qemu_capabilities.c 166.0 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
              "scsi-disk.device_id",
528 529
    );

530

531 532 533 534
struct virQEMUCapsMachineType {
    char *name;
    char *alias;
    unsigned int maxCpus;
535
    bool hotplugCpus;
536
    bool qemuDefault;
537
};
538 539 540 541 542 543 544 545 546 547 548

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;
549 550
    /* Migratable host CPU definition used for updating guest CPU. */
    virCPUDefPtr migratable;
551 552 553 554
    /* 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;
555 556
};

557 558 559 560 561 562
/*
 * 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.
563 564
 *
 * And don't forget to update virQEMUCapsNewCopy.
565
 */
566
struct _virQEMUCaps {
567
    virObject parent;
568

569
    bool usedQMP;
570
    bool kvmSupportsNesting;
571

572
    char *binary;
573
    time_t ctime;
574
    time_t libvirtCtime;
575

576
    virBitmapPtr flags;
577 578 579

    unsigned int version;
    unsigned int kvmVersion;
580
    unsigned int libvirtVersion;
581
    unsigned int microcodeVersion;
582
    char *package;
583
    char *kernelVersion;
584

585
    virArch arch;
586

587 588
    virDomainCapsCPUModelsPtr kvmCPUModels;
    virDomainCapsCPUModelsPtr tcgCPUModels;
589 590

    size_t nmachineTypes;
591
    struct virQEMUCapsMachineType *machineTypes;
A
Andrea Bolognani 已提交
592 593 594

    size_t ngicCapabilities;
    virGICCapability *gicCapabilities;
595

596 597
    virSEVCapability *sevCapabilities;

598 599
    virQEMUCapsHostCPUData kvmCPU;
    virQEMUCapsHostCPUData tcgCPU;
600 601
};

602 603
struct virQEMUCapsSearchData {
    virArch arch;
604
    const char *binaryFilter;
605 606
};

607

608 609
static virClassPtr virQEMUCapsClass;
static void virQEMUCapsDispose(void *obj);
610

611
static int virQEMUCapsOnceInit(void)
612
{
613
    if (!VIR_CLASS_NEW(virQEMUCaps, virClassForObject()))
614 615 616 617 618
        return -1;

    return 0;
}

619
VIR_ONCE_GLOBAL_INIT(virQEMUCaps);
620

621
static virArch virQEMUCapsArchFromString(const char *arch)
622 623 624 625 626
{
    if (STREQ(arch, "i386"))
        return VIR_ARCH_I686;
    if (STREQ(arch, "arm"))
        return VIR_ARCH_ARMV7L;
627 628
    if (STREQ(arch, "or32"))
        return VIR_ARCH_OR32;
629 630 631 632 633

    return virArchFromString(arch);
}


634
static const char *virQEMUCapsArchToString(virArch arch)
635 636 637
{
    if (arch == VIR_ARCH_I686)
        return "i386";
S
Stefan Schallenberg 已提交
638
    else if (arch == VIR_ARCH_ARMV6L || arch == VIR_ARCH_ARMV7L)
639
        return "arm";
640 641
    else if (arch == VIR_ARCH_OR32)
        return "or32";
642 643 644 645

    return virArchToString(arch);
}

646 647 648

/* Checks whether a domain with @guest arch can run natively on @host.
 */
649
bool
650 651 652
virQEMUCapsGuestIsNative(virArch host,
                         virArch guest)
{
653
    /* host & guest arches match */
654 655 656
    if (host == guest)
        return true;

657
    /* hostarch is x86_64 and guest arch is i686 (needs -cpu qemu32) */
658 659 660
    if (host == VIR_ARCH_X86_64 && guest == VIR_ARCH_I686)
        return true;

661
    /* hostarch is aarch64 and guest arch is armv7l (needs -cpu aarch64=off) */
662 663 664
    if (host == VIR_ARCH_AARCH64 && guest == VIR_ARCH_ARMV7L)
        return true;

665
    /* hostarch and guestarch are both ppc64 */
666 667 668 669 670 671 672
    if (ARCH_IS_PPC64(host) && ARCH_IS_PPC64(guest))
        return true;

    return false;
}


673 674 675 676 677 678 679 680 681
/* 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)
{
682 683 684
    if (virQEMUCapsGuestIsNative(hostarch, guestarch))
        guestarch = hostarch;

685 686 687 688 689 690
    /* Both ppc64 and ppc64le guests can use the ppc64 target */
    if (ARCH_IS_PPC64(guestarch))
        guestarch = VIR_ARCH_PPC64;

    return guestarch;
}
691

692

693
static void
694 695
virQEMUCapsSetDefaultMachine(virQEMUCapsPtr qemuCaps,
                             size_t defIdx)
696
{
697
    struct virQEMUCapsMachineType tmp = qemuCaps->machineTypes[defIdx];
698 699 700 701

    memmove(qemuCaps->machineTypes + 1,
            qemuCaps->machineTypes,
            sizeof(qemuCaps->machineTypes[0]) * defIdx);
702 703

    qemuCaps->machineTypes[0] = tmp;
704 705
}

706

707
static char *
708 709
virQEMUCapsFindBinary(const char *format,
                      const char *archstr)
710
{
711 712
    char *ret = NULL;
    char *binary = NULL;
713

714
    if (virAsprintf(&binary, format, archstr) < 0)
715
        return NULL;
716 717 718

    ret = virFindFileInPath(binary);
    VIR_FREE(binary);
719 720 721 722 723 724 725 726 727 728 729
    return ret;
}

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

730 731 732 733 734 735 736 737 738
    /* 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;
    }

739 740 741 742 743 744 745 746 747 748 749
    /* 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;
750
    }
751

752
 out:
753 754 755
    return ret;
}

756
static int
757
virQEMUCapsInitGuest(virCapsPtr caps,
758
                     virFileCachePtr cache,
759 760
                     virArch hostarch,
                     virArch guestarch)
761 762
{
    char *binary = NULL;
763
    virQEMUCapsPtr qemuCaps = NULL;
764 765
    int ret = -1;

J
Ján Tomko 已提交
766
    /* Check for existence of base emulator, or alternate base
767 768
     * which can be used with magic cpu choice
     */
769
    binary = virQEMUCapsFindBinaryForArch(hostarch, guestarch);
770

771 772 773 774 775 776 777
    /* 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;
    }

778
    /* Ignore binary if extracting version info fails */
779
    if (binary) {
780
        if (!(qemuCaps = virQEMUCapsCacheLookup(cache, binary))) {
781 782 783 784
            virResetLastError();
            VIR_FREE(binary);
        }
    }
785

786
    ret = virQEMUCapsInitGuestFromBinary(caps,
787
                                         binary, qemuCaps,
788 789 790
                                         guestarch);

    VIR_FREE(binary);
791
    virObjectUnref(qemuCaps);
792 793 794 795 796 797 798

    return ret;
}

int
virQEMUCapsInitGuestFromBinary(virCapsPtr caps,
                               const char *binary,
799
                               virQEMUCapsPtr qemuCaps,
800 801 802 803 804 805 806 807
                               virArch guestarch)
{
    virCapsGuestPtr guest;
    virCapsGuestMachinePtr *machines = NULL;
    size_t nmachines = 0;
    int ret = -1;
    bool hasdisksnapshot = false;

808 809 810
    if (!binary)
        return 0;

811
    if (virQEMUCapsGetMachineTypesCaps(qemuCaps, &nmachines, &machines) < 0)
812
        goto cleanup;
813 814 815 816

    /* We register kvm as the base emulator too, since we can
     * just give -no-kvm to disable acceleration if required */
    if ((guest = virCapabilitiesAddGuest(caps,
817
                                         VIR_DOMAIN_OSTYPE_HVM,
818
                                         guestarch,
819 820 821 822
                                         binary,
                                         NULL,
                                         nmachines,
                                         machines)) == NULL)
823
        goto cleanup;
824 825 826 827

    machines = NULL;
    nmachines = 0;

A
Andrea Bolognani 已提交
828 829 830
    /* CPU selection is always available, because all QEMU versions
     * we support can use at least '-cpu host' */
    if (!virCapabilitiesAddGuestFeature(guest, "cpuselection", true, false))
831
        goto cleanup;
832

833
    if (!virCapabilitiesAddGuestFeature(guest, "deviceboot", true, false))
834
        goto cleanup;
835

836
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DISK_SNAPSHOT))
837 838
        hasdisksnapshot = true;

839 840
    if (!virCapabilitiesAddGuestFeature(guest, "disksnapshot", hasdisksnapshot,
                                        false))
841
        goto cleanup;
842

D
Daniel P. Berrange 已提交
843
    if (virCapabilitiesAddGuestDomain(guest,
844
                                      VIR_DOMAIN_VIRT_QEMU,
D
Daniel P. Berrange 已提交
845 846 847 848
                                      NULL,
                                      NULL,
                                      0,
                                      NULL) == NULL)
849
        goto cleanup;
850

851
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
852 853 854 855 856 857
        if (virCapabilitiesAddGuestDomain(guest,
                                          VIR_DOMAIN_VIRT_KVM,
                                          NULL,
                                          NULL,
                                          0,
                                          NULL) == NULL) {
858
            goto cleanup;
D
Daniel P. Berrange 已提交
859
        }
860
    }
861

862 863 864
    if ((ARCH_IS_X86(guestarch) || guestarch == VIR_ARCH_AARCH64) &&
        virCapabilitiesAddGuestFeature(guest, "acpi", true, true) == NULL) {
        goto cleanup;
865 866
    }

A
Andrea Bolognani 已提交
867
    if (ARCH_IS_X86(guestarch) &&
868
        virCapabilitiesAddGuestFeature(guest, "apic", true, false) == NULL) {
869
        goto cleanup;
870
    }
871

872
    if ((guestarch == VIR_ARCH_I686) &&
873 874
        (virCapabilitiesAddGuestFeature(guest, "pae", true, false) == NULL ||
         virCapabilitiesAddGuestFeature(guest, "nonpae", true, false) == NULL))
875
        goto cleanup;
876 877 878

    ret = 0;

879
 cleanup:
880 881 882

    virCapabilitiesFreeMachines(machines, nmachines);

883
    return ret;
884 885 886
}


887
virCPUDefPtr
888
virQEMUCapsProbeHostCPUForEmulator(virArch hostArch,
889 890 891
                                   virQEMUCapsPtr qemuCaps,
                                   virDomainVirtType type)
{
892 893
    return virCPUGetHost(hostArch, VIR_CPU_TYPE_GUEST, NULL,
                         virQEMUCapsGetCPUDefinitions(qemuCaps, type));
894 895 896
}


897 898
virCapsPtr
virQEMUCapsInit(virFileCachePtr cache)
899 900
{
    virCapsPtr caps;
901
    size_t i;
T
Tal Kain 已提交
902
    virArch hostarch = virArchFromHost();
903

T
Tal Kain 已提交
904
    if ((caps = virCapabilitiesNew(hostarch,
905
                                   true, true)) == NULL)
906
        goto error;
907

908
    /* Some machines have problematic NUMA topology causing
909 910 911
     * 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
912
    if (virCapabilitiesInitNUMA(caps) < 0) {
913
        virCapabilitiesFreeNUMAInfo(caps);
914
        VIR_WARN("Failed to query host NUMA topology, disabling NUMA capabilities");
915 916
    }

917 918 919
    if (virCapabilitiesInitCaches(caps) < 0)
        VIR_WARN("Failed to get host CPU cache info");

M
Martin Kletzander 已提交
920
    if (!(caps->host.cpu = virCPUProbeHost(caps->host.arch)))
921
        VIR_WARN("Failed to get host CPU");
922

923
    /* Add the power management features of the host */
924
    if (virNodeSuspendGetTargetMask(&caps->host.powerMgmt) < 0)
925 926
        VIR_WARN("Failed to get host power management capabilities");

927 928 929
    /* Add IOMMU info */
    virCapabilitiesHostInitIOMMU(caps);

M
Michal Privoznik 已提交
930
    /* Add huge pages info */
931
    if (virCapabilitiesInitPages(caps) < 0)
M
Michal Privoznik 已提交
932 933
        VIR_WARN("Failed to get pages info");

934 935 936
    /* Add domain migration transport URIs */
    virCapabilitiesAddHostMigrateTransport(caps, "tcp");
    virCapabilitiesAddHostMigrateTransport(caps, "rdma");
937

938 939 940 941
    /* 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
     */
942
    for (i = 0; i < VIR_ARCH_LAST; i++)
943
        if (virQEMUCapsInitGuest(caps, cache,
T
Tal Kain 已提交
944
                                 hostarch,
945
                                 i) < 0)
946
            goto error;
947 948 949

    return caps;

950
 error:
951
    virObjectUnref(caps);
952 953 954 955
    return NULL;
}


956
struct virQEMUCapsStringFlags {
957 958 959 960 961
    const char *value;
    int flag;
};


962 963
struct virQEMUCapsStringFlags virQEMUCapsCommands[] = {
    { "transaction", QEMU_CAPS_TRANSACTION },
964
    { "block-stream", QEMU_CAPS_BLOCKJOB_ASYNC },
965 966 967 968 969 970 971
    { "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 },
972
    { "change-backing-file", QEMU_CAPS_CHANGE_BACKING_FILE },
973
    { "rtc-reset-reinjection", QEMU_CAPS_RTC_RESET_REINJECTION },
974
    { "migrate-incoming", QEMU_CAPS_INCOMING_DEFER },
975
    { "query-hotpluggable-cpus", QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS },
976
    { "query-qmp-schema", QEMU_CAPS_QUERY_QMP_SCHEMA },
977 978 979
    { "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 },
980
    { "query-cpus-fast", QEMU_CAPS_QUERY_CPUS_FAST },
981
    { "qom-list-properties", QEMU_CAPS_QOM_LIST_PROPERTIES },
982
    { "blockdev-del", QEMU_CAPS_BLOCKDEV_DEL },
983 984
};

985 986 987 988
struct virQEMUCapsStringFlags virQEMUCapsMigration[] = {
    { "rdma-pin-all", QEMU_CAPS_MIGRATE_RDMA },
};

989
struct virQEMUCapsStringFlags virQEMUCapsEvents[] = {
990
    { "MIGRATION", QEMU_CAPS_MIGRATION_EVENT },
991
    { "VSERPORT_CHANGE", QEMU_CAPS_VSERPORT_CHANGE },
992
    { "BLOCK_WRITE_THRESHOLD", QEMU_CAPS_BLOCK_WRITE_THRESHOLD },
993
    { "DUMP_COMPLETED", QEMU_CAPS_DUMP_COMPLETED },
994 995
};

996
struct virQEMUCapsStringFlags virQEMUCapsObjectTypes[] = {
997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011
    { "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 },
1012
    { "virtio-blk-ccw", QEMU_CAPS_VIRTIO_CCW },
1013
    { "sclpconsole", QEMU_CAPS_DEVICE_SCLPCONSOLE },
1014
    { "lsi53c895a", QEMU_CAPS_SCSI_LSI },
1015
    { "virtio-scsi-pci", QEMU_CAPS_VIRTIO_SCSI },
1016 1017
    { "virtio-scsi-s390", QEMU_CAPS_VIRTIO_SCSI },
    { "virtio-scsi-ccw", QEMU_CAPS_VIRTIO_SCSI },
1018
    { "virtio-scsi-device", QEMU_CAPS_VIRTIO_SCSI },
1019
    { "megasas", QEMU_CAPS_SCSI_MEGASAS },
1020
    { "qxl", QEMU_CAPS_DEVICE_QXL },
1021 1022
    { "sga", QEMU_CAPS_SGA },
    { "scsi-block", QEMU_CAPS_SCSI_BLOCK },
1023 1024 1025
    { "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
1026
    { "usb-serial", QEMU_CAPS_DEVICE_USB_SERIAL },
1027
    { "virtio-rng-pci", QEMU_CAPS_DEVICE_VIRTIO_RNG },
1028 1029
    { "virtio-rng-s390", QEMU_CAPS_DEVICE_VIRTIO_RNG },
    { "virtio-rng-ccw", QEMU_CAPS_DEVICE_VIRTIO_RNG },
1030
    { "virtio-rng-device", QEMU_CAPS_DEVICE_VIRTIO_RNG },
1031
    { "rng-random", QEMU_CAPS_OBJECT_RNG_RANDOM },
1032
    { "rng-egd", QEMU_CAPS_OBJECT_RNG_EGD },
1033
    { "spapr-nvram", QEMU_CAPS_DEVICE_NVRAM },
1034
    { "pci-bridge", QEMU_CAPS_DEVICE_PCI_BRIDGE },
1035
    { "vfio-pci", QEMU_CAPS_DEVICE_VFIO_PCI },
1036
    { "i82801b11-bridge", QEMU_CAPS_DEVICE_DMI_TO_PCI_BRIDGE },
1037
    { "usb-storage", QEMU_CAPS_DEVICE_USB_STORAGE },
1038
    { "virtio-mmio", QEMU_CAPS_DEVICE_VIRTIO_MMIO },
1039
    { "ich9-intel-hda", QEMU_CAPS_DEVICE_ICH9_INTEL_HDA },
H
Hu Tao 已提交
1040
    { "pvpanic", QEMU_CAPS_DEVICE_PANIC },
L
Li Zhang 已提交
1041
    { "usb-kbd", QEMU_CAPS_DEVICE_USB_KBD },
1042
    { "memory-backend-ram", QEMU_CAPS_OBJECT_MEMORY_RAM },
1043
    { "memory-backend-file", QEMU_CAPS_OBJECT_MEMORY_FILE },
1044
    { "usb-audio", QEMU_CAPS_OBJECT_USB_AUDIO },
J
John Ferlan 已提交
1045
    { "iothread", QEMU_CAPS_OBJECT_IOTHREAD},
1046
    { "ivshmem", QEMU_CAPS_DEVICE_IVSHMEM },
1047
    { "pc-dimm", QEMU_CAPS_DEVICE_PC_DIMM },
M
Michal Privoznik 已提交
1048
    { "pci-serial", QEMU_CAPS_DEVICE_PCI_SERIAL },
1049
    { "gpex-pcihost", QEMU_CAPS_OBJECT_GPEX},
1050
    { "ioh3420", QEMU_CAPS_DEVICE_IOH3420 },
1051
    { "x3130-upstream", QEMU_CAPS_DEVICE_X3130_UPSTREAM },
1052
    { "xio3130-downstream", QEMU_CAPS_DEVICE_XIO3130_DOWNSTREAM },
1053
    { "rtl8139", QEMU_CAPS_DEVICE_RTL8139 },
1054
    { "e1000", QEMU_CAPS_DEVICE_E1000 },
1055 1056 1057 1058
    { "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
1059 1060
    { "virtio-gpu-pci", QEMU_CAPS_DEVICE_VIRTIO_GPU },
    { "virtio-gpu-device", QEMU_CAPS_DEVICE_VIRTIO_GPU },
1061
    { "virtio-vga", QEMU_CAPS_DEVICE_VIRTIO_VGA },
1062 1063 1064 1065 1066 1067
    { "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 },
1068 1069
    { "virtio-input-host-device", QEMU_CAPS_VIRTIO_INPUT_HOST },
    { "virtio-input-host-pci", QEMU_CAPS_VIRTIO_INPUT_HOST },
1070
    { "mptsas1068", QEMU_CAPS_SCSI_MPTSAS1068 },
1071
    { "secret", QEMU_CAPS_OBJECT_SECRET },
1072
    { "pxb", QEMU_CAPS_DEVICE_PXB },
1073
    { "pxb-pcie", QEMU_CAPS_DEVICE_PXB_PCIE },
1074
    { "tls-creds-x509", QEMU_CAPS_OBJECT_TLS_CREDS_X509 },
J
Ján Tomko 已提交
1075
    { "intel-iommu", QEMU_CAPS_DEVICE_INTEL_IOMMU },
1076 1077
    { "ivshmem-plain", QEMU_CAPS_DEVICE_IVSHMEM_PLAIN },
    { "ivshmem-doorbell", QEMU_CAPS_DEVICE_IVSHMEM_DOORBELL },
1078
    { "vhost-scsi", QEMU_CAPS_DEVICE_VHOST_SCSI },
1079
    { "nvdimm", QEMU_CAPS_DEVICE_NVDIMM },
1080
    { "pcie-root-port", QEMU_CAPS_DEVICE_PCIE_ROOT_PORT },
1081
    { "qemu-xhci", QEMU_CAPS_DEVICE_QEMU_XHCI },
1082
    { "spapr-pci-host-bridge", QEMU_CAPS_DEVICE_SPAPR_PCI_HOST_BRIDGE },
M
Marc-André Lureau 已提交
1083
    { "vmcoreinfo", QEMU_CAPS_DEVICE_VMCOREINFO },
1084
    { "spapr-vty", QEMU_CAPS_DEVICE_SPAPR_VTY },
1085
    { "sclplmconsole", QEMU_CAPS_DEVICE_SCLPLMCONSOLE },
1086
    { "isa-serial", QEMU_CAPS_DEVICE_ISA_SERIAL },
1087
    { "pl011", QEMU_CAPS_DEVICE_PL011 },
1088
    { "virtio-gpu-ccw", QEMU_CAPS_DEVICE_VIRTIO_GPU_CCW },
1089 1090 1091
    { "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 },
1092
    { "pcie-pci-bridge", QEMU_CAPS_DEVICE_PCIE_PCI_BRIDGE },
1093
    { "pr-manager-helper", QEMU_CAPS_PR_MANAGER_HELPER },
1094
    { "virtual-css-bridge", QEMU_CAPS_CCW },
1095
    { "vfio-ccw", QEMU_CAPS_DEVICE_VFIO_CCW },
1096
    { "hda-output", QEMU_CAPS_HDA_OUTPUT },
1097
    { "vmgenid", QEMU_CAPS_DEVICE_VMGENID },
1098
    { "vhost-vsock-device", QEMU_CAPS_DEVICE_VHOST_VSOCK },
1099
    { "mch", QEMU_CAPS_DEVICE_MCH },
1100
    { "sev-guest", QEMU_CAPS_SEV_GUEST },
B
Boris Fiuczynski 已提交
1101
    { "vfio-ap", QEMU_CAPS_DEVICE_VFIO_AP },
Y
Yi Min Zhao 已提交
1102
    { "zpci", QEMU_CAPS_DEVICE_ZPCI },
1103
    { "memory-backend-memfd", QEMU_CAPS_OBJECT_MEMORY_MEMFD },
1104 1105
};

1106
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioBalloon[] = {
1107
    { "deflate-on-oom", QEMU_CAPS_VIRTIO_BALLOON_AUTODEFLATE },
1108 1109 1110
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1111 1112
};

1113
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioBlk[] = {
1114 1115 1116 1117
    { "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 },
1118
    { "num-queues", QEMU_CAPS_VIRTIO_BLK_NUM_QUEUES },
1119
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1120 1121 1122
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1123
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1124 1125
};

1126
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioNet[] = {
1127 1128
    { "tx", QEMU_CAPS_VIRTIO_TX_ALG },
    { "event_idx", QEMU_CAPS_VIRTIO_NET_EVENT_IDX },
1129
    { "rx_queue_size", QEMU_CAPS_VIRTIO_NET_RX_QUEUE_SIZE },
1130
    { "tx_queue_size", QEMU_CAPS_VIRTIO_NET_TX_QUEUE_SIZE },
1131
    { "host_mtu", QEMU_CAPS_VIRTIO_NET_HOST_MTU },
1132 1133 1134
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1135 1136
};

1137
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsSpaprPCIHostBridge[] = {
1138 1139 1140
    { "numa_node", QEMU_CAPS_SPAPR_PCI_HOST_BRIDGE_NUMA_NODE },
};

1141
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioSCSI[] = {
1142
    { "iothread", QEMU_CAPS_VIRTIO_SCSI_IOTHREAD },
1143 1144 1145
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1146 1147
};

1148
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVfioPCI[] = {
1149
    { "display", QEMU_CAPS_VFIO_PCI_DISPLAY },
1150 1151
};

1152
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsSCSIDisk[] = {
1153 1154
    { "channel", QEMU_CAPS_SCSI_DISK_CHANNEL },
    { "wwn", QEMU_CAPS_SCSI_DISK_WWN },
1155
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1156
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1157
    { "device_id", QEMU_CAPS_SCSI_DISK_DEVICE_ID },
1158 1159
};

1160
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsIDEDrive[] = {
1161
    { "wwn", QEMU_CAPS_IDE_DRIVE_WWN },
1162
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1163
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1164 1165
};

1166
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsPiix4PM[] = {
1167 1168
    { "disable_s3", QEMU_CAPS_PIIX_DISABLE_S3 },
    { "disable_s4", QEMU_CAPS_PIIX_DISABLE_S4 },
1169 1170
};

1171
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsUSBRedir[] = {
1172
    { "filter", QEMU_CAPS_USB_REDIR_FILTER },
1173 1174
};

1175
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsI440FXPCIHost[] = {
1176 1177 1178
    { "pci-hole64-size", QEMU_CAPS_I440FX_PCI_HOLE64_SIZE },
};

1179
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsQ35PCIHost[] = {
1180 1181 1182
    { "pci-hole64-size", QEMU_CAPS_Q35_PCI_HOLE64_SIZE },
};

1183
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsUSBStorage[] = {
1184
    { "removable", QEMU_CAPS_USB_STORAGE_REMOVABLE },
1185
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1186
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1187
    { "werror", QEMU_CAPS_USB_STORAGE_WERROR },
1188 1189
};

1190
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsKVMPit[] = {
1191 1192 1193
    { "lost_tick_policy", QEMU_CAPS_KVM_PIT_TICK_POLICY },
};

1194
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVGA[] = {
1195 1196 1197
    { "vgamem_mb", QEMU_CAPS_VGA_VGAMEM },
};

1198
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVmwareSvga[] = {
1199 1200 1201
    { "vgamem_mb", QEMU_CAPS_VMWARE_SVGA_VGAMEM },
};

1202
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsQxl[] = {
1203
    { "vgamem_mb", QEMU_CAPS_QXL_VGAMEM },
1204
    { "vram64_size_mb", QEMU_CAPS_QXL_VRAM64 },
1205
    { "max_outputs", QEMU_CAPS_QXL_MAX_OUTPUTS },
1206 1207
};

1208
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioGpu[] = {
1209
    { "virgl", QEMU_CAPS_VIRTIO_GPU_VIRGL },
1210
    { "max_outputs", QEMU_CAPS_VIRTIO_GPU_MAX_OUTPUTS },
1211 1212 1213
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1214 1215
};

1216
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsICH9[] = {
1217 1218 1219 1220
    { "disable_s3", QEMU_CAPS_ICH9_DISABLE_S3 },
    { "disable_s4", QEMU_CAPS_ICH9_DISABLE_S4 },
};

1221
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsUSBNECXHCI[] = {
1222 1223 1224
    { "p3", QEMU_CAPS_NEC_USB_XHCI_PORTS },
};

1225
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsIntelIOMMU[] = {
1226
    { "intremap", QEMU_CAPS_INTEL_IOMMU_INTREMAP },
1227
    { "caching-mode", QEMU_CAPS_INTEL_IOMMU_CACHING_MODE },
1228
    { "eim", QEMU_CAPS_INTEL_IOMMU_EIM },
1229
    { "device-iotlb", QEMU_CAPS_INTEL_IOMMU_DEVICE_IOTLB },
1230 1231
};

1232 1233 1234 1235
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsVirtualCSSBridge[] = {
    { "cssid-unrestricted", QEMU_CAPS_CCW_CSSID_UNRESTRICTED },
};

1236 1237 1238 1239
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsMCH[] = {
    { "extended-tseg-mbytes", QEMU_CAPS_MCH_EXTENDED_TSEG_MBYTES },
};

1240 1241 1242 1243
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsNVDIMM[] = {
    { "unarmed", QEMU_CAPS_DEVICE_NVDIMM_UNARMED },
};

1244
/* see documentation for virQEMUQAPISchemaPathGet for the query format */
1245
static struct virQEMUCapsStringFlags virQEMUCapsQMPSchemaQueries[] = {
1246
    { "blockdev-add/arg-type/options/+gluster/debug-level", QEMU_CAPS_GLUSTER_DEBUG_LEVEL},
1247
    { "blockdev-add/arg-type/+gluster/debug", QEMU_CAPS_GLUSTER_DEBUG_LEVEL},
J
John Ferlan 已提交
1248
    { "blockdev-add/arg-type/+vxhs", QEMU_CAPS_VXHS},
1249
    { "blockdev-add/arg-type/+iscsi/password-secret", QEMU_CAPS_ISCSI_PASSWORD_SECRET },
1250
    { "blockdev-add/arg-type/+qcow2/encrypt/+luks/key-secret", QEMU_CAPS_QCOW2_LUKS },
1251
    { "nbd-server-start/arg-type/tls-creds", QEMU_CAPS_NBD_TLS },
1252
    { "screendump/arg-type/device", QEMU_CAPS_SCREENDUMP_DEVICE },
1253
    { "block-commit/arg-type/*top",  QEMU_CAPS_ACTIVE_COMMIT },
1254
    { "query-iothreads/ret-type/poll-max-ns", QEMU_CAPS_IOTHREAD_POLLING },
1255
    { "query-display-options/ret-type/+egl-headless/rendernode", QEMU_CAPS_EGL_HEADLESS_RENDERNODE },
1256 1257
};

1258 1259
typedef struct _virQEMUCapsObjectTypeProps virQEMUCapsObjectTypeProps;
struct _virQEMUCapsObjectTypeProps {
1260
    const char *type;
1261
    struct virQEMUCapsStringFlags *props;
1262
    size_t nprops;
1263
    int capsCondition;
1264 1265
};

1266 1267 1268 1269 1270
typedef int (*virQEMUCapsObjectTypePropsCB)(qemuMonitorPtr mon,
                                            const char *type,
                                            char ***props);

static virQEMUCapsObjectTypeProps virQEMUCapsDeviceProps[] = {
1271 1272
    { "virtio-blk-pci", virQEMUCapsDevicePropsVirtioBlk,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBlk),
1273
      -1 },
1274 1275
    { "virtio-net-pci", virQEMUCapsDevicePropsVirtioNet,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioNet),
1276
      QEMU_CAPS_DEVICE_VIRTIO_NET },
1277 1278
    { "virtio-scsi-pci", virQEMUCapsDevicePropsVirtioSCSI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioSCSI),
1279
      QEMU_CAPS_VIRTIO_SCSI },
1280 1281
    { "virtio-blk-ccw", virQEMUCapsDevicePropsVirtioBlk,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBlk),
1282
      QEMU_CAPS_VIRTIO_CCW },
1283 1284
    { "virtio-net-ccw", virQEMUCapsDevicePropsVirtioNet,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioNet),
1285
      QEMU_CAPS_DEVICE_VIRTIO_NET },
1286 1287
    { "virtio-scsi-ccw", virQEMUCapsDevicePropsVirtioSCSI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioSCSI),
1288
      QEMU_CAPS_VIRTIO_SCSI },
1289 1290
    { "virtio-blk-s390", virQEMUCapsDevicePropsVirtioBlk,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBlk),
1291
      QEMU_CAPS_VIRTIO_S390 },
1292 1293
    { "virtio-net-s390", virQEMUCapsDevicePropsVirtioNet,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioNet),
1294
      QEMU_CAPS_DEVICE_VIRTIO_NET },
1295 1296
    { "vfio-pci", virQEMUCapsDevicePropsVfioPCI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVfioPCI),
1297
      QEMU_CAPS_DEVICE_VFIO_PCI },
1298
    { "scsi-hd", virQEMUCapsDevicePropsSCSIDisk,
1299
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsSCSIDisk),
1300
      -1 },
1301
    { "ide-hd", virQEMUCapsDevicePropsIDEDrive,
1302
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsIDEDrive),
1303
      -1 },
1304 1305
    { "PIIX4_PM", virQEMUCapsDevicePropsPiix4PM,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsPiix4PM),
1306
      -1 },
1307 1308
    { "usb-redir", virQEMUCapsDevicePropsUSBRedir,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsUSBRedir),
1309
      QEMU_CAPS_USB_REDIR },
1310 1311
    { "i440FX-pcihost", virQEMUCapsDevicePropsI440FXPCIHost,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsI440FXPCIHost),
1312
      -1 },
1313 1314
    { "q35-pcihost", virQEMUCapsDevicePropsQ35PCIHost,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsQ35PCIHost),
1315
      -1 },
1316 1317
    { "usb-storage", virQEMUCapsDevicePropsUSBStorage,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsUSBStorage),
1318
      QEMU_CAPS_DEVICE_USB_STORAGE },
1319 1320
    { "kvm-pit", virQEMUCapsDevicePropsKVMPit,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsKVMPit),
1321
      -1 },
1322 1323
    { "VGA", virQEMUCapsDevicePropsVGA,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVGA),
1324
      QEMU_CAPS_DEVICE_VGA },
1325 1326
    { "vmware-svga", virQEMUCapsDevicePropsVmwareSvga,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVmwareSvga),
1327
      QEMU_CAPS_DEVICE_VMWARE_SVGA },
1328 1329
    { "qxl", virQEMUCapsDevicePropsQxl,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsQxl),
1330
      QEMU_CAPS_DEVICE_QXL },
1331 1332
    { "virtio-gpu-pci", virQEMUCapsDevicePropsVirtioGpu,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioGpu),
1333
      QEMU_CAPS_DEVICE_VIRTIO_GPU },
1334 1335
    { "virtio-gpu-device", virQEMUCapsDevicePropsVirtioGpu,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioGpu),
1336
      QEMU_CAPS_DEVICE_VIRTIO_GPU },
1337 1338
    { "ICH9-LPC", virQEMUCapsDevicePropsICH9,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsICH9),
1339
      -1 },
1340 1341
    { "virtio-balloon-pci", virQEMUCapsDevicePropsVirtioBalloon,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBalloon),
1342
      -1 },
1343 1344
    { "virtio-balloon-ccw", virQEMUCapsDevicePropsVirtioBalloon,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBalloon),
1345
      -1 },
1346 1347
    { "virtio-balloon-device", virQEMUCapsDevicePropsVirtioBalloon,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBalloon),
1348
      -1 },
1349 1350
    { "nec-usb-xhci", virQEMUCapsDevicePropsUSBNECXHCI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsUSBNECXHCI),
1351
      QEMU_CAPS_NEC_USB_XHCI },
1352 1353
    { "intel-iommu", virQEMUCapsDevicePropsIntelIOMMU,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsIntelIOMMU),
1354
      QEMU_CAPS_DEVICE_INTEL_IOMMU },
1355 1356
    { "spapr-pci-host-bridge", virQEMUCapsDevicePropsSpaprPCIHostBridge,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsSpaprPCIHostBridge),
1357
      QEMU_CAPS_DEVICE_SPAPR_PCI_HOST_BRIDGE },
1358 1359
    { "virtio-gpu-ccw", virQEMUCapsDevicePropsVirtioGpu,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioGpu),
1360
      QEMU_CAPS_DEVICE_VIRTIO_GPU_CCW },
1361 1362 1363
    { "virtual-css-bridge", virQEMUCapsObjectPropsVirtualCSSBridge,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsVirtualCSSBridge),
      QEMU_CAPS_CCW },
1364 1365 1366
    { "mch", virQEMUCapsDevicePropsMCH,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsMCH),
      QEMU_CAPS_DEVICE_MCH },
1367 1368 1369
    { "nvdimm", virQEMUCapsDevicePropsNVDIMM,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsNVDIMM),
      QEMU_CAPS_DEVICE_NVDIMM },
1370 1371
};

1372 1373
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsMemoryBackendFile[] = {
    { "discard-data", QEMU_CAPS_OBJECT_MEMORY_FILE_DISCARD },
1374
    { "align", QEMU_CAPS_OBJECT_MEMORY_FILE_ALIGN },
1375
    { "pmem", QEMU_CAPS_OBJECT_MEMORY_FILE_PMEM },
1376 1377
};

1378 1379 1380 1381
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsMemoryBackendMemfd[] = {
    { "hugetlb", QEMU_CAPS_OBJECT_MEMORY_MEMFD_HUGETLB },
};

1382 1383
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsSPAPRMachine[] = {
    { "cap-hpt-max-page-size", QEMU_CAPS_MACHINE_PSERIES_CAP_HPT_MAX_PAGE_SIZE },
1384
    { "cap-htm", QEMU_CAPS_MACHINE_PSERIES_CAP_HTM },
1385
    { "cap-nested-hv", QEMU_CAPS_MACHINE_PSERIES_CAP_NESTED_HV },
1386 1387
};

1388 1389 1390 1391
static virQEMUCapsObjectTypeProps virQEMUCapsObjectProps[] = {
    { "memory-backend-file", virQEMUCapsObjectPropsMemoryBackendFile,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsMemoryBackendFile),
      QEMU_CAPS_OBJECT_MEMORY_FILE },
1392 1393 1394
    { "memory-backend-memfd", virQEMUCapsObjectPropsMemoryBackendMemfd,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsMemoryBackendMemfd),
      QEMU_CAPS_OBJECT_MEMORY_MEMFD },
1395 1396 1397
    { "spapr-machine", virQEMUCapsObjectPropsSPAPRMachine,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsSPAPRMachine),
      -1 },
1398
};
1399 1400

static void
1401 1402 1403 1404 1405
virQEMUCapsProcessStringFlags(virQEMUCapsPtr qemuCaps,
                              size_t nflags,
                              struct virQEMUCapsStringFlags *flags,
                              size_t nvalues,
                              char *const*values)
1406 1407
{
    size_t i, j;
1408
    for (i = 0; i < nflags; i++) {
1409 1410 1411
        if (virQEMUCapsGet(qemuCaps, flags[i].flag))
            continue;

1412
        for (j = 0; j < nvalues; j++) {
1413
            if (STREQ(values[j], flags[i].value)) {
1414
                virQEMUCapsSet(qemuCaps, flags[i].flag);
1415 1416 1417 1418 1419 1420 1421
                break;
            }
        }
    }
}


1422
int virQEMUCapsGetDefaultVersion(virCapsPtr caps,
1423
                                 virFileCachePtr capsCache,
1424
                                 unsigned int *version)
1425
{
1426
    virQEMUCapsPtr qemucaps;
T
Tal Kain 已提交
1427
    virArch hostarch;
1428
    virCapsDomainDataPtr capsdata;
1429 1430 1431 1432

    if (*version > 0)
        return 0;

T
Tal Kain 已提交
1433
    hostarch = virArchFromHost();
1434 1435 1436
    if (!(capsdata = virCapabilitiesDomainDataLookup(caps,
            VIR_DOMAIN_OSTYPE_HVM, hostarch, VIR_DOMAIN_VIRT_QEMU,
            NULL, NULL))) {
1437
        virReportError(VIR_ERR_INTERNAL_ERROR,
1438
                       _("Cannot find suitable emulator for %s"),
T
Tal Kain 已提交
1439
                       virArchToString(hostarch));
1440 1441 1442
        return -1;
    }

1443
    qemucaps = virQEMUCapsCacheLookup(capsCache, capsdata->emulator);
1444 1445
    VIR_FREE(capsdata);
    if (!qemucaps)
1446 1447
        return -1;

1448
    *version = virQEMUCapsGetVersion(qemucaps);
1449
    virObjectUnref(qemucaps);
1450 1451
    return 0;
}
1452 1453


1454 1455


1456 1457
virQEMUCapsPtr
virQEMUCapsNew(void)
1458
{
1459
    virQEMUCapsPtr qemuCaps;
1460

1461
    if (virQEMUCapsInitialize() < 0)
1462 1463
        return NULL;

1464
    if (!(qemuCaps = virObjectNew(virQEMUCapsClass)))
1465 1466
        return NULL;

1467
    if (!(qemuCaps->flags = virBitmapNew(QEMU_CAPS_LAST)))
1468
        goto error;
1469

1470
    return qemuCaps;
1471

1472
 error:
1473
    virObjectUnref(qemuCaps);
1474
    return NULL;
1475 1476 1477
}


1478
static int
1479 1480
virQEMUCapsHostCPUDataCopy(virQEMUCapsHostCPUDataPtr dst,
                           virQEMUCapsHostCPUDataPtr src)
1481
{
1482 1483
    if (src->info &&
        !(dst->info = qemuMonitorCPUModelInfoCopy(src->info)))
1484 1485
        return -1;

1486 1487
    if (src->reported &&
        !(dst->reported = virCPUDefCopy(src->reported)))
1488 1489
        return -1;

1490 1491 1492 1493
    if (src->migratable &&
        !(dst->migratable = virCPUDefCopy(src->migratable)))
        return -1;

1494 1495 1496 1497
    if (src->full &&
        !(dst->full = virCPUDefCopy(src->full)))
        return -1;

1498 1499 1500 1501
    return 0;
}


1502
static void
1503
virQEMUCapsHostCPUDataClear(virQEMUCapsHostCPUDataPtr cpuData)
1504
{
1505
    qemuMonitorCPUModelInfoFree(cpuData->info);
1506
    virCPUDefFree(cpuData->reported);
1507
    virCPUDefFree(cpuData->migratable);
1508
    virCPUDefFree(cpuData->full);
1509 1510

    memset(cpuData, 0, sizeof(*cpuData));
1511 1512 1513
}


1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532
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;
}


1533
virQEMUCapsPtr virQEMUCapsNewCopy(virQEMUCapsPtr qemuCaps)
1534
{
1535
    virQEMUCapsPtr ret = virQEMUCapsNew();
1536 1537 1538 1539 1540
    size_t i;

    if (!ret)
        return NULL;

1541
    ret->usedQMP = qemuCaps->usedQMP;
1542
    ret->kvmSupportsNesting = qemuCaps->kvmSupportsNesting;
1543

1544 1545 1546 1547 1548
    if (VIR_STRDUP(ret->binary, qemuCaps->binary) < 0)
        goto error;

    ret->ctime = qemuCaps->ctime;

1549
    virBitmapCopy(ret->flags, qemuCaps->flags);
1550

1551 1552
    ret->version = qemuCaps->version;
    ret->kvmVersion = qemuCaps->kvmVersion;
1553
    ret->microcodeVersion = qemuCaps->microcodeVersion;
1554 1555 1556 1557

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

1558 1559 1560
    if (VIR_STRDUP(ret->kernelVersion, qemuCaps->kernelVersion) < 0)
        goto error;

1561
    ret->arch = qemuCaps->arch;
1562

1563 1564 1565 1566 1567 1568 1569 1570 1571
    if (qemuCaps->kvmCPUModels) {
        ret->kvmCPUModels = virDomainCapsCPUModelsCopy(qemuCaps->kvmCPUModels);
        if (!ret->kvmCPUModels)
            goto error;
    }

    if (qemuCaps->tcgCPUModels) {
        ret->tcgCPUModels = virDomainCapsCPUModelsCopy(qemuCaps->tcgCPUModels);
        if (!ret->tcgCPUModels)
1572
            goto error;
1573 1574
    }

1575 1576
    if (virQEMUCapsHostCPUDataCopy(&ret->kvmCPU, &qemuCaps->kvmCPU) < 0 ||
        virQEMUCapsHostCPUDataCopy(&ret->tcgCPU, &qemuCaps->tcgCPU) < 0)
1577 1578
        goto error;

1579
    if (VIR_ALLOC_N(ret->machineTypes, qemuCaps->nmachineTypes) < 0)
1580
        goto error;
1581
    ret->nmachineTypes = qemuCaps->nmachineTypes;
1582
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
1583 1584
        if (VIR_STRDUP(ret->machineTypes[i].name, qemuCaps->machineTypes[i].name) < 0 ||
            VIR_STRDUP(ret->machineTypes[i].alias, qemuCaps->machineTypes[i].alias) < 0)
1585
            goto error;
1586
        ret->machineTypes[i].maxCpus = qemuCaps->machineTypes[i].maxCpus;
1587
        ret->machineTypes[i].hotplugCpus = qemuCaps->machineTypes[i].hotplugCpus;
1588 1589
        ret->machineTypes[i].qemuDefault = qemuCaps->machineTypes[i].qemuDefault;

1590 1591
    }

1592 1593 1594 1595 1596 1597
    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];

1598 1599 1600 1601 1602
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_SEV_GUEST) &&
        virQEMUCapsSEVInfoCopy(&ret->sevCapabilities,
                               qemuCaps->sevCapabilities) < 0)
        goto error;

1603 1604
    return ret;

1605
 error:
1606 1607 1608 1609 1610
    virObjectUnref(ret);
    return NULL;
}


1611
void virQEMUCapsDispose(void *obj)
1612
{
1613
    virQEMUCapsPtr qemuCaps = obj;
1614 1615
    size_t i;

1616
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
1617 1618
        VIR_FREE(qemuCaps->machineTypes[i].name);
        VIR_FREE(qemuCaps->machineTypes[i].alias);
1619
    }
1620
    VIR_FREE(qemuCaps->machineTypes);
1621

1622 1623
    virObjectUnref(qemuCaps->kvmCPUModels);
    virObjectUnref(qemuCaps->tcgCPUModels);
1624

1625
    virBitmapFree(qemuCaps->flags);
1626

1627
    VIR_FREE(qemuCaps->package);
1628
    VIR_FREE(qemuCaps->kernelVersion);
1629
    VIR_FREE(qemuCaps->binary);
A
Andrea Bolognani 已提交
1630 1631

    VIR_FREE(qemuCaps->gicCapabilities);
1632

1633 1634
    virSEVCapabilitiesFree(qemuCaps->sevCapabilities);

1635 1636
    virQEMUCapsHostCPUDataClear(&qemuCaps->kvmCPU);
    virQEMUCapsHostCPUDataClear(&qemuCaps->tcgCPU);
1637 1638
}

1639
void
1640
virQEMUCapsSet(virQEMUCapsPtr qemuCaps,
1641
               virQEMUCapsFlags flag)
1642
{
1643
    ignore_value(virBitmapSetBit(qemuCaps->flags, flag));
1644 1645 1646 1647
}


void
1648
virQEMUCapsSetList(virQEMUCapsPtr qemuCaps, ...)
1649 1650 1651 1652
{
    va_list list;
    int flag;

1653
    va_start(list, qemuCaps);
1654
    while ((flag = va_arg(list, int)) < QEMU_CAPS_LAST)
1655
        ignore_value(virBitmapSetBit(qemuCaps->flags, flag));
1656
    va_end(list);
1657 1658 1659 1660
}


void
1661
virQEMUCapsClear(virQEMUCapsPtr qemuCaps,
1662
                 virQEMUCapsFlags flag)
1663
{
1664
    ignore_value(virBitmapClearBit(qemuCaps->flags, flag));
1665 1666 1667
}


1668
char *virQEMUCapsFlagsString(virQEMUCapsPtr qemuCaps)
1669
{
1670
    return virBitmapToString(qemuCaps->flags, true, false);
1671 1672 1673 1674
}


bool
1675
virQEMUCapsGet(virQEMUCapsPtr qemuCaps,
1676
               virQEMUCapsFlags flag)
1677
{
J
Ján Tomko 已提交
1678
    return qemuCaps && virBitmapIsBitSet(qemuCaps->flags, flag);
1679
}
1680 1681


D
Daniel P. Berrange 已提交
1682
bool virQEMUCapsHasPCIMultiBus(virQEMUCapsPtr qemuCaps,
1683
                               const virDomainDef *def)
D
Daniel P. Berrange 已提交
1684
{
1685 1686
    /* x86_64 and i686 support PCI-multibus on all machine types
     * since forever */
1687
    if (ARCH_IS_X86(def->os.arch))
D
Daniel P. Berrange 已提交
1688 1689 1690
        return true;

    if (def->os.arch == VIR_ARCH_PPC ||
1691
        ARCH_IS_PPC64(def->os.arch)) {
D
Daniel P. Berrange 已提交
1692 1693 1694 1695 1696
        /*
         * Usage of pci.0 naming:
         *
         *    ref405ep: no pci
         *       taihu: no pci
1697
         *      bamboo: 1.1.0 (<= 1.5.0, so basically forever)
D
Daniel P. Berrange 已提交
1698 1699
         *       mac99: 2.0.0
         *     g3beige: 2.0.0
1700
         *        prep: 1.4.0 (<= 1.5.0, so basically forever)
D
Daniel P. Berrange 已提交
1701 1702 1703 1704 1705 1706
         *     pseries: 2.0.0
         *   mpc8544ds: forever
         * virtex-m507: no pci
         *     ppce500: 1.6.0
         */

1707 1708 1709 1710 1711 1712 1713 1714
        /* 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
1715 1716 1717 1718 1719 1720 1721
        if (qemuCaps->version >= 2000000)
            return true;

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

1722 1723 1724
        if (STREQ(def->os.machine, "bamboo") ||
            STREQ(def->os.machine, "mpc8544ds") ||
            STREQ(def->os.machine, "prep")) {
D
Daniel P. Berrange 已提交
1725
            return true;
1726
        }
D
Daniel P. Berrange 已提交
1727 1728 1729 1730

        return false;
    }

1731 1732 1733 1734
    /* S390 supports PCI-multibus. */
    if (ARCH_IS_S390(def->os.arch))
        return true;

1735 1736 1737 1738
    /* If the virt machine, both on ARM and RISC-V, supports PCI,
     * then it also supports multibus */
    if (qemuDomainIsARMVirt(def) ||
        qemuDomainIsRISCVVirt(def)) {
1739
        return true;
1740
    }
1741

D
Daniel P. Berrange 已提交
1742 1743 1744 1745
    return false;
}


1746
const char *virQEMUCapsGetBinary(virQEMUCapsPtr qemuCaps)
1747
{
1748
    return qemuCaps->binary;
1749 1750
}

1751 1752 1753 1754 1755 1756 1757 1758 1759

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


1760
virArch virQEMUCapsGetArch(virQEMUCapsPtr qemuCaps)
1761
{
1762
    return qemuCaps->arch;
1763 1764 1765
}


1766
unsigned int virQEMUCapsGetVersion(virQEMUCapsPtr qemuCaps)
1767
{
1768
    return qemuCaps->version;
1769 1770 1771
}


1772
unsigned int virQEMUCapsGetKVMVersion(virQEMUCapsPtr qemuCaps)
1773
{
1774
    return qemuCaps->kvmVersion;
1775 1776 1777
}


1778 1779 1780 1781 1782 1783
const char *virQEMUCapsGetPackage(virQEMUCapsPtr qemuCaps)
{
    return qemuCaps->package;
}


1784 1785
int
virQEMUCapsAddCPUDefinitions(virQEMUCapsPtr qemuCaps,
1786
                             virDomainVirtType type,
1787
                             const char **name,
1788 1789
                             size_t count,
                             virDomainCapsCPUUsable usable)
1790
{
1791
    size_t i;
1792
    virDomainCapsCPUModelsPtr cpus = NULL;
1793

1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807
    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;
    }
1808 1809

    for (i = 0; i < count; i++) {
1810
        if (virDomainCapsCPUModelsAdd(cpus, name[i], -1, usable, NULL) < 0)
1811
            return -1;
1812
    }
1813

1814 1815 1816 1817
    return 0;
}


1818
virDomainCapsCPUModelsPtr
1819
virQEMUCapsGetCPUDefinitions(virQEMUCapsPtr qemuCaps,
1820
                             virDomainVirtType type)
1821
{
1822
    if (type == VIR_DOMAIN_VIRT_KVM)
1823
        return qemuCaps->kvmCPUModels;
1824
    else
1825
        return qemuCaps->tcgCPUModels;
1826 1827 1828
}


1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839
static virQEMUCapsHostCPUDataPtr
virQEMUCapsGetHostCPUData(virQEMUCapsPtr qemuCaps,
                          virDomainVirtType type)
{
    if (type == VIR_DOMAIN_VIRT_KVM)
        return &qemuCaps->kvmCPU;
    else
        return &qemuCaps->tcgCPU;
}


1840
virCPUDefPtr
1841
virQEMUCapsGetHostModel(virQEMUCapsPtr qemuCaps,
1842 1843
                        virDomainVirtType type,
                        virQEMUCapsHostCPUType cpuType)
1844
{
1845 1846
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);

1847 1848 1849
    switch (cpuType) {
    case VIR_QEMU_CAPS_HOST_CPU_REPORTED:
        return cpuData->reported;
1850 1851 1852

    case VIR_QEMU_CAPS_HOST_CPU_MIGRATABLE:
        return cpuData->migratable;
1853 1854 1855 1856 1857

    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;
1858 1859 1860
    }

    return NULL;
1861 1862 1863
}


1864 1865 1866
static void
virQEMUCapsSetHostModel(virQEMUCapsPtr qemuCaps,
                        virDomainVirtType type,
1867
                        virCPUDefPtr reported,
1868 1869
                        virCPUDefPtr migratable,
                        virCPUDefPtr full)
1870
{
1871 1872
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);

1873 1874
    cpuData->reported = reported;
    cpuData->migratable = migratable;
1875
    cpuData->full = full;
1876 1877 1878
}


1879 1880 1881 1882 1883 1884
bool
virQEMUCapsIsCPUModeSupported(virQEMUCapsPtr qemuCaps,
                              virCapsPtr caps,
                              virDomainVirtType type,
                              virCPUMode mode)
{
1885 1886
    virDomainCapsCPUModelsPtr cpus;

1887 1888 1889 1890 1891 1892
    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:
1893 1894
        return !!virQEMUCapsGetHostModel(qemuCaps, type,
                                         VIR_QEMU_CAPS_HOST_CPU_REPORTED);
1895 1896

    case VIR_CPU_MODE_CUSTOM:
1897 1898 1899 1900 1901
        if (type == VIR_DOMAIN_VIRT_KVM)
            cpus = qemuCaps->kvmCPUModels;
        else
            cpus = qemuCaps->tcgCPUModels;
        return cpus && cpus->nmodels > 0;
1902 1903 1904 1905 1906 1907 1908 1909 1910

    case VIR_CPU_MODE_LAST:
        break;
    }

    return false;
}


1911 1912 1913
int virQEMUCapsGetMachineTypesCaps(virQEMUCapsPtr qemuCaps,
                                   size_t *nmachines,
                                   virCapsGuestMachinePtr **machines)
1914 1915 1916 1917
{
    size_t i;

    *machines = NULL;
1918
    *nmachines = qemuCaps->nmachineTypes;
1919

1920 1921 1922 1923
    if (*nmachines &&
        VIR_ALLOC_N(*machines, qemuCaps->nmachineTypes) < 0)
        goto error;

1924
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
1925 1926
        virCapsGuestMachinePtr mach;
        if (VIR_ALLOC(mach) < 0)
1927
            goto error;
1928
        (*machines)[i] = mach;
1929 1930 1931
        if (qemuCaps->machineTypes[i].alias) {
            if (VIR_STRDUP(mach->name, qemuCaps->machineTypes[i].alias) < 0 ||
                VIR_STRDUP(mach->canonical, qemuCaps->machineTypes[i].name) < 0)
1932
                goto error;
1933
        } else {
1934
            if (VIR_STRDUP(mach->name, qemuCaps->machineTypes[i].name) < 0)
1935
                goto error;
1936
        }
1937
        mach->maxCpus = qemuCaps->machineTypes[i].maxCpus;
1938 1939
    }

1940 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
    /* 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++;
    }

1978 1979
    return 0;

1980
 error:
1981 1982 1983 1984 1985 1986 1987
    virCapabilitiesFreeMachines(*machines, *nmachines);
    *nmachines = 0;
    *machines = NULL;
    return -1;
}


1988 1989 1990 1991 1992 1993 1994 1995
/**
 * 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.
 */
1996 1997
const char *virQEMUCapsGetCanonicalMachine(virQEMUCapsPtr qemuCaps,
                                           const char *name)
1998 1999 2000
{
    size_t i;

2001 2002
    if (!name || !qemuCaps)
        return name;
2003

2004
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
2005
        if (!qemuCaps->machineTypes[i].alias)
2006
            continue;
2007 2008
        if (STREQ(qemuCaps->machineTypes[i].alias, name))
            return qemuCaps->machineTypes[i].name;
2009 2010 2011 2012
    }

    return name;
}
2013

2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025
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;
}
2026

2027 2028 2029 2030 2031 2032 2033 2034 2035
int virQEMUCapsGetMachineMaxCpus(virQEMUCapsPtr qemuCaps,
                                 const char *name)
{
    size_t i;

    if (!name)
        return 0;

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
2036
        if (!qemuCaps->machineTypes[i].maxCpus)
2037
            continue;
2038 2039
        if (STREQ(qemuCaps->machineTypes[i].name, name))
            return qemuCaps->machineTypes[i].maxCpus;
2040 2041 2042 2043 2044 2045
    }

    return 0;
}


2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059
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;
}


2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083
/**
 * 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;
}


2084 2085 2086 2087 2088 2089 2090
virSEVCapabilityPtr
virQEMUCapsGetSEVCapabilities(virQEMUCapsPtr qemuCaps)
{
    return qemuCaps->sevCapabilities;
}


2091
static int
2092 2093
virQEMUCapsProbeQMPCommands(virQEMUCapsPtr qemuCaps,
                            qemuMonitorPtr mon)
2094 2095 2096 2097 2098 2099 2100
{
    char **commands = NULL;
    int ncommands;

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

2101 2102 2103 2104
    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsCommands),
                                  virQEMUCapsCommands,
                                  ncommands, commands);
2105
    virStringListFreeCount(commands, ncommands);
2106

2107 2108 2109 2110
    /* 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) &&
2111 2112 2113
        qemuMonitorSupportsActiveCommit(mon))
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_ACTIVE_COMMIT);

2114 2115 2116 2117 2118
    return 0;
}


static int
2119 2120
virQEMUCapsProbeQMPEvents(virQEMUCapsPtr qemuCaps,
                          qemuMonitorPtr mon)
2121 2122 2123 2124 2125 2126 2127
{
    char **events = NULL;
    int nevents;

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

2128 2129 2130 2131
    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsEvents),
                                  virQEMUCapsEvents,
                                  nevents, events);
2132
    virStringListFreeCount(events, nevents);
2133 2134 2135 2136

    return 0;
}

2137 2138 2139 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
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;
}
2166

2167
static int
2168
virQEMUCapsProbeQMPDevices(virQEMUCapsPtr qemuCaps,
2169
                           qemuMonitorPtr mon)
2170 2171 2172 2173 2174 2175
{
    int nvalues;
    char **values;

    if ((nvalues = qemuMonitorGetObjectTypes(mon, &values)) < 0)
        return -1;
2176 2177 2178 2179
    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsObjectTypes),
                                  virQEMUCapsObjectTypes,
                                  nvalues, values);
2180
    virStringListFreeCount(values, nvalues);
2181

2182 2183 2184 2185 2186 2187
    if (virQEMUCapsProbeQMPGenericProps(qemuCaps,
                                        mon,
                                        virQEMUCapsDeviceProps,
                                        ARRAY_CARDINALITY(virQEMUCapsDeviceProps),
                                        qemuMonitorGetDeviceProps) < 0)
        return -1;
2188

2189 2190 2191 2192 2193 2194 2195 2196
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_QOM_LIST_PROPERTIES) &&
        virQEMUCapsProbeQMPGenericProps(qemuCaps,
                                        mon,
                                        virQEMUCapsObjectProps,
                                        ARRAY_CARDINALITY(virQEMUCapsObjectProps),
                                        qemuMonitorGetObjectProps) < 0)
        return -1;

2197 2198 2199 2200
    return 0;
}


2201 2202 2203 2204 2205 2206 2207 2208 2209 2210
/* 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.
 */
2211
static const char *preferredMachines[] =
2212
{
2213 2214
    NULL, /* VIR_ARCH_NONE (not a real arch :) */
    "clipper", /* VIR_ARCH_ALPHA */
S
Stefan Schallenberg 已提交
2215
    "integratorcp", /* VIR_ARCH_ARMV6L */
2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242
    "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
2243 2244
    "spike_v1.10", /* VIR_ARCH_RISCV32 */
    "spike_v1.10", /* VIR_ARCH_RISCV64 */
2245 2246 2247
    NULL, /* VIR_ARCH_S390 (no QEMU impl) */
    "s390-ccw-virtio", /* VIR_ARCH_S390X */
    "shix", /* VIR_ARCH_SH4 */
L
Lubomir Rintel 已提交
2248

2249 2250 2251 2252 2253
    "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
2254

2255 2256
    "sim", /* VIR_ARCH_XTENSA */
    "sim", /* VIR_ARCH_XTENSAEB */
2257
};
2258
verify(ARRAY_CARDINALITY(preferredMachines) == VIR_ARCH_LAST);
2259 2260


2261
static int
2262 2263
virQEMUCapsProbeQMPMachineTypes(virQEMUCapsPtr qemuCaps,
                                qemuMonitorPtr mon)
2264 2265 2266 2267 2268
{
    qemuMonitorMachineInfoPtr *machines = NULL;
    int nmachines = 0;
    int ret = -1;
    size_t i;
2269 2270 2271
    ssize_t defIdx = -1;
    ssize_t preferredIdx = -1;
    const char *preferredMachine = preferredMachines[qemuCaps->arch];
2272 2273

    if ((nmachines = qemuMonitorGetMachines(mon, &machines)) < 0)
2274
        return -1;
2275

2276
    if (VIR_ALLOC_N(qemuCaps->machineTypes, nmachines) < 0)
2277 2278
        goto cleanup;

2279
    for (i = 0; i < nmachines; i++) {
2280
        struct virQEMUCapsMachineType *mach;
2281 2282
        if (STREQ(machines[i]->name, "none"))
            continue;
2283 2284 2285 2286 2287

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

        if (VIR_STRDUP(mach->alias, machines[i]->alias) < 0 ||
            VIR_STRDUP(mach->name, machines[i]->name) < 0)
2288
            goto cleanup;
2289 2290

        mach->maxCpus = machines[i]->maxCpus;
2291
        mach->hotplugCpus = machines[i]->hotplugCpus;
2292

2293 2294 2295 2296 2297 2298
        if (preferredMachine &&
            (STREQ_NULLABLE(mach->alias, preferredMachine) ||
             STREQ(mach->name, preferredMachine))) {
            preferredIdx = qemuCaps->nmachineTypes - 1;
        }

2299 2300
        if (machines[i]->isDefault) {
            mach->qemuDefault = true;
2301
            defIdx = qemuCaps->nmachineTypes - 1;
2302
        }
2303
    }
2304

2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317
    /*
     * 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);
2318 2319 2320

    ret = 0;

2321
 cleanup:
2322
    for (i = 0; i < nmachines; i++)
2323 2324 2325 2326 2327 2328
        qemuMonitorMachineInfoFree(machines[i]);
    VIR_FREE(machines);
    return ret;
}


2329 2330
virDomainCapsCPUModelsPtr
virQEMUCapsFetchCPUDefinitions(qemuMonitorPtr mon)
2331
{
2332 2333 2334
    virDomainCapsCPUModelsPtr models = NULL;
    qemuMonitorCPUDefInfoPtr *cpus = NULL;
    int ncpus = 0;
2335
    size_t i;
2336

2337
    if ((ncpus = qemuMonitorGetCPUDefinitions(mon, &cpus)) < 0)
2338
        return NULL;
2339

2340
    if (!(models = virDomainCapsCPUModelsNew(ncpus)))
2341
        goto error;
2342

2343
    for (i = 0; i < ncpus; i++) {
2344 2345 2346 2347 2348 2349 2350
        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;

2351
        if (virDomainCapsCPUModelsAddSteal(models, &cpus[i]->name, usable,
2352
                                           &cpus[i]->blockers) < 0)
2353
            goto error;
2354 2355 2356 2357 2358 2359
    }

 cleanup:
    for (i = 0; i < ncpus; i++)
        qemuMonitorCPUDefInfoFree(cpus[i]);
    VIR_FREE(cpus);
2360 2361 2362 2363 2364 2365
    return models;

 error:
    virObjectUnref(models);
    models = NULL;
    goto cleanup;
2366 2367
}

2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390

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;
}


2391 2392
static int
virQEMUCapsProbeQMPHostCPU(virQEMUCapsPtr qemuCaps,
2393 2394
                           qemuMonitorPtr mon,
                           bool tcg)
2395
{
2396
    qemuMonitorCPUModelInfoPtr modelInfo = NULL;
2397 2398
    qemuMonitorCPUModelInfoPtr nonMigratable = NULL;
    virHashTablePtr hash = NULL;
2399
    const char *model;
2400
    qemuMonitorCPUModelExpansionType type;
2401 2402
    virDomainVirtType virtType;
    virQEMUCapsHostCPUDataPtr cpuData;
2403
    int ret = -1;
2404 2405

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_CPU_MODEL_EXPANSION))
2406 2407
        return 0;

2408
    if (tcg || !virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
2409
        virtType = VIR_DOMAIN_VIRT_QEMU;
2410 2411
        model = "max";
    } else {
2412
        virtType = VIR_DOMAIN_VIRT_KVM;
2413 2414 2415
        model = "host";
    }

2416 2417
    cpuData = virQEMUCapsGetHostCPUData(qemuCaps, virtType);

2418 2419 2420 2421 2422 2423 2424 2425 2426 2427
    /* 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;

2428 2429
    if (qemuMonitorGetCPUModelExpansion(mon, type, model, true, &modelInfo) < 0)
        goto cleanup;
2430 2431

    /* Try to check migratability of each feature. */
2432
    if (modelInfo &&
2433 2434
        qemuMonitorGetCPUModelExpansion(mon, type, model, false,
                                        &nonMigratable) < 0)
2435
        goto cleanup;
2436 2437 2438 2439 2440 2441 2442

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

        if (!(hash = virHashCreate(0, NULL)))
2443
            goto cleanup;
2444

2445 2446
        for (i = 0; i < modelInfo->nprops; i++) {
            prop = modelInfo->props + i;
2447
            if (virHashAddEntry(hash, prop->name, prop) < 0)
2448
                goto cleanup;
2449 2450 2451 2452 2453 2454 2455 2456 2457
        }

        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;

2458
            if (prop->value.boolean) {
2459
                prop->migratable = VIR_TRISTATE_BOOL_YES;
2460 2461 2462 2463
            } else if (nmProp->value.boolean) {
                prop->value.boolean = true;
                prop->migratable = VIR_TRISTATE_BOOL_NO;
            }
2464 2465
        }

2466
        modelInfo->migratability = true;
2467 2468
    }

2469
    VIR_STEAL_PTR(cpuData->info, modelInfo);
2470 2471 2472 2473 2474
    ret = 0;

 cleanup:
    virHashFree(hash);
    qemuMonitorCPUModelInfoFree(nonMigratable);
2475
    qemuMonitorCPUModelInfoFree(modelInfo);
2476 2477

    return ret;
2478 2479
}

2480 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

/**
 * 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;
}


2532 2533
struct tpmTypeToCaps {
    int type;
2534
    virQEMUCapsFlags caps;
2535 2536 2537 2538 2539 2540 2541
};

static const struct tpmTypeToCaps virQEMUCapsTPMTypesToCaps[] = {
    {
        .type = VIR_DOMAIN_TPM_TYPE_PASSTHROUGH,
        .caps = QEMU_CAPS_DEVICE_TPM_PASSTHROUGH,
    },
2542 2543 2544 2545
    {
        .type = VIR_DOMAIN_TPM_TYPE_EMULATOR,
        .caps = QEMU_CAPS_DEVICE_TPM_EMULATOR,
    },
2546 2547 2548 2549 2550 2551 2552
};

const struct tpmTypeToCaps virQEMUCapsTPMModelsToCaps[] = {
    {
        .type = VIR_DOMAIN_TPM_MODEL_TIS,
        .caps = QEMU_CAPS_DEVICE_TPM_TIS,
    },
2553 2554 2555 2556
    {
        .type = VIR_DOMAIN_TPM_MODEL_CRB,
        .caps = QEMU_CAPS_DEVICE_TPM_CRB,
    },
2557 2558 2559 2560 2561 2562
};

static int
virQEMUCapsProbeQMPTPM(virQEMUCapsPtr qemuCaps,
                       qemuMonitorPtr mon)
{
2563 2564
    int nentries;
    size_t i;
2565
    char **entries = NULL;
S
Stefan Berger 已提交
2566

2567 2568 2569 2570 2571 2572 2573
    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);
2574
            if (virStringListHasString((const char **)entries, needle))
2575 2576 2577 2578
                virQEMUCapsSet(qemuCaps,
                               virQEMUCapsTPMModelsToCaps[i].caps);
        }
    }
2579
    virStringListFree(entries);
2580 2581 2582 2583 2584 2585 2586 2587

    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);
2588
            if (virStringListHasString((const char **)entries, needle))
2589 2590 2591
                virQEMUCapsSet(qemuCaps, virQEMUCapsTPMTypesToCaps[i].caps);
        }
    }
2592
    virStringListFree(entries);
2593 2594 2595 2596

    return 0;
}

2597

2598
static int
2599 2600
virQEMUCapsProbeQMPKVMState(virQEMUCapsPtr qemuCaps,
                            qemuMonitorPtr mon)
2601 2602 2603 2604 2605 2606 2607
{
    bool enabled = false;
    bool present = false;

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

A
Andrea Bolognani 已提交
2608 2609
    if (present && enabled)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_KVM);
2610 2611 2612 2613

    return 0;
}

2614 2615 2616 2617 2618 2619 2620 2621
struct virQEMUCapsCommandLineProps {
    const char *option;
    const char *param;
    int flag;
};

static struct virQEMUCapsCommandLineProps virQEMUCapsCommandLine[] = {
    { "machine", "mem-merge", QEMU_CAPS_MEM_MERGE },
2622
    { "machine", "vmport", QEMU_CAPS_MACHINE_VMPORT_OPT },
O
Osier Yang 已提交
2623
    { "drive", "discard", QEMU_CAPS_DRIVE_DISCARD },
2624
    { "drive", "detect-zeroes", QEMU_CAPS_DRIVE_DETECT_ZEROES },
2625
    { "realtime", "mlock", QEMU_CAPS_REALTIME_MLOCK },
2626
    { "boot-opts", "strict", QEMU_CAPS_BOOT_STRICT },
2627
    { "boot-opts", "reboot-timeout", QEMU_CAPS_REBOOT_TIMEOUT },
2628
    { "boot-opts", "splash-time", QEMU_CAPS_SPLASH_TIMEOUT },
2629
    { "spice", "disable-agent-file-xfer", QEMU_CAPS_SPICE_FILE_XFER_DISABLE },
2630
    { "msg", "timestamp", QEMU_CAPS_MSG_TIMESTAMP },
2631
    { "numa", NULL, QEMU_CAPS_NUMA },
2632
    { "drive", "throttling.bps-total-max", QEMU_CAPS_DRIVE_IOTUNE_MAX},
2633 2634
    { "machine", "aes-key-wrap", QEMU_CAPS_AES_KEY_WRAP },
    { "machine", "dea-key-wrap", QEMU_CAPS_DEA_KEY_WRAP },
2635
    { "chardev", "append", QEMU_CAPS_CHARDEV_FILE_APPEND },
2636
    { "spice", "gl", QEMU_CAPS_SPICE_GL },
2637
    { "chardev", "logfile", QEMU_CAPS_CHARDEV_LOGFILE },
2638
    { "name", "debug-threads", QEMU_CAPS_NAME_DEBUG_THREADS },
2639
    { "name", "guest", QEMU_CAPS_NAME_GUEST },
2640
    { "spice", "unix", QEMU_CAPS_SPICE_UNIX },
2641
    { "drive", "throttling.bps-total-max-length", QEMU_CAPS_DRIVE_IOTUNE_MAX_LENGTH },
2642
    { "drive", "throttling.group", QEMU_CAPS_DRIVE_IOTUNE_GROUP },
2643
    { "spice", "rendernode", QEMU_CAPS_SPICE_RENDERNODE },
2644
    { "machine", "kernel_irqchip", QEMU_CAPS_MACHINE_KERNEL_IRQCHIP },
2645
    { "machine", "loadparm", QEMU_CAPS_LOADPARM },
2646
    { "vnc", "vnc", QEMU_CAPS_VNC_MULTI_SERVERS },
2647
    { "chardev", "reconnect", QEMU_CAPS_CHARDEV_RECONNECT },
2648
    { "sandbox", "enable", QEMU_CAPS_SECCOMP_SANDBOX },
2649
    { "sandbox", "elevateprivileges", QEMU_CAPS_SECCOMP_BLACKLIST },
2650
    { "chardev", "fd", QEMU_CAPS_CHARDEV_FD_PASS },
2651 2652 2653 2654 2655 2656
};

static int
virQEMUCapsProbeQMPCommandLine(virQEMUCapsPtr qemuCaps,
                               qemuMonitorPtr mon)
{
2657
    bool found = false;
2658 2659 2660 2661 2662 2663 2664
    int nvalues;
    char **values;
    size_t i, j;

    for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsCommandLine); i++) {
        if ((nvalues = qemuMonitorGetCommandLineOptionParameters(mon,
                                                                 virQEMUCapsCommandLine[i].option,
2665 2666
                                                                 &values,
                                                                 &found)) < 0)
2667
            return -1;
2668 2669 2670 2671

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

2672
        for (j = 0; j < nvalues; j++) {
2673
            if (STREQ_NULLABLE(virQEMUCapsCommandLine[i].param, values[j])) {
2674 2675 2676 2677
                virQEMUCapsSet(qemuCaps, virQEMUCapsCommandLine[i].flag);
                break;
            }
        }
2678
        virStringListFree(values);
2679 2680 2681 2682
    }

    return 0;
}
2683

2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697
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);
2698
    virStringListFreeCount(caps, ncaps);
2699 2700 2701 2702

    return 0;
}

A
Andrea Bolognani 已提交
2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722
/**
 * 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;

2723
    virQEMUCapsSetGICCapabilities(qemuCaps, caps, ncaps);
A
Andrea Bolognani 已提交
2724 2725 2726 2727

    return 0;
}

2728

2729
/* Returns -1 on error, 0 if SEV is not supported, 1 if SEV is supported */
2730 2731 2732 2733
static int
virQEMUCapsProbeQMPSEVCapabilities(virQEMUCapsPtr qemuCaps,
                                   qemuMonitorPtr mon)
{
2734
    int rc = -1;
2735 2736
    virSEVCapability *caps = NULL;

2737 2738
    if ((rc = qemuMonitorGetSEVCapabilities(mon, &caps)) <= 0)
        return rc;
2739

2740 2741
    virSEVCapabilitiesFree(qemuCaps->sevCapabilities);
    qemuCaps->sevCapabilities = caps;
2742
    return rc;
2743 2744 2745
}


2746
bool
2747
virQEMUCapsCPUFilterFeatures(const char *name,
2748
                             void *opaque)
2749
{
2750
    virArch *arch = opaque;
2751

2752
    if (!ARCH_IS_X86(*arch))
2753 2754
        return true;

2755 2756 2757 2758 2759 2760 2761 2762 2763
    if (STREQ(name, "cmt") ||
        STREQ(name, "mbm_total") ||
        STREQ(name, "mbm_local"))
        return false;

    return true;
}


2764 2765 2766
/**
 * 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,
2767
 *          2 when cpu model info is not supported for this configuration,
2768 2769 2770 2771
 *         -1 on error.
 */
static int
virQEMUCapsInitCPUModelS390(virQEMUCapsPtr qemuCaps,
2772
                            virDomainVirtType type,
2773
                            qemuMonitorCPUModelInfoPtr modelInfo,
2774 2775
                            virCPUDefPtr cpu,
                            bool migratable)
2776
{
2777
    size_t i;
2778

2779
    if (!modelInfo) {
2780 2781 2782 2783 2784 2785 2786 2787
        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;
2788
    }
J
Jiri Denemark 已提交
2789

2790 2791
    if (VIR_STRDUP(cpu->model, modelInfo->name) < 0 ||
        VIR_ALLOC_N(cpu->features, modelInfo->nprops) < 0)
2792
        return -1;
2793 2794 2795 2796 2797

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

    for (i = 0; i < modelInfo->nprops; i++) {
2798 2799
        virCPUFeatureDefPtr feature = cpu->features + cpu->nfeatures;
        qemuMonitorCPUPropertyPtr prop = modelInfo->props + i;
2800

2801 2802
        if (prop->type != QEMU_MONITOR_CPU_PROPERTY_BOOLEAN)
            continue;
2803

2804 2805
        if (VIR_STRDUP(feature->name, prop->name) < 0)
            return -1;
2806 2807 2808 2809 2810 2811

        if (!prop->value.boolean ||
            (migratable && prop->migratable == VIR_TRISTATE_BOOL_NO))
            feature->policy = VIR_CPU_FEATURE_DISABLE;
        else
            feature->policy = VIR_CPU_FEATURE_REQUIRE;
2812 2813 2814
        cpu->nfeatures++;
    }

2815 2816
    return 0;
}
2817

2818

2819 2820 2821 2822 2823 2824 2825 2826
/**
 * 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,
2827
                           qemuMonitorCPUModelInfoPtr model,
2828 2829
                           virCPUDefPtr cpu,
                           bool migratable)
2830 2831 2832 2833
{
    virCPUDataPtr data = NULL;
    unsigned long long sigFamily = 0;
    unsigned long long sigModel = 0;
2834
    unsigned long long sigStepping = 0;
2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848
    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:
2849 2850 2851 2852 2853
            if (!prop->value.boolean ||
                (migratable && prop->migratable == VIR_TRISTATE_BOOL_NO))
                continue;

            if (virCPUx86DataAddFeature(data, prop->name) < 0)
2854
                goto cleanup;
2855

2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868
            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;
2869 2870
            else if (STREQ(prop->name, "stepping"))
                sigStepping = prop->value.number;
2871 2872 2873 2874 2875 2876 2877
            break;

        case QEMU_MONITOR_CPU_PROPERTY_LAST:
            break;
        }
    }

2878
    if (virCPUx86DataSetSignature(data, sigFamily, sigModel, sigStepping) < 0)
2879 2880
        goto cleanup;

2881
    if (cpuDecode(cpu, data, virQEMUCapsGetCPUDefinitions(qemuCaps, type)) < 0)
2882 2883 2884 2885 2886 2887 2888 2889 2890 2891
        goto cleanup;

    ret = 0;

 cleanup:
    virCPUDataFree(data);
    return ret;
}


2892 2893
/**
 * Returns  0 when host CPU model provided by QEMU was filled in qemuCaps,
2894 2895
 *          1 when the caller should fall back to other methods,
 *          2 when cpu model info is not supported for this configuration,
2896 2897
 *         -1 on error.
 */
2898
int
2899
virQEMUCapsInitCPUModel(virQEMUCapsPtr qemuCaps,
2900
                        virDomainVirtType type,
2901 2902
                        virCPUDefPtr cpu,
                        bool migratable)
2903
{
2904
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);
2905 2906
    int ret = 1;

2907
    if (migratable && cpuData->info && !cpuData->info->migratability)
2908 2909
        return 1;

2910
    if (ARCH_IS_S390(qemuCaps->arch)) {
2911
        ret = virQEMUCapsInitCPUModelS390(qemuCaps, type, cpuData->info,
2912 2913 2914 2915 2916
                                          cpu, migratable);
    } else if (ARCH_IS_X86(qemuCaps->arch)) {
        ret = virQEMUCapsInitCPUModelX86(qemuCaps, type, cpuData->info,
                                         cpu, migratable);
    }
2917

2918 2919 2920
    if (ret == 0)
        cpu->fallback = VIR_CPU_FALLBACK_FORBID;

2921
    return ret;
2922 2923 2924
}


2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941
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;
}


2942 2943
void
virQEMUCapsInitHostCPUModel(virQEMUCapsPtr qemuCaps,
2944
                            virArch hostArch,
2945
                            virDomainVirtType type)
2946 2947
{
    virCPUDefPtr cpu = NULL;
2948
    virCPUDefPtr cpuExpanded = NULL;
2949
    virCPUDefPtr migCPU = NULL;
2950
    virCPUDefPtr hostCPU = NULL;
2951 2952
    virCPUDefPtr fullCPU = NULL;
    size_t i;
2953
    int rc;
2954

2955
    if (!virQEMUCapsGuestIsNative(hostArch, qemuCaps->arch))
2956 2957
        return;

2958
    if (!(cpu = virQEMUCapsNewHostCPUModel()))
2959 2960
        goto error;

2961
    if ((rc = virQEMUCapsInitCPUModel(qemuCaps, type, cpu, false)) < 0) {
2962 2963
        goto error;
    } else if (rc == 1) {
2964
        VIR_DEBUG("No host CPU model info from QEMU; probing host CPU directly");
2965

2966
        hostCPU = virQEMUCapsProbeHostCPUForEmulator(hostArch, qemuCaps, type);
2967 2968
        if (!hostCPU ||
            virCPUDefCopyModelFilter(cpu, hostCPU, true,
2969
                                     virQEMUCapsCPUFilterFeatures,
2970
                                     &qemuCaps->arch) < 0)
2971
            goto error;
2972 2973 2974 2975 2976
    } else if (rc == 2) {
        VIR_DEBUG("QEMU does not provide CPU model for arch=%s virttype=%s",
                  virArchToString(qemuCaps->arch),
                  virDomainVirtTypeToString(type));
        goto error;
2977 2978 2979
    } else if (type == VIR_DOMAIN_VIRT_KVM &&
               virCPUGetHostIsSupported(qemuCaps->arch)) {
        if (!(fullCPU = virCPUGetHost(qemuCaps->arch, VIR_CPU_TYPE_GUEST,
2980
                                      NULL, NULL)))
2981 2982
            goto error;

2983 2984 2985 2986 2987 2988 2989
        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,
2990 2991 2992
                                       VIR_CPU_FEATURE_REQUIRE) < 0)
                goto error;
        }
2993 2994
    }

2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007
    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;
    }

3008
    virQEMUCapsSetHostModel(qemuCaps, type, cpu, migCPU, fullCPU);
3009

3010
 cleanup:
3011
    virCPUDefFree(cpuExpanded);
3012
    virCPUDefFree(hostCPU);
3013 3014 3015 3016
    return;

 error:
    virCPUDefFree(cpu);
3017
    virCPUDefFree(migCPU);
3018
    virCPUDefFree(fullCPU);
3019
    virResetLastError();
3020
    goto cleanup;
3021 3022 3023
}


3024 3025 3026 3027 3028
void
virQEMUCapsSetCPUModelInfo(virQEMUCapsPtr qemuCaps,
                           virDomainVirtType type,
                           qemuMonitorCPUModelInfoPtr modelInfo)
{
3029 3030 3031
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);

    cpuData->info = modelInfo;
3032 3033 3034
}


3035 3036
static int
virQEMUCapsLoadHostCPUModelInfo(virQEMUCapsPtr qemuCaps,
3037 3038
                                xmlXPathContextPtr ctxt,
                                virDomainVirtType virtType)
3039 3040 3041
{
    char *str = NULL;
    xmlNodePtr hostCPUNode;
3042
    xmlNodePtr *nodes = NULL;
3043 3044 3045 3046 3047
    xmlNodePtr oldnode = ctxt->node;
    qemuMonitorCPUModelInfoPtr hostCPU = NULL;
    int ret = -1;
    size_t i;
    int n;
3048
    int val;
3049

3050 3051 3052 3053 3054 3055
    if (virtType == VIR_DOMAIN_VIRT_KVM)
        hostCPUNode = virXPathNode("./hostCPU[@type='kvm']", ctxt);
    else
        hostCPUNode = virXPathNode("./hostCPU[@type='tcg']", ctxt);

    if (!hostCPUNode) {
3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069
        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;
    }

3070 3071 3072 3073 3074 3075 3076 3077 3078
    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);

3079 3080
    ctxt->node = hostCPUNode;

3081
    if ((n = virXPathNodeSet("./property", ctxt, &nodes)) > 0) {
3082 3083 3084 3085 3086 3087
        if (VIR_ALLOC_N(hostCPU->props, n) < 0)
            goto cleanup;

        hostCPU->nprops = n;

        for (i = 0; i < n; i++) {
3088 3089 3090 3091 3092
            qemuMonitorCPUPropertyPtr prop = hostCPU->props + i;

            ctxt->node = nodes[i];

            if (!(prop->name = virXMLPropString(ctxt->node, "name"))) {
3093 3094
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing 'name' attribute for a host CPU"
3095
                                 " model property in QEMU capabilities cache"));
3096 3097 3098
                goto cleanup;
            }

3099
            if (!(str = virXMLPropString(ctxt->node, "type")) ||
3100
                (val = qemuMonitorCPUPropertyTypeFromString(str)) < 0) {
3101
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
3102 3103
                               _("missing or invalid CPU model property type "
                                 "in QEMU capabilities cache"));
3104 3105 3106
                goto cleanup;
            }
            VIR_FREE(str);
3107

3108
            prop->type = val;
3109 3110 3111 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
            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;
            }
3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152

            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);
            }
3153 3154 3155
        }
    }

3156
    virQEMUCapsSetCPUModelInfo(qemuCaps, virtType, hostCPU);
3157 3158 3159 3160 3161 3162
    hostCPU = NULL;
    ret = 0;

 cleanup:
    ctxt->node = oldnode;
    VIR_FREE(str);
3163
    VIR_FREE(nodes);
3164 3165 3166 3167 3168
    qemuMonitorCPUModelInfoFree(hostCPU);
    return ret;
}


3169 3170
static int
virQEMUCapsLoadCPUModels(virQEMUCapsPtr qemuCaps,
3171 3172
                         xmlXPathContextPtr ctxt,
                         virDomainVirtType type)
3173
{
3174
    virDomainCapsCPUModelsPtr cpus = NULL;
3175 3176 3177 3178 3179
    xmlNodePtr *nodes = NULL;
    char *str = NULL;
    size_t i;
    int n;
    int ret = -1;
3180 3181 3182 3183
    xmlNodePtr node;
    xmlNodePtr *blockerNodes = NULL;
    char **blockers = NULL;
    int nblockers;
3184

3185 3186 3187 3188 3189 3190
    if (type == VIR_DOMAIN_VIRT_KVM)
        n = virXPathNodeSet("./cpu[@type='kvm']", ctxt, &nodes);
    else
        n = virXPathNodeSet("./cpu[@type='tcg']", ctxt, &nodes);

    if (n < 0) {
3191 3192 3193 3194 3195 3196 3197 3198 3199 3200
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to parse qemu capabilities cpus"));
        goto cleanup;
    }

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

3201
    if (!(cpus = virDomainCapsCPUModelsNew(n)))
3202 3203
        goto cleanup;

3204 3205 3206 3207 3208
    if (type == VIR_DOMAIN_VIRT_KVM)
        qemuCaps->kvmCPUModels = cpus;
    else
        qemuCaps->tcgCPUModels = cpus;

3209
    for (i = 0; i < n; i++) {
3210 3211 3212 3213 3214 3215 3216 3217 3218 3219
        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);

3220 3221 3222 3223 3224 3225
        if (!(str = virXMLPropString(nodes[i], "name"))) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("missing cpu name in QEMU capabilities cache"));
            goto cleanup;
        }

3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250
        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;
                }
            }
3251
            VIR_FREE(blockerNodes);
3252 3253 3254
        }

        if (virDomainCapsCPUModelsAddSteal(cpus, &str, usable, &blockers) < 0)
3255 3256 3257 3258 3259 3260 3261 3262
            goto cleanup;
    }

    ret = 0;

 cleanup:
    VIR_FREE(nodes);
    VIR_FREE(str);
3263 3264
    VIR_FREE(blockerNodes);
    virStringListFree(blockers);
3265 3266 3267 3268
    return ret;
}


3269 3270 3271 3272 3273
struct _virQEMUCapsCachePriv {
    char *libDir;
    uid_t runUid;
    gid_t runGid;
    virArch hostArch;
3274
    unsigned int microcodeVersion;
3275
    char *kernelVersion;
3276 3277 3278 3279

    /* cache whether /dev/kvm is usable as runUid:runGuid */
    virTristateBool kvmUsable;
    time_t kvmCtime;
3280 3281 3282 3283 3284
};
typedef struct _virQEMUCapsCachePriv virQEMUCapsCachePriv;
typedef virQEMUCapsCachePriv *virQEMUCapsCachePrivPtr;


3285
static void
3286
virQEMUCapsCachePrivFree(void *privData)
3287
{
3288 3289
    virQEMUCapsCachePrivPtr priv = privData;

3290
    VIR_FREE(priv->libDir);
3291
    VIR_FREE(priv->kernelVersion);
3292 3293 3294 3295
    VIR_FREE(priv);
}


3296 3297 3298 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
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;
}


3348 3349 3350 3351 3352 3353
/*
 * Parsing a doc that looks like
 *
 * <qemuCaps>
 *   <qemuctime>234235253</qemuctime>
 *   <selfctime>234235253</selfctime>
3354
 *   <selfvers>1002016</selfvers>
3355 3356 3357 3358 3359 3360
 *   <usedQMP/>
 *   <flag name='foo'/>
 *   <flag name='bar'/>
 *   ...
 *   <cpu name="pentium3"/>
 *   ...
3361
 *   <machine name="pc-1.0" alias="pc" hotplugCpus='yes' maxCpus="4" default="yes"/>
3362 3363 3364
 *   ...
 * </qemuCaps>
 */
3365
int
3366
virQEMUCapsLoadCache(virArch hostArch,
3367
                     virQEMUCapsPtr qemuCaps,
3368
                     const char *filename)
3369 3370 3371 3372 3373 3374 3375
{
    xmlDocPtr doc = NULL;
    int ret = -1;
    size_t i;
    int n;
    xmlNodePtr *nodes = NULL;
    xmlXPathContextPtr ctxt = NULL;
J
Ján Tomko 已提交
3376
    char *str = NULL;
3377
    long long int l;
3378
    unsigned long lu;
3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402

    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;
    }
3403
    qemuCaps->ctime = (time_t)l;
3404 3405 3406 3407 3408 3409

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

3412
    qemuCaps->libvirtVersion = 0;
3413
    if (virXPathULong("string(./selfvers)", ctxt, &lu) == 0)
3414
        qemuCaps->libvirtVersion = lu;
3415

3416 3417 3418 3419 3420 3421 3422 3423 3424
    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);
3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436
    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;
3437
        }
3438 3439
        VIR_FREE(str);
        virQEMUCapsSet(qemuCaps, flag);
3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454
    }
    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;
    }

3455 3456 3457 3458 3459 3460 3461
    if (virXPathUInt("string(./microcodeVersion)", ctxt,
                     &qemuCaps->microcodeVersion) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing microcode version in QEMU capabilities cache"));
        goto cleanup;
    }

3462 3463 3464 3465 3466 3467
    if (virXPathBoolean("boolean(./package)", ctxt) > 0) {
        qemuCaps->package = virXPathString("string(./package)", ctxt);
        if (!qemuCaps->package &&
            VIR_STRDUP(qemuCaps->package, "") < 0)
            goto cleanup;
    }
3468

3469 3470 3471 3472 3473 3474
    if (virXPathBoolean("boolean(./kernelVersion)", ctxt) > 0) {
        qemuCaps->kernelVersion = virXPathString("string(./kernelVersion)", ctxt);
        if (!qemuCaps->kernelVersion)
            goto cleanup;
    }

3475 3476 3477 3478 3479 3480 3481 3482 3483 3484
    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
3485
    VIR_FREE(str);
3486

3487 3488
    if (virQEMUCapsLoadHostCPUModelInfo(qemuCaps, ctxt, VIR_DOMAIN_VIRT_KVM) < 0 ||
        virQEMUCapsLoadHostCPUModelInfo(qemuCaps, ctxt, VIR_DOMAIN_VIRT_QEMU) < 0)
3489 3490
        goto cleanup;

3491 3492
    if (virQEMUCapsLoadCPUModels(qemuCaps, ctxt, VIR_DOMAIN_VIRT_KVM) < 0 ||
        virQEMUCapsLoadCPUModels(qemuCaps, ctxt, VIR_DOMAIN_VIRT_QEMU) < 0)
3493 3494 3495 3496 3497 3498 3499 3500 3501
        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;
3502
        if (VIR_ALLOC_N(qemuCaps->machineTypes, qemuCaps->nmachineTypes) < 0)
3503 3504 3505
            goto cleanup;

        for (i = 0; i < n; i++) {
3506
            if (!(qemuCaps->machineTypes[i].name = virXMLPropString(nodes[i], "name"))) {
3507 3508 3509 3510
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing machine name in QEMU capabilities cache"));
                goto cleanup;
            }
3511
            qemuCaps->machineTypes[i].alias = virXMLPropString(nodes[i], "alias");
3512 3513 3514

            str = virXMLPropString(nodes[i], "maxCpus");
            if (str &&
3515
                virStrToLong_ui(str, NULL, 10, &(qemuCaps->machineTypes[i].maxCpus)) < 0) {
3516 3517 3518 3519
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("malformed machine cpu count in QEMU capabilities cache"));
                goto cleanup;
            }
J
Ján Tomko 已提交
3520
            VIR_FREE(str);
3521 3522 3523 3524 3525

            str = virXMLPropString(nodes[i], "hotplugCpus");
            if (STREQ_NULLABLE(str, "yes"))
                qemuCaps->machineTypes[i].hotplugCpus = true;
            VIR_FREE(str);
3526 3527 3528 3529 3530

            str = virXMLPropString(nodes[i], "default");
            if (STREQ_NULLABLE(str, "yes"))
                qemuCaps->machineTypes[i].qemuDefault = true;
            VIR_FREE(str);
3531 3532 3533 3534
        }
    }
    VIR_FREE(nodes);

A
Andrea Bolognani 已提交
3535 3536 3537 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
    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);

3601 3602 3603
    if (virQEMUCapsParseSEVInfo(qemuCaps, ctxt) < 0)
        goto cleanup;

3604 3605
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_QEMU);
3606

3607 3608 3609
    if (virXPathBoolean("boolean(./kvmSupportsNesting)", ctxt) > 0)
        qemuCaps->kvmSupportsNesting = true;

3610
    ret = 0;
3611
 cleanup:
J
Ján Tomko 已提交
3612
    VIR_FREE(str);
3613 3614 3615 3616 3617 3618 3619
    VIR_FREE(nodes);
    xmlXPathFreeContext(ctxt);
    xmlFreeDoc(doc);
    return ret;
}


3620 3621
static void
virQEMUCapsFormatHostCPUModelInfo(virQEMUCapsPtr qemuCaps,
3622 3623
                                  virBufferPtr buf,
                                  virDomainVirtType type)
3624
{
3625 3626 3627
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);
    qemuMonitorCPUModelInfoPtr model = cpuData->info;
    const char *typeStr = type == VIR_DOMAIN_VIRT_KVM ? "kvm" : "tcg";
3628 3629
    size_t i;

3630 3631 3632
    if (!model)
        return;

3633 3634 3635 3636
    virBufferAsprintf(buf,
                      "<hostCPU type='%s' model='%s' migratability='%s'>\n",
                      typeStr, model->name,
                      model->migratability ? "yes" : "no");
3637 3638 3639
    virBufferAdjustIndent(buf, 2);

    for (i = 0; i < model->nprops; i++) {
3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662
        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;
        }
3663 3664 3665 3666 3667

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

3668
        virBufferAddLit(buf, "/>\n");
3669 3670 3671 3672 3673 3674 3675
    }

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


3676 3677
static void
virQEMUCapsFormatCPUModels(virQEMUCapsPtr qemuCaps,
3678 3679
                           virBufferPtr buf,
                           virDomainVirtType type)
3680
{
3681 3682
    virDomainCapsCPUModelsPtr cpus;
    const char *typeStr;
3683 3684
    size_t i;

3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696
    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++) {
3697 3698
        virDomainCapsCPUModelPtr cpu = cpus->models + i;

3699
        virBufferAsprintf(buf, "<cpu type='%s' ", typeStr);
3700 3701 3702 3703 3704
        virBufferEscapeString(buf, "name='%s'", cpu->name);
        if (cpu->usable) {
            virBufferAsprintf(buf, " usable='%s'",
                              virDomainCapsCPUUsableTypeToString(cpu->usable));
        }
3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719

        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");
        }
3720 3721 3722 3723
    }
}


3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741
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");
}


3742
char *
3743
virQEMUCapsFormatCache(virQEMUCapsPtr qemuCaps)
3744 3745
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;
3746
    char *ret = NULL;
3747 3748 3749
    size_t i;

    virBufferAddLit(&buf, "<qemuCaps>\n");
3750
    virBufferAdjustIndent(&buf, 2);
3751

3752
    virBufferAsprintf(&buf, "<qemuctime>%llu</qemuctime>\n",
3753
                      (long long)qemuCaps->ctime);
3754
    virBufferAsprintf(&buf, "<selfctime>%llu</selfctime>\n",
3755
                      (long long)qemuCaps->libvirtCtime);
3756
    virBufferAsprintf(&buf, "<selfvers>%lu</selfvers>\n",
3757
                      (unsigned long)qemuCaps->libvirtVersion);
3758 3759

    if (qemuCaps->usedQMP)
3760
        virBufferAddLit(&buf, "<usedQMP/>\n");
3761 3762 3763

    for (i = 0; i < QEMU_CAPS_LAST; i++) {
        if (virQEMUCapsGet(qemuCaps, i)) {
3764
            virBufferAsprintf(&buf, "<flag name='%s'/>\n",
3765 3766 3767 3768
                              virQEMUCapsTypeToString(i));
        }
    }

3769
    virBufferAsprintf(&buf, "<version>%d</version>\n",
3770 3771
                      qemuCaps->version);

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

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

3778 3779 3780 3781
    if (qemuCaps->package)
        virBufferAsprintf(&buf, "<package>%s</package>\n",
                          qemuCaps->package);

3782 3783 3784 3785
    if (qemuCaps->kernelVersion)
        virBufferAsprintf(&buf, "<kernelVersion>%s</kernelVersion>\n",
                          qemuCaps->kernelVersion);

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

3789 3790
    virQEMUCapsFormatHostCPUModelInfo(qemuCaps, &buf, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsFormatHostCPUModelInfo(qemuCaps, &buf, VIR_DOMAIN_VIRT_QEMU);
3791

3792 3793
    virQEMUCapsFormatCPUModels(qemuCaps, &buf, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsFormatCPUModels(qemuCaps, &buf, VIR_DOMAIN_VIRT_QEMU);
3794 3795

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
3796
        virBufferEscapeString(&buf, "<machine name='%s'",
3797 3798
                              qemuCaps->machineTypes[i].name);
        if (qemuCaps->machineTypes[i].alias)
3799
            virBufferEscapeString(&buf, " alias='%s'",
3800
                              qemuCaps->machineTypes[i].alias);
3801 3802
        if (qemuCaps->machineTypes[i].hotplugCpus)
            virBufferAddLit(&buf, " hotplugCpus='yes'");
3803
        virBufferAsprintf(&buf, " maxCpus='%u'",
3804
                          qemuCaps->machineTypes[i].maxCpus);
3805 3806 3807
        if (qemuCaps->machineTypes[i].qemuDefault)
            virBufferAddLit(&buf, " default='yes'");
        virBufferAddLit(&buf, "/>\n");
3808 3809
    }

A
Andrea Bolognani 已提交
3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825
    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");
    }

3826 3827 3828
    if (qemuCaps->sevCapabilities)
        virQEMUCapsFormatSEVInfo(qemuCaps, &buf);

3829 3830 3831
    if (qemuCaps->kvmSupportsNesting)
        virBufferAddLit(&buf, "<kvmSupportsNesting/>\n");

3832
    virBufferAdjustIndent(&buf, -2);
3833 3834
    virBufferAddLit(&buf, "</qemuCaps>\n");

3835 3836 3837 3838 3839 3840 3841 3842
    if (virBufferCheckError(&buf) == 0)
        ret = virBufferContentAndReset(&buf);

    return ret;
}


static int
3843 3844 3845
virQEMUCapsSaveFile(void *data,
                    const char *filename,
                    void *privData ATTRIBUTE_UNUSED)
3846
{
3847
    virQEMUCapsPtr qemuCaps = data;
3848 3849
    char *xml = NULL;
    int ret = -1;
3850

3851
    xml = virQEMUCapsFormatCache(qemuCaps);
3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862

    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,
3863
              (long long)qemuCaps->libvirtCtime);
3864 3865 3866 3867 3868 3869 3870 3871

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


3872 3873 3874 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
/* 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;
}


3907 3908 3909 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
/* 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;
}


3955
static bool
3956 3957
virQEMUCapsIsValid(void *data,
                   void *privData)
3958
{
3959 3960
    virQEMUCapsPtr qemuCaps = data;
    virQEMUCapsCachePrivPtr priv = privData;
3961
    bool kvmUsable;
3962
    struct stat sb;
3963
    bool kvmSupportsNesting;
3964 3965 3966 3967

    if (!qemuCaps->binary)
        return true;

3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979
    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;
    }

3980 3981 3982 3983 3984 3985
    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;
3986 3987
    }

3988
    if (sb.st_ctime != qemuCaps->ctime) {
3989 3990 3991
        VIR_DEBUG("Outdated capabilities for '%s': QEMU binary changed "
                  "(%lld vs %lld)",
                  qemuCaps->binary,
3992
                  (long long)sb.st_ctime, (long long)qemuCaps->ctime);
3993 3994 3995
        return false;
    }

3996 3997 3998 3999 4000 4001 4002 4003
    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;
    }

4004
    kvmUsable = virQEMUCapsKVMUsable(priv);
4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021

    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;
    }

4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039
    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;
        }
4040 4041 4042 4043 4044 4045 4046 4047

        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;
        }
4048 4049
    }

4050 4051 4052 4053
    return true;
}


4054
static void virQEMUCapsMonitorNotify(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
4055 4056
                                     virDomainObjPtr vm ATTRIBUTE_UNUSED,
                                     void *opaque ATTRIBUTE_UNUSED)
4057 4058 4059 4060
{
}

static qemuMonitorCallbacks callbacks = {
4061 4062
    .eofNotify = virQEMUCapsMonitorNotify,
    .errorNotify = virQEMUCapsMonitorNotify,
4063 4064 4065
};


4066 4067 4068 4069 4070 4071 4072 4073
/**
 * virQEMUCapsInitQMPArch:
 * @qemuCaps: QEMU capabilities
 * @mon: QEMU monitor
 *
 * Initialize the architecture for @qemuCaps by asking @mon.
 *
 * Returns: 0 on success, <0 on failure
4074 4075
 */
static int
4076
virQEMUCapsInitQMPArch(virQEMUCapsPtr qemuCaps,
4077 4078 4079 4080 4081 4082
                            qemuMonitorPtr mon)
{
    char *archstr = NULL;
    int ret = -1;

    if (!(archstr = qemuMonitorGetTargetArch(mon)))
4083
        goto cleanup;
4084 4085 4086 4087 4088 4089 4090

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

4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104
    ret = 0;

 cleanup:
    VIR_FREE(archstr);
    return ret;
}


/**
 * virQEMUCapsInitQMPBasicArch:
 * @qemuCaps: QEMU capabilities
 *
 * Initialize @qemuCaps with basic architecture-dependent capabilities.
 */
4105
void
4106 4107
virQEMUCapsInitQMPBasicArch(virQEMUCapsPtr qemuCaps)
{
4108 4109 4110
    /* ACPI only works on x86 and aarch64 */
    if (ARCH_IS_X86(qemuCaps->arch) ||
        qemuCaps->arch == VIR_ARCH_AARCH64) {
4111
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_NO_ACPI);
4112 4113
    }

J
Ján Tomko 已提交
4114 4115
    /* HPET is x86 specific */
    if (ARCH_IS_X86(qemuCaps->arch))
J
Ján Tomko 已提交
4116
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_NO_HPET);
4117
}
4118

4119 4120 4121 4122 4123 4124

static int
virQEMUCapsProbeQMPSchemaCapabilities(virQEMUCapsPtr qemuCaps,
                                      qemuMonitorPtr mon)
{
    struct virQEMUCapsStringFlags *entry;
4125 4126
    virJSONValuePtr schemareply;
    virHashTablePtr schema = NULL;
4127 4128
    size_t i;

4129
    if (!(schemareply = qemuMonitorQueryQMPSchema(mon)))
4130 4131
        return -1;

4132 4133 4134 4135
    if (!(schema = virQEMUQAPISchemaConvert(schemareply)))
        return -1;
    schemareply = NULL;

4136 4137 4138
    for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsQMPSchemaQueries); i++) {
        entry = virQEMUCapsQMPSchemaQueries + i;

4139
        if (virQEMUQAPISchemaPathExists(entry->value, schema))
4140 4141 4142 4143 4144 4145 4146
            virQEMUCapsSet(qemuCaps, entry->flag);
    }

    virHashFree(schema);
    return 0;
}

J
Ján Tomko 已提交
4147
#define QEMU_MIN_MAJOR 1
J
Ján Tomko 已提交
4148
#define QEMU_MIN_MINOR 5
J
Ján Tomko 已提交
4149
#define QEMU_MIN_MICRO 0
4150

4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162
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",
4163
                  virGetLastErrorMessage());
4164 4165 4166 4167 4168 4169 4170 4171
        ret = 0;
        goto cleanup;
    }

    if (qemuMonitorGetVersion(mon,
                              &major, &minor, &micro,
                              &package) < 0) {
        VIR_DEBUG("Failed to query monitor version %s",
4172
                  virGetLastErrorMessage());
4173 4174 4175 4176 4177 4178 4179
        ret = 0;
        goto cleanup;
    }

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

J
Ján Tomko 已提交
4180 4181 4182 4183 4184 4185
    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);
4186 4187 4188 4189
        goto cleanup;
    }

    qemuCaps->version = major * 1000000 + minor * 1000 + micro;
4190
    qemuCaps->package = package;
4191 4192
    qemuCaps->usedQMP = true;

4193
    if (virQEMUCapsInitQMPArch(qemuCaps, mon) < 0)
4194 4195
        goto cleanup;

4196 4197
    virQEMUCapsInitQMPBasicArch(qemuCaps);

4198 4199 4200
    if (qemuCaps->version >= 1006000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_DEVICE_VIDEO_PRIMARY);

4201 4202 4203 4204
    /* vmport option is supported v2.2.0 onwards */
    if (qemuCaps->version >= 2002000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_VMPORT_OPT);

4205 4206 4207 4208 4209 4210
    /* -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);

4211 4212 4213
    /* 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
4214
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_VHOSTUSER_MULTIQUEUE);
4215

M
Michal Privoznik 已提交
4216 4217 4218 4219
    /* smm option is supported from v2.4.0 */
    if (qemuCaps->version >= 2004000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_SMM_OPT);

4220 4221 4222 4223
    /* sdl -gl option is supported from v2.4.0 (qemu commit id 0b71a5d5) */
    if (qemuCaps->version >= 2004000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_SDL_GL);

4224 4225 4226 4227
    /* Since 2.4.50 ARM virt machine supports gic-version option */
    if (qemuCaps->version >= 2004050)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACH_VIRT_GIC_VERSION);

4228 4229 4230 4231
    /* no way to query if -machine kernel_irqchip supports split */
    if (qemuCaps->version >= 2006000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_KERNEL_IRQCHIP_SPLIT);

4232 4233 4234 4235 4236 4237 4238
    /* 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);
    }

4239 4240 4241 4242 4243
    /* '-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);

4244 4245 4246 4247
    /* no way to query for -numa dist */
    if (qemuCaps->version >= 2010000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_NUMA_DIST);

4248 4249 4250 4251 4252 4253
    /* 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);
    }

4254 4255
    if (virQEMUCapsProbeQMPCommands(qemuCaps, mon) < 0)
        goto cleanup;
J
Jiri Denemark 已提交
4256 4257 4258 4259 4260

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

4261 4262
    if (virQEMUCapsProbeQMPEvents(qemuCaps, mon) < 0)
        goto cleanup;
4263
    if (virQEMUCapsProbeQMPDevices(qemuCaps, mon) < 0)
4264 4265 4266
        goto cleanup;
    if (virQEMUCapsProbeQMPMachineTypes(qemuCaps, mon) < 0)
        goto cleanup;
4267
    if (virQEMUCapsProbeQMPCPUDefinitions(qemuCaps, mon, false) < 0)
4268 4269 4270 4271 4272
        goto cleanup;
    if (virQEMUCapsProbeQMPTPM(qemuCaps, mon) < 0)
        goto cleanup;
    if (virQEMUCapsProbeQMPCommandLine(qemuCaps, mon) < 0)
        goto cleanup;
4273 4274
    if (virQEMUCapsProbeQMPMigrationCapabilities(qemuCaps, mon) < 0)
        goto cleanup;
4275 4276 4277
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_QMP_SCHEMA) &&
        virQEMUCapsProbeQMPSchemaCapabilities(qemuCaps, mon) < 0)
        goto cleanup;
4278
    if (virQEMUCapsProbeQMPHostCPU(qemuCaps, mon, false) < 0)
4279
        goto cleanup;
4280

4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291
    /* '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
4292 4293
    /* GIC capabilities, eg. available GIC versions */
    if ((qemuCaps->arch == VIR_ARCH_AARCH64 ||
S
Stefan Schallenberg 已提交
4294
         qemuCaps->arch == VIR_ARCH_ARMV6L ||
A
Andrea Bolognani 已提交
4295 4296 4297 4298
         qemuCaps->arch == VIR_ARCH_ARMV7L) &&
        virQEMUCapsProbeQMPGICCapabilities(qemuCaps, mon) < 0)
        goto cleanup;

4299 4300 4301 4302 4303 4304 4305
    /* 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);

4306 4307 4308 4309
    if (ARCH_IS_X86(qemuCaps->arch) &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_CPU_MODEL_EXPANSION))
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_CPU_CACHE);

4310 4311 4312 4313 4314
    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);
4315 4316
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CCW_CSSID_UNRESTRICTED))
            virQEMUCapsClear(qemuCaps, QEMU_CAPS_DEVICE_VFIO_CCW);
4317 4318
    }

4319 4320
    /* Probe for SEV capabilities */
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_SEV_GUEST)) {
4321 4322 4323 4324 4325 4326
        int rc = virQEMUCapsProbeQMPSEVCapabilities(qemuCaps, mon);

        if (rc < 0)
            goto cleanup;

        if (rc == 0)
4327 4328 4329
            virQEMUCapsClear(qemuCaps, QEMU_CAPS_SEV_GUEST);
    }

4330 4331 4332 4333 4334
    /* To avoid guest ABI regression, blockdev shall be enabled only when
     * we are able to pass the custom 'device_id' for SCSI disks and cdroms. */
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SCSI_DISK_DEVICE_ID))
        virQEMUCapsClear(qemuCaps, QEMU_CAPS_BLOCKDEV);

4335
    ret = 0;
4336
 cleanup:
4337 4338 4339
    return ret;
}

4340

4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353
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;
    }

4354 4355 4356
    if (virQEMUCapsProbeQMPCPUDefinitions(qemuCaps, mon, true) < 0)
        goto cleanup;

4357 4358 4359
    if (virQEMUCapsProbeQMPHostCPU(qemuCaps, mon, true) < 0)
        goto cleanup;

4360 4361 4362 4363 4364 4365
    ret = 0;
 cleanup:
    return ret;
}


4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377
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;
4378
    virDomainChrSourceDef config;
4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396
    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)
4397
        unlink(cmd->monpath);
4398 4399 4400 4401 4402 4403

    virDomainObjEndAPI(&cmd->vm);

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

4404
        VIR_DEBUG("Killing QMP caps process %lld", (long long)cmd->pid);
4405 4406
        if (virProcessKill(cmd->pid, SIGKILL) < 0 && errno != ESRCH)
            VIR_ERROR(_("Failed to kill process %lld: %s"),
4407
                      (long long)cmd->pid,
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 4449 4450
                      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;
4451

4452 4453 4454
    /* the ".sock" sufix is important to avoid a possible clash with a qemu
     * domain called "capabilities"
     */
4455 4456 4457 4458 4459
    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;
4460

4461 4462
    /* ".pidfile" suffix is used rather than ".pid" to avoid a possible clash
     * with a qemu domain called "capabilities"
4463 4464 4465
     * 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
4466
     */
4467 4468
    if (virAsprintf(&cmd->pidfile, "%s/%s", libDir, "capabilities.pidfile") < 0)
        goto error;
4469

4470
    virPidFileForceCleanupPath(cmd->pidfile);
4471

4472 4473 4474
    cmd->config.type = VIR_DOMAIN_CHR_TYPE_UNIX;
    cmd->config.data.nix.path = cmd->monpath;
    cmd->config.data.nix.listen = false;
4475

4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488
    return cmd;

 error:
    virQEMUCapsInitQMPCommandFree(cmd);
    return NULL;
}


/* Returns -1 on fatal error,
 *          0 on success,
 *          1 when probing QEMU failed
 */
static int
4489 4490
virQEMUCapsInitQMPCommandRun(virQEMUCapsInitQMPCommandPtr cmd,
                             bool forceTCG)
4491 4492
{
    virDomainXMLOptionPtr xmlopt = NULL;
4493
    const char *machine;
4494 4495 4496
    int status = 0;
    int ret = -1;

4497 4498 4499 4500 4501 4502 4503
    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);
4504

4505 4506 4507 4508 4509 4510 4511
    /*
     * 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.
     */
4512 4513 4514 4515 4516
    cmd->cmd = virCommandNewArgList(cmd->binary,
                                    "-S",
                                    "-no-user-config",
                                    "-nodefaults",
                                    "-nographic",
4517
                                    "-machine", machine,
4518 4519 4520 4521 4522 4523
                                    "-qmp", cmd->monarg,
                                    "-pidfile", cmd->pidfile,
                                    "-daemonize",
                                    NULL);
    virCommandAddEnvPassCommon(cmd->cmd);
    virCommandClearCaps(cmd->cmd);
4524 4525 4526 4527

#if WITH_CAPNG
    /* QEMU might run into permission issues, e.g. /dev/sev (0600), override
     * them just for the purpose of probing */
4528 4529
    if (geteuid() == 0)
        virCommandAllowCap(cmd->cmd, CAP_DAC_OVERRIDE);
4530 4531
#endif

4532 4533 4534 4535
    virCommandSetGID(cmd->cmd, cmd->runGid);
    virCommandSetUID(cmd->cmd, cmd->runUid);

    virCommandSetErrorBuffer(cmd->cmd, cmd->qmperr);
4536

4537
    /* Log, but otherwise ignore, non-zero status.  */
4538
    if (virCommandRun(cmd->cmd, &status) < 0)
4539 4540 4541
        goto cleanup;

    if (status != 0) {
4542
        VIR_DEBUG("QEMU %s exited with status %d: %s",
4543 4544
                  cmd->binary, status, *cmd->qmperr);
        goto ignore;
4545 4546
    }

4547 4548 4549
    if (virPidFileReadPath(cmd->pidfile, &cmd->pid) < 0) {
        VIR_DEBUG("Failed to read pidfile %s", cmd->pidfile);
        goto ignore;
4550 4551
    }

4552
    if (!(xmlopt = virDomainXMLOptionNew(NULL, NULL, NULL, NULL, NULL)) ||
4553
        !(cmd->vm = virDomainObjNew(xmlopt)))
4554 4555
        goto cleanup;

4556
    cmd->vm->pid = cmd->pid;
4557

4558
    if (!(cmd->mon = qemuMonitorOpen(cmd->vm, &cmd->config, true, true,
4559
                                     0, &callbacks, NULL)))
4560
        goto ignore;
4561

4562
    virObjectLock(cmd->mon);
4563 4564 4565

    ret = 0;

4566
 cleanup:
4567 4568
    if (!cmd->mon)
        virQEMUCapsInitQMPCommandAbort(cmd);
4569
    virObjectUnref(xmlopt);
4570

4571
    return ret;
4572

4573 4574 4575 4576
 ignore:
    ret = 1;
    goto cleanup;
}
4577

4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593

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;

4594
    if ((rc = virQEMUCapsInitQMPCommandRun(cmd, false)) != 0) {
4595 4596 4597
        if (rc == 1)
            ret = 0;
        goto cleanup;
4598
    }
4599 4600 4601 4602

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

4603 4604 4605 4606 4607 4608 4609 4610 4611 4612 4613 4614
    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;
    }

4615 4616 4617 4618
    ret = 0;

 cleanup:
    virQEMUCapsInitQMPCommandFree(cmd);
4619 4620 4621 4622
    return ret;
}


4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633
#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 },
    };

4634
    virLogMessage(&virLogSelf,
4635 4636 4637 4638
                  VIR_LOG_WARN,
                  __FILE__, __LINE__, __func__,
                  meta,
                  _("Failed to probe capabilities for %s: %s"),
4639
                  binary, virGetLastErrorMessage());
4640 4641 4642
}


4643
virQEMUCapsPtr
4644
virQEMUCapsNewForBinaryInternal(virArch hostArch,
4645
                                const char *binary,
4646 4647 4648
                                const char *libDir,
                                uid_t runUid,
                                gid_t runGid,
4649
                                unsigned int microcodeVersion,
J
Ján Tomko 已提交
4650
                                const char *kernelVersion)
4651
{
4652
    virQEMUCapsPtr qemuCaps;
4653
    struct stat sb;
4654
    char *qmperr = NULL;
4655

4656 4657 4658
    if (!(qemuCaps = virQEMUCapsNew()))
        goto error;

4659 4660
    if (VIR_STRDUP(qemuCaps->binary, binary) < 0)
        goto error;
4661 4662 4663 4664 4665 4666 4667 4668

    /* 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;
    }
4669
    qemuCaps->ctime = sb.st_ctime;
4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680

    /* 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;
    }

4681 4682
    if (virQEMUCapsInitQMP(qemuCaps, libDir, runUid, runGid, &qmperr) < 0) {
        virQEMUCapsLogProbeFailure(binary);
4683
        goto error;
4684
    }
4685

J
Ján Tomko 已提交
4686
    if (!qemuCaps->usedQMP) {
4687 4688 4689 4690 4691 4692
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Failed to probe QEMU binary with QMP: %s"),
                       qmperr ? qmperr : _("unknown error"));
        virQEMUCapsLogProbeFailure(binary);
        goto error;
    }
4693

4694 4695
    qemuCaps->libvirtCtime = virGetSelfLastChanged();
    qemuCaps->libvirtVersion = LIBVIR_VERSION_NUMBER;
4696

4697 4698
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_QEMU);
4699

4700
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
4701 4702
        qemuCaps->microcodeVersion = microcodeVersion;

4703 4704
        if (VIR_STRDUP(qemuCaps->kernelVersion, kernelVersion) < 0)
            goto error;
4705 4706

        qemuCaps->kvmSupportsNesting = virQEMUCapsKVMSupportsNesting();
4707 4708
    }

4709
 cleanup:
4710
    VIR_FREE(qmperr);
4711
    return qemuCaps;
4712

4713
 error:
4714 4715
    virObjectUnref(qemuCaps);
    qemuCaps = NULL;
4716
    goto cleanup;
4717 4718
}

4719 4720 4721
static void *
virQEMUCapsNewData(const char *binary,
                   void *privData)
4722
{
4723 4724 4725 4726 4727 4728 4729
    virQEMUCapsCachePrivPtr priv = privData;

    return virQEMUCapsNewForBinaryInternal(priv->hostArch,
                                           binary,
                                           priv->libDir,
                                           priv->runUid,
                                           priv->runGid,
4730
                                           priv->microcodeVersion,
J
Ján Tomko 已提交
4731
                                           priv->kernelVersion);
4732
}
4733 4734


4735 4736 4737 4738 4739 4740 4741
static void *
virQEMUCapsLoadFile(const char *filename,
                    const char *binary,
                    void *privData)
{
    virQEMUCapsPtr qemuCaps = virQEMUCapsNew();
    virQEMUCapsCachePrivPtr priv = privData;
4742

4743 4744
    if (!qemuCaps)
        return NULL;
4745

4746 4747 4748 4749 4750 4751 4752
    if (VIR_STRDUP(qemuCaps->binary, binary) < 0)
        goto error;

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

 cleanup:
4753 4754 4755 4756
    return qemuCaps;

 error:
    virObjectUnref(qemuCaps);
4757 4758
    qemuCaps = NULL;
    goto cleanup;
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 4793 4794
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]);
    }

4795 4796
    if (!virQEMUCapsGetMachineHotplugCpus(qemuCaps, machineType))
        virQEMUCapsClear(qemuCaps, QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS);
4797 4798 4799
}


4800 4801 4802 4803 4804 4805 4806 4807 4808 4809
virFileCacheHandlers qemuCapsCacheHandlers = {
    .isValid = virQEMUCapsIsValid,
    .newData = virQEMUCapsNewData,
    .loadFile = virQEMUCapsLoadFile,
    .saveFile = virQEMUCapsSaveFile,
    .privFree = virQEMUCapsCachePrivFree,
};


virFileCachePtr
4810
virQEMUCapsCacheNew(const char *libDir,
4811
                    const char *cacheDir,
4812
                    uid_t runUid,
4813 4814
                    gid_t runGid,
                    unsigned int microcodeVersion)
4815
{
4816 4817 4818
    char *capsCacheDir = NULL;
    virFileCachePtr cache = NULL;
    virQEMUCapsCachePrivPtr priv = NULL;
4819
    struct utsname uts;
4820

4821
    if (virAsprintf(&capsCacheDir, "%s/capabilities", cacheDir) < 0)
4822
        goto error;
4823 4824

    if (!(cache = virFileCacheNew(capsCacheDir, "xml", &qemuCapsCacheHandlers)))
4825
        goto error;
4826

4827
    if (VIR_ALLOC(priv) < 0)
4828
        goto error;
4829
    virFileCacheSetPriv(cache, priv);
4830

4831
    if (VIR_STRDUP(priv->libDir, libDir) < 0)
4832 4833
        goto error;

4834
    priv->hostArch = virArchFromHost();
4835

4836 4837
    priv->runUid = runUid;
    priv->runGid = runGid;
4838
    priv->microcodeVersion = microcodeVersion;
4839
    priv->kvmUsable = VIR_TRISTATE_BOOL_ABSENT;
4840

4841 4842 4843 4844
    if (uname(&uts) == 0 &&
        virAsprintf(&priv->kernelVersion, "%s %s", uts.release, uts.version) < 0)
        goto error;

4845 4846
 cleanup:
    VIR_FREE(capsCacheDir);
4847 4848
    return cache;

4849
 error:
4850 4851 4852
    virObjectUnref(cache);
    cache = NULL;
    goto cleanup;
4853 4854 4855
}


4856
virQEMUCapsPtr
4857
virQEMUCapsCacheLookup(virFileCachePtr cache,
4858
                       const char *binary)
4859
{
4860
    virQEMUCapsPtr ret = NULL;
4861

4862
    ret = virFileCacheLookup(cache, binary);
4863 4864

    VIR_DEBUG("Returning caps %p for %s", ret, binary);
4865 4866 4867 4868
    return ret;
}


4869
virQEMUCapsPtr
4870
virQEMUCapsCacheLookupCopy(virFileCachePtr cache,
4871
                           const char *binary,
4872
                           const char *machineType)
4873
{
4874
    virQEMUCapsPtr qemuCaps = virQEMUCapsCacheLookup(cache, binary);
4875
    virQEMUCapsPtr ret;
4876

4877
    if (!qemuCaps)
4878 4879
        return NULL;

4880 4881
    ret = virQEMUCapsNewCopy(qemuCaps);
    virObjectUnref(qemuCaps);
4882 4883 4884 4885

    if (!ret)
        return NULL;

4886
    virQEMUCapsFilterByMachineType(ret, machineType);
4887 4888 4889 4890
    return ret;
}


4891 4892 4893 4894 4895
static int
virQEMUCapsCompareArch(const void *payload,
                       const void *name ATTRIBUTE_UNUSED,
                       const void *opaque)
{
4896
    struct virQEMUCapsSearchData *data = (struct virQEMUCapsSearchData *)opaque;
4897 4898
    const virQEMUCaps *qemuCaps = payload;

4899 4900 4901 4902 4903 4904 4905 4906 4907
    if (qemuCaps->arch != data->arch)
        return false;

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

    return true;
4908 4909 4910 4911
}


virQEMUCapsPtr
4912
virQEMUCapsCacheLookupByArch(virFileCachePtr cache,
4913 4914 4915
                             virArch arch)
{
    virQEMUCapsPtr ret = NULL;
4916 4917 4918 4919
    const char *binaryFilters[] = {
        "qemu-system-",
        NULL,
    };
4920 4921 4922 4923
    virArch archs[] = {
        arch,
        virQEMUCapsFindTarget(virArchFromHost(), arch),
    };
4924
    size_t i;
4925 4926
    size_t j;

4927 4928 4929 4930 4931 4932
    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],
            };
4933

4934 4935 4936 4937
            ret = virFileCacheLookupByFunc(cache, virQEMUCapsCompareArch, &data);
            if (ret)
                goto done;
        }
4938 4939
    }

4940 4941 4942 4943 4944
    virReportError(VIR_ERR_INVALID_ARG,
                   _("unable to find any emulator to serve '%s' "
                     "architecture"), virArchToString(arch));

 done:
4945 4946
    VIR_DEBUG("Returning caps %p for arch %s", ret, virArchToString(arch));

4947 4948 4949 4950
    return ret;
}


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 5035 5036
/**
 * 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 {
5037
        machine = virQEMUCapsGetPreferredMachine(qemuCaps);
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 5067 5068
    }

    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;
}

5069 5070 5071 5072 5073 5074 5075
bool
virQEMUCapsSupportsVmport(virQEMUCapsPtr qemuCaps,
                          const virDomainDef *def)
{
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_VMPORT_OPT))
        return false;

5076 5077
    return qemuDomainIsI440FX(def) ||
        qemuDomainIsQ35(def) ||
5078 5079 5080 5081
        STREQ(def->os.machine, "isapc");
}


5082 5083 5084 5085 5086 5087 5088
bool
virQEMUCapsIsMachineSupported(virQEMUCapsPtr qemuCaps,
                              const char *canonical_machine)
{
    size_t i;

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
5089
        if (STREQ(canonical_machine, qemuCaps->machineTypes[i].name))
5090 5091 5092 5093
            return true;
    }
    return false;
}
5094 5095


5096 5097 5098 5099
/*
 * The preferred machine to use if none is listed explicitly
 * Note that this may differ from QEMU's own default machine
 */
5100
const char *
5101
virQEMUCapsGetPreferredMachine(virQEMUCapsPtr qemuCaps)
5102 5103 5104
{
    if (!qemuCaps->nmachineTypes)
        return NULL;
5105
    return qemuCaps->machineTypes[0].name;
5106
}
5107 5108


5109
static int
5110
virQEMUCapsFillDomainLoaderCaps(virDomainCapsLoaderPtr capsLoader,
5111 5112
                                virFirmwarePtr *firmwares,
                                size_t nfirmwares)
5113
{
5114 5115
    size_t i;

5116
    capsLoader->supported = true;
5117

5118
    if (VIR_ALLOC_N(capsLoader->values.values, nfirmwares) < 0)
5119 5120
        return -1;

5121 5122
    for (i = 0; i < nfirmwares; i++) {
        const char *filename = firmwares[i]->name;
5123 5124 5125 5126 5127 5128

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

5129
        if (VIR_STRDUP(capsLoader->values.values[capsLoader->values.nvalues],
5130 5131
                       filename) < 0)
            return -1;
5132
        capsLoader->values.nvalues++;
5133 5134
    }

5135
    VIR_DOMAIN_CAPS_ENUM_SET(capsLoader->type,
5136 5137
                             VIR_DOMAIN_LOADER_TYPE_ROM);

5138 5139
    VIR_DOMAIN_CAPS_ENUM_SET(capsLoader->type,
                             VIR_DOMAIN_LOADER_TYPE_PFLASH);
5140 5141


5142 5143 5144
    VIR_DOMAIN_CAPS_ENUM_SET(capsLoader->readonly,
                             VIR_TRISTATE_BOOL_YES,
                             VIR_TRISTATE_BOOL_NO);
5145
    return 0;
5146 5147 5148
}


5149
static int
5150
virQEMUCapsFillDomainOSCaps(virDomainCapsOSPtr os,
5151 5152
                            virFirmwarePtr *firmwares,
                            size_t nfirmwares)
5153
{
5154
    virDomainCapsLoaderPtr capsLoader = &os->loader;
5155

5156
    os->supported = true;
5157
    if (virQEMUCapsFillDomainLoaderCaps(capsLoader, firmwares, nfirmwares) < 0)
5158 5159
        return -1;
    return 0;
5160 5161 5162
}


5163 5164 5165 5166 5167
static int
virQEMUCapsFillDomainCPUCaps(virCapsPtr caps,
                             virQEMUCapsPtr qemuCaps,
                             virDomainCapsPtr domCaps)
{
5168 5169
    if (virQEMUCapsIsCPUModeSupported(qemuCaps, caps, domCaps->virttype,
                                      VIR_CPU_MODE_HOST_PASSTHROUGH))
5170 5171
        domCaps->cpu.hostPassthrough = true;

5172
    if (virQEMUCapsIsCPUModeSupported(qemuCaps, caps, domCaps->virttype,
5173
                                      VIR_CPU_MODE_HOST_MODEL)) {
5174 5175
        virCPUDefPtr cpu = virQEMUCapsGetHostModel(qemuCaps, domCaps->virttype,
                                                   VIR_QEMU_CAPS_HOST_CPU_REPORTED);
5176 5177
        domCaps->cpu.hostModel = virCPUDefCopy(cpu);
    }
5178 5179 5180 5181 5182

    if (virQEMUCapsIsCPUModeSupported(qemuCaps, caps, domCaps->virttype,
                                      VIR_CPU_MODE_CUSTOM)) {
        virDomainCapsCPUModelsPtr filtered = NULL;
        char **models = NULL;
5183
        const char *blacklist[] = { "host", NULL };
5184

J
Jiri Denemark 已提交
5185
        if (virCPUGetModels(domCaps->arch, &models) >= 0) {
5186 5187 5188 5189 5190 5191 5192 5193
            virDomainCapsCPUModelsPtr cpus;

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

            filtered = virDomainCapsCPUModelsFilter(cpus,
5194
                                                    (const char **)models,
5195
                                                    blacklist);
5196
            virStringListFree(models);
5197 5198
        }
        domCaps->cpu.custom = filtered;
5199
    }
5200 5201 5202 5203 5204

    return 0;
}


5205 5206 5207 5208 5209 5210 5211 5212 5213 5214
static int
virQEMUCapsFillDomainIOThreadCaps(virQEMUCapsPtr qemuCaps,
                                  virDomainCapsPtr domCaps)
{
    domCaps->iothreads = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_IOTHREAD);

    return 0;
}


5215
static int
5216
virQEMUCapsFillDomainDeviceDiskCaps(virQEMUCapsPtr qemuCaps,
5217
                                    const char *machine,
5218 5219
                                    virDomainCapsDeviceDiskPtr disk)
{
5220
    disk->supported = true;
5221 5222 5223
    /* QEMU supports all of these */
    VIR_DOMAIN_CAPS_ENUM_SET(disk->diskDevice,
                             VIR_DOMAIN_DISK_DEVICE_DISK,
5224 5225
                             VIR_DOMAIN_DISK_DEVICE_CDROM,
                             VIR_DOMAIN_DISK_DEVICE_LUN);
5226 5227

    /* PowerPC pseries based VMs do not support floppy device */
5228
    if (!qemuDomainMachineIsPSeries(machine, qemuCaps->arch))
5229
        VIR_DOMAIN_CAPS_ENUM_SET(disk->diskDevice, VIR_DOMAIN_DISK_DEVICE_FLOPPY);
5230

5231 5232 5233
    if (qemuDomainMachineHasBuiltinIDE(machine))
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_IDE);

5234 5235 5236 5237 5238
    VIR_DOMAIN_CAPS_ENUM_SET(disk->bus,
                             VIR_DOMAIN_DISK_BUS_SCSI,
                             VIR_DOMAIN_DISK_BUS_VIRTIO,
                             /* VIR_DOMAIN_DISK_BUS_SD */);

5239
    /* PowerPC pseries based VMs do not support floppy device */
5240
    if (!qemuDomainMachineIsPSeries(machine, qemuCaps->arch))
5241 5242
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_FDC);

5243 5244
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_USB_STORAGE))
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_USB);
5245 5246 5247 5248

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

5249
    return 0;
5250 5251 5252
}


5253 5254 5255 5256 5257 5258
static int
virQEMUCapsFillDomainDeviceGraphicsCaps(virQEMUCapsPtr qemuCaps,
                                        virDomainCapsDeviceGraphicsPtr dev)
{
    dev->supported = true;

J
Ján Tomko 已提交
5259
    VIR_DOMAIN_CAPS_ENUM_SET(dev->type, VIR_DOMAIN_GRAPHICS_TYPE_SDL);
5260 5261 5262 5263 5264 5265 5266 5267 5268
    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;
}


5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280
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);
5281
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_QXL))
5282 5283 5284 5285 5286 5287 5288 5289
        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;
}


5290
static int
5291 5292 5293 5294 5295 5296
virQEMUCapsFillDomainDeviceHostdevCaps(virQEMUCapsPtr qemuCaps,
                                       virDomainCapsDeviceHostdevPtr hostdev)
{
    bool supportsPassthroughKVM = qemuHostdevHostSupportsPassthroughLegacy();
    bool supportsPassthroughVFIO = qemuHostdevHostSupportsPassthroughVFIO();

5297
    hostdev->supported = true;
5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309
    /* 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,
5310 5311
                             VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI,
                             VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI);
5312 5313 5314 5315 5316 5317 5318 5319 5320 5321 5322 5323

    /* 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
5324
    if (supportsPassthroughKVM) {
5325 5326 5327 5328
        VIR_DOMAIN_CAPS_ENUM_SET(hostdev->pciBackend,
                                 VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT,
                                 VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM);
    }
5329
    return 0;
5330 5331 5332
}


5333 5334 5335 5336 5337 5338 5339
/**
 * virQEMUCapsSupportsGICVersion:
 * @qemuCaps: QEMU capabilities
 * @virtType: domain type
 * @version: GIC version
 *
 * Checks the QEMU binary with capabilities @qemuCaps supports a specific
5340 5341
 * GIC version for a domain of type @virtType. If @qemuCaps is NULL, the GIC
 * @version is considered unsupported.
5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352
 *
 * Returns: true if the binary supports the requested GIC version, false
 *          otherwise
 */
bool
virQEMUCapsSupportsGICVersion(virQEMUCapsPtr qemuCaps,
                              virDomainVirtType virtType,
                              virGICVersion version)
{
    size_t i;

5353 5354 5355
    if (!qemuCaps)
        return false;

5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367 5368 5369 5370 5371 5372 5373 5374
    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;
}


5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396 5397 5398
/**
 * 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;
5399
    virGICVersion version;
5400

5401
    if (!qemuDomainMachineIsARMVirt(domCaps->machine, domCaps->arch))
5402 5403
        return 0;

5404 5405 5406 5407 5408 5409
    for (version = VIR_GIC_VERSION_LAST - 1;
         version > VIR_GIC_VERSION_NONE;
         version--) {
        if (!virQEMUCapsSupportsGICVersion(qemuCaps,
                                           domCaps->virttype,
                                           version))
5410 5411 5412 5413
            continue;

        gic->supported = true;
        VIR_DOMAIN_CAPS_ENUM_SET(gic->version,
5414
                                 version);
5415 5416 5417 5418 5419 5420
    }

    return 0;
}


5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436
/**
 * 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;
5437
    VIR_AUTOPTR(virSEVCapability) sev = NULL;
5438 5439 5440 5441 5442 5443 5444 5445

    if (!cap)
        return 0;

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

    if (VIR_STRDUP(sev->pdh, cap->pdh) < 0)
5446
        return -1;
5447 5448

    if (VIR_STRDUP(sev->cert_chain, cap->cert_chain) < 0)
5449
        return -1;
5450 5451 5452 5453 5454

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

5455
    return 0;
5456 5457 5458
}


5459
int
5460 5461
virQEMUCapsFillDomainCaps(virCapsPtr caps,
                          virDomainCapsPtr domCaps,
5462
                          virQEMUCapsPtr qemuCaps,
5463
                          virFirmwarePtr *firmwares,
5464
                          size_t nfirmwares)
5465
{
5466
    virDomainCapsOSPtr os = &domCaps->os;
5467 5468
    virDomainCapsDeviceDiskPtr disk = &domCaps->disk;
    virDomainCapsDeviceHostdevPtr hostdev = &domCaps->hostdev;
5469
    virDomainCapsDeviceGraphicsPtr graphics = &domCaps->graphics;
5470
    virDomainCapsDeviceVideoPtr video = &domCaps->video;
5471

5472 5473
    domCaps->maxvcpus = virQEMUCapsGetMachineMaxCpus(qemuCaps,
                                                     domCaps->machine);
5474
    if (domCaps->virttype == VIR_DOMAIN_VIRT_KVM) {
5475 5476 5477 5478 5479 5480
        int hostmaxvcpus;

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

        domCaps->maxvcpus = MIN(domCaps->maxvcpus, hostmaxvcpus);
5481
    }
5482

5483 5484 5485
    domCaps->vmcoreinfo = virQEMUCapsGet(qemuCaps,
                                         QEMU_CAPS_DEVICE_VMCOREINFO);

5486 5487
    domCaps->genid = virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VMGENID);

5488
    if (virQEMUCapsFillDomainOSCaps(os, firmwares, nfirmwares) < 0 ||
5489
        virQEMUCapsFillDomainCPUCaps(caps, qemuCaps, domCaps) < 0 ||
5490
        virQEMUCapsFillDomainIOThreadCaps(qemuCaps, domCaps) < 0 ||
5491 5492 5493
        virQEMUCapsFillDomainDeviceDiskCaps(qemuCaps,
                                            domCaps->machine, disk) < 0 ||
        virQEMUCapsFillDomainDeviceGraphicsCaps(qemuCaps, graphics) < 0 ||
5494
        virQEMUCapsFillDomainDeviceVideoCaps(qemuCaps, video) < 0 ||
5495
        virQEMUCapsFillDomainDeviceHostdevCaps(qemuCaps, hostdev) < 0 ||
5496 5497
        virQEMUCapsFillDomainFeatureGICCaps(qemuCaps, domCaps) < 0 ||
        virQEMUCapsFillDomainFeatureSEVCaps(qemuCaps, domCaps) < 0)
5498
        return -1;
5499

5500
    return 0;
5501
}
5502 5503 5504 5505 5506 5507 5508 5509


void
virQEMUCapsSetMicrocodeVersion(virQEMUCapsPtr qemuCaps,
                               unsigned int microcodeVersion)
{
    qemuCaps->microcodeVersion = microcodeVersion;
}
5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524 5525 5526


/**
 * 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);
}