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",
527 528
    );

529

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

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

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

568
    bool usedQMP;
569
    bool kvmSupportsNesting;
570

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

575
    virBitmapPtr flags;
576 577 578

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

584
    virArch arch;
585

586 587
    virDomainCapsCPUModelsPtr kvmCPUModels;
    virDomainCapsCPUModelsPtr tcgCPUModels;
588 589

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

    size_t ngicCapabilities;
    virGICCapability *gicCapabilities;
594

595 596
    virSEVCapability *sevCapabilities;

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

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

606

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

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

    return 0;
}

618
VIR_ONCE_GLOBAL_INIT(virQEMUCaps);
619

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

    return virArchFromString(arch);
}


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

    return virArchToString(arch);
}

645 646 647

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

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

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

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

    return false;
}


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

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

    return guestarch;
}
690

691

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

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

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

705

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

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

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

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

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

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

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

751
 out:
752 753 754
    return ret;
}

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

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

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

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

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

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

    return ret;
}

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

807 808 809
    if (!binary)
        return 0;

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

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

    machines = NULL;
    nmachines = 0;

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

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

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

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

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

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

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

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

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

    ret = 0;

878
 cleanup:
879 880 881

    virCapabilitiesFreeMachines(machines, nmachines);

882
    return ret;
883 884 885
}


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


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

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

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

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

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

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

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

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

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

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

    return caps;

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


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


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

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

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

999
struct virQEMUCapsStringFlags virQEMUCapsObjectTypes[] = {
1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014
    { "hda-duplex", QEMU_CAPS_HDA_DUPLEX },
    { "hda-micro", QEMU_CAPS_HDA_MICRO },
    { "ccid-card-emulated", QEMU_CAPS_CCID_EMULATED },
    { "ccid-card-passthru", QEMU_CAPS_CCID_PASSTHRU },
    { "piix3-usb-uhci", QEMU_CAPS_PIIX3_USB_UHCI },
    { "piix4-usb-uhci", QEMU_CAPS_PIIX4_USB_UHCI },
    { "usb-ehci", QEMU_CAPS_USB_EHCI },
    { "ich9-usb-ehci1", QEMU_CAPS_ICH9_USB_EHCI1 },
    { "vt82c686b-usb-uhci", QEMU_CAPS_VT82C686B_USB_UHCI },
    { "pci-ohci", QEMU_CAPS_PCI_OHCI },
    { "nec-usb-xhci", QEMU_CAPS_NEC_USB_XHCI },
    { "usb-redir", QEMU_CAPS_USB_REDIR },
    { "usb-hub", QEMU_CAPS_USB_HUB },
    { "ich9-ahci", QEMU_CAPS_ICH9_AHCI },
    { "virtio-blk-s390", QEMU_CAPS_VIRTIO_S390 },
1015
    { "virtio-blk-ccw", QEMU_CAPS_VIRTIO_CCW },
1016
    { "sclpconsole", QEMU_CAPS_DEVICE_SCLPCONSOLE },
1017
    { "lsi53c895a", QEMU_CAPS_SCSI_LSI },
1018
    { "virtio-scsi-pci", QEMU_CAPS_VIRTIO_SCSI },
1019 1020
    { "virtio-scsi-s390", QEMU_CAPS_VIRTIO_SCSI },
    { "virtio-scsi-ccw", QEMU_CAPS_VIRTIO_SCSI },
1021
    { "virtio-scsi-device", QEMU_CAPS_VIRTIO_SCSI },
1022
    { "megasas", QEMU_CAPS_SCSI_MEGASAS },
1023
    { "qxl", QEMU_CAPS_DEVICE_QXL },
1024 1025 1026 1027
    { "sga", QEMU_CAPS_SGA },
    { "scsi-block", QEMU_CAPS_SCSI_BLOCK },
    { "scsi-cd", QEMU_CAPS_SCSI_CD },
    { "ide-cd", QEMU_CAPS_IDE_CD },
1028 1029 1030
    { "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
1031
    { "usb-serial", QEMU_CAPS_DEVICE_USB_SERIAL },
1032
    { "virtio-rng-pci", QEMU_CAPS_DEVICE_VIRTIO_RNG },
1033 1034
    { "virtio-rng-s390", QEMU_CAPS_DEVICE_VIRTIO_RNG },
    { "virtio-rng-ccw", QEMU_CAPS_DEVICE_VIRTIO_RNG },
1035
    { "virtio-rng-device", QEMU_CAPS_DEVICE_VIRTIO_RNG },
1036
    { "rng-random", QEMU_CAPS_OBJECT_RNG_RANDOM },
1037
    { "rng-egd", QEMU_CAPS_OBJECT_RNG_EGD },
1038
    { "spapr-nvram", QEMU_CAPS_DEVICE_NVRAM },
1039
    { "pci-bridge", QEMU_CAPS_DEVICE_PCI_BRIDGE },
1040
    { "vfio-pci", QEMU_CAPS_DEVICE_VFIO_PCI },
1041
    { "i82801b11-bridge", QEMU_CAPS_DEVICE_DMI_TO_PCI_BRIDGE },
1042
    { "usb-storage", QEMU_CAPS_DEVICE_USB_STORAGE },
1043
    { "virtio-mmio", QEMU_CAPS_DEVICE_VIRTIO_MMIO },
1044
    { "ich9-intel-hda", QEMU_CAPS_DEVICE_ICH9_INTEL_HDA },
H
Hu Tao 已提交
1045
    { "pvpanic", QEMU_CAPS_DEVICE_PANIC },
L
Li Zhang 已提交
1046
    { "usb-kbd", QEMU_CAPS_DEVICE_USB_KBD },
1047
    { "memory-backend-ram", QEMU_CAPS_OBJECT_MEMORY_RAM },
1048
    { "memory-backend-file", QEMU_CAPS_OBJECT_MEMORY_FILE },
1049
    { "usb-audio", QEMU_CAPS_OBJECT_USB_AUDIO },
J
John Ferlan 已提交
1050
    { "iothread", QEMU_CAPS_OBJECT_IOTHREAD},
1051
    { "ivshmem", QEMU_CAPS_DEVICE_IVSHMEM },
1052
    { "pc-dimm", QEMU_CAPS_DEVICE_PC_DIMM },
M
Michal Privoznik 已提交
1053
    { "pci-serial", QEMU_CAPS_DEVICE_PCI_SERIAL },
1054
    { "gpex-pcihost", QEMU_CAPS_OBJECT_GPEX},
1055
    { "ioh3420", QEMU_CAPS_DEVICE_IOH3420 },
1056
    { "x3130-upstream", QEMU_CAPS_DEVICE_X3130_UPSTREAM },
1057
    { "xio3130-downstream", QEMU_CAPS_DEVICE_XIO3130_DOWNSTREAM },
1058
    { "rtl8139", QEMU_CAPS_DEVICE_RTL8139 },
1059
    { "e1000", QEMU_CAPS_DEVICE_E1000 },
1060 1061 1062 1063
    { "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
1064 1065
    { "virtio-gpu-pci", QEMU_CAPS_DEVICE_VIRTIO_GPU },
    { "virtio-gpu-device", QEMU_CAPS_DEVICE_VIRTIO_GPU },
1066
    { "virtio-vga", QEMU_CAPS_DEVICE_VIRTIO_VGA },
1067 1068 1069 1070 1071 1072
    { "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 },
1073 1074
    { "virtio-input-host-device", QEMU_CAPS_VIRTIO_INPUT_HOST },
    { "virtio-input-host-pci", QEMU_CAPS_VIRTIO_INPUT_HOST },
1075
    { "mptsas1068", QEMU_CAPS_SCSI_MPTSAS1068 },
1076
    { "secret", QEMU_CAPS_OBJECT_SECRET },
1077
    { "pxb", QEMU_CAPS_DEVICE_PXB },
1078
    { "pxb-pcie", QEMU_CAPS_DEVICE_PXB_PCIE },
1079
    { "tls-creds-x509", QEMU_CAPS_OBJECT_TLS_CREDS_X509 },
J
Ján Tomko 已提交
1080
    { "intel-iommu", QEMU_CAPS_DEVICE_INTEL_IOMMU },
1081 1082
    { "ivshmem-plain", QEMU_CAPS_DEVICE_IVSHMEM_PLAIN },
    { "ivshmem-doorbell", QEMU_CAPS_DEVICE_IVSHMEM_DOORBELL },
1083
    { "vhost-scsi", QEMU_CAPS_DEVICE_VHOST_SCSI },
1084
    { "nvdimm", QEMU_CAPS_DEVICE_NVDIMM },
1085
    { "pcie-root-port", QEMU_CAPS_DEVICE_PCIE_ROOT_PORT },
1086
    { "qemu-xhci", QEMU_CAPS_DEVICE_QEMU_XHCI },
1087
    { "spapr-pci-host-bridge", QEMU_CAPS_DEVICE_SPAPR_PCI_HOST_BRIDGE },
M
Marc-André Lureau 已提交
1088
    { "vmcoreinfo", QEMU_CAPS_DEVICE_VMCOREINFO },
1089
    { "spapr-vty", QEMU_CAPS_DEVICE_SPAPR_VTY },
1090
    { "sclplmconsole", QEMU_CAPS_DEVICE_SCLPLMCONSOLE },
1091
    { "isa-serial", QEMU_CAPS_DEVICE_ISA_SERIAL },
1092
    { "pl011", QEMU_CAPS_DEVICE_PL011 },
1093
    { "virtio-gpu-ccw", QEMU_CAPS_DEVICE_VIRTIO_GPU_CCW },
1094 1095 1096
    { "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 },
1097
    { "pcie-pci-bridge", QEMU_CAPS_DEVICE_PCIE_PCI_BRIDGE },
1098
    { "pr-manager-helper", QEMU_CAPS_PR_MANAGER_HELPER },
1099
    { "virtual-css-bridge", QEMU_CAPS_CCW },
1100
    { "vfio-ccw", QEMU_CAPS_DEVICE_VFIO_CCW },
1101
    { "hda-output", QEMU_CAPS_HDA_OUTPUT },
1102
    { "vmgenid", QEMU_CAPS_DEVICE_VMGENID },
1103
    { "vhost-vsock-device", QEMU_CAPS_DEVICE_VHOST_VSOCK },
1104
    { "mch", QEMU_CAPS_DEVICE_MCH },
1105
    { "sev-guest", QEMU_CAPS_SEV_GUEST },
B
Boris Fiuczynski 已提交
1106
    { "vfio-ap", QEMU_CAPS_DEVICE_VFIO_AP },
Y
Yi Min Zhao 已提交
1107
    { "zpci", QEMU_CAPS_DEVICE_ZPCI },
1108
    { "memory-backend-memfd", QEMU_CAPS_OBJECT_MEMORY_MEMFD },
1109 1110
};

1111
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioBalloon[] = {
1112
    { "deflate-on-oom", QEMU_CAPS_VIRTIO_BALLOON_AUTODEFLATE },
1113 1114 1115
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1116 1117
};

1118
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioBlk[] = {
1119 1120 1121 1122
    { "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 },
1123
    { "num-queues", QEMU_CAPS_VIRTIO_BLK_NUM_QUEUES },
1124
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1125 1126 1127
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1128
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1129 1130
};

1131
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioNet[] = {
1132 1133
    { "tx", QEMU_CAPS_VIRTIO_TX_ALG },
    { "event_idx", QEMU_CAPS_VIRTIO_NET_EVENT_IDX },
1134
    { "rx_queue_size", QEMU_CAPS_VIRTIO_NET_RX_QUEUE_SIZE },
1135
    { "tx_queue_size", QEMU_CAPS_VIRTIO_NET_TX_QUEUE_SIZE },
1136
    { "host_mtu", QEMU_CAPS_VIRTIO_NET_HOST_MTU },
1137 1138 1139
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1140 1141
};

1142
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsSpaprPCIHostBridge[] = {
1143 1144 1145
    { "numa_node", QEMU_CAPS_SPAPR_PCI_HOST_BRIDGE_NUMA_NODE },
};

1146
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioSCSI[] = {
1147
    { "iothread", QEMU_CAPS_VIRTIO_SCSI_IOTHREAD },
1148 1149 1150
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1151 1152
};

1153
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVfioPCI[] = {
1154
    { "display", QEMU_CAPS_VFIO_PCI_DISPLAY },
1155 1156
};

1157
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsSCSIDisk[] = {
1158 1159
    { "channel", QEMU_CAPS_SCSI_DISK_CHANNEL },
    { "wwn", QEMU_CAPS_SCSI_DISK_WWN },
1160
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1161
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1162 1163
};

1164
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsIDEDrive[] = {
1165
    { "wwn", QEMU_CAPS_IDE_DRIVE_WWN },
1166
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1167
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1168 1169
};

1170
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsPiix4PM[] = {
1171 1172
    { "disable_s3", QEMU_CAPS_PIIX_DISABLE_S3 },
    { "disable_s4", QEMU_CAPS_PIIX_DISABLE_S4 },
1173 1174
};

1175
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsUSBRedir[] = {
1176
    { "filter", QEMU_CAPS_USB_REDIR_FILTER },
1177 1178
};

1179
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsI440FXPCIHost[] = {
1180 1181 1182
    { "pci-hole64-size", QEMU_CAPS_I440FX_PCI_HOLE64_SIZE },
};

1183
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsQ35PCIHost[] = {
1184 1185 1186
    { "pci-hole64-size", QEMU_CAPS_Q35_PCI_HOLE64_SIZE },
};

1187
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsUSBStorage[] = {
1188
    { "removable", QEMU_CAPS_USB_STORAGE_REMOVABLE },
1189
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1190
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1191
    { "werror", QEMU_CAPS_USB_STORAGE_WERROR },
1192 1193
};

1194
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsKVMPit[] = {
1195 1196 1197
    { "lost_tick_policy", QEMU_CAPS_KVM_PIT_TICK_POLICY },
};

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

1202
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVmwareSvga[] = {
1203 1204 1205
    { "vgamem_mb", QEMU_CAPS_VMWARE_SVGA_VGAMEM },
};

1206
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsQxl[] = {
1207
    { "vgamem_mb", QEMU_CAPS_QXL_VGAMEM },
1208
    { "vram64_size_mb", QEMU_CAPS_QXL_VRAM64 },
1209
    { "max_outputs", QEMU_CAPS_QXL_MAX_OUTPUTS },
1210 1211
};

1212
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioGpu[] = {
1213
    { "virgl", QEMU_CAPS_VIRTIO_GPU_VIRGL },
1214
    { "max_outputs", QEMU_CAPS_VIRTIO_GPU_MAX_OUTPUTS },
1215 1216 1217
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1218 1219
};

1220
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsICH9[] = {
1221 1222 1223 1224
    { "disable_s3", QEMU_CAPS_ICH9_DISABLE_S3 },
    { "disable_s4", QEMU_CAPS_ICH9_DISABLE_S4 },
};

1225
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsUSBNECXHCI[] = {
1226 1227 1228
    { "p3", QEMU_CAPS_NEC_USB_XHCI_PORTS },
};

1229
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsIntelIOMMU[] = {
1230
    { "intremap", QEMU_CAPS_INTEL_IOMMU_INTREMAP },
1231
    { "caching-mode", QEMU_CAPS_INTEL_IOMMU_CACHING_MODE },
1232
    { "eim", QEMU_CAPS_INTEL_IOMMU_EIM },
1233
    { "device-iotlb", QEMU_CAPS_INTEL_IOMMU_DEVICE_IOTLB },
1234 1235
};

1236 1237 1238 1239
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsVirtualCSSBridge[] = {
    { "cssid-unrestricted", QEMU_CAPS_CCW_CSSID_UNRESTRICTED },
};

1240 1241 1242 1243
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsMCH[] = {
    { "extended-tseg-mbytes", QEMU_CAPS_MCH_EXTENDED_TSEG_MBYTES },
};

1244 1245 1246 1247
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsNVDIMM[] = {
    { "unarmed", QEMU_CAPS_DEVICE_NVDIMM_UNARMED },
};

1248
/* see documentation for virQEMUQAPISchemaPathGet for the query format */
1249
static struct virQEMUCapsStringFlags virQEMUCapsQMPSchemaQueries[] = {
1250
    { "blockdev-add/arg-type/options/+gluster/debug-level", QEMU_CAPS_GLUSTER_DEBUG_LEVEL},
1251
    { "blockdev-add/arg-type/+gluster/debug", QEMU_CAPS_GLUSTER_DEBUG_LEVEL},
J
John Ferlan 已提交
1252
    { "blockdev-add/arg-type/+vxhs", QEMU_CAPS_VXHS},
1253
    { "blockdev-add/arg-type/+iscsi/password-secret", QEMU_CAPS_ISCSI_PASSWORD_SECRET },
1254
    { "blockdev-add/arg-type/+qcow2/encrypt/+luks/key-secret", QEMU_CAPS_QCOW2_LUKS },
1255
    { "nbd-server-start/arg-type/tls-creds", QEMU_CAPS_NBD_TLS },
1256
    { "screendump/arg-type/device", QEMU_CAPS_SCREENDUMP_DEVICE },
1257
    { "block-commit/arg-type/*top",  QEMU_CAPS_ACTIVE_COMMIT },
1258
    { "query-iothreads/ret-type/poll-max-ns", QEMU_CAPS_IOTHREAD_POLLING },
1259
    { "query-display-options/ret-type/+egl-headless/rendernode", QEMU_CAPS_EGL_HEADLESS_RENDERNODE },
1260 1261
};

1262 1263
typedef struct _virQEMUCapsObjectTypeProps virQEMUCapsObjectTypeProps;
struct _virQEMUCapsObjectTypeProps {
1264
    const char *type;
1265
    struct virQEMUCapsStringFlags *props;
1266
    size_t nprops;
1267
    int capsCondition;
1268 1269
};

1270 1271 1272 1273 1274
typedef int (*virQEMUCapsObjectTypePropsCB)(qemuMonitorPtr mon,
                                            const char *type,
                                            char ***props);

static virQEMUCapsObjectTypeProps virQEMUCapsDeviceProps[] = {
1275 1276
    { "virtio-blk-pci", virQEMUCapsDevicePropsVirtioBlk,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBlk),
1277
      -1 },
1278 1279
    { "virtio-net-pci", virQEMUCapsDevicePropsVirtioNet,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioNet),
1280
      QEMU_CAPS_DEVICE_VIRTIO_NET },
1281 1282
    { "virtio-scsi-pci", virQEMUCapsDevicePropsVirtioSCSI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioSCSI),
1283
      QEMU_CAPS_VIRTIO_SCSI },
1284 1285
    { "virtio-blk-ccw", virQEMUCapsDevicePropsVirtioBlk,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBlk),
1286
      QEMU_CAPS_VIRTIO_CCW },
1287 1288
    { "virtio-net-ccw", virQEMUCapsDevicePropsVirtioNet,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioNet),
1289
      QEMU_CAPS_DEVICE_VIRTIO_NET },
1290 1291
    { "virtio-scsi-ccw", virQEMUCapsDevicePropsVirtioSCSI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioSCSI),
1292
      QEMU_CAPS_VIRTIO_SCSI },
1293 1294
    { "virtio-blk-s390", virQEMUCapsDevicePropsVirtioBlk,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBlk),
1295
      QEMU_CAPS_VIRTIO_S390 },
1296 1297
    { "virtio-net-s390", virQEMUCapsDevicePropsVirtioNet,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioNet),
1298
      QEMU_CAPS_DEVICE_VIRTIO_NET },
1299 1300
    { "vfio-pci", virQEMUCapsDevicePropsVfioPCI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVfioPCI),
1301
      QEMU_CAPS_DEVICE_VFIO_PCI },
1302 1303
    { "scsi-disk", virQEMUCapsDevicePropsSCSIDisk,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsSCSIDisk),
1304
      -1 },
1305 1306
    { "ide-drive", virQEMUCapsDevicePropsIDEDrive,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsIDEDrive),
1307
      -1 },
1308 1309
    { "PIIX4_PM", virQEMUCapsDevicePropsPiix4PM,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsPiix4PM),
1310
      -1 },
1311 1312
    { "usb-redir", virQEMUCapsDevicePropsUSBRedir,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsUSBRedir),
1313
      QEMU_CAPS_USB_REDIR },
1314 1315
    { "i440FX-pcihost", virQEMUCapsDevicePropsI440FXPCIHost,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsI440FXPCIHost),
1316
      -1 },
1317 1318
    { "q35-pcihost", virQEMUCapsDevicePropsQ35PCIHost,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsQ35PCIHost),
1319
      -1 },
1320 1321
    { "usb-storage", virQEMUCapsDevicePropsUSBStorage,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsUSBStorage),
1322
      QEMU_CAPS_DEVICE_USB_STORAGE },
1323 1324
    { "kvm-pit", virQEMUCapsDevicePropsKVMPit,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsKVMPit),
1325
      -1 },
1326 1327
    { "VGA", virQEMUCapsDevicePropsVGA,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVGA),
1328
      QEMU_CAPS_DEVICE_VGA },
1329 1330
    { "vmware-svga", virQEMUCapsDevicePropsVmwareSvga,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVmwareSvga),
1331
      QEMU_CAPS_DEVICE_VMWARE_SVGA },
1332 1333
    { "qxl", virQEMUCapsDevicePropsQxl,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsQxl),
1334
      QEMU_CAPS_DEVICE_QXL },
1335 1336
    { "virtio-gpu-pci", virQEMUCapsDevicePropsVirtioGpu,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioGpu),
1337
      QEMU_CAPS_DEVICE_VIRTIO_GPU },
1338 1339
    { "virtio-gpu-device", virQEMUCapsDevicePropsVirtioGpu,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioGpu),
1340
      QEMU_CAPS_DEVICE_VIRTIO_GPU },
1341 1342
    { "ICH9-LPC", virQEMUCapsDevicePropsICH9,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsICH9),
1343
      -1 },
1344 1345
    { "virtio-balloon-pci", virQEMUCapsDevicePropsVirtioBalloon,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBalloon),
1346
      -1 },
1347 1348
    { "virtio-balloon-ccw", virQEMUCapsDevicePropsVirtioBalloon,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBalloon),
1349
      -1 },
1350 1351
    { "virtio-balloon-device", virQEMUCapsDevicePropsVirtioBalloon,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBalloon),
1352
      -1 },
1353 1354
    { "nec-usb-xhci", virQEMUCapsDevicePropsUSBNECXHCI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsUSBNECXHCI),
1355
      QEMU_CAPS_NEC_USB_XHCI },
1356 1357
    { "intel-iommu", virQEMUCapsDevicePropsIntelIOMMU,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsIntelIOMMU),
1358
      QEMU_CAPS_DEVICE_INTEL_IOMMU },
1359 1360
    { "spapr-pci-host-bridge", virQEMUCapsDevicePropsSpaprPCIHostBridge,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsSpaprPCIHostBridge),
1361
      QEMU_CAPS_DEVICE_SPAPR_PCI_HOST_BRIDGE },
1362 1363
    { "virtio-gpu-ccw", virQEMUCapsDevicePropsVirtioGpu,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioGpu),
1364
      QEMU_CAPS_DEVICE_VIRTIO_GPU_CCW },
1365 1366 1367
    { "virtual-css-bridge", virQEMUCapsObjectPropsVirtualCSSBridge,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsVirtualCSSBridge),
      QEMU_CAPS_CCW },
1368 1369 1370
    { "mch", virQEMUCapsDevicePropsMCH,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsMCH),
      QEMU_CAPS_DEVICE_MCH },
1371 1372 1373
    { "nvdimm", virQEMUCapsDevicePropsNVDIMM,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsNVDIMM),
      QEMU_CAPS_DEVICE_NVDIMM },
1374 1375
};

1376 1377
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsMemoryBackendFile[] = {
    { "discard-data", QEMU_CAPS_OBJECT_MEMORY_FILE_DISCARD },
1378
    { "align", QEMU_CAPS_OBJECT_MEMORY_FILE_ALIGN },
1379
    { "pmem", QEMU_CAPS_OBJECT_MEMORY_FILE_PMEM },
1380 1381
};

1382 1383 1384 1385
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsMemoryBackendMemfd[] = {
    { "hugetlb", QEMU_CAPS_OBJECT_MEMORY_MEMFD_HUGETLB },
};

1386 1387
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsSPAPRMachine[] = {
    { "cap-hpt-max-page-size", QEMU_CAPS_MACHINE_PSERIES_CAP_HPT_MAX_PAGE_SIZE },
1388
    { "cap-htm", QEMU_CAPS_MACHINE_PSERIES_CAP_HTM },
1389
    { "cap-nested-hv", QEMU_CAPS_MACHINE_PSERIES_CAP_NESTED_HV },
1390 1391
};

1392 1393 1394 1395
static virQEMUCapsObjectTypeProps virQEMUCapsObjectProps[] = {
    { "memory-backend-file", virQEMUCapsObjectPropsMemoryBackendFile,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsMemoryBackendFile),
      QEMU_CAPS_OBJECT_MEMORY_FILE },
1396 1397 1398
    { "memory-backend-memfd", virQEMUCapsObjectPropsMemoryBackendMemfd,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsMemoryBackendMemfd),
      QEMU_CAPS_OBJECT_MEMORY_MEMFD },
1399 1400 1401
    { "spapr-machine", virQEMUCapsObjectPropsSPAPRMachine,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsSPAPRMachine),
      -1 },
1402
};
1403 1404

static void
1405 1406 1407 1408 1409
virQEMUCapsProcessStringFlags(virQEMUCapsPtr qemuCaps,
                              size_t nflags,
                              struct virQEMUCapsStringFlags *flags,
                              size_t nvalues,
                              char *const*values)
1410 1411
{
    size_t i, j;
1412
    for (i = 0; i < nflags; i++) {
1413 1414 1415
        if (virQEMUCapsGet(qemuCaps, flags[i].flag))
            continue;

1416
        for (j = 0; j < nvalues; j++) {
1417
            if (STREQ(values[j], flags[i].value)) {
1418
                virQEMUCapsSet(qemuCaps, flags[i].flag);
1419 1420 1421 1422 1423 1424 1425
                break;
            }
        }
    }
}


1426
int virQEMUCapsGetDefaultVersion(virCapsPtr caps,
1427
                                 virFileCachePtr capsCache,
1428
                                 unsigned int *version)
1429
{
1430
    virQEMUCapsPtr qemucaps;
T
Tal Kain 已提交
1431
    virArch hostarch;
1432
    virCapsDomainDataPtr capsdata;
1433 1434 1435 1436

    if (*version > 0)
        return 0;

T
Tal Kain 已提交
1437
    hostarch = virArchFromHost();
1438 1439 1440
    if (!(capsdata = virCapabilitiesDomainDataLookup(caps,
            VIR_DOMAIN_OSTYPE_HVM, hostarch, VIR_DOMAIN_VIRT_QEMU,
            NULL, NULL))) {
1441
        virReportError(VIR_ERR_INTERNAL_ERROR,
1442
                       _("Cannot find suitable emulator for %s"),
T
Tal Kain 已提交
1443
                       virArchToString(hostarch));
1444 1445 1446
        return -1;
    }

1447
    qemucaps = virQEMUCapsCacheLookup(capsCache, capsdata->emulator);
1448 1449
    VIR_FREE(capsdata);
    if (!qemucaps)
1450 1451
        return -1;

1452
    *version = virQEMUCapsGetVersion(qemucaps);
1453
    virObjectUnref(qemucaps);
1454 1455
    return 0;
}
1456 1457


1458 1459


1460 1461
virQEMUCapsPtr
virQEMUCapsNew(void)
1462
{
1463
    virQEMUCapsPtr qemuCaps;
1464

1465
    if (virQEMUCapsInitialize() < 0)
1466 1467
        return NULL;

1468
    if (!(qemuCaps = virObjectNew(virQEMUCapsClass)))
1469 1470
        return NULL;

1471
    if (!(qemuCaps->flags = virBitmapNew(QEMU_CAPS_LAST)))
1472
        goto error;
1473

1474
    return qemuCaps;
1475

1476
 error:
1477
    virObjectUnref(qemuCaps);
1478
    return NULL;
1479 1480 1481
}


1482
static int
1483 1484
virQEMUCapsHostCPUDataCopy(virQEMUCapsHostCPUDataPtr dst,
                           virQEMUCapsHostCPUDataPtr src)
1485
{
1486 1487
    if (src->info &&
        !(dst->info = qemuMonitorCPUModelInfoCopy(src->info)))
1488 1489
        return -1;

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

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

1498 1499 1500 1501
    if (src->full &&
        !(dst->full = virCPUDefCopy(src->full)))
        return -1;

1502 1503 1504 1505
    return 0;
}


1506
static void
1507
virQEMUCapsHostCPUDataClear(virQEMUCapsHostCPUDataPtr cpuData)
1508
{
1509
    qemuMonitorCPUModelInfoFree(cpuData->info);
1510
    virCPUDefFree(cpuData->reported);
1511
    virCPUDefFree(cpuData->migratable);
1512
    virCPUDefFree(cpuData->full);
1513 1514

    memset(cpuData, 0, sizeof(*cpuData));
1515 1516 1517
}


1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536
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;
}


1537
virQEMUCapsPtr virQEMUCapsNewCopy(virQEMUCapsPtr qemuCaps)
1538
{
1539
    virQEMUCapsPtr ret = virQEMUCapsNew();
1540 1541 1542 1543 1544
    size_t i;

    if (!ret)
        return NULL;

1545
    ret->usedQMP = qemuCaps->usedQMP;
1546
    ret->kvmSupportsNesting = qemuCaps->kvmSupportsNesting;
1547

1548 1549 1550 1551 1552
    if (VIR_STRDUP(ret->binary, qemuCaps->binary) < 0)
        goto error;

    ret->ctime = qemuCaps->ctime;

1553
    virBitmapCopy(ret->flags, qemuCaps->flags);
1554

1555 1556
    ret->version = qemuCaps->version;
    ret->kvmVersion = qemuCaps->kvmVersion;
1557
    ret->microcodeVersion = qemuCaps->microcodeVersion;
1558 1559 1560 1561

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

1562 1563 1564
    if (VIR_STRDUP(ret->kernelVersion, qemuCaps->kernelVersion) < 0)
        goto error;

1565
    ret->arch = qemuCaps->arch;
1566

1567 1568 1569 1570 1571 1572 1573 1574 1575
    if (qemuCaps->kvmCPUModels) {
        ret->kvmCPUModels = virDomainCapsCPUModelsCopy(qemuCaps->kvmCPUModels);
        if (!ret->kvmCPUModels)
            goto error;
    }

    if (qemuCaps->tcgCPUModels) {
        ret->tcgCPUModels = virDomainCapsCPUModelsCopy(qemuCaps->tcgCPUModels);
        if (!ret->tcgCPUModels)
1576
            goto error;
1577 1578
    }

1579 1580
    if (virQEMUCapsHostCPUDataCopy(&ret->kvmCPU, &qemuCaps->kvmCPU) < 0 ||
        virQEMUCapsHostCPUDataCopy(&ret->tcgCPU, &qemuCaps->tcgCPU) < 0)
1581 1582
        goto error;

1583
    if (VIR_ALLOC_N(ret->machineTypes, qemuCaps->nmachineTypes) < 0)
1584
        goto error;
1585
    ret->nmachineTypes = qemuCaps->nmachineTypes;
1586
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
1587 1588
        if (VIR_STRDUP(ret->machineTypes[i].name, qemuCaps->machineTypes[i].name) < 0 ||
            VIR_STRDUP(ret->machineTypes[i].alias, qemuCaps->machineTypes[i].alias) < 0)
1589
            goto error;
1590
        ret->machineTypes[i].maxCpus = qemuCaps->machineTypes[i].maxCpus;
1591
        ret->machineTypes[i].hotplugCpus = qemuCaps->machineTypes[i].hotplugCpus;
1592 1593
        ret->machineTypes[i].qemuDefault = qemuCaps->machineTypes[i].qemuDefault;

1594 1595
    }

1596 1597 1598 1599 1600 1601
    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];

1602 1603 1604 1605 1606
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_SEV_GUEST) &&
        virQEMUCapsSEVInfoCopy(&ret->sevCapabilities,
                               qemuCaps->sevCapabilities) < 0)
        goto error;

1607 1608
    return ret;

1609
 error:
1610 1611 1612 1613 1614
    virObjectUnref(ret);
    return NULL;
}


1615
void virQEMUCapsDispose(void *obj)
1616
{
1617
    virQEMUCapsPtr qemuCaps = obj;
1618 1619
    size_t i;

1620
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
1621 1622
        VIR_FREE(qemuCaps->machineTypes[i].name);
        VIR_FREE(qemuCaps->machineTypes[i].alias);
1623
    }
1624
    VIR_FREE(qemuCaps->machineTypes);
1625

1626 1627
    virObjectUnref(qemuCaps->kvmCPUModels);
    virObjectUnref(qemuCaps->tcgCPUModels);
1628

1629
    virBitmapFree(qemuCaps->flags);
1630

1631
    VIR_FREE(qemuCaps->package);
1632
    VIR_FREE(qemuCaps->kernelVersion);
1633
    VIR_FREE(qemuCaps->binary);
A
Andrea Bolognani 已提交
1634 1635

    VIR_FREE(qemuCaps->gicCapabilities);
1636

1637 1638
    virSEVCapabilitiesFree(qemuCaps->sevCapabilities);

1639 1640
    virQEMUCapsHostCPUDataClear(&qemuCaps->kvmCPU);
    virQEMUCapsHostCPUDataClear(&qemuCaps->tcgCPU);
1641 1642
}

1643
void
1644
virQEMUCapsSet(virQEMUCapsPtr qemuCaps,
1645
               virQEMUCapsFlags flag)
1646
{
1647
    ignore_value(virBitmapSetBit(qemuCaps->flags, flag));
1648 1649 1650 1651
}


void
1652
virQEMUCapsSetList(virQEMUCapsPtr qemuCaps, ...)
1653 1654 1655 1656
{
    va_list list;
    int flag;

1657
    va_start(list, qemuCaps);
1658
    while ((flag = va_arg(list, int)) < QEMU_CAPS_LAST)
1659
        ignore_value(virBitmapSetBit(qemuCaps->flags, flag));
1660
    va_end(list);
1661 1662 1663 1664
}


void
1665
virQEMUCapsClear(virQEMUCapsPtr qemuCaps,
1666
                 virQEMUCapsFlags flag)
1667
{
1668
    ignore_value(virBitmapClearBit(qemuCaps->flags, flag));
1669 1670 1671
}


1672
char *virQEMUCapsFlagsString(virQEMUCapsPtr qemuCaps)
1673
{
1674
    return virBitmapToString(qemuCaps->flags, true, false);
1675 1676 1677 1678
}


bool
1679
virQEMUCapsGet(virQEMUCapsPtr qemuCaps,
1680
               virQEMUCapsFlags flag)
1681
{
J
Ján Tomko 已提交
1682
    return qemuCaps && virBitmapIsBitSet(qemuCaps->flags, flag);
1683
}
1684 1685


D
Daniel P. Berrange 已提交
1686
bool virQEMUCapsHasPCIMultiBus(virQEMUCapsPtr qemuCaps,
1687
                               const virDomainDef *def)
D
Daniel P. Berrange 已提交
1688
{
1689 1690
    /* x86_64 and i686 support PCI-multibus on all machine types
     * since forever */
1691
    if (ARCH_IS_X86(def->os.arch))
D
Daniel P. Berrange 已提交
1692 1693 1694
        return true;

    if (def->os.arch == VIR_ARCH_PPC ||
1695
        ARCH_IS_PPC64(def->os.arch)) {
D
Daniel P. Berrange 已提交
1696 1697 1698 1699 1700
        /*
         * Usage of pci.0 naming:
         *
         *    ref405ep: no pci
         *       taihu: no pci
1701
         *      bamboo: 1.1.0 (<= 1.5.0, so basically forever)
D
Daniel P. Berrange 已提交
1702 1703
         *       mac99: 2.0.0
         *     g3beige: 2.0.0
1704
         *        prep: 1.4.0 (<= 1.5.0, so basically forever)
D
Daniel P. Berrange 已提交
1705 1706 1707 1708 1709 1710
         *     pseries: 2.0.0
         *   mpc8544ds: forever
         * virtex-m507: no pci
         *     ppce500: 1.6.0
         */

1711 1712 1713 1714 1715 1716 1717 1718
        /* 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
1719 1720 1721 1722 1723 1724 1725
        if (qemuCaps->version >= 2000000)
            return true;

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

1726 1727 1728
        if (STREQ(def->os.machine, "bamboo") ||
            STREQ(def->os.machine, "mpc8544ds") ||
            STREQ(def->os.machine, "prep")) {
D
Daniel P. Berrange 已提交
1729
            return true;
1730
        }
D
Daniel P. Berrange 已提交
1731 1732 1733 1734

        return false;
    }

1735 1736 1737 1738
    /* S390 supports PCI-multibus. */
    if (ARCH_IS_S390(def->os.arch))
        return true;

1739 1740 1741 1742
    /* If the virt machine, both on ARM and RISC-V, supports PCI,
     * then it also supports multibus */
    if (qemuDomainIsARMVirt(def) ||
        qemuDomainIsRISCVVirt(def)) {
1743
        return true;
1744
    }
1745

D
Daniel P. Berrange 已提交
1746 1747 1748 1749
    return false;
}


1750
const char *virQEMUCapsGetBinary(virQEMUCapsPtr qemuCaps)
1751
{
1752
    return qemuCaps->binary;
1753 1754
}

1755 1756 1757 1758 1759 1760 1761 1762 1763

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


1764
virArch virQEMUCapsGetArch(virQEMUCapsPtr qemuCaps)
1765
{
1766
    return qemuCaps->arch;
1767 1768 1769
}


1770
unsigned int virQEMUCapsGetVersion(virQEMUCapsPtr qemuCaps)
1771
{
1772
    return qemuCaps->version;
1773 1774 1775
}


1776
unsigned int virQEMUCapsGetKVMVersion(virQEMUCapsPtr qemuCaps)
1777
{
1778
    return qemuCaps->kvmVersion;
1779 1780 1781
}


1782 1783 1784 1785 1786 1787
const char *virQEMUCapsGetPackage(virQEMUCapsPtr qemuCaps)
{
    return qemuCaps->package;
}


1788 1789
int
virQEMUCapsAddCPUDefinitions(virQEMUCapsPtr qemuCaps,
1790
                             virDomainVirtType type,
1791
                             const char **name,
1792 1793
                             size_t count,
                             virDomainCapsCPUUsable usable)
1794
{
1795
    size_t i;
1796
    virDomainCapsCPUModelsPtr cpus = NULL;
1797

1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811
    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;
    }
1812 1813

    for (i = 0; i < count; i++) {
1814
        if (virDomainCapsCPUModelsAdd(cpus, name[i], -1, usable, NULL) < 0)
1815
            return -1;
1816
    }
1817

1818 1819 1820 1821
    return 0;
}


1822
virDomainCapsCPUModelsPtr
1823
virQEMUCapsGetCPUDefinitions(virQEMUCapsPtr qemuCaps,
1824
                             virDomainVirtType type)
1825
{
1826
    if (type == VIR_DOMAIN_VIRT_KVM)
1827
        return qemuCaps->kvmCPUModels;
1828
    else
1829
        return qemuCaps->tcgCPUModels;
1830 1831 1832
}


1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843
static virQEMUCapsHostCPUDataPtr
virQEMUCapsGetHostCPUData(virQEMUCapsPtr qemuCaps,
                          virDomainVirtType type)
{
    if (type == VIR_DOMAIN_VIRT_KVM)
        return &qemuCaps->kvmCPU;
    else
        return &qemuCaps->tcgCPU;
}


1844
virCPUDefPtr
1845
virQEMUCapsGetHostModel(virQEMUCapsPtr qemuCaps,
1846 1847
                        virDomainVirtType type,
                        virQEMUCapsHostCPUType cpuType)
1848
{
1849 1850
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);

1851 1852 1853
    switch (cpuType) {
    case VIR_QEMU_CAPS_HOST_CPU_REPORTED:
        return cpuData->reported;
1854 1855 1856

    case VIR_QEMU_CAPS_HOST_CPU_MIGRATABLE:
        return cpuData->migratable;
1857 1858 1859 1860 1861

    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;
1862 1863 1864
    }

    return NULL;
1865 1866 1867
}


1868 1869 1870
static void
virQEMUCapsSetHostModel(virQEMUCapsPtr qemuCaps,
                        virDomainVirtType type,
1871
                        virCPUDefPtr reported,
1872 1873
                        virCPUDefPtr migratable,
                        virCPUDefPtr full)
1874
{
1875 1876
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);

1877 1878
    cpuData->reported = reported;
    cpuData->migratable = migratable;
1879
    cpuData->full = full;
1880 1881 1882
}


1883 1884 1885 1886 1887 1888
bool
virQEMUCapsIsCPUModeSupported(virQEMUCapsPtr qemuCaps,
                              virCapsPtr caps,
                              virDomainVirtType type,
                              virCPUMode mode)
{
1889 1890
    virDomainCapsCPUModelsPtr cpus;

1891 1892 1893 1894 1895 1896
    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:
1897 1898
        return !!virQEMUCapsGetHostModel(qemuCaps, type,
                                         VIR_QEMU_CAPS_HOST_CPU_REPORTED);
1899 1900

    case VIR_CPU_MODE_CUSTOM:
1901 1902 1903 1904 1905
        if (type == VIR_DOMAIN_VIRT_KVM)
            cpus = qemuCaps->kvmCPUModels;
        else
            cpus = qemuCaps->tcgCPUModels;
        return cpus && cpus->nmodels > 0;
1906 1907 1908 1909 1910 1911 1912 1913 1914

    case VIR_CPU_MODE_LAST:
        break;
    }

    return false;
}


1915 1916 1917
int virQEMUCapsGetMachineTypesCaps(virQEMUCapsPtr qemuCaps,
                                   size_t *nmachines,
                                   virCapsGuestMachinePtr **machines)
1918 1919 1920 1921
{
    size_t i;

    *machines = NULL;
1922
    *nmachines = qemuCaps->nmachineTypes;
1923

1924 1925 1926 1927
    if (*nmachines &&
        VIR_ALLOC_N(*machines, qemuCaps->nmachineTypes) < 0)
        goto error;

1928
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
1929 1930
        virCapsGuestMachinePtr mach;
        if (VIR_ALLOC(mach) < 0)
1931
            goto error;
1932
        (*machines)[i] = mach;
1933 1934 1935
        if (qemuCaps->machineTypes[i].alias) {
            if (VIR_STRDUP(mach->name, qemuCaps->machineTypes[i].alias) < 0 ||
                VIR_STRDUP(mach->canonical, qemuCaps->machineTypes[i].name) < 0)
1936
                goto error;
1937
        } else {
1938
            if (VIR_STRDUP(mach->name, qemuCaps->machineTypes[i].name) < 0)
1939
                goto error;
1940
        }
1941
        mach->maxCpus = qemuCaps->machineTypes[i].maxCpus;
1942 1943
    }

1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981
    /* 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++;
    }

1982 1983
    return 0;

1984
 error:
1985 1986 1987 1988 1989 1990 1991
    virCapabilitiesFreeMachines(*machines, *nmachines);
    *nmachines = 0;
    *machines = NULL;
    return -1;
}


1992 1993 1994 1995 1996 1997 1998 1999
/**
 * 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.
 */
2000 2001
const char *virQEMUCapsGetCanonicalMachine(virQEMUCapsPtr qemuCaps,
                                           const char *name)
2002 2003 2004
{
    size_t i;

2005 2006
    if (!name || !qemuCaps)
        return name;
2007

2008
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
2009
        if (!qemuCaps->machineTypes[i].alias)
2010
            continue;
2011 2012
        if (STREQ(qemuCaps->machineTypes[i].alias, name))
            return qemuCaps->machineTypes[i].name;
2013 2014 2015 2016
    }

    return name;
}
2017

2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029
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;
}
2030

2031 2032 2033 2034 2035 2036 2037 2038 2039
int virQEMUCapsGetMachineMaxCpus(virQEMUCapsPtr qemuCaps,
                                 const char *name)
{
    size_t i;

    if (!name)
        return 0;

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
2040
        if (!qemuCaps->machineTypes[i].maxCpus)
2041
            continue;
2042 2043
        if (STREQ(qemuCaps->machineTypes[i].name, name))
            return qemuCaps->machineTypes[i].maxCpus;
2044 2045 2046 2047 2048 2049
    }

    return 0;
}


2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063
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;
}


2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087
/**
 * 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;
}


2088 2089 2090 2091 2092 2093 2094
virSEVCapabilityPtr
virQEMUCapsGetSEVCapabilities(virQEMUCapsPtr qemuCaps)
{
    return qemuCaps->sevCapabilities;
}


2095
static int
2096 2097
virQEMUCapsProbeQMPCommands(virQEMUCapsPtr qemuCaps,
                            qemuMonitorPtr mon)
2098 2099 2100 2101 2102 2103 2104
{
    char **commands = NULL;
    int ncommands;

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

2105 2106 2107 2108
    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsCommands),
                                  virQEMUCapsCommands,
                                  ncommands, commands);
2109
    virStringListFreeCount(commands, ncommands);
2110

2111 2112 2113 2114
    /* 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) &&
2115 2116 2117
        qemuMonitorSupportsActiveCommit(mon))
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_ACTIVE_COMMIT);

2118 2119 2120 2121 2122
    return 0;
}


static int
2123 2124
virQEMUCapsProbeQMPEvents(virQEMUCapsPtr qemuCaps,
                          qemuMonitorPtr mon)
2125 2126 2127 2128 2129 2130 2131
{
    char **events = NULL;
    int nevents;

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

2132 2133 2134 2135
    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsEvents),
                                  virQEMUCapsEvents,
                                  nevents, events);
2136
    virStringListFreeCount(events, nevents);
2137 2138 2139 2140

    return 0;
}

2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169
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;
}
2170

2171
static int
2172
virQEMUCapsProbeQMPDevices(virQEMUCapsPtr qemuCaps,
2173
                           qemuMonitorPtr mon)
2174 2175 2176 2177 2178 2179
{
    int nvalues;
    char **values;

    if ((nvalues = qemuMonitorGetObjectTypes(mon, &values)) < 0)
        return -1;
2180 2181 2182 2183
    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsObjectTypes),
                                  virQEMUCapsObjectTypes,
                                  nvalues, values);
2184
    virStringListFreeCount(values, nvalues);
2185

2186 2187 2188 2189 2190 2191
    if (virQEMUCapsProbeQMPGenericProps(qemuCaps,
                                        mon,
                                        virQEMUCapsDeviceProps,
                                        ARRAY_CARDINALITY(virQEMUCapsDeviceProps),
                                        qemuMonitorGetDeviceProps) < 0)
        return -1;
2192

2193 2194 2195 2196 2197 2198 2199 2200
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_QOM_LIST_PROPERTIES) &&
        virQEMUCapsProbeQMPGenericProps(qemuCaps,
                                        mon,
                                        virQEMUCapsObjectProps,
                                        ARRAY_CARDINALITY(virQEMUCapsObjectProps),
                                        qemuMonitorGetObjectProps) < 0)
        return -1;

2201 2202 2203 2204
    return 0;
}


2205 2206 2207 2208 2209 2210 2211 2212 2213 2214
/* 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.
 */
2215
static const char *preferredMachines[] =
2216
{
2217 2218
    NULL, /* VIR_ARCH_NONE (not a real arch :) */
    "clipper", /* VIR_ARCH_ALPHA */
S
Stefan Schallenberg 已提交
2219
    "integratorcp", /* VIR_ARCH_ARMV6L */
2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246
    "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
2247 2248
    "spike_v1.10", /* VIR_ARCH_RISCV32 */
    "spike_v1.10", /* VIR_ARCH_RISCV64 */
2249 2250 2251
    NULL, /* VIR_ARCH_S390 (no QEMU impl) */
    "s390-ccw-virtio", /* VIR_ARCH_S390X */
    "shix", /* VIR_ARCH_SH4 */
L
Lubomir Rintel 已提交
2252

2253 2254 2255 2256 2257
    "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
2258

2259 2260
    "sim", /* VIR_ARCH_XTENSA */
    "sim", /* VIR_ARCH_XTENSAEB */
2261
};
2262
verify(ARRAY_CARDINALITY(preferredMachines) == VIR_ARCH_LAST);
2263 2264


2265
static int
2266 2267
virQEMUCapsProbeQMPMachineTypes(virQEMUCapsPtr qemuCaps,
                                qemuMonitorPtr mon)
2268 2269 2270 2271 2272
{
    qemuMonitorMachineInfoPtr *machines = NULL;
    int nmachines = 0;
    int ret = -1;
    size_t i;
2273 2274 2275
    ssize_t defIdx = -1;
    ssize_t preferredIdx = -1;
    const char *preferredMachine = preferredMachines[qemuCaps->arch];
2276 2277

    if ((nmachines = qemuMonitorGetMachines(mon, &machines)) < 0)
2278
        return -1;
2279

2280
    if (VIR_ALLOC_N(qemuCaps->machineTypes, nmachines) < 0)
2281 2282
        goto cleanup;

2283
    for (i = 0; i < nmachines; i++) {
2284
        struct virQEMUCapsMachineType *mach;
2285 2286
        if (STREQ(machines[i]->name, "none"))
            continue;
2287 2288 2289 2290 2291

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

        if (VIR_STRDUP(mach->alias, machines[i]->alias) < 0 ||
            VIR_STRDUP(mach->name, machines[i]->name) < 0)
2292
            goto cleanup;
2293 2294

        mach->maxCpus = machines[i]->maxCpus;
2295
        mach->hotplugCpus = machines[i]->hotplugCpus;
2296

2297 2298 2299 2300 2301 2302
        if (preferredMachine &&
            (STREQ_NULLABLE(mach->alias, preferredMachine) ||
             STREQ(mach->name, preferredMachine))) {
            preferredIdx = qemuCaps->nmachineTypes - 1;
        }

2303 2304
        if (machines[i]->isDefault) {
            mach->qemuDefault = true;
2305
            defIdx = qemuCaps->nmachineTypes - 1;
2306
        }
2307
    }
2308

2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321
    /*
     * 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);
2322 2323 2324

    ret = 0;

2325
 cleanup:
2326
    for (i = 0; i < nmachines; i++)
2327 2328 2329 2330 2331 2332
        qemuMonitorMachineInfoFree(machines[i]);
    VIR_FREE(machines);
    return ret;
}


2333 2334
virDomainCapsCPUModelsPtr
virQEMUCapsFetchCPUDefinitions(qemuMonitorPtr mon)
2335
{
2336 2337 2338
    virDomainCapsCPUModelsPtr models = NULL;
    qemuMonitorCPUDefInfoPtr *cpus = NULL;
    int ncpus = 0;
2339
    size_t i;
2340

2341
    if ((ncpus = qemuMonitorGetCPUDefinitions(mon, &cpus)) < 0)
2342
        return NULL;
2343

2344
    if (!(models = virDomainCapsCPUModelsNew(ncpus)))
2345
        goto error;
2346

2347
    for (i = 0; i < ncpus; i++) {
2348 2349 2350 2351 2352 2353 2354
        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;

2355
        if (virDomainCapsCPUModelsAddSteal(models, &cpus[i]->name, usable,
2356
                                           &cpus[i]->blockers) < 0)
2357
            goto error;
2358 2359 2360 2361 2362 2363
    }

 cleanup:
    for (i = 0; i < ncpus; i++)
        qemuMonitorCPUDefInfoFree(cpus[i]);
    VIR_FREE(cpus);
2364 2365 2366 2367 2368 2369
    return models;

 error:
    virObjectUnref(models);
    models = NULL;
    goto cleanup;
2370 2371
}

2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394

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;
}


2395 2396
static int
virQEMUCapsProbeQMPHostCPU(virQEMUCapsPtr qemuCaps,
2397 2398
                           qemuMonitorPtr mon,
                           bool tcg)
2399
{
2400
    qemuMonitorCPUModelInfoPtr modelInfo = NULL;
2401 2402
    qemuMonitorCPUModelInfoPtr nonMigratable = NULL;
    virHashTablePtr hash = NULL;
2403
    const char *model;
2404
    qemuMonitorCPUModelExpansionType type;
2405 2406
    virDomainVirtType virtType;
    virQEMUCapsHostCPUDataPtr cpuData;
2407
    int ret = -1;
2408 2409

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_CPU_MODEL_EXPANSION))
2410 2411
        return 0;

2412
    if (tcg || !virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
2413
        virtType = VIR_DOMAIN_VIRT_QEMU;
2414 2415
        model = "max";
    } else {
2416
        virtType = VIR_DOMAIN_VIRT_KVM;
2417 2418 2419
        model = "host";
    }

2420 2421
    cpuData = virQEMUCapsGetHostCPUData(qemuCaps, virtType);

2422 2423 2424 2425 2426 2427 2428 2429 2430 2431
    /* 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;

2432 2433
    if (qemuMonitorGetCPUModelExpansion(mon, type, model, true, &modelInfo) < 0)
        goto cleanup;
2434 2435

    /* Try to check migratability of each feature. */
2436
    if (modelInfo &&
2437 2438
        qemuMonitorGetCPUModelExpansion(mon, type, model, false,
                                        &nonMigratable) < 0)
2439
        goto cleanup;
2440 2441 2442 2443 2444 2445 2446

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

        if (!(hash = virHashCreate(0, NULL)))
2447
            goto cleanup;
2448

2449 2450
        for (i = 0; i < modelInfo->nprops; i++) {
            prop = modelInfo->props + i;
2451
            if (virHashAddEntry(hash, prop->name, prop) < 0)
2452
                goto cleanup;
2453 2454 2455 2456 2457 2458 2459 2460 2461
        }

        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;

2462
            if (prop->value.boolean) {
2463
                prop->migratable = VIR_TRISTATE_BOOL_YES;
2464 2465 2466 2467
            } else if (nmProp->value.boolean) {
                prop->value.boolean = true;
                prop->migratable = VIR_TRISTATE_BOOL_NO;
            }
2468 2469
        }

2470
        modelInfo->migratability = true;
2471 2472
    }

2473
    VIR_STEAL_PTR(cpuData->info, modelInfo);
2474 2475 2476 2477 2478
    ret = 0;

 cleanup:
    virHashFree(hash);
    qemuMonitorCPUModelInfoFree(nonMigratable);
2479
    qemuMonitorCPUModelInfoFree(modelInfo);
2480 2481

    return ret;
2482 2483
}

2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535

/**
 * 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;
}


2536 2537
struct tpmTypeToCaps {
    int type;
2538
    virQEMUCapsFlags caps;
2539 2540 2541 2542 2543 2544 2545
};

static const struct tpmTypeToCaps virQEMUCapsTPMTypesToCaps[] = {
    {
        .type = VIR_DOMAIN_TPM_TYPE_PASSTHROUGH,
        .caps = QEMU_CAPS_DEVICE_TPM_PASSTHROUGH,
    },
2546 2547 2548 2549
    {
        .type = VIR_DOMAIN_TPM_TYPE_EMULATOR,
        .caps = QEMU_CAPS_DEVICE_TPM_EMULATOR,
    },
2550 2551 2552 2553 2554 2555 2556
};

const struct tpmTypeToCaps virQEMUCapsTPMModelsToCaps[] = {
    {
        .type = VIR_DOMAIN_TPM_MODEL_TIS,
        .caps = QEMU_CAPS_DEVICE_TPM_TIS,
    },
2557 2558 2559 2560
    {
        .type = VIR_DOMAIN_TPM_MODEL_CRB,
        .caps = QEMU_CAPS_DEVICE_TPM_CRB,
    },
2561 2562 2563 2564 2565 2566
};

static int
virQEMUCapsProbeQMPTPM(virQEMUCapsPtr qemuCaps,
                       qemuMonitorPtr mon)
{
2567 2568
    int nentries;
    size_t i;
2569
    char **entries = NULL;
S
Stefan Berger 已提交
2570

2571 2572 2573 2574 2575 2576 2577
    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);
2578
            if (virStringListHasString((const char **)entries, needle))
2579 2580 2581 2582
                virQEMUCapsSet(qemuCaps,
                               virQEMUCapsTPMModelsToCaps[i].caps);
        }
    }
2583
    virStringListFree(entries);
2584 2585 2586 2587 2588 2589 2590 2591

    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);
2592
            if (virStringListHasString((const char **)entries, needle))
2593 2594 2595
                virQEMUCapsSet(qemuCaps, virQEMUCapsTPMTypesToCaps[i].caps);
        }
    }
2596
    virStringListFree(entries);
2597 2598 2599 2600

    return 0;
}

2601

2602
static int
2603 2604
virQEMUCapsProbeQMPKVMState(virQEMUCapsPtr qemuCaps,
                            qemuMonitorPtr mon)
2605 2606 2607 2608 2609 2610 2611
{
    bool enabled = false;
    bool present = false;

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

A
Andrea Bolognani 已提交
2612 2613
    if (present && enabled)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_KVM);
2614 2615 2616 2617

    return 0;
}

2618 2619 2620 2621 2622 2623 2624 2625
struct virQEMUCapsCommandLineProps {
    const char *option;
    const char *param;
    int flag;
};

static struct virQEMUCapsCommandLineProps virQEMUCapsCommandLine[] = {
    { "machine", "mem-merge", QEMU_CAPS_MEM_MERGE },
2626
    { "machine", "vmport", QEMU_CAPS_MACHINE_VMPORT_OPT },
O
Osier Yang 已提交
2627
    { "drive", "discard", QEMU_CAPS_DRIVE_DISCARD },
2628
    { "drive", "detect-zeroes", QEMU_CAPS_DRIVE_DETECT_ZEROES },
2629
    { "realtime", "mlock", QEMU_CAPS_REALTIME_MLOCK },
2630
    { "boot-opts", "strict", QEMU_CAPS_BOOT_STRICT },
2631
    { "boot-opts", "reboot-timeout", QEMU_CAPS_REBOOT_TIMEOUT },
2632
    { "boot-opts", "splash-time", QEMU_CAPS_SPLASH_TIMEOUT },
2633
    { "spice", "disable-agent-file-xfer", QEMU_CAPS_SPICE_FILE_XFER_DISABLE },
2634
    { "msg", "timestamp", QEMU_CAPS_MSG_TIMESTAMP },
2635
    { "numa", NULL, QEMU_CAPS_NUMA },
2636
    { "drive", "throttling.bps-total-max", QEMU_CAPS_DRIVE_IOTUNE_MAX},
2637 2638
    { "machine", "aes-key-wrap", QEMU_CAPS_AES_KEY_WRAP },
    { "machine", "dea-key-wrap", QEMU_CAPS_DEA_KEY_WRAP },
2639
    { "chardev", "append", QEMU_CAPS_CHARDEV_FILE_APPEND },
2640
    { "spice", "gl", QEMU_CAPS_SPICE_GL },
2641
    { "chardev", "logfile", QEMU_CAPS_CHARDEV_LOGFILE },
2642
    { "name", "debug-threads", QEMU_CAPS_NAME_DEBUG_THREADS },
2643
    { "name", "guest", QEMU_CAPS_NAME_GUEST },
2644
    { "spice", "unix", QEMU_CAPS_SPICE_UNIX },
2645
    { "drive", "throttling.bps-total-max-length", QEMU_CAPS_DRIVE_IOTUNE_MAX_LENGTH },
2646
    { "drive", "throttling.group", QEMU_CAPS_DRIVE_IOTUNE_GROUP },
2647
    { "spice", "rendernode", QEMU_CAPS_SPICE_RENDERNODE },
2648
    { "machine", "kernel_irqchip", QEMU_CAPS_MACHINE_KERNEL_IRQCHIP },
2649
    { "machine", "loadparm", QEMU_CAPS_LOADPARM },
2650
    { "vnc", "vnc", QEMU_CAPS_VNC_MULTI_SERVERS },
2651
    { "chardev", "reconnect", QEMU_CAPS_CHARDEV_RECONNECT },
2652
    { "sandbox", "enable", QEMU_CAPS_SECCOMP_SANDBOX },
2653
    { "sandbox", "elevateprivileges", QEMU_CAPS_SECCOMP_BLACKLIST },
2654
    { "chardev", "fd", QEMU_CAPS_CHARDEV_FD_PASS },
2655 2656 2657 2658 2659 2660
};

static int
virQEMUCapsProbeQMPCommandLine(virQEMUCapsPtr qemuCaps,
                               qemuMonitorPtr mon)
{
2661
    bool found = false;
2662 2663 2664 2665 2666 2667 2668
    int nvalues;
    char **values;
    size_t i, j;

    for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsCommandLine); i++) {
        if ((nvalues = qemuMonitorGetCommandLineOptionParameters(mon,
                                                                 virQEMUCapsCommandLine[i].option,
2669 2670
                                                                 &values,
                                                                 &found)) < 0)
2671
            return -1;
2672 2673 2674 2675

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

2676
        for (j = 0; j < nvalues; j++) {
2677
            if (STREQ_NULLABLE(virQEMUCapsCommandLine[i].param, values[j])) {
2678 2679 2680 2681
                virQEMUCapsSet(qemuCaps, virQEMUCapsCommandLine[i].flag);
                break;
            }
        }
2682
        virStringListFree(values);
2683 2684 2685 2686
    }

    return 0;
}
2687

2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701
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);
2702
    virStringListFreeCount(caps, ncaps);
2703 2704 2705 2706

    return 0;
}

A
Andrea Bolognani 已提交
2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726
/**
 * 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;

2727
    virQEMUCapsSetGICCapabilities(qemuCaps, caps, ncaps);
A
Andrea Bolognani 已提交
2728 2729 2730 2731

    return 0;
}

2732

2733
/* Returns -1 on error, 0 if SEV is not supported, 1 if SEV is supported */
2734 2735 2736 2737
static int
virQEMUCapsProbeQMPSEVCapabilities(virQEMUCapsPtr qemuCaps,
                                   qemuMonitorPtr mon)
{
2738
    int rc = -1;
2739 2740
    virSEVCapability *caps = NULL;

2741 2742
    if ((rc = qemuMonitorGetSEVCapabilities(mon, &caps)) <= 0)
        return rc;
2743

2744 2745
    virSEVCapabilitiesFree(qemuCaps->sevCapabilities);
    qemuCaps->sevCapabilities = caps;
2746
    return rc;
2747 2748 2749
}


2750
bool
2751
virQEMUCapsCPUFilterFeatures(const char *name,
2752
                             void *opaque)
2753
{
2754
    virArch *arch = opaque;
2755

2756
    if (!ARCH_IS_X86(*arch))
2757 2758
        return true;

2759 2760 2761 2762 2763 2764 2765 2766 2767
    if (STREQ(name, "cmt") ||
        STREQ(name, "mbm_total") ||
        STREQ(name, "mbm_local"))
        return false;

    return true;
}


2768 2769 2770
/**
 * 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,
2771
 *          2 when cpu model info is not supported for this configuration,
2772 2773 2774 2775
 *         -1 on error.
 */
static int
virQEMUCapsInitCPUModelS390(virQEMUCapsPtr qemuCaps,
2776
                            virDomainVirtType type,
2777
                            qemuMonitorCPUModelInfoPtr modelInfo,
2778 2779
                            virCPUDefPtr cpu,
                            bool migratable)
2780
{
2781
    size_t i;
2782

2783
    if (!modelInfo) {
2784 2785 2786 2787 2788 2789 2790 2791
        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;
2792
    }
J
Jiri Denemark 已提交
2793

2794 2795
    if (VIR_STRDUP(cpu->model, modelInfo->name) < 0 ||
        VIR_ALLOC_N(cpu->features, modelInfo->nprops) < 0)
2796
        return -1;
2797 2798 2799 2800 2801

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

    for (i = 0; i < modelInfo->nprops; i++) {
2802 2803
        virCPUFeatureDefPtr feature = cpu->features + cpu->nfeatures;
        qemuMonitorCPUPropertyPtr prop = modelInfo->props + i;
2804

2805 2806
        if (prop->type != QEMU_MONITOR_CPU_PROPERTY_BOOLEAN)
            continue;
2807

2808 2809
        if (VIR_STRDUP(feature->name, prop->name) < 0)
            return -1;
2810 2811 2812 2813 2814 2815

        if (!prop->value.boolean ||
            (migratable && prop->migratable == VIR_TRISTATE_BOOL_NO))
            feature->policy = VIR_CPU_FEATURE_DISABLE;
        else
            feature->policy = VIR_CPU_FEATURE_REQUIRE;
2816 2817 2818
        cpu->nfeatures++;
    }

2819 2820
    return 0;
}
2821

2822

2823 2824 2825 2826 2827 2828 2829 2830
/**
 * 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,
2831
                           qemuMonitorCPUModelInfoPtr model,
2832 2833
                           virCPUDefPtr cpu,
                           bool migratable)
2834 2835 2836 2837
{
    virCPUDataPtr data = NULL;
    unsigned long long sigFamily = 0;
    unsigned long long sigModel = 0;
2838
    unsigned long long sigStepping = 0;
2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852
    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:
2853 2854 2855 2856 2857
            if (!prop->value.boolean ||
                (migratable && prop->migratable == VIR_TRISTATE_BOOL_NO))
                continue;

            if (virCPUx86DataAddFeature(data, prop->name) < 0)
2858
                goto cleanup;
2859

2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872
            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;
2873 2874
            else if (STREQ(prop->name, "stepping"))
                sigStepping = prop->value.number;
2875 2876 2877 2878 2879 2880 2881
            break;

        case QEMU_MONITOR_CPU_PROPERTY_LAST:
            break;
        }
    }

2882
    if (virCPUx86DataSetSignature(data, sigFamily, sigModel, sigStepping) < 0)
2883 2884
        goto cleanup;

2885
    if (cpuDecode(cpu, data, virQEMUCapsGetCPUDefinitions(qemuCaps, type)) < 0)
2886 2887 2888 2889 2890 2891 2892 2893 2894 2895
        goto cleanup;

    ret = 0;

 cleanup:
    virCPUDataFree(data);
    return ret;
}


2896 2897
/**
 * Returns  0 when host CPU model provided by QEMU was filled in qemuCaps,
2898 2899
 *          1 when the caller should fall back to other methods,
 *          2 when cpu model info is not supported for this configuration,
2900 2901
 *         -1 on error.
 */
2902
int
2903
virQEMUCapsInitCPUModel(virQEMUCapsPtr qemuCaps,
2904
                        virDomainVirtType type,
2905 2906
                        virCPUDefPtr cpu,
                        bool migratable)
2907
{
2908
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);
2909 2910
    int ret = 1;

2911
    if (migratable && cpuData->info && !cpuData->info->migratability)
2912 2913
        return 1;

2914
    if (ARCH_IS_S390(qemuCaps->arch)) {
2915
        ret = virQEMUCapsInitCPUModelS390(qemuCaps, type, cpuData->info,
2916 2917 2918 2919 2920
                                          cpu, migratable);
    } else if (ARCH_IS_X86(qemuCaps->arch)) {
        ret = virQEMUCapsInitCPUModelX86(qemuCaps, type, cpuData->info,
                                         cpu, migratable);
    }
2921

2922 2923 2924
    if (ret == 0)
        cpu->fallback = VIR_CPU_FALLBACK_FORBID;

2925
    return ret;
2926 2927 2928
}


2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945
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;
}


2946 2947
void
virQEMUCapsInitHostCPUModel(virQEMUCapsPtr qemuCaps,
2948
                            virArch hostArch,
2949
                            virDomainVirtType type)
2950 2951
{
    virCPUDefPtr cpu = NULL;
2952
    virCPUDefPtr cpuExpanded = NULL;
2953
    virCPUDefPtr migCPU = NULL;
2954
    virCPUDefPtr hostCPU = NULL;
2955 2956
    virCPUDefPtr fullCPU = NULL;
    size_t i;
2957
    int rc;
2958

2959
    if (!virQEMUCapsGuestIsNative(hostArch, qemuCaps->arch))
2960 2961
        return;

2962
    if (!(cpu = virQEMUCapsNewHostCPUModel()))
2963 2964
        goto error;

2965
    if ((rc = virQEMUCapsInitCPUModel(qemuCaps, type, cpu, false)) < 0) {
2966 2967
        goto error;
    } else if (rc == 1) {
2968
        VIR_DEBUG("No host CPU model info from QEMU; probing host CPU directly");
2969

2970
        hostCPU = virQEMUCapsProbeHostCPUForEmulator(hostArch, qemuCaps, type);
2971 2972
        if (!hostCPU ||
            virCPUDefCopyModelFilter(cpu, hostCPU, true,
2973
                                     virQEMUCapsCPUFilterFeatures,
2974
                                     &qemuCaps->arch) < 0)
2975
            goto error;
2976 2977 2978 2979 2980
    } else if (rc == 2) {
        VIR_DEBUG("QEMU does not provide CPU model for arch=%s virttype=%s",
                  virArchToString(qemuCaps->arch),
                  virDomainVirtTypeToString(type));
        goto error;
2981 2982 2983
    } else if (type == VIR_DOMAIN_VIRT_KVM &&
               virCPUGetHostIsSupported(qemuCaps->arch)) {
        if (!(fullCPU = virCPUGetHost(qemuCaps->arch, VIR_CPU_TYPE_GUEST,
2984
                                      NULL, NULL)))
2985 2986
            goto error;

2987 2988 2989 2990 2991 2992 2993
        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,
2994 2995 2996
                                       VIR_CPU_FEATURE_REQUIRE) < 0)
                goto error;
        }
2997 2998
    }

2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011
    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;
    }

3012
    virQEMUCapsSetHostModel(qemuCaps, type, cpu, migCPU, fullCPU);
3013

3014
 cleanup:
3015
    virCPUDefFree(cpuExpanded);
3016
    virCPUDefFree(hostCPU);
3017 3018 3019 3020
    return;

 error:
    virCPUDefFree(cpu);
3021
    virCPUDefFree(migCPU);
3022
    virCPUDefFree(fullCPU);
3023
    virResetLastError();
3024
    goto cleanup;
3025 3026 3027
}


3028 3029 3030 3031 3032
void
virQEMUCapsSetCPUModelInfo(virQEMUCapsPtr qemuCaps,
                           virDomainVirtType type,
                           qemuMonitorCPUModelInfoPtr modelInfo)
{
3033 3034 3035
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);

    cpuData->info = modelInfo;
3036 3037 3038
}


3039 3040
static int
virQEMUCapsLoadHostCPUModelInfo(virQEMUCapsPtr qemuCaps,
3041 3042
                                xmlXPathContextPtr ctxt,
                                virDomainVirtType virtType)
3043 3044 3045
{
    char *str = NULL;
    xmlNodePtr hostCPUNode;
3046
    xmlNodePtr *nodes = NULL;
3047 3048 3049 3050 3051
    xmlNodePtr oldnode = ctxt->node;
    qemuMonitorCPUModelInfoPtr hostCPU = NULL;
    int ret = -1;
    size_t i;
    int n;
3052
    int val;
3053

3054 3055 3056 3057 3058 3059
    if (virtType == VIR_DOMAIN_VIRT_KVM)
        hostCPUNode = virXPathNode("./hostCPU[@type='kvm']", ctxt);
    else
        hostCPUNode = virXPathNode("./hostCPU[@type='tcg']", ctxt);

    if (!hostCPUNode) {
3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073
        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;
    }

3074 3075 3076 3077 3078 3079 3080 3081 3082
    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);

3083 3084
    ctxt->node = hostCPUNode;

3085
    if ((n = virXPathNodeSet("./property", ctxt, &nodes)) > 0) {
3086 3087 3088 3089 3090 3091
        if (VIR_ALLOC_N(hostCPU->props, n) < 0)
            goto cleanup;

        hostCPU->nprops = n;

        for (i = 0; i < n; i++) {
3092 3093 3094 3095 3096
            qemuMonitorCPUPropertyPtr prop = hostCPU->props + i;

            ctxt->node = nodes[i];

            if (!(prop->name = virXMLPropString(ctxt->node, "name"))) {
3097 3098
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing 'name' attribute for a host CPU"
3099
                                 " model property in QEMU capabilities cache"));
3100 3101 3102
                goto cleanup;
            }

3103
            if (!(str = virXMLPropString(ctxt->node, "type")) ||
3104
                (val = qemuMonitorCPUPropertyTypeFromString(str)) < 0) {
3105
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
3106 3107
                               _("missing or invalid CPU model property type "
                                 "in QEMU capabilities cache"));
3108 3109 3110
                goto cleanup;
            }
            VIR_FREE(str);
3111

3112
            prop->type = val;
3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143
            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;
            }
3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156

            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);
            }
3157 3158 3159
        }
    }

3160
    virQEMUCapsSetCPUModelInfo(qemuCaps, virtType, hostCPU);
3161 3162 3163 3164 3165 3166
    hostCPU = NULL;
    ret = 0;

 cleanup:
    ctxt->node = oldnode;
    VIR_FREE(str);
3167
    VIR_FREE(nodes);
3168 3169 3170 3171 3172
    qemuMonitorCPUModelInfoFree(hostCPU);
    return ret;
}


3173 3174
static int
virQEMUCapsLoadCPUModels(virQEMUCapsPtr qemuCaps,
3175 3176
                         xmlXPathContextPtr ctxt,
                         virDomainVirtType type)
3177
{
3178
    virDomainCapsCPUModelsPtr cpus = NULL;
3179 3180 3181 3182 3183
    xmlNodePtr *nodes = NULL;
    char *str = NULL;
    size_t i;
    int n;
    int ret = -1;
3184 3185 3186 3187
    xmlNodePtr node;
    xmlNodePtr *blockerNodes = NULL;
    char **blockers = NULL;
    int nblockers;
3188

3189 3190 3191 3192 3193 3194
    if (type == VIR_DOMAIN_VIRT_KVM)
        n = virXPathNodeSet("./cpu[@type='kvm']", ctxt, &nodes);
    else
        n = virXPathNodeSet("./cpu[@type='tcg']", ctxt, &nodes);

    if (n < 0) {
3195 3196 3197 3198 3199 3200 3201 3202 3203 3204
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to parse qemu capabilities cpus"));
        goto cleanup;
    }

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

3205
    if (!(cpus = virDomainCapsCPUModelsNew(n)))
3206 3207
        goto cleanup;

3208 3209 3210 3211 3212
    if (type == VIR_DOMAIN_VIRT_KVM)
        qemuCaps->kvmCPUModels = cpus;
    else
        qemuCaps->tcgCPUModels = cpus;

3213
    for (i = 0; i < n; i++) {
3214 3215 3216 3217 3218 3219 3220 3221 3222 3223
        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);

3224 3225 3226 3227 3228 3229
        if (!(str = virXMLPropString(nodes[i], "name"))) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("missing cpu name in QEMU capabilities cache"));
            goto cleanup;
        }

3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254
        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;
                }
            }
3255
            VIR_FREE(blockerNodes);
3256 3257 3258
        }

        if (virDomainCapsCPUModelsAddSteal(cpus, &str, usable, &blockers) < 0)
3259 3260 3261 3262 3263 3264 3265 3266
            goto cleanup;
    }

    ret = 0;

 cleanup:
    VIR_FREE(nodes);
    VIR_FREE(str);
3267 3268
    VIR_FREE(blockerNodes);
    virStringListFree(blockers);
3269 3270 3271 3272
    return ret;
}


3273 3274 3275 3276 3277
struct _virQEMUCapsCachePriv {
    char *libDir;
    uid_t runUid;
    gid_t runGid;
    virArch hostArch;
3278
    unsigned int microcodeVersion;
3279
    char *kernelVersion;
3280 3281 3282 3283

    /* cache whether /dev/kvm is usable as runUid:runGuid */
    virTristateBool kvmUsable;
    time_t kvmCtime;
3284 3285 3286 3287 3288
};
typedef struct _virQEMUCapsCachePriv virQEMUCapsCachePriv;
typedef virQEMUCapsCachePriv *virQEMUCapsCachePrivPtr;


3289
static void
3290
virQEMUCapsCachePrivFree(void *privData)
3291
{
3292 3293
    virQEMUCapsCachePrivPtr priv = privData;

3294
    VIR_FREE(priv->libDir);
3295
    VIR_FREE(priv->kernelVersion);
3296 3297 3298 3299
    VIR_FREE(priv);
}


3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351
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;
}


3352 3353 3354 3355 3356 3357
/*
 * Parsing a doc that looks like
 *
 * <qemuCaps>
 *   <qemuctime>234235253</qemuctime>
 *   <selfctime>234235253</selfctime>
3358
 *   <selfvers>1002016</selfvers>
3359 3360 3361 3362 3363 3364
 *   <usedQMP/>
 *   <flag name='foo'/>
 *   <flag name='bar'/>
 *   ...
 *   <cpu name="pentium3"/>
 *   ...
3365
 *   <machine name="pc-1.0" alias="pc" hotplugCpus='yes' maxCpus="4" default="yes"/>
3366 3367 3368
 *   ...
 * </qemuCaps>
 */
3369
int
3370
virQEMUCapsLoadCache(virArch hostArch,
3371
                     virQEMUCapsPtr qemuCaps,
3372
                     const char *filename)
3373 3374 3375 3376 3377 3378 3379
{
    xmlDocPtr doc = NULL;
    int ret = -1;
    size_t i;
    int n;
    xmlNodePtr *nodes = NULL;
    xmlXPathContextPtr ctxt = NULL;
J
Ján Tomko 已提交
3380
    char *str = NULL;
3381
    long long int l;
3382
    unsigned long lu;
3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406

    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;
    }
3407
    qemuCaps->ctime = (time_t)l;
3408 3409 3410 3411 3412 3413

    if (virXPathLongLong("string(./selfctime)", ctxt, &l) < 0) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing selfctime in QEMU capabilities XML"));
        goto cleanup;
    }
3414
    qemuCaps->libvirtCtime = (time_t)l;
3415

3416
    qemuCaps->libvirtVersion = 0;
3417
    if (virXPathULong("string(./selfvers)", ctxt, &lu) == 0)
3418
        qemuCaps->libvirtVersion = lu;
3419

3420 3421 3422 3423 3424 3425 3426 3427 3428
    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);
3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440
    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;
3441
        }
3442 3443
        VIR_FREE(str);
        virQEMUCapsSet(qemuCaps, flag);
3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458
    }
    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;
    }

3459 3460 3461 3462 3463 3464 3465
    if (virXPathUInt("string(./microcodeVersion)", ctxt,
                     &qemuCaps->microcodeVersion) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing microcode version in QEMU capabilities cache"));
        goto cleanup;
    }

3466 3467 3468 3469 3470 3471
    if (virXPathBoolean("boolean(./package)", ctxt) > 0) {
        qemuCaps->package = virXPathString("string(./package)", ctxt);
        if (!qemuCaps->package &&
            VIR_STRDUP(qemuCaps->package, "") < 0)
            goto cleanup;
    }
3472

3473 3474 3475 3476 3477 3478
    if (virXPathBoolean("boolean(./kernelVersion)", ctxt) > 0) {
        qemuCaps->kernelVersion = virXPathString("string(./kernelVersion)", ctxt);
        if (!qemuCaps->kernelVersion)
            goto cleanup;
    }

3479 3480 3481 3482 3483 3484 3485 3486 3487 3488
    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
3489
    VIR_FREE(str);
3490

3491 3492
    if (virQEMUCapsLoadHostCPUModelInfo(qemuCaps, ctxt, VIR_DOMAIN_VIRT_KVM) < 0 ||
        virQEMUCapsLoadHostCPUModelInfo(qemuCaps, ctxt, VIR_DOMAIN_VIRT_QEMU) < 0)
3493 3494
        goto cleanup;

3495 3496
    if (virQEMUCapsLoadCPUModels(qemuCaps, ctxt, VIR_DOMAIN_VIRT_KVM) < 0 ||
        virQEMUCapsLoadCPUModels(qemuCaps, ctxt, VIR_DOMAIN_VIRT_QEMU) < 0)
3497 3498 3499 3500 3501 3502 3503 3504 3505
        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;
3506
        if (VIR_ALLOC_N(qemuCaps->machineTypes, qemuCaps->nmachineTypes) < 0)
3507 3508 3509
            goto cleanup;

        for (i = 0; i < n; i++) {
3510
            if (!(qemuCaps->machineTypes[i].name = virXMLPropString(nodes[i], "name"))) {
3511 3512 3513 3514
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing machine name in QEMU capabilities cache"));
                goto cleanup;
            }
3515
            qemuCaps->machineTypes[i].alias = virXMLPropString(nodes[i], "alias");
3516 3517 3518

            str = virXMLPropString(nodes[i], "maxCpus");
            if (str &&
3519
                virStrToLong_ui(str, NULL, 10, &(qemuCaps->machineTypes[i].maxCpus)) < 0) {
3520 3521 3522 3523
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("malformed machine cpu count in QEMU capabilities cache"));
                goto cleanup;
            }
J
Ján Tomko 已提交
3524
            VIR_FREE(str);
3525 3526 3527 3528 3529

            str = virXMLPropString(nodes[i], "hotplugCpus");
            if (STREQ_NULLABLE(str, "yes"))
                qemuCaps->machineTypes[i].hotplugCpus = true;
            VIR_FREE(str);
3530 3531 3532 3533 3534

            str = virXMLPropString(nodes[i], "default");
            if (STREQ_NULLABLE(str, "yes"))
                qemuCaps->machineTypes[i].qemuDefault = true;
            VIR_FREE(str);
3535 3536 3537 3538
        }
    }
    VIR_FREE(nodes);

A
Andrea Bolognani 已提交
3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604
    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);

3605 3606 3607
    if (virQEMUCapsParseSEVInfo(qemuCaps, ctxt) < 0)
        goto cleanup;

3608 3609
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_QEMU);
3610

3611 3612 3613
    if (virXPathBoolean("boolean(./kvmSupportsNesting)", ctxt) > 0)
        qemuCaps->kvmSupportsNesting = true;

3614
    ret = 0;
3615
 cleanup:
J
Ján Tomko 已提交
3616
    VIR_FREE(str);
3617 3618 3619 3620 3621 3622 3623
    VIR_FREE(nodes);
    xmlXPathFreeContext(ctxt);
    xmlFreeDoc(doc);
    return ret;
}


3624 3625
static void
virQEMUCapsFormatHostCPUModelInfo(virQEMUCapsPtr qemuCaps,
3626 3627
                                  virBufferPtr buf,
                                  virDomainVirtType type)
3628
{
3629 3630 3631
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);
    qemuMonitorCPUModelInfoPtr model = cpuData->info;
    const char *typeStr = type == VIR_DOMAIN_VIRT_KVM ? "kvm" : "tcg";
3632 3633
    size_t i;

3634 3635 3636
    if (!model)
        return;

3637 3638 3639 3640
    virBufferAsprintf(buf,
                      "<hostCPU type='%s' model='%s' migratability='%s'>\n",
                      typeStr, model->name,
                      model->migratability ? "yes" : "no");
3641 3642 3643
    virBufferAdjustIndent(buf, 2);

    for (i = 0; i < model->nprops; i++) {
3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666
        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;
        }
3667 3668 3669 3670 3671

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

3672
        virBufferAddLit(buf, "/>\n");
3673 3674 3675 3676 3677 3678 3679
    }

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


3680 3681
static void
virQEMUCapsFormatCPUModels(virQEMUCapsPtr qemuCaps,
3682 3683
                           virBufferPtr buf,
                           virDomainVirtType type)
3684
{
3685 3686
    virDomainCapsCPUModelsPtr cpus;
    const char *typeStr;
3687 3688
    size_t i;

3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700
    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++) {
3701 3702
        virDomainCapsCPUModelPtr cpu = cpus->models + i;

3703
        virBufferAsprintf(buf, "<cpu type='%s' ", typeStr);
3704 3705 3706 3707 3708
        virBufferEscapeString(buf, "name='%s'", cpu->name);
        if (cpu->usable) {
            virBufferAsprintf(buf, " usable='%s'",
                              virDomainCapsCPUUsableTypeToString(cpu->usable));
        }
3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723

        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");
        }
3724 3725 3726 3727
    }
}


3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745
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");
}


3746
char *
3747
virQEMUCapsFormatCache(virQEMUCapsPtr qemuCaps)
3748 3749
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;
3750
    char *ret = NULL;
3751 3752 3753
    size_t i;

    virBufferAddLit(&buf, "<qemuCaps>\n");
3754
    virBufferAdjustIndent(&buf, 2);
3755

3756
    virBufferAsprintf(&buf, "<qemuctime>%llu</qemuctime>\n",
3757
                      (long long)qemuCaps->ctime);
3758
    virBufferAsprintf(&buf, "<selfctime>%llu</selfctime>\n",
3759
                      (long long)qemuCaps->libvirtCtime);
3760
    virBufferAsprintf(&buf, "<selfvers>%lu</selfvers>\n",
3761
                      (unsigned long)qemuCaps->libvirtVersion);
3762 3763

    if (qemuCaps->usedQMP)
3764
        virBufferAddLit(&buf, "<usedQMP/>\n");
3765 3766 3767

    for (i = 0; i < QEMU_CAPS_LAST; i++) {
        if (virQEMUCapsGet(qemuCaps, i)) {
3768
            virBufferAsprintf(&buf, "<flag name='%s'/>\n",
3769 3770 3771 3772
                              virQEMUCapsTypeToString(i));
        }
    }

3773
    virBufferAsprintf(&buf, "<version>%d</version>\n",
3774 3775
                      qemuCaps->version);

3776
    virBufferAsprintf(&buf, "<kvmVersion>%d</kvmVersion>\n",
3777 3778
                      qemuCaps->kvmVersion);

3779 3780 3781
    virBufferAsprintf(&buf, "<microcodeVersion>%u</microcodeVersion>\n",
                      qemuCaps->microcodeVersion);

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

3786 3787 3788 3789
    if (qemuCaps->kernelVersion)
        virBufferAsprintf(&buf, "<kernelVersion>%s</kernelVersion>\n",
                          qemuCaps->kernelVersion);

3790
    virBufferAsprintf(&buf, "<arch>%s</arch>\n",
3791 3792
                      virArchToString(qemuCaps->arch));

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

3796 3797
    virQEMUCapsFormatCPUModels(qemuCaps, &buf, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsFormatCPUModels(qemuCaps, &buf, VIR_DOMAIN_VIRT_QEMU);
3798 3799

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
3800
        virBufferEscapeString(&buf, "<machine name='%s'",
3801 3802
                              qemuCaps->machineTypes[i].name);
        if (qemuCaps->machineTypes[i].alias)
3803
            virBufferEscapeString(&buf, " alias='%s'",
3804
                              qemuCaps->machineTypes[i].alias);
3805 3806
        if (qemuCaps->machineTypes[i].hotplugCpus)
            virBufferAddLit(&buf, " hotplugCpus='yes'");
3807
        virBufferAsprintf(&buf, " maxCpus='%u'",
3808
                          qemuCaps->machineTypes[i].maxCpus);
3809 3810 3811
        if (qemuCaps->machineTypes[i].qemuDefault)
            virBufferAddLit(&buf, " default='yes'");
        virBufferAddLit(&buf, "/>\n");
3812 3813
    }

A
Andrea Bolognani 已提交
3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829
    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");
    }

3830 3831 3832
    if (qemuCaps->sevCapabilities)
        virQEMUCapsFormatSEVInfo(qemuCaps, &buf);

3833 3834 3835
    if (qemuCaps->kvmSupportsNesting)
        virBufferAddLit(&buf, "<kvmSupportsNesting/>\n");

3836
    virBufferAdjustIndent(&buf, -2);
3837 3838
    virBufferAddLit(&buf, "</qemuCaps>\n");

3839 3840 3841 3842 3843 3844 3845 3846
    if (virBufferCheckError(&buf) == 0)
        ret = virBufferContentAndReset(&buf);

    return ret;
}


static int
3847 3848 3849
virQEMUCapsSaveFile(void *data,
                    const char *filename,
                    void *privData ATTRIBUTE_UNUSED)
3850
{
3851
    virQEMUCapsPtr qemuCaps = data;
3852 3853
    char *xml = NULL;
    int ret = -1;
3854

3855
    xml = virQEMUCapsFormatCache(qemuCaps);
3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866

    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,
3867
              (long long)qemuCaps->libvirtCtime);
3868 3869 3870 3871 3872 3873 3874 3875

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


3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910
/* 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;
}


3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958
/* 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;
}


3959
static bool
3960 3961
virQEMUCapsIsValid(void *data,
                   void *privData)
3962
{
3963 3964
    virQEMUCapsPtr qemuCaps = data;
    virQEMUCapsCachePrivPtr priv = privData;
3965
    bool kvmUsable;
3966
    struct stat sb;
3967
    bool kvmSupportsNesting;
3968 3969 3970 3971

    if (!qemuCaps->binary)
        return true;

3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983
    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;
    }

3984 3985 3986 3987 3988 3989
    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;
3990 3991
    }

3992
    if (sb.st_ctime != qemuCaps->ctime) {
3993 3994 3995
        VIR_DEBUG("Outdated capabilities for '%s': QEMU binary changed "
                  "(%lld vs %lld)",
                  qemuCaps->binary,
3996
                  (long long)sb.st_ctime, (long long)qemuCaps->ctime);
3997 3998 3999
        return false;
    }

4000 4001 4002 4003 4004 4005 4006 4007
    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;
    }

4008
    kvmUsable = virQEMUCapsKVMUsable(priv);
4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025

    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;
    }

4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043
    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;
        }
4044 4045 4046 4047 4048 4049 4050 4051

        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;
        }
4052 4053
    }

4054 4055 4056 4057
    return true;
}


4058
static void virQEMUCapsMonitorNotify(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
4059 4060
                                     virDomainObjPtr vm ATTRIBUTE_UNUSED,
                                     void *opaque ATTRIBUTE_UNUSED)
4061 4062 4063 4064
{
}

static qemuMonitorCallbacks callbacks = {
4065 4066
    .eofNotify = virQEMUCapsMonitorNotify,
    .errorNotify = virQEMUCapsMonitorNotify,
4067 4068 4069
};


4070 4071 4072 4073 4074 4075 4076 4077
/**
 * virQEMUCapsInitQMPArch:
 * @qemuCaps: QEMU capabilities
 * @mon: QEMU monitor
 *
 * Initialize the architecture for @qemuCaps by asking @mon.
 *
 * Returns: 0 on success, <0 on failure
4078 4079
 */
static int
4080
virQEMUCapsInitQMPArch(virQEMUCapsPtr qemuCaps,
4081 4082 4083 4084 4085 4086
                            qemuMonitorPtr mon)
{
    char *archstr = NULL;
    int ret = -1;

    if (!(archstr = qemuMonitorGetTargetArch(mon)))
4087
        goto cleanup;
4088 4089 4090 4091 4092 4093 4094

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

4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108
    ret = 0;

 cleanup:
    VIR_FREE(archstr);
    return ret;
}


/**
 * virQEMUCapsInitQMPBasicArch:
 * @qemuCaps: QEMU capabilities
 *
 * Initialize @qemuCaps with basic architecture-dependent capabilities.
 */
4109
void
4110 4111
virQEMUCapsInitQMPBasicArch(virQEMUCapsPtr qemuCaps)
{
4112 4113 4114
    /* ACPI only works on x86 and aarch64 */
    if (ARCH_IS_X86(qemuCaps->arch) ||
        qemuCaps->arch == VIR_ARCH_AARCH64) {
4115
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_NO_ACPI);
4116 4117
    }

J
Ján Tomko 已提交
4118 4119
    /* HPET is x86 specific */
    if (ARCH_IS_X86(qemuCaps->arch))
J
Ján Tomko 已提交
4120
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_NO_HPET);
4121
}
4122

4123 4124 4125 4126 4127 4128

static int
virQEMUCapsProbeQMPSchemaCapabilities(virQEMUCapsPtr qemuCaps,
                                      qemuMonitorPtr mon)
{
    struct virQEMUCapsStringFlags *entry;
4129 4130
    virJSONValuePtr schemareply;
    virHashTablePtr schema = NULL;
4131 4132
    size_t i;

4133
    if (!(schemareply = qemuMonitorQueryQMPSchema(mon)))
4134 4135
        return -1;

4136 4137 4138 4139
    if (!(schema = virQEMUQAPISchemaConvert(schemareply)))
        return -1;
    schemareply = NULL;

4140 4141 4142
    for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsQMPSchemaQueries); i++) {
        entry = virQEMUCapsQMPSchemaQueries + i;

4143
        if (virQEMUQAPISchemaPathExists(entry->value, schema))
4144 4145 4146 4147 4148 4149 4150
            virQEMUCapsSet(qemuCaps, entry->flag);
    }

    virHashFree(schema);
    return 0;
}

J
Ján Tomko 已提交
4151
#define QEMU_MIN_MAJOR 1
J
Ján Tomko 已提交
4152
#define QEMU_MIN_MINOR 5
J
Ján Tomko 已提交
4153
#define QEMU_MIN_MICRO 0
4154

4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166
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",
4167
                  virGetLastErrorMessage());
4168 4169 4170 4171 4172 4173 4174 4175
        ret = 0;
        goto cleanup;
    }

    if (qemuMonitorGetVersion(mon,
                              &major, &minor, &micro,
                              &package) < 0) {
        VIR_DEBUG("Failed to query monitor version %s",
4176
                  virGetLastErrorMessage());
4177 4178 4179 4180 4181 4182 4183
        ret = 0;
        goto cleanup;
    }

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

J
Ján Tomko 已提交
4184 4185 4186 4187 4188 4189
    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);
4190 4191 4192 4193
        goto cleanup;
    }

    qemuCaps->version = major * 1000000 + minor * 1000 + micro;
4194
    qemuCaps->package = package;
4195 4196
    qemuCaps->usedQMP = true;

4197
    if (virQEMUCapsInitQMPArch(qemuCaps, mon) < 0)
4198 4199
        goto cleanup;

4200 4201
    virQEMUCapsInitQMPBasicArch(qemuCaps);

4202 4203 4204
    if (qemuCaps->version >= 1006000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_DEVICE_VIDEO_PRIMARY);

4205 4206 4207 4208
    /* vmport option is supported v2.2.0 onwards */
    if (qemuCaps->version >= 2002000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_VMPORT_OPT);

4209 4210 4211 4212 4213 4214
    /* -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);

4215 4216 4217
    /* 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
4218
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_VHOSTUSER_MULTIQUEUE);
4219

M
Michal Privoznik 已提交
4220 4221 4222 4223
    /* smm option is supported from v2.4.0 */
    if (qemuCaps->version >= 2004000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_SMM_OPT);

4224 4225 4226 4227
    /* sdl -gl option is supported from v2.4.0 (qemu commit id 0b71a5d5) */
    if (qemuCaps->version >= 2004000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_SDL_GL);

4228 4229 4230 4231
    /* Since 2.4.50 ARM virt machine supports gic-version option */
    if (qemuCaps->version >= 2004050)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACH_VIRT_GIC_VERSION);

4232 4233 4234 4235
    /* no way to query if -machine kernel_irqchip supports split */
    if (qemuCaps->version >= 2006000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_KERNEL_IRQCHIP_SPLIT);

4236 4237 4238 4239 4240 4241 4242
    /* 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);
    }

4243 4244 4245 4246 4247
    /* '-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);

4248 4249 4250 4251
    /* no way to query for -numa dist */
    if (qemuCaps->version >= 2010000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_NUMA_DIST);

4252 4253 4254 4255 4256 4257
    /* 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);
    }

4258 4259
    if (virQEMUCapsProbeQMPCommands(qemuCaps, mon) < 0)
        goto cleanup;
J
Jiri Denemark 已提交
4260 4261 4262 4263 4264

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

4265 4266
    if (virQEMUCapsProbeQMPEvents(qemuCaps, mon) < 0)
        goto cleanup;
4267
    if (virQEMUCapsProbeQMPDevices(qemuCaps, mon) < 0)
4268 4269 4270
        goto cleanup;
    if (virQEMUCapsProbeQMPMachineTypes(qemuCaps, mon) < 0)
        goto cleanup;
4271
    if (virQEMUCapsProbeQMPCPUDefinitions(qemuCaps, mon, false) < 0)
4272 4273 4274 4275 4276
        goto cleanup;
    if (virQEMUCapsProbeQMPTPM(qemuCaps, mon) < 0)
        goto cleanup;
    if (virQEMUCapsProbeQMPCommandLine(qemuCaps, mon) < 0)
        goto cleanup;
4277 4278
    if (virQEMUCapsProbeQMPMigrationCapabilities(qemuCaps, mon) < 0)
        goto cleanup;
4279 4280 4281
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_QMP_SCHEMA) &&
        virQEMUCapsProbeQMPSchemaCapabilities(qemuCaps, mon) < 0)
        goto cleanup;
4282
    if (virQEMUCapsProbeQMPHostCPU(qemuCaps, mon, false) < 0)
4283
        goto cleanup;
4284

4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295
    /* '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
4296 4297
    /* GIC capabilities, eg. available GIC versions */
    if ((qemuCaps->arch == VIR_ARCH_AARCH64 ||
S
Stefan Schallenberg 已提交
4298
         qemuCaps->arch == VIR_ARCH_ARMV6L ||
A
Andrea Bolognani 已提交
4299 4300 4301 4302
         qemuCaps->arch == VIR_ARCH_ARMV7L) &&
        virQEMUCapsProbeQMPGICCapabilities(qemuCaps, mon) < 0)
        goto cleanup;

4303 4304 4305 4306 4307 4308 4309
    /* 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);

4310 4311 4312 4313
    if (ARCH_IS_X86(qemuCaps->arch) &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_CPU_MODEL_EXPANSION))
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_CPU_CACHE);

4314 4315 4316 4317 4318
    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);
4319 4320
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CCW_CSSID_UNRESTRICTED))
            virQEMUCapsClear(qemuCaps, QEMU_CAPS_DEVICE_VFIO_CCW);
4321 4322
    }

4323 4324
    /* Probe for SEV capabilities */
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_SEV_GUEST)) {
4325 4326 4327 4328 4329 4330
        int rc = virQEMUCapsProbeQMPSEVCapabilities(qemuCaps, mon);

        if (rc < 0)
            goto cleanup;

        if (rc == 0)
4331 4332 4333
            virQEMUCapsClear(qemuCaps, QEMU_CAPS_SEV_GUEST);
    }

4334
    ret = 0;
4335
 cleanup:
4336 4337 4338
    return ret;
}

4339

4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352
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;
    }

4353 4354 4355
    if (virQEMUCapsProbeQMPCPUDefinitions(qemuCaps, mon, true) < 0)
        goto cleanup;

4356 4357 4358
    if (virQEMUCapsProbeQMPHostCPU(qemuCaps, mon, true) < 0)
        goto cleanup;

4359 4360 4361 4362 4363 4364
    ret = 0;
 cleanup:
    return ret;
}


4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376
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;
4377
    virDomainChrSourceDef config;
4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395
    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)
4396
        unlink(cmd->monpath);
4397 4398 4399 4400 4401 4402

    virDomainObjEndAPI(&cmd->vm);

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

4403
        VIR_DEBUG("Killing QMP caps process %lld", (long long)cmd->pid);
4404 4405
        if (virProcessKill(cmd->pid, SIGKILL) < 0 && errno != ESRCH)
            VIR_ERROR(_("Failed to kill process %lld: %s"),
4406
                      (long long)cmd->pid,
4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449
                      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;
4450

4451 4452 4453
    /* the ".sock" sufix is important to avoid a possible clash with a qemu
     * domain called "capabilities"
     */
4454 4455 4456 4457 4458
    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;
4459

4460 4461
    /* ".pidfile" suffix is used rather than ".pid" to avoid a possible clash
     * with a qemu domain called "capabilities"
4462 4463 4464
     * 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
4465
     */
4466 4467
    if (virAsprintf(&cmd->pidfile, "%s/%s", libDir, "capabilities.pidfile") < 0)
        goto error;
4468

4469
    virPidFileForceCleanupPath(cmd->pidfile);
4470

4471 4472 4473
    cmd->config.type = VIR_DOMAIN_CHR_TYPE_UNIX;
    cmd->config.data.nix.path = cmd->monpath;
    cmd->config.data.nix.listen = false;
4474

4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487
    return cmd;

 error:
    virQEMUCapsInitQMPCommandFree(cmd);
    return NULL;
}


/* Returns -1 on fatal error,
 *          0 on success,
 *          1 when probing QEMU failed
 */
static int
4488 4489
virQEMUCapsInitQMPCommandRun(virQEMUCapsInitQMPCommandPtr cmd,
                             bool forceTCG)
4490 4491
{
    virDomainXMLOptionPtr xmlopt = NULL;
4492
    const char *machine;
4493 4494 4495
    int status = 0;
    int ret = -1;

4496 4497 4498 4499 4500 4501 4502
    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);
4503

4504 4505 4506 4507 4508 4509 4510
    /*
     * 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.
     */
4511 4512 4513 4514 4515
    cmd->cmd = virCommandNewArgList(cmd->binary,
                                    "-S",
                                    "-no-user-config",
                                    "-nodefaults",
                                    "-nographic",
4516
                                    "-machine", machine,
4517 4518 4519 4520 4521 4522
                                    "-qmp", cmd->monarg,
                                    "-pidfile", cmd->pidfile,
                                    "-daemonize",
                                    NULL);
    virCommandAddEnvPassCommon(cmd->cmd);
    virCommandClearCaps(cmd->cmd);
4523 4524 4525 4526

#if WITH_CAPNG
    /* QEMU might run into permission issues, e.g. /dev/sev (0600), override
     * them just for the purpose of probing */
4527 4528
    if (geteuid() == 0)
        virCommandAllowCap(cmd->cmd, CAP_DAC_OVERRIDE);
4529 4530
#endif

4531 4532 4533 4534
    virCommandSetGID(cmd->cmd, cmd->runGid);
    virCommandSetUID(cmd->cmd, cmd->runUid);

    virCommandSetErrorBuffer(cmd->cmd, cmd->qmperr);
4535

4536
    /* Log, but otherwise ignore, non-zero status.  */
4537
    if (virCommandRun(cmd->cmd, &status) < 0)
4538 4539 4540
        goto cleanup;

    if (status != 0) {
4541
        VIR_DEBUG("QEMU %s exited with status %d: %s",
4542 4543
                  cmd->binary, status, *cmd->qmperr);
        goto ignore;
4544 4545
    }

4546 4547 4548
    if (virPidFileReadPath(cmd->pidfile, &cmd->pid) < 0) {
        VIR_DEBUG("Failed to read pidfile %s", cmd->pidfile);
        goto ignore;
4549 4550
    }

4551
    if (!(xmlopt = virDomainXMLOptionNew(NULL, NULL, NULL, NULL, NULL)) ||
4552
        !(cmd->vm = virDomainObjNew(xmlopt)))
4553 4554
        goto cleanup;

4555
    cmd->vm->pid = cmd->pid;
4556

4557
    if (!(cmd->mon = qemuMonitorOpen(cmd->vm, &cmd->config, true, true,
4558
                                     0, &callbacks, NULL)))
4559
        goto ignore;
4560

4561
    virObjectLock(cmd->mon);
4562 4563 4564

    ret = 0;

4565
 cleanup:
4566 4567
    if (!cmd->mon)
        virQEMUCapsInitQMPCommandAbort(cmd);
4568
    virObjectUnref(xmlopt);
4569

4570
    return ret;
4571

4572 4573 4574 4575
 ignore:
    ret = 1;
    goto cleanup;
}
4576

4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592

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;

4593
    if ((rc = virQEMUCapsInitQMPCommandRun(cmd, false)) != 0) {
4594 4595 4596
        if (rc == 1)
            ret = 0;
        goto cleanup;
4597
    }
4598 4599 4600 4601

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

4602 4603 4604 4605 4606 4607 4608 4609 4610 4611 4612 4613
    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;
    }

4614 4615 4616 4617
    ret = 0;

 cleanup:
    virQEMUCapsInitQMPCommandFree(cmd);
4618 4619 4620 4621
    return ret;
}


4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632
#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 },
    };

4633
    virLogMessage(&virLogSelf,
4634 4635 4636 4637
                  VIR_LOG_WARN,
                  __FILE__, __LINE__, __func__,
                  meta,
                  _("Failed to probe capabilities for %s: %s"),
4638
                  binary, virGetLastErrorMessage());
4639 4640 4641
}


4642
virQEMUCapsPtr
4643
virQEMUCapsNewForBinaryInternal(virArch hostArch,
4644
                                const char *binary,
4645 4646 4647
                                const char *libDir,
                                uid_t runUid,
                                gid_t runGid,
4648
                                unsigned int microcodeVersion,
J
Ján Tomko 已提交
4649
                                const char *kernelVersion)
4650
{
4651
    virQEMUCapsPtr qemuCaps;
4652
    struct stat sb;
4653
    char *qmperr = NULL;
4654

4655 4656 4657
    if (!(qemuCaps = virQEMUCapsNew()))
        goto error;

4658 4659
    if (VIR_STRDUP(qemuCaps->binary, binary) < 0)
        goto error;
4660 4661 4662 4663 4664 4665 4666 4667

    /* 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;
    }
4668
    qemuCaps->ctime = sb.st_ctime;
4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679

    /* 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;
    }

4680 4681
    if (virQEMUCapsInitQMP(qemuCaps, libDir, runUid, runGid, &qmperr) < 0) {
        virQEMUCapsLogProbeFailure(binary);
4682
        goto error;
4683
    }
4684

J
Ján Tomko 已提交
4685
    if (!qemuCaps->usedQMP) {
4686 4687 4688 4689 4690 4691
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Failed to probe QEMU binary with QMP: %s"),
                       qmperr ? qmperr : _("unknown error"));
        virQEMUCapsLogProbeFailure(binary);
        goto error;
    }
4692

4693 4694
    qemuCaps->libvirtCtime = virGetSelfLastChanged();
    qemuCaps->libvirtVersion = LIBVIR_VERSION_NUMBER;
4695

4696 4697
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_QEMU);
4698

4699
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
4700 4701
        qemuCaps->microcodeVersion = microcodeVersion;

4702 4703
        if (VIR_STRDUP(qemuCaps->kernelVersion, kernelVersion) < 0)
            goto error;
4704 4705

        qemuCaps->kvmSupportsNesting = virQEMUCapsKVMSupportsNesting();
4706 4707
    }

4708
 cleanup:
4709
    VIR_FREE(qmperr);
4710
    return qemuCaps;
4711

4712
 error:
4713 4714
    virObjectUnref(qemuCaps);
    qemuCaps = NULL;
4715
    goto cleanup;
4716 4717
}

4718 4719 4720
static void *
virQEMUCapsNewData(const char *binary,
                   void *privData)
4721
{
4722 4723 4724 4725 4726 4727 4728
    virQEMUCapsCachePrivPtr priv = privData;

    return virQEMUCapsNewForBinaryInternal(priv->hostArch,
                                           binary,
                                           priv->libDir,
                                           priv->runUid,
                                           priv->runGid,
4729
                                           priv->microcodeVersion,
J
Ján Tomko 已提交
4730
                                           priv->kernelVersion);
4731
}
4732 4733


4734 4735 4736 4737 4738 4739 4740
static void *
virQEMUCapsLoadFile(const char *filename,
                    const char *binary,
                    void *privData)
{
    virQEMUCapsPtr qemuCaps = virQEMUCapsNew();
    virQEMUCapsCachePrivPtr priv = privData;
4741

4742 4743
    if (!qemuCaps)
        return NULL;
4744

4745 4746 4747 4748 4749 4750 4751
    if (VIR_STRDUP(qemuCaps->binary, binary) < 0)
        goto error;

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

 cleanup:
4752 4753 4754 4755
    return qemuCaps;

 error:
    virObjectUnref(qemuCaps);
4756 4757
    qemuCaps = NULL;
    goto cleanup;
4758 4759
}

4760

4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793
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]);
    }

4794 4795
    if (!virQEMUCapsGetMachineHotplugCpus(qemuCaps, machineType))
        virQEMUCapsClear(qemuCaps, QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS);
4796 4797 4798
}


4799 4800 4801 4802 4803 4804 4805 4806 4807 4808
virFileCacheHandlers qemuCapsCacheHandlers = {
    .isValid = virQEMUCapsIsValid,
    .newData = virQEMUCapsNewData,
    .loadFile = virQEMUCapsLoadFile,
    .saveFile = virQEMUCapsSaveFile,
    .privFree = virQEMUCapsCachePrivFree,
};


virFileCachePtr
4809
virQEMUCapsCacheNew(const char *libDir,
4810
                    const char *cacheDir,
4811
                    uid_t runUid,
4812 4813
                    gid_t runGid,
                    unsigned int microcodeVersion)
4814
{
4815 4816 4817
    char *capsCacheDir = NULL;
    virFileCachePtr cache = NULL;
    virQEMUCapsCachePrivPtr priv = NULL;
4818
    struct utsname uts;
4819

4820
    if (virAsprintf(&capsCacheDir, "%s/capabilities", cacheDir) < 0)
4821
        goto error;
4822 4823

    if (!(cache = virFileCacheNew(capsCacheDir, "xml", &qemuCapsCacheHandlers)))
4824
        goto error;
4825

4826
    if (VIR_ALLOC(priv) < 0)
4827
        goto error;
4828
    virFileCacheSetPriv(cache, priv);
4829

4830
    if (VIR_STRDUP(priv->libDir, libDir) < 0)
4831 4832
        goto error;

4833
    priv->hostArch = virArchFromHost();
4834

4835 4836
    priv->runUid = runUid;
    priv->runGid = runGid;
4837
    priv->microcodeVersion = microcodeVersion;
4838
    priv->kvmUsable = VIR_TRISTATE_BOOL_ABSENT;
4839

4840 4841 4842 4843
    if (uname(&uts) == 0 &&
        virAsprintf(&priv->kernelVersion, "%s %s", uts.release, uts.version) < 0)
        goto error;

4844 4845
 cleanup:
    VIR_FREE(capsCacheDir);
4846 4847
    return cache;

4848
 error:
4849 4850 4851
    virObjectUnref(cache);
    cache = NULL;
    goto cleanup;
4852 4853 4854
}


4855
virQEMUCapsPtr
4856
virQEMUCapsCacheLookup(virFileCachePtr cache,
4857
                       const char *binary)
4858
{
4859
    virQEMUCapsPtr ret = NULL;
4860

4861
    ret = virFileCacheLookup(cache, binary);
4862 4863

    VIR_DEBUG("Returning caps %p for %s", ret, binary);
4864 4865 4866 4867
    return ret;
}


4868
virQEMUCapsPtr
4869
virQEMUCapsCacheLookupCopy(virFileCachePtr cache,
4870
                           const char *binary,
4871
                           const char *machineType)
4872
{
4873
    virQEMUCapsPtr qemuCaps = virQEMUCapsCacheLookup(cache, binary);
4874
    virQEMUCapsPtr ret;
4875

4876
    if (!qemuCaps)
4877 4878
        return NULL;

4879 4880
    ret = virQEMUCapsNewCopy(qemuCaps);
    virObjectUnref(qemuCaps);
4881 4882 4883 4884

    if (!ret)
        return NULL;

4885
    virQEMUCapsFilterByMachineType(ret, machineType);
4886 4887 4888 4889
    return ret;
}


4890 4891 4892 4893 4894
static int
virQEMUCapsCompareArch(const void *payload,
                       const void *name ATTRIBUTE_UNUSED,
                       const void *opaque)
{
4895
    struct virQEMUCapsSearchData *data = (struct virQEMUCapsSearchData *)opaque;
4896 4897
    const virQEMUCaps *qemuCaps = payload;

4898 4899 4900 4901 4902 4903 4904 4905 4906
    if (qemuCaps->arch != data->arch)
        return false;

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

    return true;
4907 4908 4909 4910
}


virQEMUCapsPtr
4911
virQEMUCapsCacheLookupByArch(virFileCachePtr cache,
4912 4913 4914
                             virArch arch)
{
    virQEMUCapsPtr ret = NULL;
4915 4916 4917 4918
    const char *binaryFilters[] = {
        "qemu-system-",
        NULL,
    };
4919 4920 4921 4922
    virArch archs[] = {
        arch,
        virQEMUCapsFindTarget(virArchFromHost(), arch),
    };
4923
    size_t i;
4924 4925
    size_t j;

4926 4927 4928 4929 4930 4931
    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],
            };
4932

4933 4934 4935 4936
            ret = virFileCacheLookupByFunc(cache, virQEMUCapsCompareArch, &data);
            if (ret)
                goto done;
        }
4937 4938
    }

4939 4940 4941 4942 4943
    virReportError(VIR_ERR_INVALID_ARG,
                   _("unable to find any emulator to serve '%s' "
                     "architecture"), virArchToString(arch));

 done:
4944 4945
    VIR_DEBUG("Returning caps %p for arch %s", ret, virArchToString(arch));

4946 4947 4948 4949
    return ret;
}


4950 4951 4952 4953 4954 4955 4956 4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035
/**
 * 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 {
5036
        machine = virQEMUCapsGetPreferredMachine(qemuCaps);
5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067
    }

    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;
}

5068 5069 5070 5071 5072 5073 5074
bool
virQEMUCapsSupportsVmport(virQEMUCapsPtr qemuCaps,
                          const virDomainDef *def)
{
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_VMPORT_OPT))
        return false;

5075 5076
    return qemuDomainIsI440FX(def) ||
        qemuDomainIsQ35(def) ||
5077 5078 5079 5080
        STREQ(def->os.machine, "isapc");
}


5081 5082 5083 5084 5085 5086 5087
bool
virQEMUCapsIsMachineSupported(virQEMUCapsPtr qemuCaps,
                              const char *canonical_machine)
{
    size_t i;

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
5088
        if (STREQ(canonical_machine, qemuCaps->machineTypes[i].name))
5089 5090 5091 5092
            return true;
    }
    return false;
}
5093 5094


5095 5096 5097 5098
/*
 * The preferred machine to use if none is listed explicitly
 * Note that this may differ from QEMU's own default machine
 */
5099
const char *
5100
virQEMUCapsGetPreferredMachine(virQEMUCapsPtr qemuCaps)
5101 5102 5103
{
    if (!qemuCaps->nmachineTypes)
        return NULL;
5104
    return qemuCaps->machineTypes[0].name;
5105
}
5106 5107


5108
static int
5109
virQEMUCapsFillDomainLoaderCaps(virDomainCapsLoaderPtr capsLoader,
5110 5111
                                virFirmwarePtr *firmwares,
                                size_t nfirmwares)
5112
{
5113 5114
    size_t i;

5115
    capsLoader->supported = true;
5116

5117
    if (VIR_ALLOC_N(capsLoader->values.values, nfirmwares) < 0)
5118 5119
        return -1;

5120 5121
    for (i = 0; i < nfirmwares; i++) {
        const char *filename = firmwares[i]->name;
5122 5123 5124 5125 5126 5127

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

5128
        if (VIR_STRDUP(capsLoader->values.values[capsLoader->values.nvalues],
5129 5130
                       filename) < 0)
            return -1;
5131
        capsLoader->values.nvalues++;
5132 5133
    }

5134
    VIR_DOMAIN_CAPS_ENUM_SET(capsLoader->type,
5135 5136
                             VIR_DOMAIN_LOADER_TYPE_ROM);

5137 5138
    VIR_DOMAIN_CAPS_ENUM_SET(capsLoader->type,
                             VIR_DOMAIN_LOADER_TYPE_PFLASH);
5139 5140


5141 5142 5143
    VIR_DOMAIN_CAPS_ENUM_SET(capsLoader->readonly,
                             VIR_TRISTATE_BOOL_YES,
                             VIR_TRISTATE_BOOL_NO);
5144
    return 0;
5145 5146 5147
}


5148
static int
5149
virQEMUCapsFillDomainOSCaps(virDomainCapsOSPtr os,
5150 5151
                            virFirmwarePtr *firmwares,
                            size_t nfirmwares)
5152
{
5153
    virDomainCapsLoaderPtr capsLoader = &os->loader;
5154

5155
    os->supported = true;
5156
    if (virQEMUCapsFillDomainLoaderCaps(capsLoader, firmwares, nfirmwares) < 0)
5157 5158
        return -1;
    return 0;
5159 5160 5161
}


5162 5163 5164 5165 5166
static int
virQEMUCapsFillDomainCPUCaps(virCapsPtr caps,
                             virQEMUCapsPtr qemuCaps,
                             virDomainCapsPtr domCaps)
{
5167 5168
    if (virQEMUCapsIsCPUModeSupported(qemuCaps, caps, domCaps->virttype,
                                      VIR_CPU_MODE_HOST_PASSTHROUGH))
5169 5170
        domCaps->cpu.hostPassthrough = true;

5171
    if (virQEMUCapsIsCPUModeSupported(qemuCaps, caps, domCaps->virttype,
5172
                                      VIR_CPU_MODE_HOST_MODEL)) {
5173 5174
        virCPUDefPtr cpu = virQEMUCapsGetHostModel(qemuCaps, domCaps->virttype,
                                                   VIR_QEMU_CAPS_HOST_CPU_REPORTED);
5175 5176
        domCaps->cpu.hostModel = virCPUDefCopy(cpu);
    }
5177 5178 5179 5180 5181

    if (virQEMUCapsIsCPUModeSupported(qemuCaps, caps, domCaps->virttype,
                                      VIR_CPU_MODE_CUSTOM)) {
        virDomainCapsCPUModelsPtr filtered = NULL;
        char **models = NULL;
5182
        const char *blacklist[] = { "host", NULL };
5183

J
Jiri Denemark 已提交
5184
        if (virCPUGetModels(domCaps->arch, &models) >= 0) {
5185 5186 5187 5188 5189 5190 5191 5192
            virDomainCapsCPUModelsPtr cpus;

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

            filtered = virDomainCapsCPUModelsFilter(cpus,
5193
                                                    (const char **)models,
5194
                                                    blacklist);
5195
            virStringListFree(models);
5196 5197
        }
        domCaps->cpu.custom = filtered;
5198
    }
5199 5200 5201 5202 5203

    return 0;
}


5204 5205 5206 5207 5208 5209 5210 5211 5212 5213
static int
virQEMUCapsFillDomainIOThreadCaps(virQEMUCapsPtr qemuCaps,
                                  virDomainCapsPtr domCaps)
{
    domCaps->iothreads = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_IOTHREAD);

    return 0;
}


5214
static int
5215
virQEMUCapsFillDomainDeviceDiskCaps(virQEMUCapsPtr qemuCaps,
5216
                                    const char *machine,
5217 5218
                                    virDomainCapsDeviceDiskPtr disk)
{
5219
    disk->supported = true;
5220 5221 5222
    /* QEMU supports all of these */
    VIR_DOMAIN_CAPS_ENUM_SET(disk->diskDevice,
                             VIR_DOMAIN_DISK_DEVICE_DISK,
5223 5224
                             VIR_DOMAIN_DISK_DEVICE_CDROM,
                             VIR_DOMAIN_DISK_DEVICE_LUN);
5225 5226

    /* PowerPC pseries based VMs do not support floppy device */
5227
    if (!qemuDomainMachineIsPSeries(machine, qemuCaps->arch))
5228
        VIR_DOMAIN_CAPS_ENUM_SET(disk->diskDevice, VIR_DOMAIN_DISK_DEVICE_FLOPPY);
5229

5230 5231 5232
    if (qemuDomainMachineHasBuiltinIDE(machine))
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_IDE);

5233 5234 5235 5236 5237
    VIR_DOMAIN_CAPS_ENUM_SET(disk->bus,
                             VIR_DOMAIN_DISK_BUS_SCSI,
                             VIR_DOMAIN_DISK_BUS_VIRTIO,
                             /* VIR_DOMAIN_DISK_BUS_SD */);

5238
    /* PowerPC pseries based VMs do not support floppy device */
5239
    if (!qemuDomainMachineIsPSeries(machine, qemuCaps->arch))
5240 5241
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_FDC);

5242 5243
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_USB_STORAGE))
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_USB);
5244 5245 5246 5247

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

5248
    return 0;
5249 5250 5251
}


5252 5253 5254 5255 5256 5257
static int
virQEMUCapsFillDomainDeviceGraphicsCaps(virQEMUCapsPtr qemuCaps,
                                        virDomainCapsDeviceGraphicsPtr dev)
{
    dev->supported = true;

J
Ján Tomko 已提交
5258
    VIR_DOMAIN_CAPS_ENUM_SET(dev->type, VIR_DOMAIN_GRAPHICS_TYPE_SDL);
5259 5260 5261 5262 5263 5264 5265 5266 5267
    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;
}


5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279
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);
5280
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_QXL))
5281 5282 5283 5284 5285 5286 5287 5288
        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;
}


5289
static int
5290 5291 5292 5293 5294 5295
virQEMUCapsFillDomainDeviceHostdevCaps(virQEMUCapsPtr qemuCaps,
                                       virDomainCapsDeviceHostdevPtr hostdev)
{
    bool supportsPassthroughKVM = qemuHostdevHostSupportsPassthroughLegacy();
    bool supportsPassthroughVFIO = qemuHostdevHostSupportsPassthroughVFIO();

5296
    hostdev->supported = true;
5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308
    /* 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,
5309 5310
                             VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI,
                             VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI);
5311 5312 5313 5314 5315 5316 5317 5318 5319 5320 5321 5322

    /* 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
5323
    if (supportsPassthroughKVM) {
5324 5325 5326 5327
        VIR_DOMAIN_CAPS_ENUM_SET(hostdev->pciBackend,
                                 VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT,
                                 VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM);
    }
5328
    return 0;
5329 5330 5331
}


5332 5333 5334 5335 5336 5337 5338
/**
 * virQEMUCapsSupportsGICVersion:
 * @qemuCaps: QEMU capabilities
 * @virtType: domain type
 * @version: GIC version
 *
 * Checks the QEMU binary with capabilities @qemuCaps supports a specific
5339 5340
 * GIC version for a domain of type @virtType. If @qemuCaps is NULL, the GIC
 * @version is considered unsupported.
5341 5342 5343 5344 5345 5346 5347 5348 5349 5350 5351
 *
 * Returns: true if the binary supports the requested GIC version, false
 *          otherwise
 */
bool
virQEMUCapsSupportsGICVersion(virQEMUCapsPtr qemuCaps,
                              virDomainVirtType virtType,
                              virGICVersion version)
{
    size_t i;

5352 5353 5354
    if (!qemuCaps)
        return false;

5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367 5368 5369 5370 5371 5372 5373
    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;
}


5374 5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396 5397
/**
 * 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;
5398
    virGICVersion version;
5399

5400
    if (!qemuDomainMachineIsARMVirt(domCaps->machine, domCaps->arch))
5401 5402
        return 0;

5403 5404 5405 5406 5407 5408
    for (version = VIR_GIC_VERSION_LAST - 1;
         version > VIR_GIC_VERSION_NONE;
         version--) {
        if (!virQEMUCapsSupportsGICVersion(qemuCaps,
                                           domCaps->virttype,
                                           version))
5409 5410 5411 5412
            continue;

        gic->supported = true;
        VIR_DOMAIN_CAPS_ENUM_SET(gic->version,
5413
                                 version);
5414 5415 5416 5417 5418 5419
    }

    return 0;
}


5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435
/**
 * 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;
5436
    VIR_AUTOPTR(virSEVCapability) sev = NULL;
5437 5438 5439 5440 5441 5442 5443 5444

    if (!cap)
        return 0;

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

    if (VIR_STRDUP(sev->pdh, cap->pdh) < 0)
5445
        return -1;
5446 5447

    if (VIR_STRDUP(sev->cert_chain, cap->cert_chain) < 0)
5448
        return -1;
5449 5450 5451 5452 5453

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

5454
    return 0;
5455 5456 5457
}


5458
int
5459 5460
virQEMUCapsFillDomainCaps(virCapsPtr caps,
                          virDomainCapsPtr domCaps,
5461
                          virQEMUCapsPtr qemuCaps,
5462
                          virFirmwarePtr *firmwares,
5463
                          size_t nfirmwares)
5464
{
5465
    virDomainCapsOSPtr os = &domCaps->os;
5466 5467
    virDomainCapsDeviceDiskPtr disk = &domCaps->disk;
    virDomainCapsDeviceHostdevPtr hostdev = &domCaps->hostdev;
5468
    virDomainCapsDeviceGraphicsPtr graphics = &domCaps->graphics;
5469
    virDomainCapsDeviceVideoPtr video = &domCaps->video;
5470

5471 5472
    domCaps->maxvcpus = virQEMUCapsGetMachineMaxCpus(qemuCaps,
                                                     domCaps->machine);
5473
    if (domCaps->virttype == VIR_DOMAIN_VIRT_KVM) {
5474 5475 5476 5477 5478 5479
        int hostmaxvcpus;

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

        domCaps->maxvcpus = MIN(domCaps->maxvcpus, hostmaxvcpus);
5480
    }
5481

5482 5483 5484
    domCaps->vmcoreinfo = virQEMUCapsGet(qemuCaps,
                                         QEMU_CAPS_DEVICE_VMCOREINFO);

5485 5486
    domCaps->genid = virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VMGENID);

5487
    if (virQEMUCapsFillDomainOSCaps(os, firmwares, nfirmwares) < 0 ||
5488
        virQEMUCapsFillDomainCPUCaps(caps, qemuCaps, domCaps) < 0 ||
5489
        virQEMUCapsFillDomainIOThreadCaps(qemuCaps, domCaps) < 0 ||
5490 5491 5492
        virQEMUCapsFillDomainDeviceDiskCaps(qemuCaps,
                                            domCaps->machine, disk) < 0 ||
        virQEMUCapsFillDomainDeviceGraphicsCaps(qemuCaps, graphics) < 0 ||
5493
        virQEMUCapsFillDomainDeviceVideoCaps(qemuCaps, video) < 0 ||
5494
        virQEMUCapsFillDomainDeviceHostdevCaps(qemuCaps, hostdev) < 0 ||
5495 5496
        virQEMUCapsFillDomainFeatureGICCaps(qemuCaps, domCaps) < 0 ||
        virQEMUCapsFillDomainFeatureSEVCaps(qemuCaps, domCaps) < 0)
5497
        return -1;
5498

5499
    return 0;
5500
}
5501 5502 5503 5504 5505 5506 5507 5508


void
virQEMUCapsSetMicrocodeVersion(virQEMUCapsPtr qemuCaps,
                               unsigned int microcodeVersion)
{
    qemuCaps->microcodeVersion = microcodeVersion;
}
5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524 5525


/**
 * 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);
}