
              "kvm-pit-lost-tick-policy",
292

293 294
              /* 160 */
              "boot-strict",
295 296
              "pvpanic",
              "enable-fips",
297 298
              "spice-file-xfer-disable",
              "spiceport",
L
Li Zhang 已提交
299

300 301
              /* 165 */
              "usb-kbd",
302
              "host-pci-multidomain",
303
              "msg-timestamp",
304
              "active-commit",
305
              "change-backing-file",
306

307 308
              /* 170 */
              "memory-backend-ram",
309
              "numa",
310
              "memory-backend-file",
311
              "usb-audio",
312
              "rtc-reset-reinjection",
313

314 315
              /* 175 */
              "splash-timeout",
J
John Ferlan 已提交
316
              "iothread",
317
              "migrate-rdma",
318
              "ivshmem",
319
              "drive-iotune-max",
320

321 322
              /* 180 */
              "VGA.vgamem_mb",
323 324 325
              "vmware-svga.vgamem_mb",
              "qxl.vgamem_mb",
              "qxl-vga.vgamem_mb",
326
              "pc-dimm",
327

328 329
              /* 185 */
              "machine-vmport-opt",
330 331
              "aes-key-wrap",
              "dea-key-wrap",
M
Michal Privoznik 已提交
332
              "pci-serial",
333
              "aarch64-off",
334

335 336
              /* 190 */
              "vhost-user-multiqueue",
337
              "migration-event",
338
              "gpex-pcihost",
339
              "ioh3420",
340
              "x3130-upstream",
341

342 343
              /* 195 */
              "xio3130-downstream",
344
              "rtl8139",
345
              "e1000",
346
              "virtio-net",
347
              "gic-version",
348

349 350
              /* 200 */
              "incoming-defer",
M
Marc-André Lureau 已提交
351
              "virtio-gpu",
352
              "virtio-gpu.virgl",
353 354 355
              "virtio-keyboard",
              "virtio-mouse",

356 357
              /* 205 */
              "virtio-tablet",
358
              "virtio-input-host",
359
              "chardev-file-append",
360 361
              "ich9-disable-s3",
              "ich9-disable-s4",
362

363 364
              /* 210 */
              "vserport-change-event",
365
              "virtio-balloon-pci.deflate-on-oom",
366
              "mptsas1068",
367
              "spice-gl",
368 369
              "qxl.vram64_size_mb",

370 371
              /* 215 */
              "qxl-vga.vram64_size_mb",
372
              "chardev-logfile",
373
              "debug-threads",
374
              "secret",
375
              "pxb",
376

377 378
              /* 220 */
              "pxb-pcie",
379
              "device-tray-moved-event",
380
              "nec-usb-xhci-ports",
381
              "virtio-scsi-pci.iothread",
382
              "name-guest",
383

384 385
              /* 225 */
              "qxl.max_outputs",
386
              "qxl-vga.max_outputs",
387
              "spice-unix",
388
              "drive-detect-zeroes",
B
Boris Fiuczynski 已提交
389
              "tls-creds-x509",
390

391 392
              /* 230 */
              "display",
J
Ján Tomko 已提交
393
              "intel-iommu",
M
Michal Privoznik 已提交
394
              "smm",
395
              "virtio-pci-disable-legacy",
396
              "query-hotpluggable-cpus",
397

398 399
              /* 235 */
              "virtio-net.rx_queue_size",
400
              "machine-iommu",
401
              "virtio-vga",
402
              "drive-iotune-max-length",
403 404
              "ivshmem-plain",

405 406
              /* 240 */
              "ivshmem-doorbell",
407
              "query-qmp-schema",
408
              "gluster.debug_level",
409
              "vhost-scsi",
410
              "drive-iotune-group",
411

412 413
              /* 245 */
              "query-cpu-model-expansion",
414
              "virtio-net.host_mtu",
415
              "spice-rendernode",
416
              "nvdimm",
417
              "pcie-root-port",
418

419 420
              /* 250 */
              "query-cpu-definitions",
421
              "block-write-threshold",
422
              "query-named-block-nodes",
423
              "cpu-cache",
424
              "qemu-xhci",
425

426 427
              /* 255 */
              "kernel-irqchip",
428
              "kernel-irqchip.split",
429
              "intel-iommu.intremap",
430
              "intel-iommu.caching-mode",
431
              "intel-iommu.eim",
432

433 434
              /* 260 */
              "intel-iommu.device-iotlb",
435 436
              "virtio.iommu_platform",
              "virtio.ats",
437
              "loadparm",
438
              "spapr-pci-host-bridge",
439 440

              /* 265 */
441
              "spapr-pci-host-bridge.numa_node",
442 443
              "vnc-multi-servers",
              "virtio-net.tx_queue_size",
444
              "chardev-reconnect",
445
              "virtio-gpu.max_outputs",
J
John Ferlan 已提交
446 447 448

              /* 270 */
              "vxhs",
449
              "virtio-blk.num-queues",
450
              "machine.pseries.resize-hpt",
M
Marc-André Lureau 已提交
451
              "vmcoreinfo",
452
              "spapr-vty",
453 454 455

              /* 275 */
              "sclplmconsole",
456
              "numa.dist",
457
              "disk-share-rw",
458
              "iscsi.password-secret",
459
              "isa-serial",
460 461 462

              /* 280 */
              "pl011",
463
              "machine.pseries.max-cpu-compat",
464
              "dump-completed",
465
              "virtio-gpu-ccw",
466 467 468 469 470
              "virtio-keyboard-ccw",

              /* 285 */
              "virtio-mouse-ccw",
              "virtio-tablet-ccw",
471
              "qcow2-luks",
472
              "pcie-pci-bridge",
473
              "seccomp-blacklist",
474 475 476

              /* 290 */
              "query-cpus-fast",
477
              "disk-write-cache",
478
              "nbd-tls",
479
              "tpm-crb",
480
              "pr-manager-helper",
481 482 483

              /* 295 */
              "qom-list-properties",
484
              "memory-backend-file.discard-data",
485
              "virtual-css-bridge",
486 487
              "virtual-css-bridge.cssid-unrestricted",
              "vfio-ccw",
488 489 490

              /* 300 */
              "sdl-gl",
491
              "screendump_device",
492
              "hda-output",
493
              "blockdev-del",
494
              "vmgenid",
495 496 497

              /* 305 */
              "vhost-vsock",
498
              "chardev-fd-pass",
499
              "tpm-emulator",
500 501
              "mch",
              "mch.extended-tseg-mbytes",
502 503 504

              /* 310 */
              "sev-guest",
505
              "machine.pseries.cap-hpt-max-page-size",
506
              "machine.pseries.cap-htm",
507
              "usb-storage.werror",
508
              "egl-headless",
509 510 511

              /* 315 */
              "vfio-pci.display",
512
              "blockdev",
B
Boris Fiuczynski 已提交
513
              "vfio-ap",
Y
Yi Min Zhao 已提交
514
              "zpci",
515
              "memory-backend-memfd",
516 517 518

              /* 320 */
              "memory-backend-memfd.hugetlb",
519
              "iothread.poll-max-ns",
520
              "machine.pseries.cap-nested-hv",
521 522
              "egl-headless.rendernode",
              "memory-backend-file.align",
523 524 525

              /* 325 */
              "memory-backend-file.pmem",
526
              "nvdimm.unarmed",
527 528
    );

529

530 531 532 533
struct virQEMUCapsMachineType {
    char *name;
    char *alias;
    unsigned int maxCpus;
534
    bool hotplugCpus;
535
    bool qemuDefault;
536
};
537 538 539 540 541 542 543 544 545 546 547

typedef struct _virQEMUCapsHostCPUData virQEMUCapsHostCPUData;
typedef virQEMUCapsHostCPUData *virQEMUCapsHostCPUDataPtr;
struct _virQEMUCapsHostCPUData {
    /* Only the "info" part is stored in the capabilities cache, the rest is
     * re-computed from other fields and external data sources everytime we
     * probe QEMU or load the cache.
     */
    qemuMonitorCPUModelInfoPtr info;
    /* Host CPU definition reported in domain capabilities. */
    virCPUDefPtr reported;
548 549
    /* Migratable host CPU definition used for updating guest CPU. */
    virCPUDefPtr migratable;
550 551 552 553
    /* CPU definition with features detected by libvirt using virCPUGetHost
     * combined with features reported by QEMU. This is used for backward
     * compatible comparison between a guest CPU and a host CPU. */
    virCPUDefPtr full;
554 555
};

556 557 558 559 560 561
/*
 * Update the XML parser/formatter when adding more
 * information to this struct so that it gets cached
 * correctly. It does not have to be ABI-stable, as
 * the cache will be discarded & repopulated if the
 * timestamp on the libvirtd binary changes.
562 563
 *
 * And don't forget to update virQEMUCapsNewCopy.
564
 */
565
struct _virQEMUCaps {
566
    virObject parent;
567

568
    bool usedQMP;
569
    bool kvmSupportsNesting;
570

571
    char *binary;
572
    time_t ctime;
573
    time_t libvirtCtime;
574

575
    virBitmapPtr flags;
576 577 578

    unsigned int version;
    unsigned int kvmVersion;
579
    unsigned int libvirtVersion;
580
    unsigned int microcodeVersion;
581
    char *package;
582
    char *kernelVersion;
583

584
    virArch arch;
585

586 587
    virDomainCapsCPUModelsPtr kvmCPUModels;
    virDomainCapsCPUModelsPtr tcgCPUModels;
588 589

    size_t nmachineTypes;
590
    struct virQEMUCapsMachineType *machineTypes;
A
Andrea Bolognani 已提交
591 592 593

    size_t ngicCapabilities;
    virGICCapability *gicCapabilities;
594

595 596
    virSEVCapability *sevCapabilities;

597 598
    virQEMUCapsHostCPUData kvmCPU;
    virQEMUCapsHostCPUData tcgCPU;
599 600
};

601 602
struct virQEMUCapsSearchData {
    virArch arch;
603
    const char *binaryFilter;
604 605
};

606

607 608
static virClassPtr virQEMUCapsClass;
static void virQEMUCapsDispose(void *obj);
609

610
static int virQEMUCapsOnceInit(void)
611
{
612
    if (!VIR_CLASS_NEW(virQEMUCaps, virClassForObject()))
613 614 615 616 617
        return -1;

    return 0;
}

618
VIR_ONCE_GLOBAL_INIT(virQEMUCaps);
619

620
static virArch virQEMUCapsArchFromString(const char *arch)
621 622 623 624 625
{
    if (STREQ(arch, "i386"))
        return VIR_ARCH_I686;
    if (STREQ(arch, "arm"))
        return VIR_ARCH_ARMV7L;
626 627
    if (STREQ(arch, "or32"))
        return VIR_ARCH_OR32;
628 629 630 631 632

    return virArchFromString(arch);
}


633
static const char *virQEMUCapsArchToString(virArch arch)
634 635 636
{
    if (arch == VIR_ARCH_I686)
        return "i386";
S
Stefan Schallenberg 已提交
637
    else if (arch == VIR_ARCH_ARMV6L || arch == VIR_ARCH_ARMV7L)
638
        return "arm";
639 640
    else if (arch == VIR_ARCH_OR32)
        return "or32";
641 642 643 644

    return virArchToString(arch);
}

645 646 647

/* Checks whether a domain with @guest arch can run natively on @host.
 */
648
bool
649 650 651
virQEMUCapsGuestIsNative(virArch host,
                         virArch guest)
{
652
    /* host & guest arches match */
653 654 655
    if (host == guest)
        return true;

656
    /* hostarch is x86_64 and guest arch is i686 (needs -cpu qemu32) */
657 658 659
    if (host == VIR_ARCH_X86_64 && guest == VIR_ARCH_I686)
        return true;

660
    /* hostarch is aarch64 and guest arch is armv7l (needs -cpu aarch64=off) */
661 662 663
    if (host == VIR_ARCH_AARCH64 && guest == VIR_ARCH_ARMV7L)
        return true;

664
    /* hostarch and guestarch are both ppc64 */
665 666 667 668 669 670 671
    if (ARCH_IS_PPC64(host) && ARCH_IS_PPC64(guest))
        return true;

    return false;
}


672 673 674 675 676 677 678 679 680
/* Given a host and guest architectures, find a suitable QEMU target.
 *
 * This is meant to be used as a second attempt if qemu-system-$guestarch
 * can't be found, eg. on a x86_64 host you want to use qemu-system-i386,
 * if available, instead of qemu-system-x86_64 to run i686 guests */
static virArch
virQEMUCapsFindTarget(virArch hostarch,
                      virArch guestarch)
{
681 682 683
    if (virQEMUCapsGuestIsNative(hostarch, guestarch))
        guestarch = hostarch;

684 685 686 687 688 689
    /* Both ppc64 and ppc64le guests can use the ppc64 target */
    if (ARCH_IS_PPC64(guestarch))
        guestarch = VIR_ARCH_PPC64;

    return guestarch;
}
690

691

692
static void
693 694
virQEMUCapsSetDefaultMachine(virQEMUCapsPtr qemuCaps,
                             size_t defIdx)
695
{
696
    struct virQEMUCapsMachineType tmp = qemuCaps->machineTypes[defIdx];
697 698 699 700

    memmove(qemuCaps->machineTypes + 1,
            qemuCaps->machineTypes,
            sizeof(qemuCaps->machineTypes[0]) * defIdx);
701 702

    qemuCaps->machineTypes[0] = tmp;
703 704
}

705

706
static char *
707 708
virQEMUCapsFindBinary(const char *format,
                      const char *archstr)
709
{
710 711
    char *ret = NULL;
    char *binary = NULL;
712

713
    if (virAsprintf(&binary, format, archstr) < 0)
714
        return NULL;
715 716 717

    ret = virFindFileInPath(binary);
    VIR_FREE(binary);
718 719 720 721 722 723 724 725 726 727 728
    return ret;
}

static char *
virQEMUCapsFindBinaryForArch(virArch hostarch,
                             virArch guestarch)
{
    char *ret = NULL;
    const char *archstr;
    virArch target;

729 730 731 732 733 734 735 736 737
    /* armv7l guests can only take advantage of KVM on aarch64 hosts by
     * using the qemu-system-aarch64 binary, so look for that one first
     * to avoid using qemu-system-arm (and thus TCG) instead */
    if (hostarch == VIR_ARCH_AARCH64 && guestarch == VIR_ARCH_ARMV7L) {
        archstr = virQEMUCapsArchToString(hostarch);
        if ((ret = virQEMUCapsFindBinary("qemu-system-%s", archstr)) != NULL)
            goto out;
    }

738 739 740 741 742 743 744 745 746 747 748
    /* First attempt: try the guest architecture as it is */
    archstr = virQEMUCapsArchToString(guestarch);
    if ((ret = virQEMUCapsFindBinary("qemu-system-%s", archstr)) != NULL)
        goto out;

    /* Second attempt: try looking up by target instead */
    target = virQEMUCapsFindTarget(hostarch, guestarch);
    if (target != guestarch) {
        archstr = virQEMUCapsArchToString(target);
        if ((ret = virQEMUCapsFindBinary("qemu-system-%s", archstr)) != NULL)
            goto out;
749
    }
750

751
 out:
752 753 754
    return ret;
}

755
static int
756
virQEMUCapsInitGuest(virCapsPtr caps,
757
                     virFileCachePtr cache,
758 759
                     virArch hostarch,
                     virArch guestarch)
760 761
{
    char *binary = NULL;
762
    virQEMUCapsPtr qemuCaps = NULL;
763 764
    int ret = -1;

J
Ján Tomko 已提交
765
    /* Check for existence of base emulator, or alternate base
766 767
     * which can be used with magic cpu choice
     */
768
    binary = virQEMUCapsFindBinaryForArch(hostarch, guestarch);
769

770 771 772 773 774 775 776
    /* RHEL doesn't follow the usual naming for QEMU binaries and ships
     * a single binary named qemu-kvm outside of $PATH instead */
    if (virQEMUCapsGuestIsNative(hostarch, guestarch) && !binary) {
        if (VIR_STRDUP(binary, "/usr/libexec/qemu-kvm") < 0)
            return -1;
    }

777
    /* Ignore binary if extracting version info fails */
778
    if (binary) {
779
        if (!(qemuCaps = virQEMUCapsCacheLookup(cache, binary))) {
780 781 782 783
            virResetLastError();
            VIR_FREE(binary);
        }
    }
784

785
    ret = virQEMUCapsInitGuestFromBinary(caps,
786
                                         binary, qemuCaps,
787 788 789
                                         guestarch);

    VIR_FREE(binary);
790
    virObjectUnref(qemuCaps);
791 792 793 794 795 796 797

    return ret;
}

int
virQEMUCapsInitGuestFromBinary(virCapsPtr caps,
                               const char *binary,
798
                               virQEMUCapsPtr qemuCaps,
799 800 801 802 803 804 805 806
                               virArch guestarch)
{
    virCapsGuestPtr guest;
    virCapsGuestMachinePtr *machines = NULL;
    size_t nmachines = 0;
    int ret = -1;
    bool hasdisksnapshot = false;

807 808 809
    if (!binary)
        return 0;

810
    if (virQEMUCapsGetMachineTypesCaps(qemuCaps, &nmachines, &machines) < 0)
811
        goto cleanup;
812 813 814 815

    /* We register kvm as the base emulator too, since we can
     * just give -no-kvm to disable acceleration if required */
    if ((guest = virCapabilitiesAddGuest(caps,
816
                                         VIR_DOMAIN_OSTYPE_HVM,
817
                                         guestarch,
818 819 820 821
                                         binary,
                                         NULL,
                                         nmachines,
                                         machines)) == NULL)
822
        goto cleanup;
823 824 825 826

    machines = NULL;
    nmachines = 0;

A
Andrea Bolognani 已提交
827 828 829
    /* CPU selection is always available, because all QEMU versions
     * we support can use at least '-cpu host' */
    if (!virCapabilitiesAddGuestFeature(guest, "cpuselection", true, false))
830
        goto cleanup;
831

832
    if (!virCapabilitiesAddGuestFeature(guest, "deviceboot", true, false))
833
        goto cleanup;
834

835
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DISK_SNAPSHOT))
836 837
        hasdisksnapshot = true;

838 839
    if (!virCapabilitiesAddGuestFeature(guest, "disksnapshot", hasdisksnapshot,
                                        false))
840
        goto cleanup;
841

D
Daniel P. Berrange 已提交
842
    if (virCapabilitiesAddGuestDomain(guest,
843
                                      VIR_DOMAIN_VIRT_QEMU,
D
Daniel P. Berrange 已提交
844 845 846 847
                                      NULL,
                                      NULL,
                                      0,
                                      NULL) == NULL)
848
        goto cleanup;
849

850
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
851 852 853 854 855 856
        if (virCapabilitiesAddGuestDomain(guest,
                                          VIR_DOMAIN_VIRT_KVM,
                                          NULL,
                                          NULL,
                                          0,
                                          NULL) == NULL) {
857
            goto cleanup;
D
Daniel P. Berrange 已提交
858
        }
859
    }
860

861 862 863
    if ((ARCH_IS_X86(guestarch) || guestarch == VIR_ARCH_AARCH64) &&
        virCapabilitiesAddGuestFeature(guest, "acpi", true, true) == NULL) {
        goto cleanup;
864 865
    }

A
Andrea Bolognani 已提交
866
    if (ARCH_IS_X86(guestarch) &&
867
        virCapabilitiesAddGuestFeature(guest, "apic", true, false) == NULL) {
868
        goto cleanup;
869
    }
870

871
    if ((guestarch == VIR_ARCH_I686) &&
872 873
        (virCapabilitiesAddGuestFeature(guest, "pae", true, false) == NULL ||
         virCapabilitiesAddGuestFeature(guest, "nonpae", true, false) == NULL))
874
        goto cleanup;
875 876 877

    ret = 0;

878
 cleanup:
879 880 881

    virCapabilitiesFreeMachines(machines, nmachines);

882
    return ret;
883 884 885
}


886
virCPUDefPtr
887
virQEMUCapsProbeHostCPUForEmulator(virArch hostArch,
888 889 890
                                   virQEMUCapsPtr qemuCaps,
                                   virDomainVirtType type)
{
891 892
    return virCPUGetHost(hostArch, VIR_CPU_TYPE_GUEST, NULL,
                         virQEMUCapsGetCPUDefinitions(qemuCaps, type));
893 894 895
}


896 897
virCapsPtr
virQEMUCapsInit(virFileCachePtr cache)
898 899
{
    virCapsPtr caps;
900
    size_t i;
T
Tal Kain 已提交
901
    virArch hostarch = virArchFromHost();
902

T
Tal Kain 已提交
903
    if ((caps = virCapabilitiesNew(hostarch,
904
                                   true, true)) == NULL)
905
        goto error;
906

907
    /* Some machines have problematic NUMA topology causing
908 909 910
     * unexpected failures. We don't want to break the QEMU
     * driver in this scenario, so log errors & carry on
     */
M
Martin Kletzander 已提交
911
    if (virCapabilitiesInitNUMA(caps) < 0) {
912
        virCapabilitiesFreeNUMAInfo(caps);
913
        VIR_WARN("Failed to query host NUMA topology, disabling NUMA capabilities");
914 915
    }

916 917 918
    if (virCapabilitiesInitCaches(caps) < 0)
        VIR_WARN("Failed to get host CPU cache info");

M
Martin Kletzander 已提交
919
    if (!(caps->host.cpu = virCPUProbeHost(caps->host.arch)))
920
        VIR_WARN("Failed to get host CPU");
921

922
    /* Add the power management features of the host */
923
    if (virNodeSuspendGetTargetMask(&caps->host.powerMgmt) < 0)
924 925
        VIR_WARN("Failed to get host power management capabilities");

926 927 928
    /* Add IOMMU info */
    virCapabilitiesHostInitIOMMU(caps);

M
Michal Privoznik 已提交
929
    /* Add huge pages info */
930
    if (virCapabilitiesInitPages(caps) < 0)
M
Michal Privoznik 已提交
931 932
        VIR_WARN("Failed to get pages info");

933 934 935
    /* Add domain migration transport URIs */
    virCapabilitiesAddHostMigrateTransport(caps, "tcp");
    virCapabilitiesAddHostMigrateTransport(caps, "rdma");
936

937 938 939 940
    /* QEMU can support pretty much every arch that exists,
     * so just probe for them all - we gracefully fail
     * if a qemu-system-$ARCH binary can't be found
     */
941
    for (i = 0; i < VIR_ARCH_LAST; i++)
942
        if (virQEMUCapsInitGuest(caps, cache,
T
Tal Kain 已提交
943
                                 hostarch,
944
                                 i) < 0)
945
            goto error;
946 947 948

    return caps;

949
 error:
950
    virObjectUnref(caps);
951 952 953 954
    return NULL;
}


955
struct virQEMUCapsStringFlags {
956 957 958 959 960
    const char *value;
    int flag;
};


961 962
struct virQEMUCapsStringFlags virQEMUCapsCommands[] = {
    { "transaction", QEMU_CAPS_TRANSACTION },
963
    { "block-stream", QEMU_CAPS_BLOCKJOB_ASYNC },
964 965 966 967 968 969 970
    { "dump-guest-memory", QEMU_CAPS_DUMP_GUEST_MEMORY },
    { "query-spice", QEMU_CAPS_SPICE },
    { "block-commit", QEMU_CAPS_BLOCK_COMMIT },
    { "query-vnc", QEMU_CAPS_VNC },
    { "drive-mirror", QEMU_CAPS_DRIVE_MIRROR },
    { "blockdev-snapshot-sync", QEMU_CAPS_DISK_SNAPSHOT },
    { "nbd-server-start", QEMU_CAPS_NBD_SERVER },
971
    { "change-backing-file", QEMU_CAPS_CHANGE_BACKING_FILE },
972
    { "rtc-reset-reinjection", QEMU_CAPS_RTC_RESET_REINJECTION },
973
    { "migrate-incoming", QEMU_CAPS_INCOMING_DEFER },
974
    { "query-hotpluggable-cpus", QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS },
975
    { "query-qmp-schema", QEMU_CAPS_QUERY_QMP_SCHEMA },
976 977 978
    { "query-cpu-model-expansion", QEMU_CAPS_QUERY_CPU_MODEL_EXPANSION },
    { "query-cpu-definitions", QEMU_CAPS_QUERY_CPU_DEFINITIONS },
    { "query-named-block-nodes", QEMU_CAPS_QUERY_NAMED_BLOCK_NODES },
979
    { "query-cpus-fast", QEMU_CAPS_QUERY_CPUS_FAST },
980
    { "qom-list-properties", QEMU_CAPS_QOM_LIST_PROPERTIES },
981
    { "blockdev-del", QEMU_CAPS_BLOCKDEV_DEL },
982 983
};

984 985 986 987
struct virQEMUCapsStringFlags virQEMUCapsMigration[] = {
    { "rdma-pin-all", QEMU_CAPS_MIGRATE_RDMA },
};

988 989 990
struct virQEMUCapsStringFlags virQEMUCapsEvents[] = {
    { "BALLOON_CHANGE", QEMU_CAPS_BALLOON_EVENT },
    { "SPICE_MIGRATE_COMPLETED", QEMU_CAPS_SEAMLESS_MIGRATION },
991
    { "DEVICE_DELETED", QEMU_CAPS_DEVICE_DEL_EVENT },
992
    { "MIGRATION", QEMU_CAPS_MIGRATION_EVENT },
993
    { "VSERPORT_CHANGE", QEMU_CAPS_VSERPORT_CHANGE },
994
    { "DEVICE_TRAY_MOVED", QEMU_CAPS_DEVICE_TRAY_MOVED },
995
    { "BLOCK_WRITE_THRESHOLD", QEMU_CAPS_BLOCK_WRITE_THRESHOLD },
996
    { "DUMP_COMPLETED", QEMU_CAPS_DUMP_COMPLETED },
997 998
};

999
struct virQEMUCapsStringFlags virQEMUCapsObjectTypes[] = {
1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014
    { "hda-duplex", QEMU_CAPS_HDA_DUPLEX },
    { "hda-micro", QEMU_CAPS_HDA_MICRO },
    { "ccid-card-emulated", QEMU_CAPS_CCID_EMULATED },
    { "ccid-card-passthru", QEMU_CAPS_CCID_PASSTHRU },
    { "piix3-usb-uhci", QEMU_CAPS_PIIX3_USB_UHCI },
    { "piix4-usb-uhci", QEMU_CAPS_PIIX4_USB_UHCI },
    { "usb-ehci", QEMU_CAPS_USB_EHCI },
    { "ich9-usb-ehci1", QEMU_CAPS_ICH9_USB_EHCI1 },
    { "vt82c686b-usb-uhci", QEMU_CAPS_VT82C686B_USB_UHCI },
    { "pci-ohci", QEMU_CAPS_PCI_OHCI },
    { "nec-usb-xhci", QEMU_CAPS_NEC_USB_XHCI },
    { "usb-redir", QEMU_CAPS_USB_REDIR },
    { "usb-hub", QEMU_CAPS_USB_HUB },
    { "ich9-ahci", QEMU_CAPS_ICH9_AHCI },
    { "virtio-blk-s390", QEMU_CAPS_VIRTIO_S390 },
1015
    { "virtio-blk-ccw", QEMU_CAPS_VIRTIO_CCW },
1016
    { "sclpconsole", QEMU_CAPS_DEVICE_SCLPCONSOLE },
1017
    { "lsi53c895a", QEMU_CAPS_SCSI_LSI },
1018
    { "virtio-scsi-pci", QEMU_CAPS_VIRTIO_SCSI },
1019 1020
    { "virtio-scsi-s390", QEMU_CAPS_VIRTIO_SCSI },
    { "virtio-scsi-ccw", QEMU_CAPS_VIRTIO_SCSI },
1021
    { "virtio-scsi-device", QEMU_CAPS_VIRTIO_SCSI },
1022
    { "megasas", QEMU_CAPS_SCSI_MEGASAS },
1023
    { "qxl", QEMU_CAPS_DEVICE_QXL },
1024 1025 1026
    { "sga", QEMU_CAPS_SGA },
    { "scsi-block", QEMU_CAPS_SCSI_BLOCK },
    { "ide-cd", QEMU_CAPS_IDE_CD },
1027 1028 1029
    { "VGA", QEMU_CAPS_DEVICE_VGA },
    { "cirrus-vga", QEMU_CAPS_DEVICE_CIRRUS_VGA },
    { "vmware-svga", QEMU_CAPS_DEVICE_VMWARE_SVGA },
H
Han Cheng 已提交
1030
    { "usb-serial", QEMU_CAPS_DEVICE_USB_SERIAL },
1031
    { "virtio-rng-pci", QEMU_CAPS_DEVICE_VIRTIO_RNG },
1032 1033
    { "virtio-rng-s390", QEMU_CAPS_DEVICE_VIRTIO_RNG },
    { "virtio-rng-ccw", QEMU_CAPS_DEVICE_VIRTIO_RNG },
1034
    { "virtio-rng-device", QEMU_CAPS_DEVICE_VIRTIO_RNG },
1035
    { "rng-random", QEMU_CAPS_OBJECT_RNG_RANDOM },
1036
    { "rng-egd", QEMU_CAPS_OBJECT_RNG_EGD },
1037
    { "spapr-nvram", QEMU_CAPS_DEVICE_NVRAM },
1038
    { "pci-bridge", QEMU_CAPS_DEVICE_PCI_BRIDGE },
1039
    { "vfio-pci", QEMU_CAPS_DEVICE_VFIO_PCI },
1040
    { "i82801b11-bridge", QEMU_CAPS_DEVICE_DMI_TO_PCI_BRIDGE },
1041
    { "usb-storage", QEMU_CAPS_DEVICE_USB_STORAGE },
1042
    { "virtio-mmio", QEMU_CAPS_DEVICE_VIRTIO_MMIO },
1043
    { "ich9-intel-hda", QEMU_CAPS_DEVICE_ICH9_INTEL_HDA },
H
Hu Tao 已提交
1044
    { "pvpanic", QEMU_CAPS_DEVICE_PANIC },
L
Li Zhang 已提交
1045
    { "usb-kbd", QEMU_CAPS_DEVICE_USB_KBD },
1046
    { "memory-backend-ram", QEMU_CAPS_OBJECT_MEMORY_RAM },
1047
    { "memory-backend-file", QEMU_CAPS_OBJECT_MEMORY_FILE },
1048
    { "usb-audio", QEMU_CAPS_OBJECT_USB_AUDIO },
J
John Ferlan 已提交
1049
    { "iothread", QEMU_CAPS_OBJECT_IOTHREAD},
1050
    { "ivshmem", QEMU_CAPS_DEVICE_IVSHMEM },
1051
    { "pc-dimm", QEMU_CAPS_DEVICE_PC_DIMM },
M
Michal Privoznik 已提交
1052
    { "pci-serial", QEMU_CAPS_DEVICE_PCI_SERIAL },
1053
    { "gpex-pcihost", QEMU_CAPS_OBJECT_GPEX},
1054
    { "ioh3420", QEMU_CAPS_DEVICE_IOH3420 },
1055
    { "x3130-upstream", QEMU_CAPS_DEVICE_X3130_UPSTREAM },
1056
    { "xio3130-downstream", QEMU_CAPS_DEVICE_XIO3130_DOWNSTREAM },
1057
    { "rtl8139", QEMU_CAPS_DEVICE_RTL8139 },
1058
    { "e1000", QEMU_CAPS_DEVICE_E1000 },
1059 1060 1061 1062
    { "virtio-net-pci", QEMU_CAPS_DEVICE_VIRTIO_NET },
    { "virtio-net-ccw", QEMU_CAPS_DEVICE_VIRTIO_NET },
    { "virtio-net-s390", QEMU_CAPS_DEVICE_VIRTIO_NET },
    { "virtio-net-device", QEMU_CAPS_DEVICE_VIRTIO_NET },
M
Marc-André Lureau 已提交
1063 1064
    { "virtio-gpu-pci", QEMU_CAPS_DEVICE_VIRTIO_GPU },
    { "virtio-gpu-device", QEMU_CAPS_DEVICE_VIRTIO_GPU },
1065
    { "virtio-vga", QEMU_CAPS_DEVICE_VIRTIO_VGA },
1066 1067 1068 1069 1070 1071
    { "virtio-keyboard-device", QEMU_CAPS_VIRTIO_KEYBOARD },
    { "virtio-keyboard-pci", QEMU_CAPS_VIRTIO_KEYBOARD },
    { "virtio-mouse-device", QEMU_CAPS_VIRTIO_MOUSE },
    { "virtio-mouse-pci", QEMU_CAPS_VIRTIO_MOUSE },
    { "virtio-tablet-device", QEMU_CAPS_VIRTIO_TABLET },
    { "virtio-tablet-pci", QEMU_CAPS_VIRTIO_TABLET },
1072 1073
    { "virtio-input-host-device", QEMU_CAPS_VIRTIO_INPUT_HOST },
    { "virtio-input-host-pci", QEMU_CAPS_VIRTIO_INPUT_HOST },
1074
    { "mptsas1068", QEMU_CAPS_SCSI_MPTSAS1068 },
1075
    { "secret", QEMU_CAPS_OBJECT_SECRET },
1076
    { "pxb", QEMU_CAPS_DEVICE_PXB },
1077
    { "pxb-pcie", QEMU_CAPS_DEVICE_PXB_PCIE },
1078
    { "tls-creds-x509", QEMU_CAPS_OBJECT_TLS_CREDS_X509 },
J
Ján Tomko 已提交
1079
    { "intel-iommu", QEMU_CAPS_DEVICE_INTEL_IOMMU },
1080 1081
    { "ivshmem-plain", QEMU_CAPS_DEVICE_IVSHMEM_PLAIN },
    { "ivshmem-doorbell", QEMU_CAPS_DEVICE_IVSHMEM_DOORBELL },
1082
    { "vhost-scsi", QEMU_CAPS_DEVICE_VHOST_SCSI },
1083
    { "nvdimm", QEMU_CAPS_DEVICE_NVDIMM },
1084
    { "pcie-root-port", QEMU_CAPS_DEVICE_PCIE_ROOT_PORT },
1085
    { "qemu-xhci", QEMU_CAPS_DEVICE_QEMU_XHCI },
1086
    { "spapr-pci-host-bridge", QEMU_CAPS_DEVICE_SPAPR_PCI_HOST_BRIDGE },
M
Marc-André Lureau 已提交
1087
    { "vmcoreinfo", QEMU_CAPS_DEVICE_VMCOREINFO },
1088
    { "spapr-vty", QEMU_CAPS_DEVICE_SPAPR_VTY },
1089
    { "sclplmconsole", QEMU_CAPS_DEVICE_SCLPLMCONSOLE },
1090
    { "isa-serial", QEMU_CAPS_DEVICE_ISA_SERIAL },
1091
    { "pl011", QEMU_CAPS_DEVICE_PL011 },
1092
    { "virtio-gpu-ccw", QEMU_CAPS_DEVICE_VIRTIO_GPU_CCW },
1093 1094 1095
    { "virtio-keyboard-ccw", QEMU_CAPS_DEVICE_VIRTIO_KEYBOARD_CCW },
    { "virtio-mouse-ccw", QEMU_CAPS_DEVICE_VIRTIO_MOUSE_CCW },
    { "virtio-tablet-ccw", QEMU_CAPS_DEVICE_VIRTIO_TABLET_CCW },
1096
    { "pcie-pci-bridge", QEMU_CAPS_DEVICE_PCIE_PCI_BRIDGE },
1097
    { "pr-manager-helper", QEMU_CAPS_PR_MANAGER_HELPER },
1098
    { "virtual-css-bridge", QEMU_CAPS_CCW },
1099
    { "vfio-ccw", QEMU_CAPS_DEVICE_VFIO_CCW },
1100
    { "hda-output", QEMU_CAPS_HDA_OUTPUT },
1101
    { "vmgenid", QEMU_CAPS_DEVICE_VMGENID },
1102
    { "vhost-vsock-device", QEMU_CAPS_DEVICE_VHOST_VSOCK },
1103
    { "mch", QEMU_CAPS_DEVICE_MCH },
1104
    { "sev-guest", QEMU_CAPS_SEV_GUEST },
B
Boris Fiuczynski 已提交
1105
    { "vfio-ap", QEMU_CAPS_DEVICE_VFIO_AP },
Y
Yi Min Zhao 已提交
1106
    { "zpci", QEMU_CAPS_DEVICE_ZPCI },
1107
    { "memory-backend-memfd", QEMU_CAPS_OBJECT_MEMORY_MEMFD },
1108 1109
};

1110
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioBalloon[] = {
1111
    { "deflate-on-oom", QEMU_CAPS_VIRTIO_BALLOON_AUTODEFLATE },
1112 1113 1114
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1115 1116
};

1117
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioBlk[] = {
1118 1119 1120 1121
    { "ioeventfd", QEMU_CAPS_VIRTIO_IOEVENTFD },
    { "event_idx", QEMU_CAPS_VIRTIO_BLK_EVENT_IDX },
    { "scsi", QEMU_CAPS_VIRTIO_BLK_SCSI },
    { "logical_block_size", QEMU_CAPS_BLOCKIO },
1122
    { "num-queues", QEMU_CAPS_VIRTIO_BLK_NUM_QUEUES },
1123
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1124 1125 1126
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1127
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1128 1129
};

1130
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioNet[] = {
1131 1132
    { "tx", QEMU_CAPS_VIRTIO_TX_ALG },
    { "event_idx", QEMU_CAPS_VIRTIO_NET_EVENT_IDX },
1133
    { "rx_queue_size", QEMU_CAPS_VIRTIO_NET_RX_QUEUE_SIZE },
1134
    { "tx_queue_size", QEMU_CAPS_VIRTIO_NET_TX_QUEUE_SIZE },
1135
    { "host_mtu", QEMU_CAPS_VIRTIO_NET_HOST_MTU },
1136 1137 1138
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1139 1140
};

1141
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsSpaprPCIHostBridge[] = {
1142 1143 1144
    { "numa_node", QEMU_CAPS_SPAPR_PCI_HOST_BRIDGE_NUMA_NODE },
};

1145
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioSCSI[] = {
1146
    { "iothread", QEMU_CAPS_VIRTIO_SCSI_IOTHREAD },
1147 1148 1149
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1150 1151
};

1152
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVfioPCI[] = {
1153
    { "display", QEMU_CAPS_VFIO_PCI_DISPLAY },
1154 1155
};

1156
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsSCSIDisk[] = {
1157 1158
    { "channel", QEMU_CAPS_SCSI_DISK_CHANNEL },
    { "wwn", QEMU_CAPS_SCSI_DISK_WWN },
1159
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1160
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1161 1162
};

1163
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsIDEDrive[] = {
1164
    { "wwn", QEMU_CAPS_IDE_DRIVE_WWN },
1165
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1166
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1167 1168
};

1169
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsPiix4PM[] = {
1170 1171
    { "disable_s3", QEMU_CAPS_PIIX_DISABLE_S3 },
    { "disable_s4", QEMU_CAPS_PIIX_DISABLE_S4 },
1172 1173
};

1174
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsUSBRedir[] = {
1175
    { "filter", QEMU_CAPS_USB_REDIR_FILTER },
1176 1177
};

1178
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsI440FXPCIHost[] = {
1179 1180 1181
    { "pci-hole64-size", QEMU_CAPS_I440FX_PCI_HOLE64_SIZE },
};

1182
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsQ35PCIHost[] = {
1183 1184 1185
    { "pci-hole64-size", QEMU_CAPS_Q35_PCI_HOLE64_SIZE },
};

1186
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsUSBStorage[] = {
1187
    { "removable", QEMU_CAPS_USB_STORAGE_REMOVABLE },
1188
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1189
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1190
    { "werror", QEMU_CAPS_USB_STORAGE_WERROR },
1191 1192
};

1193
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsKVMPit[] = {
1194 1195 1196
    { "lost_tick_policy", QEMU_CAPS_KVM_PIT_TICK_POLICY },
};

1197
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVGA[] = {
1198 1199 1200
    { "vgamem_mb", QEMU_CAPS_VGA_VGAMEM },
};

1201
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVmwareSvga[] = {
1202 1203 1204
    { "vgamem_mb", QEMU_CAPS_VMWARE_SVGA_VGAMEM },
};

1205
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsQxl[] = {
1206
    { "vgamem_mb", QEMU_CAPS_QXL_VGAMEM },
1207
    { "vram64_size_mb", QEMU_CAPS_QXL_VRAM64 },
1208
    { "max_outputs", QEMU_CAPS_QXL_MAX_OUTPUTS },
1209 1210
};

1211
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioGpu[] = {
1212
    { "virgl", QEMU_CAPS_VIRTIO_GPU_VIRGL },
1213
    { "max_outputs", QEMU_CAPS_VIRTIO_GPU_MAX_OUTPUTS },
1214 1215 1216
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1217 1218
};

1219
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsICH9[] = {
1220 1221 1222 1223
    { "disable_s3", QEMU_CAPS_ICH9_DISABLE_S3 },
    { "disable_s4", QEMU_CAPS_ICH9_DISABLE_S4 },
};

1224
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsUSBNECXHCI[] = {
1225 1226 1227
    { "p3", QEMU_CAPS_NEC_USB_XHCI_PORTS },
};

1228
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsIntelIOMMU[] = {
1229
    { "intremap", QEMU_CAPS_INTEL_IOMMU_INTREMAP },
1230
    { "caching-mode", QEMU_CAPS_INTEL_IOMMU_CACHING_MODE },
1231
    { "eim", QEMU_CAPS_INTEL_IOMMU_EIM },
1232
    { "device-iotlb", QEMU_CAPS_INTEL_IOMMU_DEVICE_IOTLB },
1233 1234
};

1235 1236 1237 1238
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsVirtualCSSBridge[] = {
    { "cssid-unrestricted", QEMU_CAPS_CCW_CSSID_UNRESTRICTED },
};

1239 1240 1241 1242
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsMCH[] = {
    { "extended-tseg-mbytes", QEMU_CAPS_MCH_EXTENDED_TSEG_MBYTES },
};

1243 1244 1245 1246
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsNVDIMM[] = {
    { "unarmed", QEMU_CAPS_DEVICE_NVDIMM_UNARMED },
};

1247
/* see documentation for virQEMUQAPISchemaPathGet for the query format */
1248
static struct virQEMUCapsStringFlags virQEMUCapsQMPSchemaQueries[] = {
1249
    { "blockdev-add/arg-type/options/+gluster/debug-level", QEMU_CAPS_GLUSTER_DEBUG_LEVEL},
1250
    { "blockdev-add/arg-type/+gluster/debug", QEMU_CAPS_GLUSTER_DEBUG_LEVEL},
J
John Ferlan 已提交
1251
    { "blockdev-add/arg-type/+vxhs", QEMU_CAPS_VXHS},
1252
    { "blockdev-add/arg-type/+iscsi/password-secret", QEMU_CAPS_ISCSI_PASSWORD_SECRET },
1253
    { "blockdev-add/arg-type/+qcow2/encrypt/+luks/key-secret", QEMU_CAPS_QCOW2_LUKS },
1254
    { "nbd-server-start/arg-type/tls-creds", QEMU_CAPS_NBD_TLS },
1255
    { "screendump/arg-type/device", QEMU_CAPS_SCREENDUMP_DEVICE },
1256
    { "block-commit/arg-type/*top",  QEMU_CAPS_ACTIVE_COMMIT },
1257
    { "query-iothreads/ret-type/poll-max-ns", QEMU_CAPS_IOTHREAD_POLLING },
1258
    { "query-display-options/ret-type/+egl-headless/rendernode", QEMU_CAPS_EGL_HEADLESS_RENDERNODE },
1259 1260
};

1261 1262
typedef struct _virQEMUCapsObjectTypeProps virQEMUCapsObjectTypeProps;
struct _virQEMUCapsObjectTypeProps {
1263
    const char *type;
1264
    struct virQEMUCapsStringFlags *props;
1265
    size_t nprops;
1266
    int capsCondition;
1267 1268
};

1269 1270 1271 1272 1273
typedef int (*virQEMUCapsObjectTypePropsCB)(qemuMonitorPtr mon,
                                            const char *type,
                                            char ***props);

static virQEMUCapsObjectTypeProps virQEMUCapsDeviceProps[] = {
1274 1275
    { "virtio-blk-pci", virQEMUCapsDevicePropsVirtioBlk,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBlk),
1276
      -1 },
1277 1278
    { "virtio-net-pci", virQEMUCapsDevicePropsVirtioNet,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioNet),
1279
      QEMU_CAPS_DEVICE_VIRTIO_NET },
1280 1281
    { "virtio-scsi-pci", virQEMUCapsDevicePropsVirtioSCSI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioSCSI),
1282
      QEMU_CAPS_VIRTIO_SCSI },
1283 1284
    { "virtio-blk-ccw", virQEMUCapsDevicePropsVirtioBlk,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBlk),
1285
      QEMU_CAPS_VIRTIO_CCW },
1286 1287
    { "virtio-net-ccw", virQEMUCapsDevicePropsVirtioNet,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioNet),
1288
      QEMU_CAPS_DEVICE_VIRTIO_NET },
1289 1290
    { "virtio-scsi-ccw", virQEMUCapsDevicePropsVirtioSCSI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioSCSI),
1291
      QEMU_CAPS_VIRTIO_SCSI },
1292 1293
    { "virtio-blk-s390", virQEMUCapsDevicePropsVirtioBlk,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBlk),
1294
      QEMU_CAPS_VIRTIO_S390 },
1295 1296
    { "virtio-net-s390", virQEMUCapsDevicePropsVirtioNet,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioNet),
1297
      QEMU_CAPS_DEVICE_VIRTIO_NET },
1298 1299
    { "vfio-pci", virQEMUCapsDevicePropsVfioPCI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVfioPCI),
1300
      QEMU_CAPS_DEVICE_VFIO_PCI },
1301
    { "scsi-hd", virQEMUCapsDevicePropsSCSIDisk,
1302
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsSCSIDisk),
1303
      -1 },
1304
    { "ide-hd", virQEMUCapsDevicePropsIDEDrive,
1305
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsIDEDrive),
1306
      -1 },
1307 1308
    { "PIIX4_PM", virQEMUCapsDevicePropsPiix4PM,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsPiix4PM),
1309
      -1 },
1310 1311
    { "usb-redir", virQEMUCapsDevicePropsUSBRedir,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsUSBRedir),
1312
      QEMU_CAPS_USB_REDIR },
1313 1314
    { "i440FX-pcihost", virQEMUCapsDevicePropsI440FXPCIHost,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsI440FXPCIHost),
1315
      -1 },
1316 1317
    { "q35-pcihost", virQEMUCapsDevicePropsQ35PCIHost,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsQ35PCIHost),
1318
      -1 },
1319 1320
    { "usb-storage", virQEMUCapsDevicePropsUSBStorage,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsUSBStorage),
1321
      QEMU_CAPS_DEVICE_USB_STORAGE },
1322 1323
    { "kvm-pit", virQEMUCapsDevicePropsKVMPit,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsKVMPit),
1324
      -1 },
1325 1326
    { "VGA", virQEMUCapsDevicePropsVGA,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVGA),
1327
      QEMU_CAPS_DEVICE_VGA },
1328 1329
    { "vmware-svga", virQEMUCapsDevicePropsVmwareSvga,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVmwareSvga),
1330
      QEMU_CAPS_DEVICE_VMWARE_SVGA },
1331 1332
    { "qxl", virQEMUCapsDevicePropsQxl,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsQxl),
1333
      QEMU_CAPS_DEVICE_QXL },
1334 1335
    { "virtio-gpu-pci", virQEMUCapsDevicePropsVirtioGpu,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioGpu),
1336
      QEMU_CAPS_DEVICE_VIRTIO_GPU },
1337 1338
    { "virtio-gpu-device", virQEMUCapsDevicePropsVirtioGpu,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioGpu),
1339
      QEMU_CAPS_DEVICE_VIRTIO_GPU },
1340 1341
    { "ICH9-LPC", virQEMUCapsDevicePropsICH9,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsICH9),
1342
      -1 },
1343 1344
    { "virtio-balloon-pci", virQEMUCapsDevicePropsVirtioBalloon,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBalloon),
1345
      -1 },
1346 1347
    { "virtio-balloon-ccw", virQEMUCapsDevicePropsVirtioBalloon,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBalloon),
1348
      -1 },
1349 1350
    { "virtio-balloon-device", virQEMUCapsDevicePropsVirtioBalloon,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBalloon),
1351
      -1 },
1352 1353
    { "nec-usb-xhci", virQEMUCapsDevicePropsUSBNECXHCI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsUSBNECXHCI),
1354
      QEMU_CAPS_NEC_USB_XHCI },
1355 1356
    { "intel-iommu", virQEMUCapsDevicePropsIntelIOMMU,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsIntelIOMMU),
1357
      QEMU_CAPS_DEVICE_INTEL_IOMMU },
1358 1359
    { "spapr-pci-host-bridge", virQEMUCapsDevicePropsSpaprPCIHostBridge,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsSpaprPCIHostBridge),
1360
      QEMU_CAPS_DEVICE_SPAPR_PCI_HOST_BRIDGE },
1361 1362
    { "virtio-gpu-ccw", virQEMUCapsDevicePropsVirtioGpu,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioGpu),
1363
      QEMU_CAPS_DEVICE_VIRTIO_GPU_CCW },
1364 1365 1366
    { "virtual-css-bridge", virQEMUCapsObjectPropsVirtualCSSBridge,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsVirtualCSSBridge),
      QEMU_CAPS_CCW },
1367 1368 1369
    { "mch", virQEMUCapsDevicePropsMCH,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsMCH),
      QEMU_CAPS_DEVICE_MCH },
1370 1371 1372
    { "nvdimm", virQEMUCapsDevicePropsNVDIMM,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsNVDIMM),
      QEMU_CAPS_DEVICE_NVDIMM },
1373 1374
};

1375 1376
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsMemoryBackendFile[] = {
    { "discard-data", QEMU_CAPS_OBJECT_MEMORY_FILE_DISCARD },
1377
    { "align", QEMU_CAPS_OBJECT_MEMORY_FILE_ALIGN },
1378
    { "pmem", QEMU_CAPS_OBJECT_MEMORY_FILE_PMEM },
1379 1380
};

1381 1382 1383 1384
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsMemoryBackendMemfd[] = {
    { "hugetlb", QEMU_CAPS_OBJECT_MEMORY_MEMFD_HUGETLB },
};

1385 1386
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsSPAPRMachine[] = {
    { "cap-hpt-max-page-size", QEMU_CAPS_MACHINE_PSERIES_CAP_HPT_MAX_PAGE_SIZE },
1387
    { "cap-htm", QEMU_CAPS_MACHINE_PSERIES_CAP_HTM },
1388
    { "cap-nested-hv", QEMU_CAPS_MACHINE_PSERIES_CAP_NESTED_HV },
1389 1390
};

1391 1392 1393 1394
static virQEMUCapsObjectTypeProps virQEMUCapsObjectProps[] = {
    { "memory-backend-file", virQEMUCapsObjectPropsMemoryBackendFile,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsMemoryBackendFile),
      QEMU_CAPS_OBJECT_MEMORY_FILE },
1395 1396 1397
    { "memory-backend-memfd", virQEMUCapsObjectPropsMemoryBackendMemfd,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsMemoryBackendMemfd),
      QEMU_CAPS_OBJECT_MEMORY_MEMFD },
1398 1399 1400
    { "spapr-machine", virQEMUCapsObjectPropsSPAPRMachine,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsSPAPRMachine),
      -1 },
1401
};
1402 1403

static void
1404 1405 1406 1407 1408
virQEMUCapsProcessStringFlags(virQEMUCapsPtr qemuCaps,
                              size_t nflags,
                              struct virQEMUCapsStringFlags *flags,
                              size_t nvalues,
                              char *const*values)
1409 1410
{
    size_t i, j;
1411
    for (i = 0; i < nflags; i++) {
1412 1413 1414
        if (virQEMUCapsGet(qemuCaps, flags[i].flag))
            continue;

1415
        for (j = 0; j < nvalues; j++) {
1416
            if (STREQ(values[j], flags[i].value)) {
1417
                virQEMUCapsSet(qemuCaps, flags[i].flag);
1418 1419 1420 1421 1422 1423 1424
                break;
            }
        }
    }
}


1425
int virQEMUCapsGetDefaultVersion(virCapsPtr caps,
1426
                                 virFileCachePtr capsCache,
1427
                                 unsigned int *version)
1428
{
1429
    virQEMUCapsPtr qemucaps;
T
Tal Kain 已提交
1430
    virArch hostarch;
1431
    virCapsDomainDataPtr capsdata;
1432 1433 1434 1435

    if (*version > 0)
        return 0;

T
Tal Kain 已提交
1436
    hostarch = virArchFromHost();
1437 1438 1439
    if (!(capsdata = virCapabilitiesDomainDataLookup(caps,
            VIR_DOMAIN_OSTYPE_HVM, hostarch, VIR_DOMAIN_VIRT_QEMU,
            NULL, NULL))) {
1440
        virReportError(VIR_ERR_INTERNAL_ERROR,
1441
                       _("Cannot find suitable emulator for %s"),
T
Tal Kain 已提交
1442
                       virArchToString(hostarch));
1443 1444 1445
        return -1;
    }

1446
    qemucaps = virQEMUCapsCacheLookup(capsCache, capsdata->emulator);
1447 1448
    VIR_FREE(capsdata);
    if (!qemucaps)
1449 1450
        return -1;

1451
    *version = virQEMUCapsGetVersion(qemucaps);
1452
    virObjectUnref(qemucaps);
1453 1454
    return 0;
}
1455 1456


1457 1458


1459 1460
virQEMUCapsPtr
virQEMUCapsNew(void)
1461
{
1462
    virQEMUCapsPtr qemuCaps;
1463

1464
    if (virQEMUCapsInitialize() < 0)
1465 1466
        return NULL;

1467
    if (!(qemuCaps = virObjectNew(virQEMUCapsClass)))
1468 1469
        return NULL;

1470
    if (!(qemuCaps->flags = virBitmapNew(QEMU_CAPS_LAST)))
1471
        goto error;
1472

1473
    return qemuCaps;
1474

1475
 error:
1476
    virObjectUnref(qemuCaps);
1477
    return NULL;
1478 1479 1480
}


1481
static int
1482 1483
virQEMUCapsHostCPUDataCopy(virQEMUCapsHostCPUDataPtr dst,
                           virQEMUCapsHostCPUDataPtr src)
1484
{
1485 1486
    if (src->info &&
        !(dst->info = qemuMonitorCPUModelInfoCopy(src->info)))
1487 1488
        return -1;

1489 1490
    if (src->reported &&
        !(dst->reported = virCPUDefCopy(src->reported)))
1491 1492
        return -1;

1493 1494 1495 1496
    if (src->migratable &&
        !(dst->migratable = virCPUDefCopy(src->migratable)))
        return -1;

1497 1498 1499 1500
    if (src->full &&
        !(dst->full = virCPUDefCopy(src->full)))
        return -1;

1501 1502 1503 1504
    return 0;
}


1505
static void
1506
virQEMUCapsHostCPUDataClear(virQEMUCapsHostCPUDataPtr cpuData)
1507
{
1508
    qemuMonitorCPUModelInfoFree(cpuData->info);
1509
    virCPUDefFree(cpuData->reported);
1510
    virCPUDefFree(cpuData->migratable);
1511
    virCPUDefFree(cpuData->full);
1512 1513

    memset(cpuData, 0, sizeof(*cpuData));
1514 1515 1516
}


1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535
static int
virQEMUCapsSEVInfoCopy(virSEVCapabilityPtr *dst,
                       virSEVCapabilityPtr src)
{
    VIR_AUTOPTR(virSEVCapability) tmp = NULL;

    if (VIR_ALLOC(tmp) < 0 ||
        VIR_STRDUP(tmp->pdh, src->pdh) < 0 ||
        VIR_STRDUP(tmp->cert_chain, src->cert_chain) < 0)
        return -1;

    tmp->cbitpos = src->cbitpos;
    tmp->reduced_phys_bits = src->reduced_phys_bits;

    VIR_STEAL_PTR(*dst, tmp);
    return 0;
}


1536
virQEMUCapsPtr virQEMUCapsNewCopy(virQEMUCapsPtr qemuCaps)
1537
{
1538
    virQEMUCapsPtr ret = virQEMUCapsNew();
1539 1540 1541 1542 1543
    size_t i;

    if (!ret)
        return NULL;

1544
    ret->usedQMP = qemuCaps->usedQMP;
1545
    ret->kvmSupportsNesting = qemuCaps->kvmSupportsNesting;
1546

1547 1548 1549 1550 1551
    if (VIR_STRDUP(ret->binary, qemuCaps->binary) < 0)
        goto error;

    ret->ctime = qemuCaps->ctime;

1552
    virBitmapCopy(ret->flags, qemuCaps->flags);
1553

1554 1555
    ret->version = qemuCaps->version;
    ret->kvmVersion = qemuCaps->kvmVersion;
1556
    ret->microcodeVersion = qemuCaps->microcodeVersion;
1557 1558 1559 1560

    if (VIR_STRDUP(ret->package, qemuCaps->package) < 0)
        goto error;

1561 1562 1563
    if (VIR_STRDUP(ret->kernelVersion, qemuCaps->kernelVersion) < 0)
        goto error;

1564
    ret->arch = qemuCaps->arch;
1565

1566 1567 1568 1569 1570 1571 1572 1573 1574
    if (qemuCaps->kvmCPUModels) {
        ret->kvmCPUModels = virDomainCapsCPUModelsCopy(qemuCaps->kvmCPUModels);
        if (!ret->kvmCPUModels)
            goto error;
    }

    if (qemuCaps->tcgCPUModels) {
        ret->tcgCPUModels = virDomainCapsCPUModelsCopy(qemuCaps->tcgCPUModels);
        if (!ret->tcgCPUModels)
1575
            goto error;
1576 1577
    }

1578 1579
    if (virQEMUCapsHostCPUDataCopy(&ret->kvmCPU, &qemuCaps->kvmCPU) < 0 ||
        virQEMUCapsHostCPUDataCopy(&ret->tcgCPU, &qemuCaps->tcgCPU) < 0)
1580 1581
        goto error;

1582
    if (VIR_ALLOC_N(ret->machineTypes, qemuCaps->nmachineTypes) < 0)
1583
        goto error;
1584
    ret->nmachineTypes = qemuCaps->nmachineTypes;
1585
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
1586 1587
        if (VIR_STRDUP(ret->machineTypes[i].name, qemuCaps->machineTypes[i].name) < 0 ||
            VIR_STRDUP(ret->machineTypes[i].alias, qemuCaps->machineTypes[i].alias) < 0)
1588
            goto error;
1589
        ret->machineTypes[i].maxCpus = qemuCaps->machineTypes[i].maxCpus;
1590
        ret->machineTypes[i].hotplugCpus = qemuCaps->machineTypes[i].hotplugCpus;
1591 1592
        ret->machineTypes[i].qemuDefault = qemuCaps->machineTypes[i].qemuDefault;

1593 1594
    }

1595 1596 1597 1598 1599 1600
    if (VIR_ALLOC_N(ret->gicCapabilities, qemuCaps->ngicCapabilities) < 0)
        goto error;
    ret->ngicCapabilities = qemuCaps->ngicCapabilities;
    for (i = 0; i < qemuCaps->ngicCapabilities; i++)
        ret->gicCapabilities[i] = qemuCaps->gicCapabilities[i];

1601 1602 1603 1604 1605
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_SEV_GUEST) &&
        virQEMUCapsSEVInfoCopy(&ret->sevCapabilities,
                               qemuCaps->sevCapabilities) < 0)
        goto error;

1606 1607
    return ret;

1608
 error:
1609 1610 1611 1612 1613
    virObjectUnref(ret);
    return NULL;
}


1614
void virQEMUCapsDispose(void *obj)
1615
{
1616
    virQEMUCapsPtr qemuCaps = obj;
1617 1618
    size_t i;

1619
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
1620 1621
        VIR_FREE(qemuCaps->machineTypes[i].name);
        VIR_FREE(qemuCaps->machineTypes[i].alias);
1622
    }
1623
    VIR_FREE(qemuCaps->machineTypes);
1624

1625 1626
    virObjectUnref(qemuCaps->kvmCPUModels);
    virObjectUnref(qemuCaps->tcgCPUModels);
1627

1628
    virBitmapFree(qemuCaps->flags);
1629

1630
    VIR_FREE(qemuCaps->package);
1631
    VIR_FREE(qemuCaps->kernelVersion);
1632
    VIR_FREE(qemuCaps->binary);
A
Andrea Bolognani 已提交
1633 1634

    VIR_FREE(qemuCaps->gicCapabilities);
1635

1636 1637
    virSEVCapabilitiesFree(qemuCaps->sevCapabilities);

1638 1639
    virQEMUCapsHostCPUDataClear(&qemuCaps->kvmCPU);
    virQEMUCapsHostCPUDataClear(&qemuCaps->tcgCPU);
1640 1641
}

1642
void
1643
virQEMUCapsSet(virQEMUCapsPtr qemuCaps,
1644
               virQEMUCapsFlags flag)
1645
{
1646
    ignore_value(virBitmapSetBit(qemuCaps->flags, flag));
1647 1648 1649 1650
}


void
1651
virQEMUCapsSetList(virQEMUCapsPtr qemuCaps, ...)
1652 1653 1654 1655
{
    va_list list;
    int flag;

1656
    va_start(list, qemuCaps);
1657
    while ((flag = va_arg(list, int)) < QEMU_CAPS_LAST)
1658
        ignore_value(virBitmapSetBit(qemuCaps->flags, flag));
1659
    va_end(list);
1660 1661 1662 1663
}


void
1664
virQEMUCapsClear(virQEMUCapsPtr qemuCaps,
1665
                 virQEMUCapsFlags flag)
1666
{
1667
    ignore_value(virBitmapClearBit(qemuCaps->flags, flag));
1668 1669 1670
}


1671
char *virQEMUCapsFlagsString(virQEMUCapsPtr qemuCaps)
1672
{
1673
    return virBitmapToString(qemuCaps->flags, true, false);
1674 1675 1676 1677
}


bool
1678
virQEMUCapsGet(virQEMUCapsPtr qemuCaps,
1679
               virQEMUCapsFlags flag)
1680
{
J
Ján Tomko 已提交
1681
    return qemuCaps && virBitmapIsBitSet(qemuCaps->flags, flag);
1682
}
1683 1684


D
Daniel P. Berrange 已提交
1685
bool virQEMUCapsHasPCIMultiBus(virQEMUCapsPtr qemuCaps,
1686
                               const virDomainDef *def)
D
Daniel P. Berrange 已提交
1687
{
1688 1689
    /* x86_64 and i686 support PCI-multibus on all machine types
     * since forever */
1690
    if (ARCH_IS_X86(def->os.arch))
D
Daniel P. Berrange 已提交
1691 1692 1693
        return true;

    if (def->os.arch == VIR_ARCH_PPC ||
1694
        ARCH_IS_PPC64(def->os.arch)) {
D
Daniel P. Berrange 已提交
1695 1696 1697 1698 1699
        /*
         * Usage of pci.0 naming:
         *
         *    ref405ep: no pci
         *       taihu: no pci
1700
         *      bamboo: 1.1.0 (<= 1.5.0, so basically forever)
D
Daniel P. Berrange 已提交
1701 1702
         *       mac99: 2.0.0
         *     g3beige: 2.0.0
1703
         *        prep: 1.4.0 (<= 1.5.0, so basically forever)
D
Daniel P. Berrange 已提交
1704 1705 1706 1707 1708 1709
         *     pseries: 2.0.0
         *   mpc8544ds: forever
         * virtex-m507: no pci
         *     ppce500: 1.6.0
         */

1710 1711 1712 1713 1714 1715 1716 1717
        /* We do not store the qemu version in domain status XML.
         * Hope the user is using a QEMU new enough to use 'pci.0',
         * otherwise the results of this function will be wrong
         * for domains already running at the time of daemon
         * restart */
        if (qemuCaps->version == 0)
            return true;

D
Daniel P. Berrange 已提交
1718 1719 1720 1721 1722 1723 1724
        if (qemuCaps->version >= 2000000)
            return true;

        if (qemuCaps->version >= 1006000 &&
            STREQ(def->os.machine, "ppce500"))
            return true;

1725 1726 1727
        if (STREQ(def->os.machine, "bamboo") ||
            STREQ(def->os.machine, "mpc8544ds") ||
            STREQ(def->os.machine, "prep")) {
D
Daniel P. Berrange 已提交
1728
            return true;
1729
        }
D
Daniel P. Berrange 已提交
1730 1731 1732 1733

        return false;
    }

1734 1735 1736 1737
    /* S390 supports PCI-multibus. */
    if (ARCH_IS_S390(def->os.arch))
        return true;

1738 1739 1740 1741
    /* If the virt machine, both on ARM and RISC-V, supports PCI,
     * then it also supports multibus */
    if (qemuDomainIsARMVirt(def) ||
        qemuDomainIsRISCVVirt(def)) {
1742
        return true;
1743
    }
1744

D
Daniel P. Berrange 已提交
1745 1746 1747 1748
    return false;
}


1749
const char *virQEMUCapsGetBinary(virQEMUCapsPtr qemuCaps)
1750
{
1751
    return qemuCaps->binary;
1752 1753
}

1754 1755 1756 1757 1758 1759 1760 1761 1762

void
virQEMUCapsSetArch(virQEMUCapsPtr qemuCaps,
                   virArch arch)
{
    qemuCaps->arch = arch;
}


1763
virArch virQEMUCapsGetArch(virQEMUCapsPtr qemuCaps)
1764
{
1765
    return qemuCaps->arch;
1766 1767 1768
}


1769
unsigned int virQEMUCapsGetVersion(virQEMUCapsPtr qemuCaps)
1770
{
1771
    return qemuCaps->version;
1772 1773 1774
}


1775
unsigned int virQEMUCapsGetKVMVersion(virQEMUCapsPtr qemuCaps)
1776
{
1777
    return qemuCaps->kvmVersion;
1778 1779 1780
}


1781 1782 1783 1784 1785 1786
const char *virQEMUCapsGetPackage(virQEMUCapsPtr qemuCaps)
{
    return qemuCaps->package;
}


1787 1788
int
virQEMUCapsAddCPUDefinitions(virQEMUCapsPtr qemuCaps,
1789
                             virDomainVirtType type,
1790
                             const char **name,
1791 1792
                             size_t count,
                             virDomainCapsCPUUsable usable)
1793
{
1794
    size_t i;
1795
    virDomainCapsCPUModelsPtr cpus = NULL;
1796

1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810
    if (type == VIR_DOMAIN_VIRT_KVM && qemuCaps->kvmCPUModels)
        cpus = qemuCaps->kvmCPUModels;
    else if (type == VIR_DOMAIN_VIRT_QEMU && qemuCaps->tcgCPUModels)
        cpus = qemuCaps->tcgCPUModels;

    if (!cpus) {
        if (!(cpus = virDomainCapsCPUModelsNew(count)))
            return -1;

        if (type == VIR_DOMAIN_VIRT_KVM)
            qemuCaps->kvmCPUModels = cpus;
        else
            qemuCaps->tcgCPUModels = cpus;
    }
1811 1812

    for (i = 0; i < count; i++) {
1813
        if (virDomainCapsCPUModelsAdd(cpus, name[i], -1, usable, NULL) < 0)
1814
            return -1;
1815
    }
1816

1817 1818 1819 1820
    return 0;
}


1821
virDomainCapsCPUModelsPtr
1822
virQEMUCapsGetCPUDefinitions(virQEMUCapsPtr qemuCaps,
1823
                             virDomainVirtType type)
1824
{
1825
    if (type == VIR_DOMAIN_VIRT_KVM)
1826
        return qemuCaps->kvmCPUModels;
1827
    else
1828
        return qemuCaps->tcgCPUModels;
1829 1830 1831
}


1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842
static virQEMUCapsHostCPUDataPtr
virQEMUCapsGetHostCPUData(virQEMUCapsPtr qemuCaps,
                          virDomainVirtType type)
{
    if (type == VIR_DOMAIN_VIRT_KVM)
        return &qemuCaps->kvmCPU;
    else
        return &qemuCaps->tcgCPU;
}


1843
virCPUDefPtr
1844
virQEMUCapsGetHostModel(virQEMUCapsPtr qemuCaps,
1845 1846
                        virDomainVirtType type,
                        virQEMUCapsHostCPUType cpuType)
1847
{
1848 1849
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);

1850 1851 1852
    switch (cpuType) {
    case VIR_QEMU_CAPS_HOST_CPU_REPORTED:
        return cpuData->reported;
1853 1854 1855

    case VIR_QEMU_CAPS_HOST_CPU_MIGRATABLE:
        return cpuData->migratable;
1856 1857 1858 1859 1860

    case VIR_QEMU_CAPS_HOST_CPU_FULL:
        /* 'full' is non-NULL only if we have data from both QEMU and
         * virCPUGetHost */
        return cpuData->full ? cpuData->full : cpuData->reported;
1861 1862 1863
    }

    return NULL;
1864 1865 1866
}


1867 1868 1869
static void
virQEMUCapsSetHostModel(virQEMUCapsPtr qemuCaps,
                        virDomainVirtType type,
1870
                        virCPUDefPtr reported,
1871 1872
                        virCPUDefPtr migratable,
                        virCPUDefPtr full)
1873
{
1874 1875
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);

1876 1877
    cpuData->reported = reported;
    cpuData->migratable = migratable;
1878
    cpuData->full = full;
1879 1880 1881
}


1882 1883 1884 1885 1886 1887
bool
virQEMUCapsIsCPUModeSupported(virQEMUCapsPtr qemuCaps,
                              virCapsPtr caps,
                              virDomainVirtType type,
                              virCPUMode mode)
{
1888 1889
    virDomainCapsCPUModelsPtr cpus;

1890 1891 1892 1893 1894 1895
    switch (mode) {
    case VIR_CPU_MODE_HOST_PASSTHROUGH:
        return type == VIR_DOMAIN_VIRT_KVM &&
               virQEMUCapsGuestIsNative(caps->host.arch, qemuCaps->arch);

    case VIR_CPU_MODE_HOST_MODEL:
1896 1897
        return !!virQEMUCapsGetHostModel(qemuCaps, type,
                                         VIR_QEMU_CAPS_HOST_CPU_REPORTED);
1898 1899

    case VIR_CPU_MODE_CUSTOM:
1900 1901 1902 1903 1904
        if (type == VIR_DOMAIN_VIRT_KVM)
            cpus = qemuCaps->kvmCPUModels;
        else
            cpus = qemuCaps->tcgCPUModels;
        return cpus && cpus->nmodels > 0;
1905 1906 1907 1908 1909 1910 1911 1912 1913

    case VIR_CPU_MODE_LAST:
        break;
    }

    return false;
}


1914 1915 1916
int virQEMUCapsGetMachineTypesCaps(virQEMUCapsPtr qemuCaps,
                                   size_t *nmachines,
                                   virCapsGuestMachinePtr **machines)
1917 1918 1919 1920
{
    size_t i;

    *machines = NULL;
1921
    *nmachines = qemuCaps->nmachineTypes;
1922

1923 1924 1925 1926
    if (*nmachines &&
        VIR_ALLOC_N(*machines, qemuCaps->nmachineTypes) < 0)
        goto error;

1927
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
1928 1929
        virCapsGuestMachinePtr mach;
        if (VIR_ALLOC(mach) < 0)
1930
            goto error;
1931
        (*machines)[i] = mach;
1932 1933 1934
        if (qemuCaps->machineTypes[i].alias) {
            if (VIR_STRDUP(mach->name, qemuCaps->machineTypes[i].alias) < 0 ||
                VIR_STRDUP(mach->canonical, qemuCaps->machineTypes[i].name) < 0)
1935
                goto error;
1936
        } else {
1937
            if (VIR_STRDUP(mach->name, qemuCaps->machineTypes[i].name) < 0)
1938
                goto error;
1939
        }
1940
        mach->maxCpus = qemuCaps->machineTypes[i].maxCpus;
1941 1942
    }

1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980
    /* Make sure all canonical machine types also have their own entry so that
     * /capabilities/guest/arch[@name='...']/machine/text() XPath selects all
     * supported machine types.
     */
    i = 0;
    while (i < *nmachines) {
        size_t j;
        bool found = false;
        virCapsGuestMachinePtr machine = (*machines)[i];

        if (!machine->canonical) {
            i++;
            continue;
        }

        for (j = 0; j < *nmachines; j++) {
            if (STREQ(machine->canonical, (*machines)[j]->name)) {
                found = true;
                break;
            }
        }

        if (!found) {
            virCapsGuestMachinePtr mach;
            if (VIR_ALLOC(mach) < 0)
                goto error;
            if (VIR_INSERT_ELEMENT_COPY(*machines, i, *nmachines, mach) < 0) {
                VIR_FREE(mach);
                goto error;
            }
            if (VIR_STRDUP(mach->name, machine->canonical) < 0)
                goto error;
            mach->maxCpus = machine->maxCpus;
            i++;
        }
        i++;
    }

1981 1982
    return 0;

1983
 error:
1984 1985 1986 1987 1988 1989 1990
    virCapabilitiesFreeMachines(*machines, *nmachines);
    *nmachines = 0;
    *machines = NULL;
    return -1;
}


1991 1992 1993 1994 1995 1996 1997 1998
/**
 * virQEMUCapsGetCanonicalMachine:
 * @qemuCaps: qemu capabilities object
 * @name: machine name
 *
 * Resolves aliased machine names to the actual machine name. If qemuCaps isn't
 * present @name is returned.
 */
1999 2000
const char *virQEMUCapsGetCanonicalMachine(virQEMUCapsPtr qemuCaps,
                                           const char *name)
2001 2002 2003
{
    size_t i;

2004 2005
    if (!name || !qemuCaps)
        return name;
2006

2007
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
2008
        if (!qemuCaps->machineTypes[i].alias)
2009
            continue;
2010 2011
        if (STREQ(qemuCaps->machineTypes[i].alias, name))
            return qemuCaps->machineTypes[i].name;
2012 2013 2014 2015
    }

    return name;
}
2016

2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
const char *
virQEMUCapsGetDefaultMachine(virQEMUCapsPtr qemuCaps)
{
    size_t i;

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
        if (qemuCaps->machineTypes[i].qemuDefault)
            return qemuCaps->machineTypes[i].name;
    }

    return NULL;
}
2029

2030 2031 2032 2033 2034 2035 2036 2037 2038
int virQEMUCapsGetMachineMaxCpus(virQEMUCapsPtr qemuCaps,
                                 const char *name)
{
    size_t i;

    if (!name)
        return 0;

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
2039
        if (!qemuCaps->machineTypes[i].maxCpus)
2040
            continue;
2041 2042
        if (STREQ(qemuCaps->machineTypes[i].name, name))
            return qemuCaps->machineTypes[i].maxCpus;
2043 2044 2045 2046 2047 2048
    }

    return 0;
}


2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062
bool virQEMUCapsGetMachineHotplugCpus(virQEMUCapsPtr qemuCaps,
                                      const char *name)
{
    size_t i;

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
        if (STREQ_NULLABLE(qemuCaps->machineTypes[i].name, name))
            return qemuCaps->machineTypes[i].hotplugCpus;
    }

    return false;
}


2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086
/**
 * virQEMUCapsSetGICCapabilities:
 * @qemuCaps: QEMU capabilities
 * @capabilities: GIC capabilities
 * @ncapabilities: number of GIC capabilities
 *
 * Set the GIC capabilities for @qemuCaps.
 *
 * The ownership of @capabilities is taken away from the caller, ie. this
 * function will not make a copy of @capabilities, so releasing that memory
 * after it's been called is a bug.
 */
void
virQEMUCapsSetGICCapabilities(virQEMUCapsPtr qemuCaps,
                              virGICCapability *capabilities,
                              size_t ncapabilities)
{
    VIR_FREE(qemuCaps->gicCapabilities);

    qemuCaps->gicCapabilities = capabilities;
    qemuCaps->ngicCapabilities = ncapabilities;
}


2087 2088 2089 2090 2091 2092 2093
virSEVCapabilityPtr
virQEMUCapsGetSEVCapabilities(virQEMUCapsPtr qemuCaps)
{
    return qemuCaps->sevCapabilities;
}


2094
static int
2095 2096
virQEMUCapsProbeQMPCommands(virQEMUCapsPtr qemuCaps,
                            qemuMonitorPtr mon)
2097 2098 2099 2100 2101 2102 2103
{
    char **commands = NULL;
    int ncommands;

    if ((ncommands = qemuMonitorGetCommands(mon, &commands)) < 0)
        return -1;

2104 2105 2106 2107
    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsCommands),
                                  virQEMUCapsCommands,
                                  ncommands, commands);
2108
    virStringListFreeCount(commands, ncommands);
2109

2110 2111 2112 2113
    /* Probe for active commit of qemu 2.1. We don't need to query directly
     * if we have QMP schema support */
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_QMP_SCHEMA) &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_BLOCK_COMMIT) &&
2114 2115 2116
        qemuMonitorSupportsActiveCommit(mon))
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_ACTIVE_COMMIT);

2117 2118 2119 2120 2121
    return 0;
}


static int
2122 2123
virQEMUCapsProbeQMPEvents(virQEMUCapsPtr qemuCaps,
                          qemuMonitorPtr mon)
2124 2125 2126 2127 2128 2129 2130
{
    char **events = NULL;
    int nevents;

    if ((nevents = qemuMonitorGetEvents(mon, &events)) < 0)
        return -1;

2131 2132 2133 2134
    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsEvents),
                                  virQEMUCapsEvents,
                                  nevents, events);
2135
    virStringListFreeCount(events, nevents);
2136 2137 2138 2139

    return 0;
}

2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168
static int
virQEMUCapsProbeQMPGenericProps(virQEMUCapsPtr qemuCaps,
                                qemuMonitorPtr mon,
                                virQEMUCapsObjectTypeProps *props,
                                size_t nprops,
                                virQEMUCapsObjectTypePropsCB propsGetCB)
{
    int nvalues;
    char **values;
    size_t i;

    for (i = 0; i < nprops; i++) {
        const char *type = props[i].type;
        int cap = props[i].capsCondition;

        if (cap >= 0 && !virQEMUCapsGet(qemuCaps, cap))
            continue;

        if ((nvalues = propsGetCB(mon, type, &values)) < 0)
            return -1;
        virQEMUCapsProcessStringFlags(qemuCaps,
                                      props[i].nprops,
                                      props[i].props,
                                      nvalues, values);
        virStringListFreeCount(values, nvalues);
    }

    return 0;
}
2169

2170
static int
2171
virQEMUCapsProbeQMPDevices(virQEMUCapsPtr qemuCaps,
2172
                           qemuMonitorPtr mon)
2173 2174 2175 2176 2177 2178
{
    int nvalues;
    char **values;

    if ((nvalues = qemuMonitorGetObjectTypes(mon, &values)) < 0)
        return -1;
2179 2180 2181 2182
    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsObjectTypes),
                                  virQEMUCapsObjectTypes,
                                  nvalues, values);
2183
    virStringListFreeCount(values, nvalues);
2184

2185 2186 2187 2188 2189 2190
    if (virQEMUCapsProbeQMPGenericProps(qemuCaps,
                                        mon,
                                        virQEMUCapsDeviceProps,
                                        ARRAY_CARDINALITY(virQEMUCapsDeviceProps),
                                        qemuMonitorGetDeviceProps) < 0)
        return -1;
2191

2192 2193 2194 2195 2196 2197 2198 2199
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_QOM_LIST_PROPERTIES) &&
        virQEMUCapsProbeQMPGenericProps(qemuCaps,
                                        mon,
                                        virQEMUCapsObjectProps,
                                        ARRAY_CARDINALITY(virQEMUCapsObjectProps),
                                        qemuMonitorGetObjectProps) < 0)
        return -1;

2200 2201 2202 2203
    return 0;
}


2204 2205 2206 2207 2208 2209 2210 2211 2212 2213
/* Historically QEMU x86 targets defaulted to 'pc' machine type but
 * in future x86_64 might switch to 'q35'. Such a change is considered
 * an ABI break from libvirt's POV. Other QEMU targets may not declare
 * a default machine at all, causing libvirt to use the first reported
 * machine in the list.
 *
 * Here we record a preferred default machine for all arches, so
 * that we're not vulnerable to changes in QEMU defaults or machine
 * list ordering.
 */
2214
static const char *preferredMachines[] =
2215
{
2216 2217
    NULL, /* VIR_ARCH_NONE (not a real arch :) */
    "clipper", /* VIR_ARCH_ALPHA */
S
Stefan Schallenberg 已提交
2218
    "integratorcp", /* VIR_ARCH_ARMV6L */
2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245
    "integratorcp", /* VIR_ARCH_ARMV7L */
    "integratorcp", /* VIR_ARCH_ARMV7B */

    "integratorcp", /* VIR_ARCH_AARCH64 */
    "axis-dev88", /* VIR_ARCH_CRIS */
    "pc", /* VIR_ARCH_I686 */
    NULL, /* VIR_ARCH_ITANIUM (doesn't exist in QEMU any more) */
    "lm32-evr", /* VIR_ARCH_LM32 */

    "mcf5208evb", /* VIR_ARCH_M68K */
    "petalogix-s3adsp1800", /* VIR_ARCH_MICROBLAZE */
    "petalogix-s3adsp1800", /* VIR_ARCH_MICROBLAZEEL */
    "malta", /* VIR_ARCH_MIPS */
    "malta", /* VIR_ARCH_MIPSEL */

    "malta", /* VIR_ARCH_MIPS64 */
    "malta", /* VIR_ARCH_MIPS64EL */
    "or1k-sim", /* VIR_ARCH_OR32 */
    NULL, /* VIR_ARCH_PARISC (no QEMU impl) */
    NULL, /* VIR_ARCH_PARISC64 (no QEMU impl) */

    "g3beige", /* VIR_ARCH_PPC */
    "g3beige", /* VIR_ARCH_PPCLE */
    "pseries", /* VIR_ARCH_PPC64 */
    "pseries", /* VIR_ARCH_PPC64LE */
    "bamboo", /* VIR_ARCH_PPCEMB */

L
Lubomir Rintel 已提交
2246 2247
    "spike_v1.10", /* VIR_ARCH_RISCV32 */
    "spike_v1.10", /* VIR_ARCH_RISCV64 */
2248 2249 2250
    NULL, /* VIR_ARCH_S390 (no QEMU impl) */
    "s390-ccw-virtio", /* VIR_ARCH_S390X */
    "shix", /* VIR_ARCH_SH4 */
L
Lubomir Rintel 已提交
2251

2252 2253 2254 2255 2256
    "shix", /* VIR_ARCH_SH4EB */
    "SS-5", /* VIR_ARCH_SPARC */
    "sun4u", /* VIR_ARCH_SPARC64 */
    "puv3", /* VIR_ARCH_UNICORE32 */
    "pc", /* VIR_ARCH_X86_64 */
L
Lubomir Rintel 已提交
2257

2258 2259
    "sim", /* VIR_ARCH_XTENSA */
    "sim", /* VIR_ARCH_XTENSAEB */
2260
};
2261
verify(ARRAY_CARDINALITY(preferredMachines) == VIR_ARCH_LAST);
2262 2263


2264
static int
2265 2266
virQEMUCapsProbeQMPMachineTypes(virQEMUCapsPtr qemuCaps,
                                qemuMonitorPtr mon)
2267 2268 2269 2270 2271
{
    qemuMonitorMachineInfoPtr *machines = NULL;
    int nmachines = 0;
    int ret = -1;
    size_t i;
2272 2273 2274
    ssize_t defIdx = -1;
    ssize_t preferredIdx = -1;
    const char *preferredMachine = preferredMachines[qemuCaps->arch];
2275 2276

    if ((nmachines = qemuMonitorGetMachines(mon, &machines)) < 0)
2277
        return -1;
2278

2279
    if (VIR_ALLOC_N(qemuCaps->machineTypes, nmachines) < 0)
2280 2281
        goto cleanup;

2282
    for (i = 0; i < nmachines; i++) {
2283
        struct virQEMUCapsMachineType *mach;
2284 2285
        if (STREQ(machines[i]->name, "none"))
            continue;
2286 2287 2288 2289 2290

        mach = &(qemuCaps->machineTypes[qemuCaps->nmachineTypes++]);

        if (VIR_STRDUP(mach->alias, machines[i]->alias) < 0 ||
            VIR_STRDUP(mach->name, machines[i]->name) < 0)
2291
            goto cleanup;
2292 2293

        mach->maxCpus = machines[i]->maxCpus;
2294
        mach->hotplugCpus = machines[i]->hotplugCpus;
2295

2296 2297 2298 2299 2300 2301
        if (preferredMachine &&
            (STREQ_NULLABLE(mach->alias, preferredMachine) ||
             STREQ(mach->name, preferredMachine))) {
            preferredIdx = qemuCaps->nmachineTypes - 1;
        }

2302 2303
        if (machines[i]->isDefault) {
            mach->qemuDefault = true;
2304
            defIdx = qemuCaps->nmachineTypes - 1;
2305
        }
2306
    }
2307

2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320
    /*
     * We'll prefer to use our own historical default machine
     * to avoid mgmt apps seeing semantics changes when QEMU
     * alters its defaults.
     *
     * Our preferred machine might have been compiled out of
     * QEMU at build time though, so we still fallback to honouring
     * QEMU's reported default in that case
     */
    if (preferredIdx == -1)
        preferredIdx = defIdx;
    if (preferredIdx != -1)
        virQEMUCapsSetDefaultMachine(qemuCaps, preferredIdx);
2321 2322 2323

    ret = 0;

2324
 cleanup:
2325
    for (i = 0; i < nmachines; i++)
2326 2327 2328 2329 2330 2331
        qemuMonitorMachineInfoFree(machines[i]);
    VIR_FREE(machines);
    return ret;
}


2332 2333
virDomainCapsCPUModelsPtr
virQEMUCapsFetchCPUDefinitions(qemuMonitorPtr mon)
2334
{
2335 2336 2337
    virDomainCapsCPUModelsPtr models = NULL;
    qemuMonitorCPUDefInfoPtr *cpus = NULL;
    int ncpus = 0;
2338
    size_t i;
2339

2340
    if ((ncpus = qemuMonitorGetCPUDefinitions(mon, &cpus)) < 0)
2341
        return NULL;
2342

2343
    if (!(models = virDomainCapsCPUModelsNew(ncpus)))
2344
        goto error;
2345

2346
    for (i = 0; i < ncpus; i++) {
2347 2348 2349 2350 2351 2352 2353
        virDomainCapsCPUUsable usable = VIR_DOMCAPS_CPU_USABLE_UNKNOWN;

        if (cpus[i]->usable == VIR_TRISTATE_BOOL_YES)
            usable = VIR_DOMCAPS_CPU_USABLE_YES;
        else if (cpus[i]->usable == VIR_TRISTATE_BOOL_NO)
            usable = VIR_DOMCAPS_CPU_USABLE_NO;

2354
        if (virDomainCapsCPUModelsAddSteal(models, &cpus[i]->name, usable,
2355
                                           &cpus[i]->blockers) < 0)
2356
            goto error;
2357 2358 2359 2360 2361 2362
    }

 cleanup:
    for (i = 0; i < ncpus; i++)
        qemuMonitorCPUDefInfoFree(cpus[i]);
    VIR_FREE(cpus);
2363 2364 2365 2366 2367 2368
    return models;

 error:
    virObjectUnref(models);
    models = NULL;
    goto cleanup;
2369 2370
}

2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393

int
virQEMUCapsProbeQMPCPUDefinitions(virQEMUCapsPtr qemuCaps,
                                  qemuMonitorPtr mon,
                                  bool tcg)
{
    virDomainCapsCPUModelsPtr models = NULL;

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_CPU_DEFINITIONS))
        return 0;

    if (!(models = virQEMUCapsFetchCPUDefinitions(mon)))
        return -1;

    if (tcg || !virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM))
        qemuCaps->tcgCPUModels = models;
    else
        qemuCaps->kvmCPUModels = models;

    return 0;
}


2394 2395
static int
virQEMUCapsProbeQMPHostCPU(virQEMUCapsPtr qemuCaps,
2396 2397
                           qemuMonitorPtr mon,
                           bool tcg)
2398
{
2399
    qemuMonitorCPUModelInfoPtr modelInfo = NULL;
2400 2401
    qemuMonitorCPUModelInfoPtr nonMigratable = NULL;
    virHashTablePtr hash = NULL;
2402
    const char *model;
2403
    qemuMonitorCPUModelExpansionType type;
2404 2405
    virDomainVirtType virtType;
    virQEMUCapsHostCPUDataPtr cpuData;
2406
    int ret = -1;
2407 2408

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_CPU_MODEL_EXPANSION))
2409 2410
        return 0;

2411
    if (tcg || !virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
2412
        virtType = VIR_DOMAIN_VIRT_QEMU;
2413 2414
        model = "max";
    } else {
2415
        virtType = VIR_DOMAIN_VIRT_KVM;
2416 2417 2418
        model = "host";
    }

2419 2420
    cpuData = virQEMUCapsGetHostCPUData(qemuCaps, virtType);

2421 2422 2423 2424 2425 2426 2427 2428 2429 2430
    /* Some x86_64 features defined in cpu_map.xml use spelling which differ
     * from the one preferred by QEMU. Static expansion would give us only the
     * preferred spelling, thus we need to do a full expansion on the result of
     * the initial static expansion to get all variants of all features.
     */
    if (ARCH_IS_X86(qemuCaps->arch))
        type = QEMU_MONITOR_CPU_MODEL_EXPANSION_STATIC_FULL;
    else
        type = QEMU_MONITOR_CPU_MODEL_EXPANSION_STATIC;

2431 2432
    if (qemuMonitorGetCPUModelExpansion(mon, type, model, true, &modelInfo) < 0)
        goto cleanup;
2433 2434

    /* Try to check migratability of each feature. */
2435
    if (modelInfo &&
2436 2437
        qemuMonitorGetCPUModelExpansion(mon, type, model, false,
                                        &nonMigratable) < 0)
2438
        goto cleanup;
2439 2440 2441 2442 2443 2444 2445

    if (nonMigratable) {
        qemuMonitorCPUPropertyPtr prop;
        qemuMonitorCPUPropertyPtr nmProp;
        size_t i;

        if (!(hash = virHashCreate(0, NULL)))
2446
            goto cleanup;
2447

2448 2449
        for (i = 0; i < modelInfo->nprops; i++) {
            prop = modelInfo->props + i;
2450
            if (virHashAddEntry(hash, prop->name, prop) < 0)
2451
                goto cleanup;
2452 2453 2454 2455 2456 2457 2458 2459 2460
        }

        for (i = 0; i < nonMigratable->nprops; i++) {
            nmProp = nonMigratable->props + i;
            if (!(prop = virHashLookup(hash, nmProp->name)) ||
                prop->type != QEMU_MONITOR_CPU_PROPERTY_BOOLEAN ||
                prop->type != nmProp->type)
                continue;

2461
            if (prop->value.boolean) {
2462
                prop->migratable = VIR_TRISTATE_BOOL_YES;
2463 2464 2465 2466
            } else if (nmProp->value.boolean) {
                prop->value.boolean = true;
                prop->migratable = VIR_TRISTATE_BOOL_NO;
            }
2467 2468
        }

2469
        modelInfo->migratability = true;
2470 2471
    }

2472
    VIR_STEAL_PTR(cpuData->info, modelInfo);
2473 2474 2475 2476 2477
    ret = 0;

 cleanup:
    virHashFree(hash);
    qemuMonitorCPUModelInfoFree(nonMigratable);
2478
    qemuMonitorCPUModelInfoFree(modelInfo);
2479 2480

    return ret;
2481 2482
}

2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534

/**
 * Get NULL terminated list of features supported by QEMU.
 *
 * Returns -1 on error,
 *          0 on success (@features will be NULL if QEMU does not support this),
 *          1 when @features is filled in, but migratability info is not available.
 */
int
virQEMUCapsGetCPUFeatures(virQEMUCapsPtr qemuCaps,
                          virDomainVirtType virtType,
                          bool migratable,
                          char ***features)
{
    virQEMUCapsHostCPUDataPtr data;
    char **list;
    size_t i;
    size_t n;
    int ret = -1;

    *features = NULL;
    data = virQEMUCapsGetHostCPUData(qemuCaps, virtType);

    if (!data->info)
        return 0;

    if (VIR_ALLOC_N(list, data->info->nprops + 1) < 0)
        return -1;

    n = 0;
    for (i = 0; i < data->info->nprops; i++) {
        qemuMonitorCPUPropertyPtr prop = data->info->props + i;

        if (migratable && prop->migratable == VIR_TRISTATE_BOOL_NO)
            continue;

        if (VIR_STRDUP(list[n++], prop->name) < 0)
            goto cleanup;
    }

    VIR_STEAL_PTR(*features, list);
    if (migratable && !data->info->migratability)
        ret = 1;
    else
        ret = 0;

 cleanup:
    virStringListFree(list);
    return ret;
}


2535 2536
struct tpmTypeToCaps {
    int type;
2537
    virQEMUCapsFlags caps;
2538 2539 2540 2541 2542 2543 2544
};

static const struct tpmTypeToCaps virQEMUCapsTPMTypesToCaps[] = {
    {
        .type = VIR_DOMAIN_TPM_TYPE_PASSTHROUGH,
        .caps = QEMU_CAPS_DEVICE_TPM_PASSTHROUGH,
    },
2545 2546 2547 2548
    {
        .type = VIR_DOMAIN_TPM_TYPE_EMULATOR,
        .caps = QEMU_CAPS_DEVICE_TPM_EMULATOR,
    },
2549 2550 2551 2552 2553 2554 2555
};

const struct tpmTypeToCaps virQEMUCapsTPMModelsToCaps[] = {
    {
        .type = VIR_DOMAIN_TPM_MODEL_TIS,
        .caps = QEMU_CAPS_DEVICE_TPM_TIS,
    },
2556 2557 2558 2559
    {
        .type = VIR_DOMAIN_TPM_MODEL_CRB,
        .caps = QEMU_CAPS_DEVICE_TPM_CRB,
    },
2560 2561 2562 2563 2564 2565
};

static int
virQEMUCapsProbeQMPTPM(virQEMUCapsPtr qemuCaps,
                       qemuMonitorPtr mon)
{
2566 2567
    int nentries;
    size_t i;
2568
    char **entries = NULL;
S
Stefan Berger 已提交
2569

2570 2571 2572 2573 2574 2575 2576
    if ((nentries = qemuMonitorGetTPMModels(mon, &entries)) < 0)
        return -1;

    if (nentries > 0) {
        for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsTPMModelsToCaps); i++) {
            const char *needle = virDomainTPMModelTypeToString(
                virQEMUCapsTPMModelsToCaps[i].type);
2577
            if (virStringListHasString((const char **)entries, needle))
2578 2579 2580 2581
                virQEMUCapsSet(qemuCaps,
                               virQEMUCapsTPMModelsToCaps[i].caps);
        }
    }
2582
    virStringListFree(entries);
2583 2584 2585 2586 2587 2588 2589 2590

    if ((nentries = qemuMonitorGetTPMTypes(mon, &entries)) < 0)
        return -1;

    if (nentries > 0) {
        for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsTPMTypesToCaps); i++) {
            const char *needle = virDomainTPMBackendTypeToString(
                virQEMUCapsTPMTypesToCaps[i].type);
2591
            if (virStringListHasString((const char **)entries, needle))
2592 2593 2594
                virQEMUCapsSet(qemuCaps, virQEMUCapsTPMTypesToCaps[i].caps);
        }
    }
2595
    virStringListFree(entries);
2596 2597 2598 2599

    return 0;
}

2600

2601
static int
2602 2603
virQEMUCapsProbeQMPKVMState(virQEMUCapsPtr qemuCaps,
                            qemuMonitorPtr mon)
2604 2605 2606 2607 2608 2609 2610
{
    bool enabled = false;
    bool present = false;

    if (qemuMonitorGetKVMState(mon, &enabled, &present) < 0)
        return -1;

A
Andrea Bolognani 已提交
2611 2612
    if (present && enabled)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_KVM);
2613 2614 2615 2616

    return 0;
}

2617 2618 2619 2620 2621 2622 2623 2624
struct virQEMUCapsCommandLineProps {
    const char *option;
    const char *param;
    int flag;
};

static struct virQEMUCapsCommandLineProps virQEMUCapsCommandLine[] = {
    { "machine", "mem-merge", QEMU_CAPS_MEM_MERGE },
2625
    { "machine", "vmport", QEMU_CAPS_MACHINE_VMPORT_OPT },
O
Osier Yang 已提交
2626
    { "drive", "discard", QEMU_CAPS_DRIVE_DISCARD },
2627
    { "drive", "detect-zeroes", QEMU_CAPS_DRIVE_DETECT_ZEROES },
2628
    { "realtime", "mlock", QEMU_CAPS_REALTIME_MLOCK },
2629
    { "boot-opts", "strict", QEMU_CAPS_BOOT_STRICT },
2630
    { "boot-opts", "reboot-timeout", QEMU_CAPS_REBOOT_TIMEOUT },
2631
    { "boot-opts", "splash-time", QEMU_CAPS_SPLASH_TIMEOUT },
2632
    { "spice", "disable-agent-file-xfer", QEMU_CAPS_SPICE_FILE_XFER_DISABLE },
2633
    { "msg", "timestamp", QEMU_CAPS_MSG_TIMESTAMP },
2634
    { "numa", NULL, QEMU_CAPS_NUMA },
2635
    { "drive", "throttling.bps-total-max", QEMU_CAPS_DRIVE_IOTUNE_MAX},
2636 2637
    { "machine", "aes-key-wrap", QEMU_CAPS_AES_KEY_WRAP },
    { "machine", "dea-key-wrap", QEMU_CAPS_DEA_KEY_WRAP },
2638
    { "chardev", "append", QEMU_CAPS_CHARDEV_FILE_APPEND },
2639
    { "spice", "gl", QEMU_CAPS_SPICE_GL },
2640
    { "chardev", "logfile", QEMU_CAPS_CHARDEV_LOGFILE },
2641
    { "name", "debug-threads", QEMU_CAPS_NAME_DEBUG_THREADS },
2642
    { "name", "guest", QEMU_CAPS_NAME_GUEST },
2643
    { "spice", "unix", QEMU_CAPS_SPICE_UNIX },
2644
    { "drive", "throttling.bps-total-max-length", QEMU_CAPS_DRIVE_IOTUNE_MAX_LENGTH },
2645
    { "drive", "throttling.group", QEMU_CAPS_DRIVE_IOTUNE_GROUP },
2646
    { "spice", "rendernode", QEMU_CAPS_SPICE_RENDERNODE },
2647
    { "machine", "kernel_irqchip", QEMU_CAPS_MACHINE_KERNEL_IRQCHIP },
2648
    { "machine", "loadparm", QEMU_CAPS_LOADPARM },
2649
    { "vnc", "vnc", QEMU_CAPS_VNC_MULTI_SERVERS },
2650
    { "chardev", "reconnect", QEMU_CAPS_CHARDEV_RECONNECT },
2651
    { "sandbox", "enable", QEMU_CAPS_SECCOMP_SANDBOX },
2652
    { "sandbox", "elevateprivileges", QEMU_CAPS_SECCOMP_BLACKLIST },
2653
    { "chardev", "fd", QEMU_CAPS_CHARDEV_FD_PASS },
2654 2655 2656 2657 2658 2659
};

static int
virQEMUCapsProbeQMPCommandLine(virQEMUCapsPtr qemuCaps,
                               qemuMonitorPtr mon)
{
2660
    bool found = false;
2661 2662 2663 2664 2665 2666 2667
    int nvalues;
    char **values;
    size_t i, j;

    for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsCommandLine); i++) {
        if ((nvalues = qemuMonitorGetCommandLineOptionParameters(mon,
                                                                 virQEMUCapsCommandLine[i].option,
2668 2669
                                                                 &values,
                                                                 &found)) < 0)
2670
            return -1;
2671 2672 2673 2674

        if (found && !virQEMUCapsCommandLine[i].param)
            virQEMUCapsSet(qemuCaps, virQEMUCapsCommandLine[i].flag);

2675
        for (j = 0; j < nvalues; j++) {
2676
            if (STREQ_NULLABLE(virQEMUCapsCommandLine[i].param, values[j])) {
2677 2678 2679 2680
                virQEMUCapsSet(qemuCaps, virQEMUCapsCommandLine[i].flag);
                break;
            }
        }
2681
        virStringListFree(values);
2682 2683 2684 2685
    }

    return 0;
}
2686

2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700
static int
virQEMUCapsProbeQMPMigrationCapabilities(virQEMUCapsPtr qemuCaps,
                                         qemuMonitorPtr mon)
{
    char **caps = NULL;
    int ncaps;

    if ((ncaps = qemuMonitorGetMigrationCapabilities(mon, &caps)) < 0)
        return -1;

    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsMigration),
                                  virQEMUCapsMigration,
                                  ncaps, caps);
2701
    virStringListFreeCount(caps, ncaps);
2702 2703 2704 2705

    return 0;
}

A
Andrea Bolognani 已提交
2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725
/**
 * virQEMUCapsProbeQMPGICCapabilities:
 * @qemuCaps: QEMU binary capabilities
 * @mon: QEMU monitor
 *
 * Use @mon to obtain information about the GIC capabilities for the
 * corresponding QEMU binary, and store them in @qemuCaps.
 *
 * Returns: 0 on success, <0 on failure
 */
static int
virQEMUCapsProbeQMPGICCapabilities(virQEMUCapsPtr qemuCaps,
                                   qemuMonitorPtr mon)
{
    virGICCapability *caps = NULL;
    int ncaps;

    if ((ncaps = qemuMonitorGetGICCapabilities(mon, &caps)) < 0)
        return -1;

2726
    virQEMUCapsSetGICCapabilities(qemuCaps, caps, ncaps);
A
Andrea Bolognani 已提交
2727 2728 2729 2730

    return 0;
}

2731

2732
/* Returns -1 on error, 0 if SEV is not supported, 1 if SEV is supported */
2733 2734 2735 2736
static int
virQEMUCapsProbeQMPSEVCapabilities(virQEMUCapsPtr qemuCaps,
                                   qemuMonitorPtr mon)
{
2737
    int rc = -1;
2738 2739
    virSEVCapability *caps = NULL;

2740 2741
    if ((rc = qemuMonitorGetSEVCapabilities(mon, &caps)) <= 0)
        return rc;
2742

2743 2744
    virSEVCapabilitiesFree(qemuCaps->sevCapabilities);
    qemuCaps->sevCapabilities = caps;
2745
    return rc;
2746 2747 2748
}


2749
bool
2750
virQEMUCapsCPUFilterFeatures(const char *name,
2751
                             void *opaque)
2752
{
2753
    virArch *arch = opaque;
2754

2755
    if (!ARCH_IS_X86(*arch))
2756 2757
        return true;

2758 2759 2760 2761 2762 2763 2764 2765 2766
    if (STREQ(name, "cmt") ||
        STREQ(name, "mbm_total") ||
        STREQ(name, "mbm_local"))
        return false;

    return true;
}


2767 2768 2769
/**
 * Returns  0 when host CPU model provided by QEMU was filled in qemuCaps,
 *          1 when the caller should fall back to using virCapsPtr->host.cpu,
2770
 *          2 when cpu model info is not supported for this configuration,
2771 2772 2773 2774
 *         -1 on error.
 */
static int
virQEMUCapsInitCPUModelS390(virQEMUCapsPtr qemuCaps,
2775
                            virDomainVirtType type,
2776
                            qemuMonitorCPUModelInfoPtr modelInfo,
2777 2778
                            virCPUDefPtr cpu,
                            bool migratable)
2779
{
2780
    size_t i;
2781

2782
    if (!modelInfo) {
2783 2784 2785 2786 2787 2788 2789 2790
        if (type == VIR_DOMAIN_VIRT_KVM) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("missing host CPU model info from QEMU "
                             "capabilities for binary %s"),
                           qemuCaps->binary);
            return -1;
        }
        return 2;
2791
    }
J
Jiri Denemark 已提交
2792

2793 2794
    if (VIR_STRDUP(cpu->model, modelInfo->name) < 0 ||
        VIR_ALLOC_N(cpu->features, modelInfo->nprops) < 0)
2795
        return -1;
2796 2797 2798 2799 2800

    cpu->nfeatures_max = modelInfo->nprops;
    cpu->nfeatures = 0;

    for (i = 0; i < modelInfo->nprops; i++) {
2801 2802
        virCPUFeatureDefPtr feature = cpu->features + cpu->nfeatures;
        qemuMonitorCPUPropertyPtr prop = modelInfo->props + i;
2803

2804 2805
        if (prop->type != QEMU_MONITOR_CPU_PROPERTY_BOOLEAN)
            continue;
2806

2807 2808
        if (VIR_STRDUP(feature->name, prop->name) < 0)
            return -1;
2809 2810 2811 2812 2813 2814

        if (!prop->value.boolean ||
            (migratable && prop->migratable == VIR_TRISTATE_BOOL_NO))
            feature->policy = VIR_CPU_FEATURE_DISABLE;
        else
            feature->policy = VIR_CPU_FEATURE_REQUIRE;
2815 2816 2817
        cpu->nfeatures++;
    }

2818 2819
    return 0;
}
2820

2821

2822 2823 2824 2825 2826 2827 2828 2829
/**
 * Returns  0 when host CPU model provided by QEMU was filled in qemuCaps,
 *          1 when the caller should fall back to using virCapsPtr->host.cpu,
 *         -1 on error.
 */
static int
virQEMUCapsInitCPUModelX86(virQEMUCapsPtr qemuCaps,
                           virDomainVirtType type,
2830
                           qemuMonitorCPUModelInfoPtr model,
2831 2832
                           virCPUDefPtr cpu,
                           bool migratable)
2833 2834 2835 2836
{
    virCPUDataPtr data = NULL;
    unsigned long long sigFamily = 0;
    unsigned long long sigModel = 0;
2837
    unsigned long long sigStepping = 0;
2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851
    int ret = -1;
    size_t i;

    if (!model)
        return 1;

    if (!(data = virCPUDataNew(VIR_ARCH_X86_64)))
        goto cleanup;

    for (i = 0; i < model->nprops; i++) {
        qemuMonitorCPUPropertyPtr prop = model->props + i;

        switch (prop->type) {
        case QEMU_MONITOR_CPU_PROPERTY_BOOLEAN:
2852 2853 2854 2855 2856
            if (!prop->value.boolean ||
                (migratable && prop->migratable == VIR_TRISTATE_BOOL_NO))
                continue;

            if (virCPUx86DataAddFeature(data, prop->name) < 0)
2857
                goto cleanup;
2858

2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871
            break;

        case QEMU_MONITOR_CPU_PROPERTY_STRING:
            if (STREQ(prop->name, "vendor") &&
                virCPUx86DataSetVendor(data, prop->value.string) < 0)
                goto cleanup;
            break;

        case QEMU_MONITOR_CPU_PROPERTY_NUMBER:
            if (STREQ(prop->name, "family"))
                sigFamily = prop->value.number;
            else if (STREQ(prop->name, "model"))
                sigModel = prop->value.number;
2872 2873
            else if (STREQ(prop->name, "stepping"))
                sigStepping = prop->value.number;
2874 2875 2876 2877 2878 2879 2880
            break;

        case QEMU_MONITOR_CPU_PROPERTY_LAST:
            break;
        }
    }

2881
    if (virCPUx86DataSetSignature(data, sigFamily, sigModel, sigStepping) < 0)
2882 2883
        goto cleanup;

2884
    if (cpuDecode(cpu, data, virQEMUCapsGetCPUDefinitions(qemuCaps, type)) < 0)
2885 2886 2887 2888 2889 2890 2891 2892 2893 2894
        goto cleanup;

    ret = 0;

 cleanup:
    virCPUDataFree(data);
    return ret;
}


2895 2896
/**
 * Returns  0 when host CPU model provided by QEMU was filled in qemuCaps,
2897 2898
 *          1 when the caller should fall back to other methods,
 *          2 when cpu model info is not supported for this configuration,
2899 2900
 *         -1 on error.
 */
2901
int
2902
virQEMUCapsInitCPUModel(virQEMUCapsPtr qemuCaps,
2903
                        virDomainVirtType type,
2904 2905
                        virCPUDefPtr cpu,
                        bool migratable)
2906
{
2907
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);
2908 2909
    int ret = 1;

2910
    if (migratable && cpuData->info && !cpuData->info->migratability)
2911 2912
        return 1;

2913
    if (ARCH_IS_S390(qemuCaps->arch)) {
2914
        ret = virQEMUCapsInitCPUModelS390(qemuCaps, type, cpuData->info,
2915 2916 2917 2918 2919
                                          cpu, migratable);
    } else if (ARCH_IS_X86(qemuCaps->arch)) {
        ret = virQEMUCapsInitCPUModelX86(qemuCaps, type, cpuData->info,
                                         cpu, migratable);
    }
2920

2921 2922 2923
    if (ret == 0)
        cpu->fallback = VIR_CPU_FALLBACK_FORBID;

2924
    return ret;
2925 2926 2927
}


2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944
static virCPUDefPtr
virQEMUCapsNewHostCPUModel(void)
{
    virCPUDefPtr cpu;

    if (VIR_ALLOC(cpu) < 0)
        return NULL;

    cpu->type = VIR_CPU_TYPE_GUEST;
    cpu->mode = VIR_CPU_MODE_CUSTOM;
    cpu->match = VIR_CPU_MATCH_EXACT;
    cpu->fallback = VIR_CPU_FALLBACK_ALLOW;

    return cpu;
}


2945 2946
void
virQEMUCapsInitHostCPUModel(virQEMUCapsPtr qemuCaps,
2947
                            virArch hostArch,
2948
                            virDomainVirtType type)
2949 2950
{
    virCPUDefPtr cpu = NULL;
2951
    virCPUDefPtr cpuExpanded = NULL;
2952
    virCPUDefPtr migCPU = NULL;
2953
    virCPUDefPtr hostCPU = NULL;
2954 2955
    virCPUDefPtr fullCPU = NULL;
    size_t i;
2956
    int rc;
2957

2958
    if (!virQEMUCapsGuestIsNative(hostArch, qemuCaps->arch))
2959 2960
        return;

2961
    if (!(cpu = virQEMUCapsNewHostCPUModel()))
2962 2963
        goto error;

2964
    if ((rc = virQEMUCapsInitCPUModel(qemuCaps, type, cpu, false)) < 0) {
2965 2966
        goto error;
    } else if (rc == 1) {
2967
        VIR_DEBUG("No host CPU model info from QEMU; probing host CPU directly");
2968

2969
        hostCPU = virQEMUCapsProbeHostCPUForEmulator(hostArch, qemuCaps, type);
2970 2971
        if (!hostCPU ||
            virCPUDefCopyModelFilter(cpu, hostCPU, true,
2972
                                     virQEMUCapsCPUFilterFeatures,
2973
                                     &qemuCaps->arch) < 0)
2974
            goto error;
2975 2976 2977 2978 2979
    } else if (rc == 2) {
        VIR_DEBUG("QEMU does not provide CPU model for arch=%s virttype=%s",
                  virArchToString(qemuCaps->arch),
                  virDomainVirtTypeToString(type));
        goto error;
2980 2981 2982
    } else if (type == VIR_DOMAIN_VIRT_KVM &&
               virCPUGetHostIsSupported(qemuCaps->arch)) {
        if (!(fullCPU = virCPUGetHost(qemuCaps->arch, VIR_CPU_TYPE_GUEST,
2983
                                      NULL, NULL)))
2984 2985
            goto error;

2986 2987 2988 2989 2990 2991 2992
        if (!(cpuExpanded = virCPUDefCopy(cpu)) ||
            virCPUExpandFeatures(qemuCaps->arch, cpuExpanded) < 0)
            goto error;

        for (i = 0; i < cpuExpanded->nfeatures; i++) {
            if (cpuExpanded->features[i].policy == VIR_CPU_FEATURE_REQUIRE &&
                virCPUDefUpdateFeature(fullCPU, cpuExpanded->features[i].name,
2993 2994 2995
                                       VIR_CPU_FEATURE_REQUIRE) < 0)
                goto error;
        }
2996 2997
    }

2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010
    if (!(migCPU = virQEMUCapsNewHostCPUModel()))
        goto error;

    if ((rc = virQEMUCapsInitCPUModel(qemuCaps, type, migCPU, true)) < 0) {
        goto error;
    } else if (rc == 1) {
        VIR_DEBUG("CPU migratability not provided by QEMU");

        virCPUDefFree(migCPU);
        if (!(migCPU = virCPUCopyMigratable(qemuCaps->arch, cpu)))
            goto error;
    }

3011
    virQEMUCapsSetHostModel(qemuCaps, type, cpu, migCPU, fullCPU);
3012

3013
 cleanup:
3014
    virCPUDefFree(cpuExpanded);
3015
    virCPUDefFree(hostCPU);
3016 3017 3018 3019
    return;

 error:
    virCPUDefFree(cpu);
3020
    virCPUDefFree(migCPU);
3021
    virCPUDefFree(fullCPU);
3022
    virResetLastError();
3023
    goto cleanup;
3024 3025 3026
}


3027 3028 3029 3030 3031
void
virQEMUCapsSetCPUModelInfo(virQEMUCapsPtr qemuCaps,
                           virDomainVirtType type,
                           qemuMonitorCPUModelInfoPtr modelInfo)
{
3032 3033 3034
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);

    cpuData->info = modelInfo;
3035 3036 3037
}


3038 3039
static int
virQEMUCapsLoadHostCPUModelInfo(virQEMUCapsPtr qemuCaps,
3040 3041
                                xmlXPathContextPtr ctxt,
                                virDomainVirtType virtType)
3042 3043 3044
{
    char *str = NULL;
    xmlNodePtr hostCPUNode;
3045
    xmlNodePtr *nodes = NULL;
3046 3047 3048 3049 3050
    xmlNodePtr oldnode = ctxt->node;
    qemuMonitorCPUModelInfoPtr hostCPU = NULL;
    int ret = -1;
    size_t i;
    int n;
3051
    int val;
3052

3053 3054 3055 3056 3057 3058
    if (virtType == VIR_DOMAIN_VIRT_KVM)
        hostCPUNode = virXPathNode("./hostCPU[@type='kvm']", ctxt);
    else
        hostCPUNode = virXPathNode("./hostCPU[@type='tcg']", ctxt);

    if (!hostCPUNode) {
3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072
        ret = 0;
        goto cleanup;
    }

    if (VIR_ALLOC(hostCPU) < 0)
        goto cleanup;

    if (!(hostCPU->name = virXMLPropString(hostCPUNode, "model"))) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing host CPU model name in QEMU "
                         "capabilities cache"));
        goto cleanup;
    }

3073 3074 3075 3076 3077 3078 3079 3080 3081
    if (!(str = virXMLPropString(hostCPUNode, "migratability")) ||
        (val = virTristateBoolTypeFromString(str)) <= 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("invalid migratability value for host CPU model"));
        goto cleanup;
    }
    hostCPU->migratability = val == VIR_TRISTATE_BOOL_YES;
    VIR_FREE(str);

3082 3083
    ctxt->node = hostCPUNode;

3084
    if ((n = virXPathNodeSet("./property", ctxt, &nodes)) > 0) {
3085 3086 3087 3088 3089 3090
        if (VIR_ALLOC_N(hostCPU->props, n) < 0)
            goto cleanup;

        hostCPU->nprops = n;

        for (i = 0; i < n; i++) {
3091 3092 3093 3094 3095
            qemuMonitorCPUPropertyPtr prop = hostCPU->props + i;

            ctxt->node = nodes[i];

            if (!(prop->name = virXMLPropString(ctxt->node, "name"))) {
3096 3097
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing 'name' attribute for a host CPU"
3098
                                 " model property in QEMU capabilities cache"));
3099 3100 3101
                goto cleanup;
            }

3102
            if (!(str = virXMLPropString(ctxt->node, "type")) ||
3103
                (val = qemuMonitorCPUPropertyTypeFromString(str)) < 0) {
3104
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
3105 3106
                               _("missing or invalid CPU model property type "
                                 "in QEMU capabilities cache"));
3107 3108 3109
                goto cleanup;
            }
            VIR_FREE(str);
3110

3111
            prop->type = val;
3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142
            switch (prop->type) {
            case QEMU_MONITOR_CPU_PROPERTY_BOOLEAN:
                if (virXPathBoolean("./@value='true'", ctxt))
                    prop->value.boolean = true;
                break;

            case QEMU_MONITOR_CPU_PROPERTY_STRING:
                prop->value.string = virXMLPropString(ctxt->node, "value");
                if (!prop->value.string) {
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("invalid string value for '%s' host CPU "
                                     "model property in QEMU capabilities cache"),
                                   prop->name);
                    goto cleanup;
                }
                break;

            case QEMU_MONITOR_CPU_PROPERTY_NUMBER:
                if (virXPathLongLong("string(./@value)", ctxt,
                                     &prop->value.number) < 0) {
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("invalid number value for '%s' host CPU "
                                     "model property in QEMU capabilities cache"),
                                   prop->name);
                    goto cleanup;
                }
                break;

            case QEMU_MONITOR_CPU_PROPERTY_LAST:
                break;
            }
3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155

            if ((str = virXMLPropString(ctxt->node, "migratable"))) {
                if ((val = virTristateBoolTypeFromString(str)) <= 0) {
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("unknown migratable value for '%s' host "
                                     "CPU model property"),
                                   prop->name);
                    goto cleanup;
                }

                prop->migratable = val;
                VIR_FREE(str);
            }
3156 3157 3158
        }
    }

3159
    virQEMUCapsSetCPUModelInfo(qemuCaps, virtType, hostCPU);
3160 3161 3162 3163 3164 3165
    hostCPU = NULL;
    ret = 0;

 cleanup:
    ctxt->node = oldnode;
    VIR_FREE(str);
3166
    VIR_FREE(nodes);
3167 3168 3169 3170 3171
    qemuMonitorCPUModelInfoFree(hostCPU);
    return ret;
}


3172 3173
static int
virQEMUCapsLoadCPUModels(virQEMUCapsPtr qemuCaps,
3174 3175
                         xmlXPathContextPtr ctxt,
                         virDomainVirtType type)
3176
{
3177
    virDomainCapsCPUModelsPtr cpus = NULL;
3178 3179 3180 3181 3182
    xmlNodePtr *nodes = NULL;
    char *str = NULL;
    size_t i;
    int n;
    int ret = -1;
3183 3184 3185 3186
    xmlNodePtr node;
    xmlNodePtr *blockerNodes = NULL;
    char **blockers = NULL;
    int nblockers;
3187

3188 3189 3190 3191 3192 3193
    if (type == VIR_DOMAIN_VIRT_KVM)
        n = virXPathNodeSet("./cpu[@type='kvm']", ctxt, &nodes);
    else
        n = virXPathNodeSet("./cpu[@type='tcg']", ctxt, &nodes);

    if (n < 0) {
3194 3195 3196 3197 3198 3199 3200 3201 3202 3203
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to parse qemu capabilities cpus"));
        goto cleanup;
    }

    if (n == 0) {
        ret = 0;
        goto cleanup;
    }

3204
    if (!(cpus = virDomainCapsCPUModelsNew(n)))
3205 3206
        goto cleanup;

3207 3208 3209 3210 3211
    if (type == VIR_DOMAIN_VIRT_KVM)
        qemuCaps->kvmCPUModels = cpus;
    else
        qemuCaps->tcgCPUModels = cpus;

3212
    for (i = 0; i < n; i++) {
3213 3214 3215 3216 3217 3218 3219 3220 3221 3222
        int usable = VIR_DOMCAPS_CPU_USABLE_UNKNOWN;

        if ((str = virXMLPropString(nodes[i], "usable")) &&
            (usable = virDomainCapsCPUUsableTypeFromString(str)) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unknown value '%s' in attribute 'usable'"), str);
            goto cleanup;
        }
        VIR_FREE(str);

3223 3224 3225 3226 3227 3228
        if (!(str = virXMLPropString(nodes[i], "name"))) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("missing cpu name in QEMU capabilities cache"));
            goto cleanup;
        }

3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253
        node = ctxt->node;
        ctxt->node = nodes[i];
        nblockers = virXPathNodeSet("./blocker", ctxt, &blockerNodes);
        ctxt->node = node;

        if (nblockers < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("failed to parse CPU blockers in QEMU capabilities"));
            goto cleanup;
        }

        if (nblockers > 0) {
            size_t j;

            if (VIR_ALLOC_N(blockers, nblockers + 1) < 0)
                goto cleanup;

            for (j = 0; j < nblockers; j++) {
                if (!(blockers[j] = virXMLPropString(blockerNodes[j], "name"))) {
                    virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                                   _("missing blocker name in QEMU "
                                     "capabilities cache"));
                    goto cleanup;
                }
            }
3254
            VIR_FREE(blockerNodes);
3255 3256 3257
        }

        if (virDomainCapsCPUModelsAddSteal(cpus, &str, usable, &blockers) < 0)
3258 3259 3260 3261 3262 3263 3264 3265
            goto cleanup;
    }

    ret = 0;

 cleanup:
    VIR_FREE(nodes);
    VIR_FREE(str);
3266 3267
    VIR_FREE(blockerNodes);
    virStringListFree(blockers);
3268 3269 3270 3271
    return ret;
}


3272 3273 3274 3275 3276
struct _virQEMUCapsCachePriv {
    char *libDir;
    uid_t runUid;
    gid_t runGid;
    virArch hostArch;
3277
    unsigned int microcodeVersion;
3278
    char *kernelVersion;
3279 3280 3281 3282

    /* cache whether /dev/kvm is usable as runUid:runGuid */
    virTristateBool kvmUsable;
    time_t kvmCtime;
3283 3284 3285 3286 3287
};
typedef struct _virQEMUCapsCachePriv virQEMUCapsCachePriv;
typedef virQEMUCapsCachePriv *virQEMUCapsCachePrivPtr;


3288
static void
3289
virQEMUCapsCachePrivFree(void *privData)
3290
{
3291 3292
    virQEMUCapsCachePrivPtr priv = privData;

3293
    VIR_FREE(priv->libDir);
3294
    VIR_FREE(priv->kernelVersion);
3295 3296 3297 3298
    VIR_FREE(priv);
}


3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350
static int
virQEMUCapsParseSEVInfo(virQEMUCapsPtr qemuCaps, xmlXPathContextPtr ctxt)
{
    VIR_AUTOPTR(virSEVCapability) sev = NULL;

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SEV_GUEST))
        return 0;

    if (virXPathBoolean("boolean(./sev)", ctxt) == 0) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing SEV platform data in QEMU "
                         "capabilities cache"));
        return -1;
    }

    if (VIR_ALLOC(sev) < 0)
        return -1;

    if (virXPathUInt("string(./sev/cbitpos)", ctxt, &sev->cbitpos) < 0) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing or malformed SEV cbitpos information "
                         "in QEMU capabilities cache"));
        return -1;
    }

    if (virXPathUInt("string(./sev/reducedPhysBits)", ctxt,
                     &sev->reduced_phys_bits) < 0) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing or malformed SEV reducedPhysBits information "
                         "in QEMU capabilities cache"));
        return -1;
    }

    if (!(sev->pdh = virXPathString("string(./sev/pdh)", ctxt)))  {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing SEV pdh information "
                         "in QEMU capabilities cache"));
        return -1;
    }

    if (!(sev->cert_chain = virXPathString("string(./sev/certChain)", ctxt))) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing SEV certChain information "
                         "in QEMU capabilities cache"));
        return -1;
    }

    VIR_STEAL_PTR(qemuCaps->sevCapabilities, sev);
    return 0;
}


3351 3352 3353 3354 3355 3356
/*
 * Parsing a doc that looks like
 *
 * <qemuCaps>
 *   <qemuctime>234235253</qemuctime>
 *   <selfctime>234235253</selfctime>
3357
 *   <selfvers>1002016</selfvers>
3358 3359 3360 3361 3362 3363
 *   <usedQMP/>
 *   <flag name='foo'/>
 *   <flag name='bar'/>
 *   ...
 *   <cpu name="pentium3"/>
 *   ...
3364
 *   <machine name="pc-1.0" alias="pc" hotplugCpus='yes' maxCpus="4" default="yes"/>
3365 3366 3367
 *   ...
 * </qemuCaps>
 */
3368
int
3369
virQEMUCapsLoadCache(virArch hostArch,
3370
                     virQEMUCapsPtr qemuCaps,
3371
                     const char *filename)
3372 3373 3374 3375 3376 3377 3378
{
    xmlDocPtr doc = NULL;
    int ret = -1;
    size_t i;
    int n;
    xmlNodePtr *nodes = NULL;
    xmlXPathContextPtr ctxt = NULL;
J
Ján Tomko 已提交
3379
    char *str = NULL;
3380
    long long int l;
3381
    unsigned long lu;
3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405

    if (!(doc = virXMLParseFile(filename)))
        goto cleanup;

    if (!(ctxt = xmlXPathNewContext(doc))) {
        virReportOOMError();
        goto cleanup;
    }

    ctxt->node = xmlDocGetRootElement(doc);

    if (STRNEQ((const char *)ctxt->node->name, "qemuCaps")) {
        virReportError(VIR_ERR_XML_ERROR,
                       _("unexpected root element <%s>, "
                         "expecting <qemuCaps>"),
                       ctxt->node->name);
        goto cleanup;
    }

    if (virXPathLongLong("string(./qemuctime)", ctxt, &l) < 0) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing qemuctime in QEMU capabilities XML"));
        goto cleanup;
    }
3406
    qemuCaps->ctime = (time_t)l;
3407 3408 3409 3410 3411 3412

    if (virXPathLongLong("string(./selfctime)", ctxt, &l) < 0) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing selfctime in QEMU capabilities XML"));
        goto cleanup;
    }
3413
    qemuCaps->libvirtCtime = (time_t)l;
3414

3415
    qemuCaps->libvirtVersion = 0;
3416
    if (virXPathULong("string(./selfvers)", ctxt, &lu) == 0)
3417
        qemuCaps->libvirtVersion = lu;
3418

3419 3420 3421 3422 3423 3424 3425 3426 3427
    qemuCaps->usedQMP = virXPathBoolean("count(./usedQMP) > 0",
                                        ctxt) > 0;

    if ((n = virXPathNodeSet("./flag", ctxt, &nodes)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to parse qemu capabilities flags"));
        goto cleanup;
    }
    VIR_DEBUG("Got flags %d", n);
3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439
    for (i = 0; i < n; i++) {
        int flag;
        if (!(str = virXMLPropString(nodes[i], "name"))) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("missing flag name in QEMU capabilities cache"));
            goto cleanup;
        }
        flag = virQEMUCapsTypeFromString(str);
        if (flag < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unknown qemu capabilities flag %s"), str);
            goto cleanup;
3440
        }
3441 3442
        VIR_FREE(str);
        virQEMUCapsSet(qemuCaps, flag);
3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457
    }
    VIR_FREE(nodes);

    if (virXPathUInt("string(./version)", ctxt, &qemuCaps->version) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing version in QEMU capabilities cache"));
        goto cleanup;
    }

    if (virXPathUInt("string(./kvmVersion)", ctxt, &qemuCaps->kvmVersion) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing version in QEMU capabilities cache"));
        goto cleanup;
    }

3458 3459 3460 3461 3462 3463 3464
    if (virXPathUInt("string(./microcodeVersion)", ctxt,
                     &qemuCaps->microcodeVersion) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing microcode version in QEMU capabilities cache"));
        goto cleanup;
    }

3465 3466 3467 3468 3469 3470
    if (virXPathBoolean("boolean(./package)", ctxt) > 0) {
        qemuCaps->package = virXPathString("string(./package)", ctxt);
        if (!qemuCaps->package &&
            VIR_STRDUP(qemuCaps->package, "") < 0)
            goto cleanup;
    }
3471

3472 3473 3474 3475 3476 3477
    if (virXPathBoolean("boolean(./kernelVersion)", ctxt) > 0) {
        qemuCaps->kernelVersion = virXPathString("string(./kernelVersion)", ctxt);
        if (!qemuCaps->kernelVersion)
            goto cleanup;
    }

3478 3479 3480 3481 3482 3483 3484 3485 3486 3487
    if (!(str = virXPathString("string(./arch)", ctxt))) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing arch in QEMU capabilities cache"));
        goto cleanup;
    }
    if (!(qemuCaps->arch = virArchFromString(str))) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unknown arch %s in QEMU capabilities cache"), str);
        goto cleanup;
    }
J
Ján Tomko 已提交
3488
    VIR_FREE(str);
3489

3490 3491
    if (virQEMUCapsLoadHostCPUModelInfo(qemuCaps, ctxt, VIR_DOMAIN_VIRT_KVM) < 0 ||
        virQEMUCapsLoadHostCPUModelInfo(qemuCaps, ctxt, VIR_DOMAIN_VIRT_QEMU) < 0)
3492 3493
        goto cleanup;

3494 3495
    if (virQEMUCapsLoadCPUModels(qemuCaps, ctxt, VIR_DOMAIN_VIRT_KVM) < 0 ||
        virQEMUCapsLoadCPUModels(qemuCaps, ctxt, VIR_DOMAIN_VIRT_QEMU) < 0)
3496 3497 3498 3499 3500 3501 3502 3503 3504
        goto cleanup;

    if ((n = virXPathNodeSet("./machine", ctxt, &nodes)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to parse qemu capabilities machines"));
        goto cleanup;
    }
    if (n > 0) {
        qemuCaps->nmachineTypes = n;
3505
        if (VIR_ALLOC_N(qemuCaps->machineTypes, qemuCaps->nmachineTypes) < 0)
3506 3507 3508
            goto cleanup;

        for (i = 0; i < n; i++) {
3509
            if (!(qemuCaps->machineTypes[i].name = virXMLPropString(nodes[i], "name"))) {
3510 3511 3512 3513
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing machine name in QEMU capabilities cache"));
                goto cleanup;
            }
3514
            qemuCaps->machineTypes[i].alias = virXMLPropString(nodes[i], "alias");
3515 3516 3517

            str = virXMLPropString(nodes[i], "maxCpus");
            if (str &&
3518
                virStrToLong_ui(str, NULL, 10, &(qemuCaps->machineTypes[i].maxCpus)) < 0) {
3519 3520 3521 3522
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("malformed machine cpu count in QEMU capabilities cache"));
                goto cleanup;
            }
J
Ján Tomko 已提交
3523
            VIR_FREE(str);
3524 3525 3526 3527 3528

            str = virXMLPropString(nodes[i], "hotplugCpus");
            if (STREQ_NULLABLE(str, "yes"))
                qemuCaps->machineTypes[i].hotplugCpus = true;
            VIR_FREE(str);
3529 3530 3531 3532 3533

            str = virXMLPropString(nodes[i], "default");
            if (STREQ_NULLABLE(str, "yes"))
                qemuCaps->machineTypes[i].qemuDefault = true;
            VIR_FREE(str);
3534 3535 3536 3537
        }
    }
    VIR_FREE(nodes);

A
Andrea Bolognani 已提交
3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603
    if ((n = virXPathNodeSet("./gic", ctxt, &nodes)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to parse qemu capabilities gic"));
        goto cleanup;
    }
    if (n > 0) {
        unsigned int uintValue;
        bool boolValue;

        qemuCaps->ngicCapabilities = n;
        if (VIR_ALLOC_N(qemuCaps->gicCapabilities, n) < 0)
            goto cleanup;

        for (i = 0; i < n; i++) {
            virGICCapabilityPtr cap = &qemuCaps->gicCapabilities[i];

            if (!(str = virXMLPropString(nodes[i], "version"))) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing GIC version "
                                 "in QEMU capabilities cache"));
                goto cleanup;
            }
            if (virStrToLong_ui(str, NULL, 10, &uintValue) < 0) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("malformed GIC version "
                                 "in QEMU capabilities cache"));
                goto cleanup;
            }
            cap->version = uintValue;
            VIR_FREE(str);

            if (!(str = virXMLPropString(nodes[i], "kernel"))) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing in-kernel GIC information "
                                 "in QEMU capabilities cache"));
                goto cleanup;
            }
            if (!(boolValue = STREQ(str, "yes")) && STRNEQ(str, "no")) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("malformed in-kernel GIC information "
                                 "in QEMU capabilities cache"));
                goto cleanup;
            }
            if (boolValue)
                cap->implementation |= VIR_GIC_IMPLEMENTATION_KERNEL;
            VIR_FREE(str);

            if (!(str = virXMLPropString(nodes[i], "emulated"))) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing emulated GIC information "
                                 "in QEMU capabilities cache"));
                goto cleanup;
            }
            if (!(boolValue = STREQ(str, "yes")) && STRNEQ(str, "no")) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("malformed emulated GIC information "
                                 "in QEMU capabilities cache"));
                goto cleanup;
            }
            if (boolValue)
                cap->implementation |= VIR_GIC_IMPLEMENTATION_EMULATED;
            VIR_FREE(str);
        }
    }
    VIR_FREE(nodes);

3604 3605 3606
    if (virQEMUCapsParseSEVInfo(qemuCaps, ctxt) < 0)
        goto cleanup;

3607 3608
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_QEMU);
3609

3610 3611 3612
    if (virXPathBoolean("boolean(./kvmSupportsNesting)", ctxt) > 0)
        qemuCaps->kvmSupportsNesting = true;

3613
    ret = 0;
3614
 cleanup:
J
Ján Tomko 已提交
3615
    VIR_FREE(str);
3616 3617 3618 3619 3620 3621 3622
    VIR_FREE(nodes);
    xmlXPathFreeContext(ctxt);
    xmlFreeDoc(doc);
    return ret;
}


3623 3624
static void
virQEMUCapsFormatHostCPUModelInfo(virQEMUCapsPtr qemuCaps,
3625 3626
                                  virBufferPtr buf,
                                  virDomainVirtType type)
3627
{
3628 3629 3630
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);
    qemuMonitorCPUModelInfoPtr model = cpuData->info;
    const char *typeStr = type == VIR_DOMAIN_VIRT_KVM ? "kvm" : "tcg";
3631 3632
    size_t i;

3633 3634 3635
    if (!model)
        return;

3636 3637 3638 3639
    virBufferAsprintf(buf,
                      "<hostCPU type='%s' model='%s' migratability='%s'>\n",
                      typeStr, model->name,
                      model->migratability ? "yes" : "no");
3640 3641 3642
    virBufferAdjustIndent(buf, 2);

    for (i = 0; i < model->nprops; i++) {
3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665
        qemuMonitorCPUPropertyPtr prop = model->props + i;

        virBufferAsprintf(buf, "<property name='%s' type='%s' ",
                          prop->name,
                          qemuMonitorCPUPropertyTypeToString(prop->type));

        switch (prop->type) {
        case QEMU_MONITOR_CPU_PROPERTY_BOOLEAN:
            virBufferAsprintf(buf, "value='%s'",
                              prop->value.boolean ? "true" : "false");
            break;

        case QEMU_MONITOR_CPU_PROPERTY_STRING:
            virBufferEscapeString(buf, "value='%s'", prop->value.string);
            break;

        case QEMU_MONITOR_CPU_PROPERTY_NUMBER:
            virBufferAsprintf(buf, "value='%lld'", prop->value.number);
            break;

        case QEMU_MONITOR_CPU_PROPERTY_LAST:
            break;
        }
3666 3667 3668 3669 3670

        if (prop->migratable > 0)
            virBufferAsprintf(buf, " migratable='%s'",
                              virTristateBoolTypeToString(prop->migratable));

3671
        virBufferAddLit(buf, "/>\n");
3672 3673 3674 3675 3676 3677 3678
    }

    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</hostCPU>\n");
}


3679 3680
static void
virQEMUCapsFormatCPUModels(virQEMUCapsPtr qemuCaps,
3681 3682
                           virBufferPtr buf,
                           virDomainVirtType type)
3683
{
3684 3685
    virDomainCapsCPUModelsPtr cpus;
    const char *typeStr;
3686 3687
    size_t i;

3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699
    if (type == VIR_DOMAIN_VIRT_KVM) {
        typeStr = "kvm";
        cpus = qemuCaps->kvmCPUModels;
    } else {
        typeStr = "tcg";
        cpus = qemuCaps->tcgCPUModels;
    }

    if (!cpus)
        return;

    for (i = 0; i < cpus->nmodels; i++) {
3700 3701
        virDomainCapsCPUModelPtr cpu = cpus->models + i;

3702
        virBufferAsprintf(buf, "<cpu type='%s' ", typeStr);
3703 3704 3705 3706 3707
        virBufferEscapeString(buf, "name='%s'", cpu->name);
        if (cpu->usable) {
            virBufferAsprintf(buf, " usable='%s'",
                              virDomainCapsCPUUsableTypeToString(cpu->usable));
        }
3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722

        if (cpu->blockers) {
            size_t j;

            virBufferAddLit(buf, ">\n");
            virBufferAdjustIndent(buf, 2);

            for (j = 0; cpu->blockers[j]; j++)
                virBufferAsprintf(buf, "<blocker name='%s'/>\n", cpu->blockers[j]);

            virBufferAdjustIndent(buf, -2);
            virBufferAddLit(buf, "</cpu>\n");
        } else {
            virBufferAddLit(buf, "/>\n");
        }
3723 3724 3725 3726
    }
}


3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744
static void
virQEMUCapsFormatSEVInfo(virQEMUCapsPtr qemuCaps, virBufferPtr buf)
{
    virSEVCapabilityPtr sev = virQEMUCapsGetSEVCapabilities(qemuCaps);

    virBufferAddLit(buf, "<sev>\n");
    virBufferAdjustIndent(buf, 2);
    virBufferAsprintf(buf, "<cbitpos>%u</cbitpos>\n", sev->cbitpos);
    virBufferAsprintf(buf, "<reducedPhysBits>%u</reducedPhysBits>\n",
                      sev->reduced_phys_bits);
    virBufferEscapeString(buf, "<pdh>%s</pdh>\n", sev->pdh);
    virBufferEscapeString(buf, "<certChain>%s</certChain>\n",
                          sev->cert_chain);
    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</sev>\n");
}


3745
char *
3746
virQEMUCapsFormatCache(virQEMUCapsPtr qemuCaps)
3747 3748
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;
3749
    char *ret = NULL;
3750 3751 3752
    size_t i;

    virBufferAddLit(&buf, "<qemuCaps>\n");
3753
    virBufferAdjustIndent(&buf, 2);
3754

3755
    virBufferAsprintf(&buf, "<qemuctime>%llu</qemuctime>\n",
3756
                      (long long)qemuCaps->ctime);
3757
    virBufferAsprintf(&buf, "<selfctime>%llu</selfctime>\n",
3758
                      (long long)qemuCaps->libvirtCtime);
3759
    virBufferAsprintf(&buf, "<selfvers>%lu</selfvers>\n",
3760
                      (unsigned long)qemuCaps->libvirtVersion);
3761 3762

    if (qemuCaps->usedQMP)
3763
        virBufferAddLit(&buf, "<usedQMP/>\n");
3764 3765 3766

    for (i = 0; i < QEMU_CAPS_LAST; i++) {
        if (virQEMUCapsGet(qemuCaps, i)) {
3767
            virBufferAsprintf(&buf, "<flag name='%s'/>\n",
3768 3769 3770 3771
                              virQEMUCapsTypeToString(i));
        }
    }

3772
    virBufferAsprintf(&buf, "<version>%d</version>\n",
3773 3774
                      qemuCaps->version);

3775
    virBufferAsprintf(&buf, "<kvmVersion>%d</kvmVersion>\n",
3776 3777
                      qemuCaps->kvmVersion);

3778 3779 3780
    virBufferAsprintf(&buf, "<microcodeVersion>%u</microcodeVersion>\n",
                      qemuCaps->microcodeVersion);

3781 3782 3783 3784
    if (qemuCaps->package)
        virBufferAsprintf(&buf, "<package>%s</package>\n",
                          qemuCaps->package);

3785 3786 3787 3788
    if (qemuCaps->kernelVersion)
        virBufferAsprintf(&buf, "<kernelVersion>%s</kernelVersion>\n",
                          qemuCaps->kernelVersion);

3789
    virBufferAsprintf(&buf, "<arch>%s</arch>\n",
3790 3791
                      virArchToString(qemuCaps->arch));

3792 3793
    virQEMUCapsFormatHostCPUModelInfo(qemuCaps, &buf, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsFormatHostCPUModelInfo(qemuCaps, &buf, VIR_DOMAIN_VIRT_QEMU);
3794

3795 3796
    virQEMUCapsFormatCPUModels(qemuCaps, &buf, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsFormatCPUModels(qemuCaps, &buf, VIR_DOMAIN_VIRT_QEMU);
3797 3798

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
3799
        virBufferEscapeString(&buf, "<machine name='%s'",
3800 3801
                              qemuCaps->machineTypes[i].name);
        if (qemuCaps->machineTypes[i].alias)
3802
            virBufferEscapeString(&buf, " alias='%s'",
3803
                              qemuCaps->machineTypes[i].alias);
3804 3805
        if (qemuCaps->machineTypes[i].hotplugCpus)
            virBufferAddLit(&buf, " hotplugCpus='yes'");
3806
        virBufferAsprintf(&buf, " maxCpus='%u'",
3807
                          qemuCaps->machineTypes[i].maxCpus);
3808 3809 3810
        if (qemuCaps->machineTypes[i].qemuDefault)
            virBufferAddLit(&buf, " default='yes'");
        virBufferAddLit(&buf, "/>\n");
3811 3812
    }

A
Andrea Bolognani 已提交
3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828
    for (i = 0; i < qemuCaps->ngicCapabilities; i++) {
        virGICCapabilityPtr cap;
        bool kernel;
        bool emulated;

        cap = &qemuCaps->gicCapabilities[i];
        kernel = (cap->implementation & VIR_GIC_IMPLEMENTATION_KERNEL);
        emulated = (cap->implementation & VIR_GIC_IMPLEMENTATION_EMULATED);

        virBufferAsprintf(&buf,
                          "<gic version='%d' kernel='%s' emulated='%s'/>\n",
                          cap->version,
                          kernel ? "yes" : "no",
                          emulated ? "yes" : "no");
    }

3829 3830 3831
    if (qemuCaps->sevCapabilities)
        virQEMUCapsFormatSEVInfo(qemuCaps, &buf);

3832 3833 3834
    if (qemuCaps->kvmSupportsNesting)
        virBufferAddLit(&buf, "<kvmSupportsNesting/>\n");

3835
    virBufferAdjustIndent(&buf, -2);
3836 3837
    virBufferAddLit(&buf, "</qemuCaps>\n");

3838 3839 3840 3841 3842 3843 3844 3845
    if (virBufferCheckError(&buf) == 0)
        ret = virBufferContentAndReset(&buf);

    return ret;
}


static int
3846 3847 3848
virQEMUCapsSaveFile(void *data,
                    const char *filename,
                    void *privData ATTRIBUTE_UNUSED)
3849
{
3850
    virQEMUCapsPtr qemuCaps = data;
3851 3852
    char *xml = NULL;
    int ret = -1;
3853

3854
    xml = virQEMUCapsFormatCache(qemuCaps);
3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865

    if (virFileWriteStr(filename, xml, 0600) < 0) {
        virReportSystemError(errno,
                             _("Failed to save '%s' for '%s'"),
                             filename, qemuCaps->binary);
        goto cleanup;
    }

    VIR_DEBUG("Saved caps '%s' for '%s' with (%lld, %lld)",
              filename, qemuCaps->binary,
              (long long)qemuCaps->ctime,
3866
              (long long)qemuCaps->libvirtCtime);
3867 3868 3869 3870 3871 3872 3873 3874

    ret = 0;
 cleanup:
    VIR_FREE(xml);
    return ret;
}


3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909
/* Check the kernel module parameters 'nested' file to determine if enabled
 *
 *   Intel: 'kvm_intel' uses 'Y'
 *   AMD:   'kvm_amd' uses '1'
 *   PPC64: 'kvm_hv' uses 'Y'
 *   S390:  'kvm' uses '1'
 */
static bool
virQEMUCapsKVMSupportsNesting(void)
{
    static char const * const kmod[] = {"kvm_intel", "kvm_amd",
                                        "kvm_hv", "kvm"};
    VIR_AUTOFREE(char *) value = NULL;
    int rc;
    size_t i;

    for (i = 0; i < ARRAY_CARDINALITY(kmod); i++) {
        VIR_FREE(value);
        rc = virFileReadValueString(&value, "/sys/module/%s/parameters/nested",
                                    kmod[i]);
        if (rc == -2)
            continue;
        if (rc < 0) {
            virResetLastError();
            return false;
        }

        if (value[0] == 'Y' || value[0] == 'y' || value[0] == '1')
            return true;
    }

    return false;
}


3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957
/* Determine whether '/dev/kvm' is usable as QEMU user:QEMU group. */
static bool
virQEMUCapsKVMUsable(virQEMUCapsCachePrivPtr priv)
{
    struct stat sb;
    static const char *kvm_device = "/dev/kvm";
    virTristateBool value;
    virTristateBool cached_value = priv->kvmUsable;
    time_t kvm_ctime;
    time_t cached_kvm_ctime = priv->kvmCtime;

    if (stat(kvm_device, &sb) < 0) {
        if (errno != ENOENT) {
            virReportSystemError(errno,
                                 _("Failed to stat %s"), kvm_device);
        }
        return false;
    }
    kvm_ctime = sb.st_ctime;

    if (kvm_ctime != cached_kvm_ctime) {
        VIR_DEBUG("%s has changed (%lld vs %lld)", kvm_device,
                  (long long)kvm_ctime, (long long)cached_kvm_ctime);
        cached_value = VIR_TRISTATE_BOOL_ABSENT;
    }

    if (cached_value != VIR_TRISTATE_BOOL_ABSENT)
        return cached_value == VIR_TRISTATE_BOOL_YES;

    if (virFileAccessibleAs(kvm_device, R_OK | W_OK,
                            priv->runUid, priv->runGid) == 0) {
        value = VIR_TRISTATE_BOOL_YES;
    } else {
        value = VIR_TRISTATE_BOOL_NO;
    }

    /* There is a race window between 'stat' and
     * 'virFileAccessibleAs'. However, since we're only interested in
     * detecting changes *after* the virFileAccessibleAs check, we can
     * neglect this here.
     */
    priv->kvmCtime = kvm_ctime;
    priv->kvmUsable = value;

    return value == VIR_TRISTATE_BOOL_YES;
}


3958
static bool
3959 3960
virQEMUCapsIsValid(void *data,
                   void *privData)
3961
{
3962 3963
    virQEMUCapsPtr qemuCaps = data;
    virQEMUCapsCachePrivPtr priv = privData;
3964
    bool kvmUsable;
3965
    struct stat sb;
3966
    bool kvmSupportsNesting;
3967 3968 3969 3970

    if (!qemuCaps->binary)
        return true;

3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982
    if (qemuCaps->libvirtCtime != virGetSelfLastChanged() ||
        qemuCaps->libvirtVersion != LIBVIR_VERSION_NUMBER) {
        VIR_DEBUG("Outdated capabilities for '%s': libvirt changed "
                  "(%lld vs %lld, %lu vs %lu)",
                  qemuCaps->binary,
                  (long long)qemuCaps->libvirtCtime,
                  (long long)virGetSelfLastChanged(),
                  (unsigned long)qemuCaps->libvirtVersion,
                  (unsigned long)LIBVIR_VERSION_NUMBER);
        return false;
    }

3983 3984 3985 3986 3987 3988
    if (stat(qemuCaps->binary, &sb) < 0) {
        char ebuf[1024];
        VIR_DEBUG("Failed to stat QEMU binary '%s': %s",
                  qemuCaps->binary,
                  virStrerror(errno, ebuf, sizeof(ebuf)));
        return false;
3989 3990
    }

3991
    if (sb.st_ctime != qemuCaps->ctime) {
3992 3993 3994
        VIR_DEBUG("Outdated capabilities for '%s': QEMU binary changed "
                  "(%lld vs %lld)",
                  qemuCaps->binary,
3995
                  (long long)sb.st_ctime, (long long)qemuCaps->ctime);
3996 3997 3998
        return false;
    }

3999 4000 4001 4002 4003 4004 4005 4006
    if (!virQEMUCapsGuestIsNative(priv->hostArch, qemuCaps->arch)) {
        VIR_DEBUG("Guest arch (%s) is not native to host arch (%s), "
                  "skipping KVM-related checks",
                  virArchToString(qemuCaps->arch),
                  virArchToString(priv->hostArch));
        return true;
    }

4007
    kvmUsable = virQEMUCapsKVMUsable(priv);
4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM) &&
        kvmUsable) {
        VIR_DEBUG("KVM was not enabled when probing '%s', "
                  "but it should be usable now",
                  qemuCaps->binary);
        return false;
    }

    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM) &&
        !kvmUsable) {
        VIR_DEBUG("KVM was enabled when probing '%s', "
                  "but it is not available now",
                  qemuCaps->binary);
        return false;
    }

4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
        if (priv->microcodeVersion != qemuCaps->microcodeVersion) {
            VIR_DEBUG("Outdated capabilities for '%s': microcode version "
                      "changed (%u vs %u)",
                      qemuCaps->binary,
                      priv->microcodeVersion,
                      qemuCaps->microcodeVersion);
            return false;
        }

        if (STRNEQ_NULLABLE(priv->kernelVersion, qemuCaps->kernelVersion)) {
            VIR_DEBUG("Outdated capabilities for '%s': kernel version changed "
                      "('%s' vs '%s')",
                      qemuCaps->binary,
                      priv->kernelVersion,
                      qemuCaps->kernelVersion);
            return false;
        }
4043 4044 4045 4046 4047 4048 4049 4050

        kvmSupportsNesting = virQEMUCapsKVMSupportsNesting();
        if (kvmSupportsNesting != qemuCaps->kvmSupportsNesting) {
            VIR_DEBUG("Outdated capabilities for '%s': kvm kernel nested "
                      "value changed from %d",
                     qemuCaps->binary, qemuCaps->kvmSupportsNesting);
            return false;
        }
4051 4052
    }

4053 4054 4055 4056
    return true;
}


4057
static void virQEMUCapsMonitorNotify(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
4058 4059
                                     virDomainObjPtr vm ATTRIBUTE_UNUSED,
                                     void *opaque ATTRIBUTE_UNUSED)
4060 4061 4062 4063
{
}

static qemuMonitorCallbacks callbacks = {
4064 4065
    .eofNotify = virQEMUCapsMonitorNotify,
    .errorNotify = virQEMUCapsMonitorNotify,
4066 4067 4068
};


4069 4070 4071 4072 4073 4074 4075 4076
/**
 * virQEMUCapsInitQMPArch:
 * @qemuCaps: QEMU capabilities
 * @mon: QEMU monitor
 *
 * Initialize the architecture for @qemuCaps by asking @mon.
 *
 * Returns: 0 on success, <0 on failure
4077 4078
 */
static int
4079
virQEMUCapsInitQMPArch(virQEMUCapsPtr qemuCaps,
4080 4081 4082 4083 4084 4085
                            qemuMonitorPtr mon)
{
    char *archstr = NULL;
    int ret = -1;

    if (!(archstr = qemuMonitorGetTargetArch(mon)))
4086
        goto cleanup;
4087 4088 4089 4090 4091 4092 4093

    if ((qemuCaps->arch = virQEMUCapsArchFromString(archstr)) == VIR_ARCH_NONE) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Unknown QEMU arch %s"), archstr);
        goto cleanup;
    }

4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107
    ret = 0;

 cleanup:
    VIR_FREE(archstr);
    return ret;
}


/**
 * virQEMUCapsInitQMPBasicArch:
 * @qemuCaps: QEMU capabilities
 *
 * Initialize @qemuCaps with basic architecture-dependent capabilities.
 */
4108
void
4109 4110
virQEMUCapsInitQMPBasicArch(virQEMUCapsPtr qemuCaps)
{
4111 4112 4113
    /* ACPI only works on x86 and aarch64 */
    if (ARCH_IS_X86(qemuCaps->arch) ||
        qemuCaps->arch == VIR_ARCH_AARCH64) {
4114
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_NO_ACPI);
4115 4116
    }

J
Ján Tomko 已提交
4117 4118
    /* HPET is x86 specific */
    if (ARCH_IS_X86(qemuCaps->arch))
J
Ján Tomko 已提交
4119
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_NO_HPET);
4120
}
4121

4122 4123 4124 4125 4126 4127

static int
virQEMUCapsProbeQMPSchemaCapabilities(virQEMUCapsPtr qemuCaps,
                                      qemuMonitorPtr mon)
{
    struct virQEMUCapsStringFlags *entry;
4128 4129
    virJSONValuePtr schemareply;
    virHashTablePtr schema = NULL;
4130 4131
    size_t i;

4132
    if (!(schemareply = qemuMonitorQueryQMPSchema(mon)))
4133 4134
        return -1;

4135 4136 4137 4138
    if (!(schema = virQEMUQAPISchemaConvert(schemareply)))
        return -1;
    schemareply = NULL;

4139 4140 4141
    for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsQMPSchemaQueries); i++) {
        entry = virQEMUCapsQMPSchemaQueries + i;

4142
        if (virQEMUQAPISchemaPathExists(entry->value, schema))
4143 4144 4145 4146 4147 4148 4149
            virQEMUCapsSet(qemuCaps, entry->flag);
    }

    virHashFree(schema);
    return 0;
}

J
Ján Tomko 已提交
4150
#define QEMU_MIN_MAJOR 1
J
Ján Tomko 已提交
4151
#define QEMU_MIN_MINOR 5
J
Ján Tomko 已提交
4152
#define QEMU_MIN_MICRO 0
4153

4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165
int
virQEMUCapsInitQMPMonitor(virQEMUCapsPtr qemuCaps,
                          qemuMonitorPtr mon)
{
    int ret = -1;
    int major, minor, micro;
    char *package = NULL;

    /* @mon is supposed to be locked by callee */

    if (qemuMonitorSetCapabilities(mon) < 0) {
        VIR_DEBUG("Failed to set monitor capabilities %s",
4166
                  virGetLastErrorMessage());
4167 4168 4169 4170 4171 4172 4173 4174
        ret = 0;
        goto cleanup;
    }

    if (qemuMonitorGetVersion(mon,
                              &major, &minor, &micro,
                              &package) < 0) {
        VIR_DEBUG("Failed to query monitor version %s",
4175
                  virGetLastErrorMessage());
4176 4177 4178 4179 4180 4181 4182
        ret = 0;
        goto cleanup;
    }

    VIR_DEBUG("Got version %d.%d.%d (%s)",
              major, minor, micro, NULLSTR(package));

J
Ján Tomko 已提交
4183 4184 4185 4186 4187 4188
    if (major < QEMU_MIN_MAJOR ||
        (major == QEMU_MIN_MAJOR && minor < QEMU_MIN_MINOR)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("QEMU version >= %d.%d.%d is required, but %d.%d.%d found"),
                       QEMU_MIN_MAJOR, QEMU_MIN_MINOR, QEMU_MIN_MICRO,
                       major, minor, micro);
4189 4190 4191 4192
        goto cleanup;
    }

    qemuCaps->version = major * 1000000 + minor * 1000 + micro;
4193
    qemuCaps->package = package;
4194 4195
    qemuCaps->usedQMP = true;

4196
    if (virQEMUCapsInitQMPArch(qemuCaps, mon) < 0)
4197 4198
        goto cleanup;

4199 4200
    virQEMUCapsInitQMPBasicArch(qemuCaps);

4201 4202 4203
    if (qemuCaps->version >= 1006000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_DEVICE_VIDEO_PRIMARY);

4204 4205 4206 4207
    /* vmport option is supported v2.2.0 onwards */
    if (qemuCaps->version >= 2002000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_VMPORT_OPT);

4208 4209 4210 4211 4212 4213
    /* -cpu ...,aarch64=off supported in v2.3.0 and onwards. But it
       isn't detectable via qmp at this point */
    if (qemuCaps->arch == VIR_ARCH_AARCH64 &&
        qemuCaps->version >= 2003000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_CPU_AARCH64_OFF);

4214 4215 4216
    /* vhost-user supports multi-queue from v2.4.0 onwards,
     * but there is no way to query for that capability */
    if (qemuCaps->version >= 2004000)
J
Ján Tomko 已提交
4217
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_VHOSTUSER_MULTIQUEUE);
4218

M
Michal Privoznik 已提交
4219 4220 4221 4222
    /* smm option is supported from v2.4.0 */
    if (qemuCaps->version >= 2004000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_SMM_OPT);

4223 4224 4225 4226
    /* sdl -gl option is supported from v2.4.0 (qemu commit id 0b71a5d5) */
    if (qemuCaps->version >= 2004000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_SDL_GL);

4227 4228 4229 4230
    /* Since 2.4.50 ARM virt machine supports gic-version option */
    if (qemuCaps->version >= 2004050)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACH_VIRT_GIC_VERSION);

4231 4232 4233 4234
    /* no way to query if -machine kernel_irqchip supports split */
    if (qemuCaps->version >= 2006000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_KERNEL_IRQCHIP_SPLIT);

4235 4236 4237 4238 4239 4240 4241
    /* HPT resizing is supported since QEMU 2.10 on ppc64; unfortunately
     * there's no sane way to probe for it */
    if (qemuCaps->version >= 2010000 &&
        ARCH_IS_PPC64(qemuCaps->arch)) {
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_PSERIES_RESIZE_HPT);
    }

4242 4243 4244 4245 4246
    /* '-display egl-headless' cmdline option is supported since QEMU 2.10, but
     * there's no way to probe it */
    if (qemuCaps->version >= 2010000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_EGL_HEADLESS);

4247 4248 4249 4250
    /* no way to query for -numa dist */
    if (qemuCaps->version >= 2010000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_NUMA_DIST);

4251 4252 4253 4254 4255 4256
    /* no way to query max-cpu-compat */
    if (qemuCaps->version >= 2010000 &&
        ARCH_IS_PPC64(qemuCaps->arch)) {
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_PSERIES_MAX_CPU_COMPAT);
    }

4257 4258
    if (virQEMUCapsProbeQMPCommands(qemuCaps, mon) < 0)
        goto cleanup;
J
Jiri Denemark 已提交
4259 4260 4261 4262 4263

    /* Some capabilities may differ depending on KVM state */
    if (virQEMUCapsProbeQMPKVMState(qemuCaps, mon) < 0)
        goto cleanup;

4264 4265
    if (virQEMUCapsProbeQMPEvents(qemuCaps, mon) < 0)
        goto cleanup;
4266
    if (virQEMUCapsProbeQMPDevices(qemuCaps, mon) < 0)
4267 4268 4269
        goto cleanup;
    if (virQEMUCapsProbeQMPMachineTypes(qemuCaps, mon) < 0)
        goto cleanup;
4270
    if (virQEMUCapsProbeQMPCPUDefinitions(qemuCaps, mon, false) < 0)
4271 4272 4273 4274 4275
        goto cleanup;
    if (virQEMUCapsProbeQMPTPM(qemuCaps, mon) < 0)
        goto cleanup;
    if (virQEMUCapsProbeQMPCommandLine(qemuCaps, mon) < 0)
        goto cleanup;
4276 4277
    if (virQEMUCapsProbeQMPMigrationCapabilities(qemuCaps, mon) < 0)
        goto cleanup;
4278 4279 4280
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_QMP_SCHEMA) &&
        virQEMUCapsProbeQMPSchemaCapabilities(qemuCaps, mon) < 0)
        goto cleanup;
4281
    if (virQEMUCapsProbeQMPHostCPU(qemuCaps, mon, false) < 0)
4282
        goto cleanup;
4283

4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294
    /* 'intel-iommu' shows up as a device since 2.2.0, but can
     * not be used with -device until 2.7.0. Before that it
     * requires -machine iommu=on. So we must clear the device
     * capability we detected on older QEMUs
     */
    if (qemuCaps->version < 2007000 &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_INTEL_IOMMU)) {
        virQEMUCapsClear(qemuCaps, QEMU_CAPS_DEVICE_INTEL_IOMMU);
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_IOMMU);
    }

A
Andrea Bolognani 已提交
4295 4296
    /* GIC capabilities, eg. available GIC versions */
    if ((qemuCaps->arch == VIR_ARCH_AARCH64 ||
S
Stefan Schallenberg 已提交
4297
         qemuCaps->arch == VIR_ARCH_ARMV6L ||
A
Andrea Bolognani 已提交
4298 4299 4300 4301
         qemuCaps->arch == VIR_ARCH_ARMV7L) &&
        virQEMUCapsProbeQMPGICCapabilities(qemuCaps, mon) < 0)
        goto cleanup;

4302 4303 4304 4305 4306 4307 4308
    /* Prealloc on NVDIMMs is broken on older QEMUs leading to
     * user data corruption. If we are dealing with such version
     * of QEMU pretend we don't know how to NVDIMM. */
    if (qemuCaps->version < 2009000 &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_NVDIMM))
        virQEMUCapsClear(qemuCaps, QEMU_CAPS_DEVICE_NVDIMM);

4309 4310 4311 4312
    if (ARCH_IS_X86(qemuCaps->arch) &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_CPU_MODEL_EXPANSION))
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_CPU_CACHE);

4313 4314 4315 4316 4317
    if (ARCH_IS_S390(qemuCaps->arch)) {
        /* Legacy assurance for QEMU_CAPS_CCW */
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CCW) &&
            virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_CCW))
            virQEMUCapsSet(qemuCaps, QEMU_CAPS_CCW);
4318 4319
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CCW_CSSID_UNRESTRICTED))
            virQEMUCapsClear(qemuCaps, QEMU_CAPS_DEVICE_VFIO_CCW);
4320 4321
    }

4322 4323
    /* Probe for SEV capabilities */
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_SEV_GUEST)) {
4324 4325 4326 4327 4328 4329
        int rc = virQEMUCapsProbeQMPSEVCapabilities(qemuCaps, mon);

        if (rc < 0)
            goto cleanup;

        if (rc == 0)
4330 4331 4332
            virQEMUCapsClear(qemuCaps, QEMU_CAPS_SEV_GUEST);
    }

4333
    ret = 0;
4334
 cleanup:
4335 4336 4337
    return ret;
}

4338

4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351
int
virQEMUCapsInitQMPMonitorTCG(virQEMUCapsPtr qemuCaps ATTRIBUTE_UNUSED,
                             qemuMonitorPtr mon)
{
    int ret = -1;

    if (qemuMonitorSetCapabilities(mon) < 0) {
        VIR_DEBUG("Failed to set monitor capabilities %s",
                  virGetLastErrorMessage());
        ret = 0;
        goto cleanup;
    }

4352 4353 4354
    if (virQEMUCapsProbeQMPCPUDefinitions(qemuCaps, mon, true) < 0)
        goto cleanup;

4355 4356 4357
    if (virQEMUCapsProbeQMPHostCPU(qemuCaps, mon, true) < 0)
        goto cleanup;

4358 4359 4360 4361 4362 4363
    ret = 0;
 cleanup:
    return ret;
}


4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375
typedef struct _virQEMUCapsInitQMPCommand virQEMUCapsInitQMPCommand;
typedef virQEMUCapsInitQMPCommand *virQEMUCapsInitQMPCommandPtr;
struct _virQEMUCapsInitQMPCommand {
    char *binary;
    uid_t runUid;
    gid_t runGid;
    char **qmperr;
    char *monarg;
    char *monpath;
    char *pidfile;
    virCommandPtr cmd;
    qemuMonitorPtr mon;
4376
    virDomainChrSourceDef config;
4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394
    pid_t pid;
    virDomainObjPtr vm;
};


static void
virQEMUCapsInitQMPCommandAbort(virQEMUCapsInitQMPCommandPtr cmd)
{
    if (cmd->mon)
        virObjectUnlock(cmd->mon);
    qemuMonitorClose(cmd->mon);
    cmd->mon = NULL;

    virCommandAbort(cmd->cmd);
    virCommandFree(cmd->cmd);
    cmd->cmd = NULL;

    if (cmd->monpath)
4395
        unlink(cmd->monpath);
4396 4397 4398 4399 4400 4401

    virDomainObjEndAPI(&cmd->vm);

    if (cmd->pid != 0) {
        char ebuf[1024];

4402
        VIR_DEBUG("Killing QMP caps process %lld", (long long)cmd->pid);
4403 4404
        if (virProcessKill(cmd->pid, SIGKILL) < 0 && errno != ESRCH)
            VIR_ERROR(_("Failed to kill process %lld: %s"),
4405
                      (long long)cmd->pid,
4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448
                      virStrerror(errno, ebuf, sizeof(ebuf)));

        VIR_FREE(*cmd->qmperr);
    }
    if (cmd->pidfile)
        unlink(cmd->pidfile);
    cmd->pid = 0;
}


static void
virQEMUCapsInitQMPCommandFree(virQEMUCapsInitQMPCommandPtr cmd)
{
    if (!cmd)
        return;

    virQEMUCapsInitQMPCommandAbort(cmd);
    VIR_FREE(cmd->binary);
    VIR_FREE(cmd->monpath);
    VIR_FREE(cmd->monarg);
    VIR_FREE(cmd->pidfile);
    VIR_FREE(cmd);
}


static virQEMUCapsInitQMPCommandPtr
virQEMUCapsInitQMPCommandNew(char *binary,
                             const char *libDir,
                             uid_t runUid,
                             gid_t runGid,
                             char **qmperr)
{
    virQEMUCapsInitQMPCommandPtr cmd = NULL;

    if (VIR_ALLOC(cmd) < 0)
        goto error;

    if (VIR_STRDUP(cmd->binary, binary) < 0)
        goto error;

    cmd->runUid = runUid;
    cmd->runGid = runGid;
    cmd->qmperr = qmperr;
4449

4450 4451 4452
    /* the ".sock" sufix is important to avoid a possible clash with a qemu
     * domain called "capabilities"
     */
4453 4454 4455 4456 4457
    if (virAsprintf(&cmd->monpath, "%s/%s", libDir,
                    "capabilities.monitor.sock") < 0)
        goto error;
    if (virAsprintf(&cmd->monarg, "unix:%s,server,nowait", cmd->monpath) < 0)
        goto error;
4458

4459 4460
    /* ".pidfile" suffix is used rather than ".pid" to avoid a possible clash
     * with a qemu domain called "capabilities"
4461 4462 4463
     * Normally we'd use runDir for pid files, but because we're using
     * -daemonize we need QEMU to be allowed to create them, rather
     * than libvirtd. So we're using libDir which QEMU can write to
4464
     */
4465 4466
    if (virAsprintf(&cmd->pidfile, "%s/%s", libDir, "capabilities.pidfile") < 0)
        goto error;
4467

4468
    virPidFileForceCleanupPath(cmd->pidfile);
4469

4470 4471 4472
    cmd->config.type = VIR_DOMAIN_CHR_TYPE_UNIX;
    cmd->config.data.nix.path = cmd->monpath;
    cmd->config.data.nix.listen = false;
4473

4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486
    return cmd;

 error:
    virQEMUCapsInitQMPCommandFree(cmd);
    return NULL;
}


/* Returns -1 on fatal error,
 *          0 on success,
 *          1 when probing QEMU failed
 */
static int
4487 4488
virQEMUCapsInitQMPCommandRun(virQEMUCapsInitQMPCommandPtr cmd,
                             bool forceTCG)
4489 4490
{
    virDomainXMLOptionPtr xmlopt = NULL;
4491
    const char *machine;
4492 4493 4494
    int status = 0;
    int ret = -1;

4495 4496 4497 4498 4499 4500 4501
    if (forceTCG)
        machine = "none,accel=tcg";
    else
        machine = "none,accel=kvm:tcg";

    VIR_DEBUG("Try to probe capabilities of '%s' via QMP, machine %s",
              cmd->binary, machine);
4502

4503 4504 4505 4506 4507 4508 4509
    /*
     * We explicitly need to use -daemonize here, rather than
     * virCommandDaemonize, because we need to synchronize
     * with QEMU creating its monitor socket API. Using
     * daemonize guarantees control won't return to libvirt
     * until the socket is present.
     */
4510 4511 4512 4513 4514
    cmd->cmd = virCommandNewArgList(cmd->binary,
                                    "-S",
                                    "-no-user-config",
                                    "-nodefaults",
                                    "-nographic",
4515
                                    "-machine", machine,
4516 4517 4518 4519 4520 4521
                                    "-qmp", cmd->monarg,
                                    "-pidfile", cmd->pidfile,
                                    "-daemonize",
                                    NULL);
    virCommandAddEnvPassCommon(cmd->cmd);
    virCommandClearCaps(cmd->cmd);
4522 4523 4524 4525

#if WITH_CAPNG
    /* QEMU might run into permission issues, e.g. /dev/sev (0600), override
     * them just for the purpose of probing */
4526 4527
    if (geteuid() == 0)
        virCommandAllowCap(cmd->cmd, CAP_DAC_OVERRIDE);
4528 4529
#endif

4530 4531 4532 4533
    virCommandSetGID(cmd->cmd, cmd->runGid);
    virCommandSetUID(cmd->cmd, cmd->runUid);

    virCommandSetErrorBuffer(cmd->cmd, cmd->qmperr);
4534

4535
    /* Log, but otherwise ignore, non-zero status.  */
4536
    if (virCommandRun(cmd->cmd, &status) < 0)
4537 4538 4539
        goto cleanup;

    if (status != 0) {
4540
        VIR_DEBUG("QEMU %s exited with status %d: %s",
4541 4542
                  cmd->binary, status, *cmd->qmperr);
        goto ignore;
4543 4544
    }

4545 4546 4547
    if (virPidFileReadPath(cmd->pidfile, &cmd->pid) < 0) {
        VIR_DEBUG("Failed to read pidfile %s", cmd->pidfile);
        goto ignore;
4548 4549
    }

4550
    if (!(xmlopt = virDomainXMLOptionNew(NULL, NULL, NULL, NULL, NULL)) ||
4551
        !(cmd->vm = virDomainObjNew(xmlopt)))
4552 4553
        goto cleanup;

4554
    cmd->vm->pid = cmd->pid;
4555

4556
    if (!(cmd->mon = qemuMonitorOpen(cmd->vm, &cmd->config, true, true,
4557
                                     0, &callbacks, NULL)))
4558
        goto ignore;
4559

4560
    virObjectLock(cmd->mon);
4561 4562 4563

    ret = 0;

4564
 cleanup:
4565 4566
    if (!cmd->mon)
        virQEMUCapsInitQMPCommandAbort(cmd);
4567
    virObjectUnref(xmlopt);
4568

4569
    return ret;
4570

4571 4572 4573 4574
 ignore:
    ret = 1;
    goto cleanup;
}
4575

4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591

static int
virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps,
                   const char *libDir,
                   uid_t runUid,
                   gid_t runGid,
                   char **qmperr)
{
    virQEMUCapsInitQMPCommandPtr cmd = NULL;
    int ret = -1;
    int rc;

    if (!(cmd = virQEMUCapsInitQMPCommandNew(qemuCaps->binary, libDir,
                                             runUid, runGid, qmperr)))
        goto cleanup;

4592
    if ((rc = virQEMUCapsInitQMPCommandRun(cmd, false)) != 0) {
4593 4594 4595
        if (rc == 1)
            ret = 0;
        goto cleanup;
4596
    }
4597 4598 4599 4600

    if (virQEMUCapsInitQMPMonitor(qemuCaps, cmd->mon) < 0)
        goto cleanup;

4601 4602 4603 4604 4605 4606 4607 4608 4609 4610 4611 4612
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
        virQEMUCapsInitQMPCommandAbort(cmd);
        if ((rc = virQEMUCapsInitQMPCommandRun(cmd, true)) != 0) {
            if (rc == 1)
                ret = 0;
            goto cleanup;
        }

        if (virQEMUCapsInitQMPMonitorTCG(qemuCaps, cmd->mon) < 0)
            goto cleanup;
    }

4613 4614 4615 4616
    ret = 0;

 cleanup:
    virQEMUCapsInitQMPCommandFree(cmd);
4617 4618 4619 4620
    return ret;
}


4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631
#define MESSAGE_ID_CAPS_PROBE_FAILURE "8ae2f3fb-2dbe-498e-8fbd-012d40afa361"

static void
virQEMUCapsLogProbeFailure(const char *binary)
{
    virLogMetadata meta[] = {
        { .key = "MESSAGE_ID", .s = MESSAGE_ID_CAPS_PROBE_FAILURE, .iv = 0 },
        { .key = "LIBVIRT_QEMU_BINARY", .s = binary, .iv = 0 },
        { .key = NULL },
    };

4632
    virLogMessage(&virLogSelf,
4633 4634 4635 4636
                  VIR_LOG_WARN,
                  __FILE__, __LINE__, __func__,
                  meta,
                  _("Failed to probe capabilities for %s: %s"),
4637
                  binary, virGetLastErrorMessage());
4638 4639 4640
}


4641
virQEMUCapsPtr
4642
virQEMUCapsNewForBinaryInternal(virArch hostArch,
4643
                                const char *binary,
4644 4645 4646
                                const char *libDir,
                                uid_t runUid,
                                gid_t runGid,
4647
                                unsigned int microcodeVersion,
J
Ján Tomko 已提交
4648
                                const char *kernelVersion)
4649
{
4650
    virQEMUCapsPtr qemuCaps;
4651
    struct stat sb;
4652
    char *qmperr = NULL;
4653

4654 4655 4656
    if (!(qemuCaps = virQEMUCapsNew()))
        goto error;

4657 4658
    if (VIR_STRDUP(qemuCaps->binary, binary) < 0)
        goto error;
4659 4660 4661 4662 4663 4664 4665 4666

    /* We would also want to check faccessat if we cared about ACLs,
     * but we don't.  */
    if (stat(binary, &sb) < 0) {
        virReportSystemError(errno, _("Cannot check QEMU binary %s"),
                             binary);
        goto error;
    }
4667
    qemuCaps->ctime = sb.st_ctime;
4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678

    /* Make sure the binary we are about to try exec'ing exists.
     * Technically we could catch the exec() failure, but that's
     * in a sub-process so it's hard to feed back a useful error.
     */
    if (!virFileIsExecutable(binary)) {
        virReportSystemError(errno, _("QEMU binary %s is not executable"),
                             binary);
        goto error;
    }

4679 4680
    if (virQEMUCapsInitQMP(qemuCaps, libDir, runUid, runGid, &qmperr) < 0) {
        virQEMUCapsLogProbeFailure(binary);
4681
        goto error;
4682
    }
4683

J
Ján Tomko 已提交
4684
    if (!qemuCaps->usedQMP) {
4685 4686 4687 4688 4689 4690
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Failed to probe QEMU binary with QMP: %s"),
                       qmperr ? qmperr : _("unknown error"));
        virQEMUCapsLogProbeFailure(binary);
        goto error;
    }
4691

4692 4693
    qemuCaps->libvirtCtime = virGetSelfLastChanged();
    qemuCaps->libvirtVersion = LIBVIR_VERSION_NUMBER;
4694

4695 4696
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_QEMU);
4697

4698
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
4699 4700
        qemuCaps->microcodeVersion = microcodeVersion;

4701 4702
        if (VIR_STRDUP(qemuCaps->kernelVersion, kernelVersion) < 0)
            goto error;
4703 4704

        qemuCaps->kvmSupportsNesting = virQEMUCapsKVMSupportsNesting();
4705 4706
    }

4707
 cleanup:
4708
    VIR_FREE(qmperr);
4709
    return qemuCaps;
4710

4711
 error:
4712 4713
    virObjectUnref(qemuCaps);
    qemuCaps = NULL;
4714
    goto cleanup;
4715 4716
}

4717 4718 4719
static void *
virQEMUCapsNewData(const char *binary,
                   void *privData)
4720
{
4721 4722 4723 4724 4725 4726 4727
    virQEMUCapsCachePrivPtr priv = privData;

    return virQEMUCapsNewForBinaryInternal(priv->hostArch,
                                           binary,
                                           priv->libDir,
                                           priv->runUid,
                                           priv->runGid,
4728
                                           priv->microcodeVersion,
J
Ján Tomko 已提交
4729
                                           priv->kernelVersion);
4730
}
4731 4732


4733 4734 4735 4736 4737 4738 4739
static void *
virQEMUCapsLoadFile(const char *filename,
                    const char *binary,
                    void *privData)
{
    virQEMUCapsPtr qemuCaps = virQEMUCapsNew();
    virQEMUCapsCachePrivPtr priv = privData;
4740

4741 4742
    if (!qemuCaps)
        return NULL;
4743

4744 4745 4746 4747 4748 4749 4750
    if (VIR_STRDUP(qemuCaps->binary, binary) < 0)
        goto error;

    if (virQEMUCapsLoadCache(priv->hostArch, qemuCaps, filename) < 0)
        goto error;

 cleanup:
4751 4752 4753 4754
    return qemuCaps;

 error:
    virObjectUnref(qemuCaps);
4755 4756
    qemuCaps = NULL;
    goto cleanup;
4757 4758
}

4759

4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792
struct virQEMUCapsMachineTypeFilter {
    const char *machineType;
    virQEMUCapsFlags *flags;
    size_t nflags;
};

static const struct virQEMUCapsMachineTypeFilter virQEMUCapsMachineFilter[] = {
    /* { "blah", virQEMUCapsMachineBLAHFilter,
         ARRAY_CARDINALITY(virQEMUCapsMachineBLAHFilter) }, */
    { "", NULL, 0 },
};


void
virQEMUCapsFilterByMachineType(virQEMUCapsPtr qemuCaps,
                               const char *machineType)
{
    size_t i;

    if (!machineType)
        return;

    for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsMachineFilter); i++) {
        const struct virQEMUCapsMachineTypeFilter *filter = &virQEMUCapsMachineFilter[i];
        size_t j;

        if (STRNEQ(filter->machineType, machineType))
            continue;

        for (j = 0; j < filter->nflags; j++)
            virQEMUCapsClear(qemuCaps, filter->flags[j]);
    }

4793 4794
    if (!virQEMUCapsGetMachineHotplugCpus(qemuCaps, machineType))
        virQEMUCapsClear(qemuCaps, QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS);
4795 4796 4797
}


4798 4799 4800 4801 4802 4803 4804 4805 4806 4807
virFileCacheHandlers qemuCapsCacheHandlers = {
    .isValid = virQEMUCapsIsValid,
    .newData = virQEMUCapsNewData,
    .loadFile = virQEMUCapsLoadFile,
    .saveFile = virQEMUCapsSaveFile,
    .privFree = virQEMUCapsCachePrivFree,
};


virFileCachePtr
4808
virQEMUCapsCacheNew(const char *libDir,
4809
                    const char *cacheDir,
4810
                    uid_t runUid,
4811 4812
                    gid_t runGid,
                    unsigned int microcodeVersion)
4813
{
4814 4815 4816
    char *capsCacheDir = NULL;
    virFileCachePtr cache = NULL;
    virQEMUCapsCachePrivPtr priv = NULL;
4817
    struct utsname uts;
4818

4819
    if (virAsprintf(&capsCacheDir, "%s/capabilities", cacheDir) < 0)
4820
        goto error;
4821 4822

    if (!(cache = virFileCacheNew(capsCacheDir, "xml", &qemuCapsCacheHandlers)))
4823
        goto error;
4824

4825
    if (VIR_ALLOC(priv) < 0)
4826
        goto error;
4827
    virFileCacheSetPriv(cache, priv);
4828

4829
    if (VIR_STRDUP(priv->libDir, libDir) < 0)
4830 4831
        goto error;

4832
    priv->hostArch = virArchFromHost();
4833

4834 4835
    priv->runUid = runUid;
    priv->runGid = runGid;
4836
    priv->microcodeVersion = microcodeVersion;
4837
    priv->kvmUsable = VIR_TRISTATE_BOOL_ABSENT;
4838

4839 4840 4841 4842
    if (uname(&uts) == 0 &&
        virAsprintf(&priv->kernelVersion, "%s %s", uts.release, uts.version) < 0)
        goto error;

4843 4844
 cleanup:
    VIR_FREE(capsCacheDir);
4845 4846
    return cache;

4847
 error:
4848 4849 4850
    virObjectUnref(cache);
    cache = NULL;
    goto cleanup;
4851 4852 4853
}


4854
virQEMUCapsPtr
4855
virQEMUCapsCacheLookup(virFileCachePtr cache,
4856
                       const char *binary)
4857
{
4858
    virQEMUCapsPtr ret = NULL;
4859

4860
    ret = virFileCacheLookup(cache, binary);
4861 4862

    VIR_DEBUG("Returning caps %p for %s", ret, binary);
4863 4864 4865 4866
    return ret;
}


4867
virQEMUCapsPtr
4868
virQEMUCapsCacheLookupCopy(virFileCachePtr cache,
4869
                           const char *binary,
4870
                           const char *machineType)
4871
{
4872
    virQEMUCapsPtr qemuCaps = virQEMUCapsCacheLookup(cache, binary);
4873
    virQEMUCapsPtr ret;
4874

4875
    if (!qemuCaps)
4876 4877
        return NULL;

4878 4879
    ret = virQEMUCapsNewCopy(qemuCaps);
    virObjectUnref(qemuCaps);
4880 4881 4882 4883

    if (!ret)
        return NULL;

4884
    virQEMUCapsFilterByMachineType(ret, machineType);
4885 4886 4887 4888
    return ret;
}


4889 4890 4891 4892 4893
static int
virQEMUCapsCompareArch(const void *payload,
                       const void *name ATTRIBUTE_UNUSED,
                       const void *opaque)
{
4894
    struct virQEMUCapsSearchData *data = (struct virQEMUCapsSearchData *)opaque;
4895 4896
    const virQEMUCaps *qemuCaps = payload;

4897 4898 4899 4900 4901 4902 4903 4904 4905
    if (qemuCaps->arch != data->arch)
        return false;

    if (data->binaryFilter &&
        !strstr(qemuCaps->binary, data->binaryFilter)) {
        return false;
    }

    return true;
4906 4907 4908 4909
}


virQEMUCapsPtr
4910
virQEMUCapsCacheLookupByArch(virFileCachePtr cache,
4911 4912 4913
                             virArch arch)
{
    virQEMUCapsPtr ret = NULL;
4914 4915 4916 4917
    const char *binaryFilters[] = {
        "qemu-system-",
        NULL,
    };
4918 4919 4920 4921
    virArch archs[] = {
        arch,
        virQEMUCapsFindTarget(virArchFromHost(), arch),
    };
4922
    size_t i;
4923 4924
    size_t j;

4925 4926 4927 4928 4929 4930
    for (i = 0; i < ARRAY_CARDINALITY(binaryFilters); i++) {
        for (j = 0; j < ARRAY_CARDINALITY(archs); j++) {
            struct virQEMUCapsSearchData data = {
                .arch = archs[j],
                .binaryFilter = binaryFilters[i],
            };
4931

4932 4933 4934 4935
            ret = virFileCacheLookupByFunc(cache, virQEMUCapsCompareArch, &data);
            if (ret)
                goto done;
        }
4936 4937
    }

4938 4939 4940 4941 4942
    virReportError(VIR_ERR_INVALID_ARG,
                   _("unable to find any emulator to serve '%s' "
                     "architecture"), virArchToString(arch));

 done:
4943 4944
    VIR_DEBUG("Returning caps %p for arch %s", ret, virArchToString(arch));

4945 4946 4947 4948
    return ret;
}


4949 4950 4951 4952 4953 4954 4955 4956 4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034
/**
 * virQEMUCapsCacheLookupDefault:
 * @cache: QEMU capabilities cache
 * @binary: optional path to QEMU binary
 * @archStr: optional guest architecture
 * @virttypeStr: optional virt type
 * @machine: optional machine type
 * @retArch: if non-NULL, guest architecture will be returned here
 * @retVirttype: if non-NULL, domain virt type will be returned here
 * @retMachine: if non-NULL, canonical machine type will be returned here
 *
 * Looks up the QEMU binary specified by @binary and @archStr, checks it can
 * provide the required @virttypeStr and @machine and returns its capabilities.
 * Sensible defaults are used for any argument which is NULL (the function can
 * even be called with all NULL arguments).
 *
 * Returns QEMU capabilities matching the requirements, NULL on error.
 */
virQEMUCapsPtr
virQEMUCapsCacheLookupDefault(virFileCachePtr cache,
                              const char *binary,
                              const char *archStr,
                              const char *virttypeStr,
                              const char *machine,
                              virArch *retArch,
                              virDomainVirtType *retVirttype,
                              const char **retMachine)
{
    int virttype = VIR_DOMAIN_VIRT_NONE;
    int arch = virArchFromHost();
    virDomainVirtType capsType;
    virQEMUCapsPtr qemuCaps = NULL;
    virQEMUCapsPtr ret = NULL;

    if (virttypeStr &&
        (virttype = virDomainVirtTypeFromString(virttypeStr)) < 0) {
        virReportError(VIR_ERR_INVALID_ARG,
                       _("unknown virttype: %s"), virttypeStr);
        goto cleanup;
    }

    if (archStr &&
        (arch = virArchFromString(archStr)) == VIR_ARCH_NONE) {
        virReportError(VIR_ERR_INVALID_ARG,
                       _("unknown architecture: %s"), archStr);
        goto cleanup;
    }

    if (binary) {
        virArch arch_from_caps;

        if (!(qemuCaps = virQEMUCapsCacheLookup(cache, binary)))
            goto cleanup;

        arch_from_caps = virQEMUCapsGetArch(qemuCaps);

        if (arch_from_caps != arch &&
            !((ARCH_IS_X86(arch) && ARCH_IS_X86(arch_from_caps)) ||
              (ARCH_IS_PPC(arch) && ARCH_IS_PPC(arch_from_caps)) ||
              (ARCH_IS_ARM(arch) && ARCH_IS_ARM(arch_from_caps)) ||
              (ARCH_IS_S390(arch) && ARCH_IS_S390(arch_from_caps)))) {
            virReportError(VIR_ERR_INVALID_ARG,
                           _("architecture from emulator '%s' doesn't "
                             "match given architecture '%s'"),
                           virArchToString(arch_from_caps),
                           virArchToString(arch));
            goto cleanup;
        }
    } else {
        if (!(qemuCaps = virQEMUCapsCacheLookupByArch(cache, arch)))
            goto cleanup;

        binary = virQEMUCapsGetBinary(qemuCaps);
    }

    if (machine) {
        /* Turn @machine into canonical name */
        machine = virQEMUCapsGetCanonicalMachine(qemuCaps, machine);

        if (!virQEMUCapsIsMachineSupported(qemuCaps, machine)) {
            virReportError(VIR_ERR_INVALID_ARG,
                           _("the machine '%s' is not supported by emulator '%s'"),
                           machine, binary);
            goto cleanup;
        }
    } else {
5035
        machine = virQEMUCapsGetPreferredMachine(qemuCaps);
5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066
    }

    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM))
        capsType = VIR_DOMAIN_VIRT_KVM;
    else
        capsType = VIR_DOMAIN_VIRT_QEMU;

    if (virttype == VIR_DOMAIN_VIRT_NONE)
        virttype = capsType;

    if (virttype == VIR_DOMAIN_VIRT_KVM && capsType == VIR_DOMAIN_VIRT_QEMU) {
        virReportError(VIR_ERR_INVALID_ARG,
                       _("KVM is not supported by '%s' on this host"),
                       binary);
        goto cleanup;
    }

    if (retArch)
        *retArch = arch;
    if (retVirttype)
        *retVirttype = virttype;
    if (retMachine)
        *retMachine = machine;

    VIR_STEAL_PTR(ret, qemuCaps);

 cleanup:
    virObjectUnref(qemuCaps);
    return ret;
}

5067 5068 5069 5070 5071 5072 5073
bool
virQEMUCapsSupportsVmport(virQEMUCapsPtr qemuCaps,
                          const virDomainDef *def)
{
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_VMPORT_OPT))
        return false;

5074 5075
    return qemuDomainIsI440FX(def) ||
        qemuDomainIsQ35(def) ||
5076 5077 5078 5079
        STREQ(def->os.machine, "isapc");
}


5080 5081 5082 5083 5084 5085 5086
bool
virQEMUCapsIsMachineSupported(virQEMUCapsPtr qemuCaps,
                              const char *canonical_machine)
{
    size_t i;

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
5087
        if (STREQ(canonical_machine, qemuCaps->machineTypes[i].name))
5088 5089 5090 5091
            return true;
    }
    return false;
}
5092 5093


5094 5095 5096 5097
/*
 * The preferred machine to use if none is listed explicitly
 * Note that this may differ from QEMU's own default machine
 */
5098
const char *
5099
virQEMUCapsGetPreferredMachine(virQEMUCapsPtr qemuCaps)
5100 5101 5102
{
    if (!qemuCaps->nmachineTypes)
        return NULL;
5103
    return qemuCaps->machineTypes[0].name;
5104
}
5105 5106


5107
static int
5108
virQEMUCapsFillDomainLoaderCaps(virDomainCapsLoaderPtr capsLoader,
5109 5110
                                virFirmwarePtr *firmwares,
                                size_t nfirmwares)
5111
{
5112 5113
    size_t i;

5114
    capsLoader->supported = true;
5115

5116
    if (VIR_ALLOC_N(capsLoader->values.values, nfirmwares) < 0)
5117 5118
        return -1;

5119 5120
    for (i = 0; i < nfirmwares; i++) {
        const char *filename = firmwares[i]->name;
5121 5122 5123 5124 5125 5126

        if (!virFileExists(filename)) {
            VIR_DEBUG("loader filename=%s does not exist", filename);
            continue;
        }

5127
        if (VIR_STRDUP(capsLoader->values.values[capsLoader->values.nvalues],
5128 5129
                       filename) < 0)
            return -1;
5130
        capsLoader->values.nvalues++;
5131 5132
    }

5133
    VIR_DOMAIN_CAPS_ENUM_SET(capsLoader->type,
5134 5135
                             VIR_DOMAIN_LOADER_TYPE_ROM);

5136 5137
    VIR_DOMAIN_CAPS_ENUM_SET(capsLoader->type,
                             VIR_DOMAIN_LOADER_TYPE_PFLASH);
5138 5139


5140 5141 5142
    VIR_DOMAIN_CAPS_ENUM_SET(capsLoader->readonly,
                             VIR_TRISTATE_BOOL_YES,
                             VIR_TRISTATE_BOOL_NO);
5143
    return 0;
5144 5145 5146
}


5147
static int
5148
virQEMUCapsFillDomainOSCaps(virDomainCapsOSPtr os,
5149 5150
                            virFirmwarePtr *firmwares,
                            size_t nfirmwares)
5151
{
5152
    virDomainCapsLoaderPtr capsLoader = &os->loader;
5153

5154
    os->supported = true;
5155
    if (virQEMUCapsFillDomainLoaderCaps(capsLoader, firmwares, nfirmwares) < 0)
5156 5157
        return -1;
    return 0;
5158 5159 5160
}


5161 5162 5163 5164 5165
static int
virQEMUCapsFillDomainCPUCaps(virCapsPtr caps,
                             virQEMUCapsPtr qemuCaps,
                             virDomainCapsPtr domCaps)
{
5166 5167
    if (virQEMUCapsIsCPUModeSupported(qemuCaps, caps, domCaps->virttype,
                                      VIR_CPU_MODE_HOST_PASSTHROUGH))
5168 5169
        domCaps->cpu.hostPassthrough = true;

5170
    if (virQEMUCapsIsCPUModeSupported(qemuCaps, caps, domCaps->virttype,
5171
                                      VIR_CPU_MODE_HOST_MODEL)) {
5172 5173
        virCPUDefPtr cpu = virQEMUCapsGetHostModel(qemuCaps, domCaps->virttype,
                                                   VIR_QEMU_CAPS_HOST_CPU_REPORTED);
5174 5175
        domCaps->cpu.hostModel = virCPUDefCopy(cpu);
    }
5176 5177 5178 5179 5180

    if (virQEMUCapsIsCPUModeSupported(qemuCaps, caps, domCaps->virttype,
                                      VIR_CPU_MODE_CUSTOM)) {
        virDomainCapsCPUModelsPtr filtered = NULL;
        char **models = NULL;
5181
        const char *blacklist[] = { "host", NULL };
5182

J
Jiri Denemark 已提交
5183
        if (virCPUGetModels(domCaps->arch, &models) >= 0) {
5184 5185 5186 5187 5188 5189 5190 5191
            virDomainCapsCPUModelsPtr cpus;

            if (domCaps->virttype == VIR_DOMAIN_VIRT_KVM)
                cpus = qemuCaps->kvmCPUModels;
            else
                cpus = qemuCaps->tcgCPUModels;

            filtered = virDomainCapsCPUModelsFilter(cpus,
5192
                                                    (const char **)models,
5193
                                                    blacklist);
5194
            virStringListFree(models);
5195 5196
        }
        domCaps->cpu.custom = filtered;
5197
    }
5198 5199 5200 5201 5202

    return 0;
}


5203 5204 5205 5206 5207 5208 5209 5210 5211 5212
static int
virQEMUCapsFillDomainIOThreadCaps(virQEMUCapsPtr qemuCaps,
                                  virDomainCapsPtr domCaps)
{
    domCaps->iothreads = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_IOTHREAD);

    return 0;
}


5213
static int
5214
virQEMUCapsFillDomainDeviceDiskCaps(virQEMUCapsPtr qemuCaps,
5215
                                    const char *machine,
5216 5217
                                    virDomainCapsDeviceDiskPtr disk)
{
5218
    disk->supported = true;
5219 5220 5221
    /* QEMU supports all of these */
    VIR_DOMAIN_CAPS_ENUM_SET(disk->diskDevice,
                             VIR_DOMAIN_DISK_DEVICE_DISK,
5222 5223
                             VIR_DOMAIN_DISK_DEVICE_CDROM,
                             VIR_DOMAIN_DISK_DEVICE_LUN);
5224 5225

    /* PowerPC pseries based VMs do not support floppy device */
5226
    if (!qemuDomainMachineIsPSeries(machine, qemuCaps->arch))
5227
        VIR_DOMAIN_CAPS_ENUM_SET(disk->diskDevice, VIR_DOMAIN_DISK_DEVICE_FLOPPY);
5228

5229 5230 5231
    if (qemuDomainMachineHasBuiltinIDE(machine))
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_IDE);

5232 5233 5234 5235 5236
    VIR_DOMAIN_CAPS_ENUM_SET(disk->bus,
                             VIR_DOMAIN_DISK_BUS_SCSI,
                             VIR_DOMAIN_DISK_BUS_VIRTIO,
                             /* VIR_DOMAIN_DISK_BUS_SD */);

5237
    /* PowerPC pseries based VMs do not support floppy device */
5238
    if (!qemuDomainMachineIsPSeries(machine, qemuCaps->arch))
5239 5240
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_FDC);

5241 5242
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_USB_STORAGE))
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_USB);
5243 5244 5245 5246

    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_ICH9_AHCI))
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_SATA);

5247
    return 0;
5248 5249 5250
}


5251 5252 5253 5254 5255 5256
static int
virQEMUCapsFillDomainDeviceGraphicsCaps(virQEMUCapsPtr qemuCaps,
                                        virDomainCapsDeviceGraphicsPtr dev)
{
    dev->supported = true;

J
Ján Tomko 已提交
5257
    VIR_DOMAIN_CAPS_ENUM_SET(dev->type, VIR_DOMAIN_GRAPHICS_TYPE_SDL);
5258 5259 5260 5261 5262 5263 5264 5265 5266
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_VNC))
        VIR_DOMAIN_CAPS_ENUM_SET(dev->type, VIR_DOMAIN_GRAPHICS_TYPE_VNC);
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_SPICE))
        VIR_DOMAIN_CAPS_ENUM_SET(dev->type, VIR_DOMAIN_GRAPHICS_TYPE_SPICE);

    return 0;
}


5267 5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278
static int
virQEMUCapsFillDomainDeviceVideoCaps(virQEMUCapsPtr qemuCaps,
                                     virDomainCapsDeviceVideoPtr dev)
{
    dev->supported = true;

    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VGA))
        VIR_DOMAIN_CAPS_ENUM_SET(dev->modelType, VIR_DOMAIN_VIDEO_TYPE_VGA);
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_CIRRUS_VGA))
        VIR_DOMAIN_CAPS_ENUM_SET(dev->modelType, VIR_DOMAIN_VIDEO_TYPE_CIRRUS);
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VMWARE_SVGA))
        VIR_DOMAIN_CAPS_ENUM_SET(dev->modelType, VIR_DOMAIN_VIDEO_TYPE_VMVGA);
5279
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_QXL))
5280 5281 5282 5283 5284 5285 5286 5287
        VIR_DOMAIN_CAPS_ENUM_SET(dev->modelType, VIR_DOMAIN_VIDEO_TYPE_QXL);
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_GPU))
        VIR_DOMAIN_CAPS_ENUM_SET(dev->modelType, VIR_DOMAIN_VIDEO_TYPE_VIRTIO);

    return 0;
}


5288
static int
5289 5290 5291 5292 5293 5294
virQEMUCapsFillDomainDeviceHostdevCaps(virQEMUCapsPtr qemuCaps,
                                       virDomainCapsDeviceHostdevPtr hostdev)
{
    bool supportsPassthroughKVM = qemuHostdevHostSupportsPassthroughLegacy();
    bool supportsPassthroughVFIO = qemuHostdevHostSupportsPassthroughVFIO();

5295
    hostdev->supported = true;
5296 5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307
    /* VIR_DOMAIN_HOSTDEV_MODE_CAPABILITIES is for containers only */
    VIR_DOMAIN_CAPS_ENUM_SET(hostdev->mode,
                             VIR_DOMAIN_HOSTDEV_MODE_SUBSYS);

    VIR_DOMAIN_CAPS_ENUM_SET(hostdev->startupPolicy,
                             VIR_DOMAIN_STARTUP_POLICY_DEFAULT,
                             VIR_DOMAIN_STARTUP_POLICY_MANDATORY,
                             VIR_DOMAIN_STARTUP_POLICY_REQUISITE,
                             VIR_DOMAIN_STARTUP_POLICY_OPTIONAL);

    VIR_DOMAIN_CAPS_ENUM_SET(hostdev->subsysType,
                             VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB,
5308 5309
                             VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI,
                             VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI);
5310 5311 5312 5313 5314 5315 5316 5317 5318 5319 5320 5321

    /* No virDomainHostdevCapsType for QEMU */
    virDomainCapsEnumClear(&hostdev->capsType);

    virDomainCapsEnumClear(&hostdev->pciBackend);
    if (supportsPassthroughVFIO &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VFIO_PCI)) {
        VIR_DOMAIN_CAPS_ENUM_SET(hostdev->pciBackend,
                                 VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT,
                                 VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO);
    }

J
Ján Tomko 已提交
5322
    if (supportsPassthroughKVM) {
5323 5324 5325 5326
        VIR_DOMAIN_CAPS_ENUM_SET(hostdev->pciBackend,
                                 VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT,
                                 VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM);
    }
5327
    return 0;
5328 5329 5330
}


5331 5332 5333 5334 5335 5336 5337
/**
 * virQEMUCapsSupportsGICVersion:
 * @qemuCaps: QEMU capabilities
 * @virtType: domain type
 * @version: GIC version
 *
 * Checks the QEMU binary with capabilities @qemuCaps supports a specific
5338 5339
 * GIC version for a domain of type @virtType. If @qemuCaps is NULL, the GIC
 * @version is considered unsupported.
5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350
 *
 * Returns: true if the binary supports the requested GIC version, false
 *          otherwise
 */
bool
virQEMUCapsSupportsGICVersion(virQEMUCapsPtr qemuCaps,
                              virDomainVirtType virtType,
                              virGICVersion version)
{
    size_t i;

5351 5352 5353
    if (!qemuCaps)
        return false;

5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367 5368 5369 5370 5371 5372
    for (i = 0; i < qemuCaps->ngicCapabilities; i++) {
        virGICCapabilityPtr cap = &(qemuCaps->gicCapabilities[i]);

        if (cap->version != version)
            continue;

        if (virtType == VIR_DOMAIN_VIRT_KVM &&
            cap->implementation & VIR_GIC_IMPLEMENTATION_KERNEL)
            return true;

        if (virtType == VIR_DOMAIN_VIRT_QEMU &&
            cap->implementation & VIR_GIC_IMPLEMENTATION_EMULATED)
            return true;
    }

    return false;
}


5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396
/**
 * virQEMUCapsFillDomainFeatureGICCaps:
 * @qemuCaps: QEMU capabilities
 * @domCaps: domain capabilities
 *
 * Take the information about GIC capabilities that has been obtained
 * using the 'query-gic-capabilities' QMP command and stored in @qemuCaps
 * and convert it to a form suitable for @domCaps.
 *
 * @qemuCaps contains complete information about the GIC capabilities for
 * the corresponding QEMU binary, stored as custom objects; @domCaps, on
 * the other hand, should only contain information about the GIC versions
 * available for the specific combination of architecture, machine type
 * and virtualization type. Moreover, a common format is used to store
 * information about enumerations in @domCaps, so further processing is
 * required.
 *
 * Returns: 0 on success, <0 on failure
 */
static int
virQEMUCapsFillDomainFeatureGICCaps(virQEMUCapsPtr qemuCaps,
                                    virDomainCapsPtr domCaps)
{
    virDomainCapsFeatureGICPtr gic = &domCaps->gic;
5397
    virGICVersion version;
5398

5399
    if (!qemuDomainMachineIsARMVirt(domCaps->machine, domCaps->arch))
5400 5401
        return 0;

5402 5403 5404 5405 5406 5407
    for (version = VIR_GIC_VERSION_LAST - 1;
         version > VIR_GIC_VERSION_NONE;
         version--) {
        if (!virQEMUCapsSupportsGICVersion(qemuCaps,
                                           domCaps->virttype,
                                           version))
5408 5409 5410 5411
            continue;

        gic->supported = true;
        VIR_DOMAIN_CAPS_ENUM_SET(gic->version,
5412
                                 version);
5413 5414 5415 5416 5417 5418
    }

    return 0;
}


5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434
/**
 * virQEMUCapsFillDomainFeatureSEVCaps:
 * @qemuCaps: QEMU capabilities
 * @domCaps: domain capabilities
 *
 * Take the information about SEV capabilities that has been obtained
 * using the 'query-sev-capabilities' QMP command and stored in @qemuCaps
 * and convert it to a form suitable for @domCaps.
 *
 * Returns: 0 on success, -1 on failure
 */
static int
virQEMUCapsFillDomainFeatureSEVCaps(virQEMUCapsPtr qemuCaps,
                                    virDomainCapsPtr domCaps)
{
    virSEVCapability *cap = qemuCaps->sevCapabilities;
5435
    VIR_AUTOPTR(virSEVCapability) sev = NULL;
5436 5437 5438 5439 5440 5441 5442 5443

    if (!cap)
        return 0;

    if (VIR_ALLOC(sev) < 0)
        return -1;

    if (VIR_STRDUP(sev->pdh, cap->pdh) < 0)
5444
        return -1;
5445 5446

    if (VIR_STRDUP(sev->cert_chain, cap->cert_chain) < 0)
5447
        return -1;
5448 5449 5450 5451 5452

    sev->cbitpos = cap->cbitpos;
    sev->reduced_phys_bits = cap->reduced_phys_bits;
    VIR_STEAL_PTR(domCaps->sev, sev);

5453
    return 0;
5454 5455 5456
}


5457
int
5458 5459
virQEMUCapsFillDomainCaps(virCapsPtr caps,
                          virDomainCapsPtr domCaps,
5460
                          virQEMUCapsPtr qemuCaps,
5461
                          virFirmwarePtr *firmwares,
5462
                          size_t nfirmwares)
5463
{
5464
    virDomainCapsOSPtr os = &domCaps->os;
5465 5466
    virDomainCapsDeviceDiskPtr disk = &domCaps->disk;
    virDomainCapsDeviceHostdevPtr hostdev = &domCaps->hostdev;
5467
    virDomainCapsDeviceGraphicsPtr graphics = &domCaps->graphics;
5468
    virDomainCapsDeviceVideoPtr video = &domCaps->video;
5469

5470 5471
    domCaps->maxvcpus = virQEMUCapsGetMachineMaxCpus(qemuCaps,
                                                     domCaps->machine);
5472
    if (domCaps->virttype == VIR_DOMAIN_VIRT_KVM) {
5473 5474 5475 5476 5477 5478
        int hostmaxvcpus;

        if ((hostmaxvcpus = virHostCPUGetKVMMaxVCPUs()) < 0)
            return -1;

        domCaps->maxvcpus = MIN(domCaps->maxvcpus, hostmaxvcpus);
5479
    }
5480

5481 5482 5483
    domCaps->vmcoreinfo = virQEMUCapsGet(qemuCaps,
                                         QEMU_CAPS_DEVICE_VMCOREINFO);

5484 5485
    domCaps->genid = virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VMGENID);

5486
    if (virQEMUCapsFillDomainOSCaps(os, firmwares, nfirmwares) < 0 ||
5487
        virQEMUCapsFillDomainCPUCaps(caps, qemuCaps, domCaps) < 0 ||
5488
        virQEMUCapsFillDomainIOThreadCaps(qemuCaps, domCaps) < 0 ||
5489 5490 5491
        virQEMUCapsFillDomainDeviceDiskCaps(qemuCaps,
                                            domCaps->machine, disk) < 0 ||
        virQEMUCapsFillDomainDeviceGraphicsCaps(qemuCaps, graphics) < 0 ||
5492
        virQEMUCapsFillDomainDeviceVideoCaps(qemuCaps, video) < 0 ||
5493
        virQEMUCapsFillDomainDeviceHostdevCaps(qemuCaps, hostdev) < 0 ||
5494 5495
        virQEMUCapsFillDomainFeatureGICCaps(qemuCaps, domCaps) < 0 ||
        virQEMUCapsFillDomainFeatureSEVCaps(qemuCaps, domCaps) < 0)
5496
        return -1;
5497

5498
    return 0;
5499
}
5500 5501 5502 5503 5504 5505 5506 5507


void
virQEMUCapsSetMicrocodeVersion(virQEMUCapsPtr qemuCaps,
                               unsigned int microcodeVersion)
{
    qemuCaps->microcodeVersion = microcodeVersion;
}
5508 5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524


/**
 * virQEMUCapsStripMachineAliases:
 * @qemuCaps: capabilities object to process
 *
 * Remove all aliases so that the tests depending on the latest capabilities
 * file can be stable when new files are added.
 */
void
virQEMUCapsStripMachineAliases(virQEMUCapsPtr qemuCaps)
{
    size_t i;

    for (i = 0; i < qemuCaps->nmachineTypes; i++)
        VIR_FREE(qemuCaps->machineTypes[i].alias);
}