qemu_capabilities.c 159.6 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
#include "qemu_process.h"
49

50
#include <fcntl.h>
51 52 53
#include <sys/stat.h>
#include <unistd.h>
#include <sys/wait.h>
54
#include <stdarg.h>
55
#include <sys/utsname.h>
56 57 58

#define VIR_FROM_THIS VIR_FROM_QEMU

59 60
VIR_LOG_INIT("qemu.qemu_capabilities");

61 62 63 64
/* While not public, these strings must not change. They
 * are used in domain status files which are read on
 * daemon restarts
 */
65
VIR_ENUM_IMPL(virQEMUCaps, QEMU_CAPS_LAST,
66 67
              /* 0 */
              "kqemu",
68 69 70 71 72
              "vnc-colon",
              "no-reboot",
              "drive",
              "drive-boot",

73 74
              /* 5 */
              "name",
75 76 77 78 79
              "uuid",
              "domid",
              "vnet-hdr",
              "migrate-kvm-stdio",

80 81
              /* 10 */
              "migrate-qemu-tcp",
82 83 84 85 86
              "migrate-qemu-exec",
              "drive-cache-v2",
              "kvm",
              "drive-format",

87 88
              /* 15 */
              "vga",
89 90 91 92 93
              "0.10",
              "pci-device",
              "mem-path",
              "drive-serial",

94 95
              /* 20 */
              "xen-domid",
96 97 98 99 100
              "migrate-qemu-unix",
              "chardev",
              "enable-kvm",
              "monitor-json",

101 102
              /* 25 */
              "balloon",
103 104 105 106 107
              "device",
              "sdl",
              "smp-topology",
              "netdev",

108 109
              /* 30 */
              "rtc",
110
              "vhost-net",
111 112 113 114
              "rtc-td-hack",
              "no-hpet",
              "no-kvm-pit",

115 116
              /* 35 */
              "tdf",
117 118 119 120 121
              "pci-configfd",
              "nodefconfig",
              "boot-menu",
              "enable-kqemu",

122 123
              /* 40 */
              "fsdev",
124 125 126 127 128
              "nesting",
              "name-process",
              "drive-readonly",
              "smbios-type",

129 130
              /* 45 */
              "vga-qxl",
131 132 133 134 135
              "spice",
              "vga-none",
              "migrate-qemu-fd",
              "boot-index",

136 137
              /* 50 */
              "hda-duplex",
138 139 140 141 142
              "drive-aio",
              "pci-multibus",
              "pci-bootindex",
              "ccid-emulated",

143 144
              /* 55 */
              "ccid-passthru",
145 146 147 148
              "chardev-spicevmc",
              "device-spicevmc",
              "virtio-tx-alg",
              "device-qxl-vga",
149

150 151
              /* 60 */
              "pci-multifunction",
152
              "virtio-blk-pci.ioeventfd",
M
Michal Privoznik 已提交
153
              "sga",
154 155
              "virtio-blk-pci.event_idx",
              "virtio-net-pci.event_idx",
156

157 158
              /* 65 */
              "cache-directsync",
159 160 161 162 163
              "piix3-usb-uhci",
              "piix4-usb-uhci",
              "usb-ehci",
              "ich9-usb-ehci1",

164 165
              /* 70 */
              "vt82c686b-usb-uhci",
166 167
              "pci-ohci",
              "usb-redir",
M
Marc-André Lureau 已提交
168
              "usb-hub",
169
              "no-shutdown",
170

171 172
              /* 75 */
              "cache-unsafe",
173
              "rombar",
J
Jim Fehlig 已提交
174
              "ich9-ahci",
175
              "no-acpi",
176
              "fsdev-readonly",
177

178 179
              /* 80 */
              "virtio-blk-pci.scsi",
180
              "blk-sg-io",
O
Osier Yang 已提交
181
              "drive-copy-on-read",
182
              "cpu-host",
183
              "fsdev-writeout",
184

185 186
              /* 85 */
              "drive-iotune",
187
              "system_wakeup",
188
              "scsi-disk.channel",
189
              "scsi-block",
190
              "transaction",
191

192 193
              /* 90 */
              "block-job-sync",
194
              "block-job-async",
195
              "scsi-cd",
196
              "ide-cd",
197
              "no-user-config",
M
Marc-André Lureau 已提交
198

199 200
              /* 95 */
              "hda-micro",
201
              "dump-guest-memory",
G
Gerd Hoffmann 已提交
202
              "nec-usb-xhci",
203
              "virtio-s390",
204
              "balloon-event",
M
Marc-André Lureau 已提交
205

206 207
              /* 100 */
              "bridge",
208 209
              "lsi",
              "virtio-scsi-pci",
V
Viktor Mihajlovski 已提交
210
              "blockio",
211
              "disable-s3",
R
Richa Marwaha 已提交
212

213 214
              /* 105 */
              "disable-s4",
215
              "usb-redir.filter",
216 217
              "ide-drive.wwn",
              "scsi-disk.wwn",
218
              "seccomp-sandbox",
219

220 221
              /* 110 */
              "reboot-timeout",
222
              "dump-guest-core",
223
              "seamless-migration",
224
              "block-commit",
225
              "vnc",
226

227 228
              /* 115 */
              "drive-mirror",
229 230
              "usb-redir.bootindex",
              "usb-host.bootindex",
231
              "blockdev-snapshot-sync",
232 233
              "qxl",

234 235
              /* 120 */
              "VGA",
236 237 238
              "cirrus-vga",
              "vmware-svga",
              "device-video-primary",
239
              "s390-sclp",
240

241 242
              /* 125 */
              "usb-serial",
G
Guannan Ren 已提交
243
              "usb-net",
244
              "add-fd",
245
              "nbd-server",
246 247
              "virtio-rng",

248 249
              /* 130 */
              "rng-random",
250
              "rng-egd",
O
Olivia Yin 已提交
251 252
              "virtio-ccw",
              "dtb",
253
              "megasas",
J
Ján Tomko 已提交
254

255 256
              /* 135 */
              "ipv6-migration",
257
              "machine-opt",
L
Li Zhang 已提交
258
              "machine-usb-opt",
S
Stefan Berger 已提交
259 260
              "tpm-passthrough",
              "tpm-tis",
261

262 263
              /* 140 */
              "nvram",
H
Han Cheng 已提交
264 265 266 267 268
              "pci-bridge",
              "vfio-pci",
              "vfio-pci.bootindex",
              "scsi-generic",

269 270
              /* 145 */
              "scsi-generic.bootindex",
271
              "mem-merge",
272
              "vnc-websocket",
O
Osier Yang 已提交
273
              "drive-discard",
274
              "mlock",
275

276 277
              /* 150 */
              "vnc-share-policy",
278
              "device-del-event",
279
              "dmi-to-pci-bridge",
280 281
              "i440fx-pci-hole64-size",
              "q35-pci-hole64-size",
282

283 284
              /* 155 */
              "usb-storage",
285
              "usb-storage.removable",
286
              "virtio-mmio",
287
              "ich9-intel-hda",
288
              "kvm-pit-lost-tick-policy",
289

290 291
              /* 160 */
              "boot-strict",
292 293
              "pvpanic",
              "enable-fips",
294 295
              "spice-file-xfer-disable",
              "spiceport",
L
Li Zhang 已提交
296

297 298
              /* 165 */
              "usb-kbd",
299
              "host-pci-multidomain",
300
              "msg-timestamp",
301
              "active-commit",
302
              "change-backing-file",
303

304 305
              /* 170 */
              "memory-backend-ram",
306
              "numa",
307
              "memory-backend-file",
308
              "usb-audio",
309
              "rtc-reset-reinjection",
310

311 312
              /* 175 */
              "splash-timeout",
J
John Ferlan 已提交
313
              "iothread",
314
              "migrate-rdma",
315
              "ivshmem",
316
              "drive-iotune-max",
317

318 319
              /* 180 */
              "VGA.vgamem_mb",
320 321 322
              "vmware-svga.vgamem_mb",
              "qxl.vgamem_mb",
              "qxl-vga.vgamem_mb",
323
              "pc-dimm",
324

325 326
              /* 185 */
              "machine-vmport-opt",
327 328
              "aes-key-wrap",
              "dea-key-wrap",
M
Michal Privoznik 已提交
329
              "pci-serial",
330
              "aarch64-off",
331

332 333
              /* 190 */
              "vhost-user-multiqueue",
334
              "migration-event",
335
              "gpex-pcihost",
336
              "ioh3420",
337
              "x3130-upstream",
338

339 340
              /* 195 */
              "xio3130-downstream",
341
              "rtl8139",
342
              "e1000",
343
              "virtio-net",
344
              "gic-version",
345

346 347
              /* 200 */
              "incoming-defer",
M
Marc-André Lureau 已提交
348
              "virtio-gpu",
349
              "virtio-gpu.virgl",
350 351 352
              "virtio-keyboard",
              "virtio-mouse",

353 354
              /* 205 */
              "virtio-tablet",
355
              "virtio-input-host",
356
              "chardev-file-append",
357 358
              "ich9-disable-s3",
              "ich9-disable-s4",
359

360 361
              /* 210 */
              "vserport-change-event",
362
              "virtio-balloon-pci.deflate-on-oom",
363
              "mptsas1068",
364
              "spice-gl",
365 366
              "qxl.vram64_size_mb",

367 368
              /* 215 */
              "qxl-vga.vram64_size_mb",
369
              "chardev-logfile",
370
              "debug-threads",
371
              "secret",
372
              "pxb",
373

374 375
              /* 220 */
              "pxb-pcie",
376
              "device-tray-moved-event",
377
              "nec-usb-xhci-ports",
378
              "virtio-scsi-pci.iothread",
379
              "name-guest",
380

381 382
              /* 225 */
              "qxl.max_outputs",
383
              "qxl-vga.max_outputs",
384
              "spice-unix",
385
              "drive-detect-zeroes",
B
Boris Fiuczynski 已提交
386
              "tls-creds-x509",
387

388 389
              /* 230 */
              "display",
J
Ján Tomko 已提交
390
              "intel-iommu",
M
Michal Privoznik 已提交
391
              "smm",
392
              "virtio-pci-disable-legacy",
393
              "query-hotpluggable-cpus",
394

395 396
              /* 235 */
              "virtio-net.rx_queue_size",
397
              "machine-iommu",
398
              "virtio-vga",
399
              "drive-iotune-max-length",
400 401
              "ivshmem-plain",

402 403
              /* 240 */
              "ivshmem-doorbell",
404
              "query-qmp-schema",
405
              "gluster.debug_level",
406
              "vhost-scsi",
407
              "drive-iotune-group",
408

409 410
              /* 245 */
              "query-cpu-model-expansion",
411
              "virtio-net.host_mtu",
412
              "spice-rendernode",
413
              "nvdimm",
414
              "pcie-root-port",
415

416 417
              /* 250 */
              "query-cpu-definitions",
418
              "block-write-threshold",
419
              "query-named-block-nodes",
420
              "cpu-cache",
421
              "qemu-xhci",
422

423 424
              /* 255 */
              "kernel-irqchip",
425
              "kernel-irqchip.split",
426
              "intel-iommu.intremap",
427
              "intel-iommu.caching-mode",
428
              "intel-iommu.eim",
429

430 431
              /* 260 */
              "intel-iommu.device-iotlb",
432 433
              "virtio.iommu_platform",
              "virtio.ats",
434
              "loadparm",
435
              "spapr-pci-host-bridge",
436 437

              /* 265 */
438
              "spapr-pci-host-bridge.numa_node",
439 440
              "vnc-multi-servers",
              "virtio-net.tx_queue_size",
441
              "chardev-reconnect",
442
              "virtio-gpu.max_outputs",
J
John Ferlan 已提交
443 444 445

              /* 270 */
              "vxhs",
446
              "virtio-blk.num-queues",
447
              "machine.pseries.resize-hpt",
M
Marc-André Lureau 已提交
448
              "vmcoreinfo",
449
              "spapr-vty",
450 451 452

              /* 275 */
              "sclplmconsole",
453
              "numa.dist",
454
              "disk-share-rw",
455
              "iscsi.password-secret",
456
              "isa-serial",
457 458 459

              /* 280 */
              "pl011",
460
              "machine.pseries.max-cpu-compat",
461
              "dump-completed",
462
              "virtio-gpu-ccw",
463 464 465 466 467
              "virtio-keyboard-ccw",

              /* 285 */
              "virtio-mouse-ccw",
              "virtio-tablet-ccw",
468
              "qcow2-luks",
469
              "pcie-pci-bridge",
470
              "seccomp-blacklist",
471 472 473

              /* 290 */
              "query-cpus-fast",
474
              "disk-write-cache",
475
              "nbd-tls",
476
              "tpm-crb",
477
              "pr-manager-helper",
478 479 480

              /* 295 */
              "qom-list-properties",
481
              "memory-backend-file.discard-data",
482
              "virtual-css-bridge",
483 484
              "virtual-css-bridge.cssid-unrestricted",
              "vfio-ccw",
485 486 487

              /* 300 */
              "sdl-gl",
488
              "screendump_device",
489
              "hda-output",
490
              "blockdev-del",
491
              "vmgenid",
492 493 494

              /* 305 */
              "vhost-vsock",
495
              "chardev-fd-pass",
496
              "tpm-emulator",
497 498
              "mch",
              "mch.extended-tseg-mbytes",
499 500 501

              /* 310 */
              "sev-guest",
502
              "machine.pseries.cap-hpt-max-page-size",
503
              "machine.pseries.cap-htm",
504
              "usb-storage.werror",
505
              "egl-headless",
506 507 508

              /* 315 */
              "vfio-pci.display",
509
              "blockdev",
B
Boris Fiuczynski 已提交
510
              "vfio-ap",
Y
Yi Min Zhao 已提交
511
              "zpci",
512
              "memory-backend-memfd",
513 514 515

              /* 320 */
              "memory-backend-memfd.hugetlb",
516
              "iothread.poll-max-ns",
517
              "machine.pseries.cap-nested-hv",
518 519
              "egl-headless.rendernode",
              "memory-backend-file.align",
520 521 522

              /* 325 */
              "memory-backend-file.pmem",
523
              "nvdimm.unarmed",
524
              "scsi-disk.device_id",
525 526
    );

527

528 529 530 531
struct virQEMUCapsMachineType {
    char *name;
    char *alias;
    unsigned int maxCpus;
532
    bool hotplugCpus;
533
    bool qemuDefault;
534
};
535 536 537 538 539 540 541 542 543 544 545

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;
546 547
    /* Migratable host CPU definition used for updating guest CPU. */
    virCPUDefPtr migratable;
548 549 550 551
    /* 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;
552 553
};

554 555 556 557 558 559
/*
 * 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.
560 561
 *
 * And don't forget to update virQEMUCapsNewCopy.
562
 */
563
struct _virQEMUCaps {
564
    virObject parent;
565

566
    bool usedQMP;
567
    bool kvmSupportsNesting;
568

569
    char *binary;
570
    time_t ctime;
571
    time_t libvirtCtime;
572

573
    virBitmapPtr flags;
574 575 576

    unsigned int version;
    unsigned int kvmVersion;
577
    unsigned int libvirtVersion;
578
    unsigned int microcodeVersion;
579
    char *package;
580
    char *kernelVersion;
581

582
    virArch arch;
583

584 585
    virDomainCapsCPUModelsPtr kvmCPUModels;
    virDomainCapsCPUModelsPtr tcgCPUModels;
586 587

    size_t nmachineTypes;
588
    struct virQEMUCapsMachineType *machineTypes;
A
Andrea Bolognani 已提交
589 590 591

    size_t ngicCapabilities;
    virGICCapability *gicCapabilities;
592

593 594
    virSEVCapability *sevCapabilities;

595 596
    virQEMUCapsHostCPUData kvmCPU;
    virQEMUCapsHostCPUData tcgCPU;
597 598
};

599 600
struct virQEMUCapsSearchData {
    virArch arch;
601
    const char *binaryFilter;
602 603
};

604

605 606
static virClassPtr virQEMUCapsClass;
static void virQEMUCapsDispose(void *obj);
607

608
static int virQEMUCapsOnceInit(void)
609
{
610
    if (!VIR_CLASS_NEW(virQEMUCaps, virClassForObject()))
611 612 613 614 615
        return -1;

    return 0;
}

616
VIR_ONCE_GLOBAL_INIT(virQEMUCaps);
617

618
static virArch virQEMUCapsArchFromString(const char *arch)
619 620 621 622 623
{
    if (STREQ(arch, "i386"))
        return VIR_ARCH_I686;
    if (STREQ(arch, "arm"))
        return VIR_ARCH_ARMV7L;
624 625
    if (STREQ(arch, "or32"))
        return VIR_ARCH_OR32;
626 627 628 629 630

    return virArchFromString(arch);
}


631
static const char *virQEMUCapsArchToString(virArch arch)
632 633 634
{
    if (arch == VIR_ARCH_I686)
        return "i386";
S
Stefan Schallenberg 已提交
635
    else if (arch == VIR_ARCH_ARMV6L || arch == VIR_ARCH_ARMV7L)
636
        return "arm";
637 638
    else if (arch == VIR_ARCH_OR32)
        return "or32";
639 640 641 642

    return virArchToString(arch);
}

643 644 645

/* Checks whether a domain with @guest arch can run natively on @host.
 */
646
bool
647 648 649
virQEMUCapsGuestIsNative(virArch host,
                         virArch guest)
{
650
    /* host & guest arches match */
651 652 653
    if (host == guest)
        return true;

654
    /* hostarch is x86_64 and guest arch is i686 (needs -cpu qemu32) */
655 656 657
    if (host == VIR_ARCH_X86_64 && guest == VIR_ARCH_I686)
        return true;

658
    /* hostarch is aarch64 and guest arch is armv7l (needs -cpu aarch64=off) */
659 660 661
    if (host == VIR_ARCH_AARCH64 && guest == VIR_ARCH_ARMV7L)
        return true;

662
    /* hostarch and guestarch are both ppc64 */
663 664 665 666 667 668 669
    if (ARCH_IS_PPC64(host) && ARCH_IS_PPC64(guest))
        return true;

    return false;
}


670 671 672 673 674 675 676 677 678
/* 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)
{
679 680 681
    if (virQEMUCapsGuestIsNative(hostarch, guestarch))
        guestarch = hostarch;

682 683 684 685 686 687
    /* Both ppc64 and ppc64le guests can use the ppc64 target */
    if (ARCH_IS_PPC64(guestarch))
        guestarch = VIR_ARCH_PPC64;

    return guestarch;
}
688

689

690
static void
691 692
virQEMUCapsSetDefaultMachine(virQEMUCapsPtr qemuCaps,
                             size_t defIdx)
693
{
694
    struct virQEMUCapsMachineType tmp = qemuCaps->machineTypes[defIdx];
695 696 697 698

    memmove(qemuCaps->machineTypes + 1,
            qemuCaps->machineTypes,
            sizeof(qemuCaps->machineTypes[0]) * defIdx);
699 700

    qemuCaps->machineTypes[0] = tmp;
701 702
}

703

704
static char *
705 706
virQEMUCapsFindBinary(const char *format,
                      const char *archstr)
707
{
708 709
    char *ret = NULL;
    char *binary = NULL;
710

711
    if (virAsprintf(&binary, format, archstr) < 0)
712
        return NULL;
713 714 715

    ret = virFindFileInPath(binary);
    VIR_FREE(binary);
716 717 718 719 720 721 722 723 724 725 726
    return ret;
}

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

727 728 729 730 731 732 733 734 735
    /* 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;
    }

736 737 738 739 740 741 742 743 744 745 746
    /* 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;
747
    }
748

749
 out:
750 751 752
    return ret;
}

753
static int
754
virQEMUCapsInitGuest(virCapsPtr caps,
755
                     virFileCachePtr cache,
756 757
                     virArch hostarch,
                     virArch guestarch)
758 759
{
    char *binary = NULL;
760
    virQEMUCapsPtr qemuCaps = NULL;
761 762
    int ret = -1;

J
Ján Tomko 已提交
763
    /* Check for existence of base emulator, or alternate base
764 765
     * which can be used with magic cpu choice
     */
766
    binary = virQEMUCapsFindBinaryForArch(hostarch, guestarch);
767

768 769 770 771 772 773 774
    /* 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;
    }

775
    /* Ignore binary if extracting version info fails */
776
    if (binary) {
777
        if (!(qemuCaps = virQEMUCapsCacheLookup(cache, binary))) {
778 779 780 781
            virResetLastError();
            VIR_FREE(binary);
        }
    }
782

783
    ret = virQEMUCapsInitGuestFromBinary(caps,
784
                                         binary, qemuCaps,
785 786 787
                                         guestarch);

    VIR_FREE(binary);
788
    virObjectUnref(qemuCaps);
789 790 791 792 793 794 795

    return ret;
}

int
virQEMUCapsInitGuestFromBinary(virCapsPtr caps,
                               const char *binary,
796
                               virQEMUCapsPtr qemuCaps,
797 798 799 800 801 802 803 804
                               virArch guestarch)
{
    virCapsGuestPtr guest;
    virCapsGuestMachinePtr *machines = NULL;
    size_t nmachines = 0;
    int ret = -1;
    bool hasdisksnapshot = false;

805 806 807
    if (!binary)
        return 0;

808
    if (virQEMUCapsGetMachineTypesCaps(qemuCaps, &nmachines, &machines) < 0)
809
        goto cleanup;
810 811 812 813

    /* We register kvm as the base emulator too, since we can
     * just give -no-kvm to disable acceleration if required */
    if ((guest = virCapabilitiesAddGuest(caps,
814
                                         VIR_DOMAIN_OSTYPE_HVM,
815
                                         guestarch,
816 817 818 819
                                         binary,
                                         NULL,
                                         nmachines,
                                         machines)) == NULL)
820
        goto cleanup;
821 822 823 824

    machines = NULL;
    nmachines = 0;

A
Andrea Bolognani 已提交
825 826 827
    /* CPU selection is always available, because all QEMU versions
     * we support can use at least '-cpu host' */
    if (!virCapabilitiesAddGuestFeature(guest, "cpuselection", true, false))
828
        goto cleanup;
829

830
    if (!virCapabilitiesAddGuestFeature(guest, "deviceboot", true, false))
831
        goto cleanup;
832

833
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DISK_SNAPSHOT))
834 835
        hasdisksnapshot = true;

836 837
    if (!virCapabilitiesAddGuestFeature(guest, "disksnapshot", hasdisksnapshot,
                                        false))
838
        goto cleanup;
839

D
Daniel P. Berrange 已提交
840
    if (virCapabilitiesAddGuestDomain(guest,
841
                                      VIR_DOMAIN_VIRT_QEMU,
D
Daniel P. Berrange 已提交
842 843 844 845
                                      NULL,
                                      NULL,
                                      0,
                                      NULL) == NULL)
846
        goto cleanup;
847

848
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
849 850 851 852 853 854
        if (virCapabilitiesAddGuestDomain(guest,
                                          VIR_DOMAIN_VIRT_KVM,
                                          NULL,
                                          NULL,
                                          0,
                                          NULL) == NULL) {
855
            goto cleanup;
D
Daniel P. Berrange 已提交
856
        }
857
    }
858

859 860 861
    if ((ARCH_IS_X86(guestarch) || guestarch == VIR_ARCH_AARCH64) &&
        virCapabilitiesAddGuestFeature(guest, "acpi", true, true) == NULL) {
        goto cleanup;
862 863
    }

A
Andrea Bolognani 已提交
864
    if (ARCH_IS_X86(guestarch) &&
865
        virCapabilitiesAddGuestFeature(guest, "apic", true, false) == NULL) {
866
        goto cleanup;
867
    }
868

869
    if ((guestarch == VIR_ARCH_I686) &&
870 871
        (virCapabilitiesAddGuestFeature(guest, "pae", true, false) == NULL ||
         virCapabilitiesAddGuestFeature(guest, "nonpae", true, false) == NULL))
872
        goto cleanup;
873 874 875

    ret = 0;

876
 cleanup:
877 878 879

    virCapabilitiesFreeMachines(machines, nmachines);

880
    return ret;
881 882 883
}


884
virCPUDefPtr
885
virQEMUCapsProbeHostCPUForEmulator(virArch hostArch,
886 887 888
                                   virQEMUCapsPtr qemuCaps,
                                   virDomainVirtType type)
{
889 890
    return virCPUGetHost(hostArch, VIR_CPU_TYPE_GUEST, NULL,
                         virQEMUCapsGetCPUDefinitions(qemuCaps, type));
891 892 893
}


894 895
virCapsPtr
virQEMUCapsInit(virFileCachePtr cache)
896 897
{
    virCapsPtr caps;
898
    size_t i;
T
Tal Kain 已提交
899
    virArch hostarch = virArchFromHost();
900

T
Tal Kain 已提交
901
    if ((caps = virCapabilitiesNew(hostarch,
902
                                   true, true)) == NULL)
903
        goto error;
904

905
    /* Some machines have problematic NUMA topology causing
906 907 908
     * 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
909
    if (virCapabilitiesInitNUMA(caps) < 0) {
910
        virCapabilitiesFreeNUMAInfo(caps);
911
        VIR_WARN("Failed to query host NUMA topology, disabling NUMA capabilities");
912 913
    }

914 915 916
    if (virCapabilitiesInitCaches(caps) < 0)
        VIR_WARN("Failed to get host CPU cache info");

M
Martin Kletzander 已提交
917
    if (!(caps->host.cpu = virCPUProbeHost(caps->host.arch)))
918
        VIR_WARN("Failed to get host CPU");
919

920
    /* Add the power management features of the host */
921
    if (virNodeSuspendGetTargetMask(&caps->host.powerMgmt) < 0)
922 923
        VIR_WARN("Failed to get host power management capabilities");

924 925 926
    /* Add IOMMU info */
    virCapabilitiesHostInitIOMMU(caps);

M
Michal Privoznik 已提交
927
    /* Add huge pages info */
928
    if (virCapabilitiesInitPages(caps) < 0)
M
Michal Privoznik 已提交
929 930
        VIR_WARN("Failed to get pages info");

931 932 933
    /* Add domain migration transport URIs */
    virCapabilitiesAddHostMigrateTransport(caps, "tcp");
    virCapabilitiesAddHostMigrateTransport(caps, "rdma");
934

935 936 937 938
    /* 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
     */
939
    for (i = 0; i < VIR_ARCH_LAST; i++)
940
        if (virQEMUCapsInitGuest(caps, cache,
T
Tal Kain 已提交
941
                                 hostarch,
942
                                 i) < 0)
943
            goto error;
944 945 946

    return caps;

947
 error:
948
    virObjectUnref(caps);
949 950 951 952
    return NULL;
}


953
struct virQEMUCapsStringFlags {
954 955 956 957 958
    const char *value;
    int flag;
};


959 960
struct virQEMUCapsStringFlags virQEMUCapsCommands[] = {
    { "transaction", QEMU_CAPS_TRANSACTION },
961
    { "block-stream", QEMU_CAPS_BLOCKJOB_ASYNC },
962 963 964 965 966 967 968
    { "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 },
969
    { "change-backing-file", QEMU_CAPS_CHANGE_BACKING_FILE },
970
    { "rtc-reset-reinjection", QEMU_CAPS_RTC_RESET_REINJECTION },
971
    { "migrate-incoming", QEMU_CAPS_INCOMING_DEFER },
972
    { "query-hotpluggable-cpus", QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS },
973
    { "query-qmp-schema", QEMU_CAPS_QUERY_QMP_SCHEMA },
974 975 976
    { "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 },
977
    { "query-cpus-fast", QEMU_CAPS_QUERY_CPUS_FAST },
978
    { "qom-list-properties", QEMU_CAPS_QOM_LIST_PROPERTIES },
979
    { "blockdev-del", QEMU_CAPS_BLOCKDEV_DEL },
980 981
};

982 983 984 985
struct virQEMUCapsStringFlags virQEMUCapsMigration[] = {
    { "rdma-pin-all", QEMU_CAPS_MIGRATE_RDMA },
};

986
/* Use virQEMUCapsQMPSchemaQueries for querying parameters of events */
987
struct virQEMUCapsStringFlags virQEMUCapsEvents[] = {
988
    { "MIGRATION", QEMU_CAPS_MIGRATION_EVENT },
989
    { "VSERPORT_CHANGE", QEMU_CAPS_VSERPORT_CHANGE },
990
    { "BLOCK_WRITE_THRESHOLD", QEMU_CAPS_BLOCK_WRITE_THRESHOLD },
991
    { "DUMP_COMPLETED", QEMU_CAPS_DUMP_COMPLETED },
992 993
};

994
struct virQEMUCapsStringFlags virQEMUCapsObjectTypes[] = {
995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009
    { "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 },
1010
    { "virtio-blk-ccw", QEMU_CAPS_VIRTIO_CCW },
1011
    { "sclpconsole", QEMU_CAPS_DEVICE_SCLPCONSOLE },
1012
    { "lsi53c895a", QEMU_CAPS_SCSI_LSI },
1013
    { "virtio-scsi-pci", QEMU_CAPS_VIRTIO_SCSI },
1014 1015
    { "virtio-scsi-s390", QEMU_CAPS_VIRTIO_SCSI },
    { "virtio-scsi-ccw", QEMU_CAPS_VIRTIO_SCSI },
1016
    { "virtio-scsi-device", QEMU_CAPS_VIRTIO_SCSI },
1017
    { "megasas", QEMU_CAPS_SCSI_MEGASAS },
1018
    { "qxl", QEMU_CAPS_DEVICE_QXL },
1019 1020
    { "sga", QEMU_CAPS_SGA },
    { "scsi-block", QEMU_CAPS_SCSI_BLOCK },
1021 1022 1023
    { "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
1024
    { "usb-serial", QEMU_CAPS_DEVICE_USB_SERIAL },
1025
    { "virtio-rng-pci", QEMU_CAPS_DEVICE_VIRTIO_RNG },
1026 1027
    { "virtio-rng-s390", QEMU_CAPS_DEVICE_VIRTIO_RNG },
    { "virtio-rng-ccw", QEMU_CAPS_DEVICE_VIRTIO_RNG },
1028
    { "virtio-rng-device", QEMU_CAPS_DEVICE_VIRTIO_RNG },
1029
    { "rng-random", QEMU_CAPS_OBJECT_RNG_RANDOM },
1030
    { "rng-egd", QEMU_CAPS_OBJECT_RNG_EGD },
1031
    { "spapr-nvram", QEMU_CAPS_DEVICE_NVRAM },
1032
    { "pci-bridge", QEMU_CAPS_DEVICE_PCI_BRIDGE },
1033
    { "vfio-pci", QEMU_CAPS_DEVICE_VFIO_PCI },
1034
    { "i82801b11-bridge", QEMU_CAPS_DEVICE_DMI_TO_PCI_BRIDGE },
1035
    { "usb-storage", QEMU_CAPS_DEVICE_USB_STORAGE },
1036
    { "virtio-mmio", QEMU_CAPS_DEVICE_VIRTIO_MMIO },
1037
    { "ich9-intel-hda", QEMU_CAPS_DEVICE_ICH9_INTEL_HDA },
H
Hu Tao 已提交
1038
    { "pvpanic", QEMU_CAPS_DEVICE_PANIC },
L
Li Zhang 已提交
1039
    { "usb-kbd", QEMU_CAPS_DEVICE_USB_KBD },
1040
    { "memory-backend-ram", QEMU_CAPS_OBJECT_MEMORY_RAM },
1041
    { "memory-backend-file", QEMU_CAPS_OBJECT_MEMORY_FILE },
1042
    { "usb-audio", QEMU_CAPS_OBJECT_USB_AUDIO },
J
John Ferlan 已提交
1043
    { "iothread", QEMU_CAPS_OBJECT_IOTHREAD},
1044
    { "ivshmem", QEMU_CAPS_DEVICE_IVSHMEM },
1045
    { "pc-dimm", QEMU_CAPS_DEVICE_PC_DIMM },
M
Michal Privoznik 已提交
1046
    { "pci-serial", QEMU_CAPS_DEVICE_PCI_SERIAL },
1047
    { "gpex-pcihost", QEMU_CAPS_OBJECT_GPEX},
1048
    { "ioh3420", QEMU_CAPS_DEVICE_IOH3420 },
1049
    { "x3130-upstream", QEMU_CAPS_DEVICE_X3130_UPSTREAM },
1050
    { "xio3130-downstream", QEMU_CAPS_DEVICE_XIO3130_DOWNSTREAM },
1051
    { "rtl8139", QEMU_CAPS_DEVICE_RTL8139 },
1052
    { "e1000", QEMU_CAPS_DEVICE_E1000 },
1053 1054 1055 1056
    { "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
1057 1058
    { "virtio-gpu-pci", QEMU_CAPS_DEVICE_VIRTIO_GPU },
    { "virtio-gpu-device", QEMU_CAPS_DEVICE_VIRTIO_GPU },
1059
    { "virtio-vga", QEMU_CAPS_DEVICE_VIRTIO_VGA },
1060 1061 1062 1063 1064 1065
    { "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 },
1066 1067
    { "virtio-input-host-device", QEMU_CAPS_VIRTIO_INPUT_HOST },
    { "virtio-input-host-pci", QEMU_CAPS_VIRTIO_INPUT_HOST },
1068
    { "mptsas1068", QEMU_CAPS_SCSI_MPTSAS1068 },
1069
    { "secret", QEMU_CAPS_OBJECT_SECRET },
1070
    { "pxb", QEMU_CAPS_DEVICE_PXB },
1071
    { "pxb-pcie", QEMU_CAPS_DEVICE_PXB_PCIE },
1072
    { "tls-creds-x509", QEMU_CAPS_OBJECT_TLS_CREDS_X509 },
J
Ján Tomko 已提交
1073
    { "intel-iommu", QEMU_CAPS_DEVICE_INTEL_IOMMU },
1074 1075
    { "ivshmem-plain", QEMU_CAPS_DEVICE_IVSHMEM_PLAIN },
    { "ivshmem-doorbell", QEMU_CAPS_DEVICE_IVSHMEM_DOORBELL },
1076
    { "vhost-scsi", QEMU_CAPS_DEVICE_VHOST_SCSI },
1077
    { "nvdimm", QEMU_CAPS_DEVICE_NVDIMM },
1078
    { "pcie-root-port", QEMU_CAPS_DEVICE_PCIE_ROOT_PORT },
1079
    { "qemu-xhci", QEMU_CAPS_DEVICE_QEMU_XHCI },
1080
    { "spapr-pci-host-bridge", QEMU_CAPS_DEVICE_SPAPR_PCI_HOST_BRIDGE },
M
Marc-André Lureau 已提交
1081
    { "vmcoreinfo", QEMU_CAPS_DEVICE_VMCOREINFO },
1082
    { "spapr-vty", QEMU_CAPS_DEVICE_SPAPR_VTY },
1083
    { "sclplmconsole", QEMU_CAPS_DEVICE_SCLPLMCONSOLE },
1084
    { "isa-serial", QEMU_CAPS_DEVICE_ISA_SERIAL },
1085
    { "pl011", QEMU_CAPS_DEVICE_PL011 },
1086
    { "virtio-gpu-ccw", QEMU_CAPS_DEVICE_VIRTIO_GPU_CCW },
1087 1088 1089
    { "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 },
1090
    { "pcie-pci-bridge", QEMU_CAPS_DEVICE_PCIE_PCI_BRIDGE },
1091
    { "pr-manager-helper", QEMU_CAPS_PR_MANAGER_HELPER },
1092
    { "virtual-css-bridge", QEMU_CAPS_CCW },
1093
    { "vfio-ccw", QEMU_CAPS_DEVICE_VFIO_CCW },
1094
    { "hda-output", QEMU_CAPS_HDA_OUTPUT },
1095
    { "vmgenid", QEMU_CAPS_DEVICE_VMGENID },
1096
    { "vhost-vsock-device", QEMU_CAPS_DEVICE_VHOST_VSOCK },
1097
    { "mch", QEMU_CAPS_DEVICE_MCH },
1098
    { "sev-guest", QEMU_CAPS_SEV_GUEST },
B
Boris Fiuczynski 已提交
1099
    { "vfio-ap", QEMU_CAPS_DEVICE_VFIO_AP },
Y
Yi Min Zhao 已提交
1100
    { "zpci", QEMU_CAPS_DEVICE_ZPCI },
1101
    { "memory-backend-memfd", QEMU_CAPS_OBJECT_MEMORY_MEMFD },
1102 1103
};

1104
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioBalloon[] = {
1105
    { "deflate-on-oom", QEMU_CAPS_VIRTIO_BALLOON_AUTODEFLATE },
1106 1107 1108
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1109 1110
};

1111
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioBlk[] = {
1112 1113 1114 1115
    { "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 },
1116
    { "num-queues", QEMU_CAPS_VIRTIO_BLK_NUM_QUEUES },
1117
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1118 1119 1120
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1121
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1122 1123
};

1124
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioNet[] = {
1125 1126
    { "tx", QEMU_CAPS_VIRTIO_TX_ALG },
    { "event_idx", QEMU_CAPS_VIRTIO_NET_EVENT_IDX },
1127
    { "rx_queue_size", QEMU_CAPS_VIRTIO_NET_RX_QUEUE_SIZE },
1128
    { "tx_queue_size", QEMU_CAPS_VIRTIO_NET_TX_QUEUE_SIZE },
1129
    { "host_mtu", QEMU_CAPS_VIRTIO_NET_HOST_MTU },
1130 1131 1132
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1133 1134
};

1135
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsSpaprPCIHostBridge[] = {
1136 1137 1138
    { "numa_node", QEMU_CAPS_SPAPR_PCI_HOST_BRIDGE_NUMA_NODE },
};

1139
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioSCSI[] = {
1140
    { "iothread", QEMU_CAPS_VIRTIO_SCSI_IOTHREAD },
1141 1142 1143
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1144 1145
};

1146
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVfioPCI[] = {
1147
    { "display", QEMU_CAPS_VFIO_PCI_DISPLAY },
1148 1149
};

1150
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsSCSIDisk[] = {
1151 1152
    { "channel", QEMU_CAPS_SCSI_DISK_CHANNEL },
    { "wwn", QEMU_CAPS_SCSI_DISK_WWN },
1153
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1154
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1155
    { "device_id", QEMU_CAPS_SCSI_DISK_DEVICE_ID },
1156 1157
};

1158
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsIDEDrive[] = {
1159
    { "wwn", QEMU_CAPS_IDE_DRIVE_WWN },
1160
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1161
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1162 1163
};

1164
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsPiix4PM[] = {
1165 1166
    { "disable_s3", QEMU_CAPS_PIIX_DISABLE_S3 },
    { "disable_s4", QEMU_CAPS_PIIX_DISABLE_S4 },
1167 1168
};

1169
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsUSBRedir[] = {
1170
    { "filter", QEMU_CAPS_USB_REDIR_FILTER },
1171 1172
};

1173
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsI440FXPCIHost[] = {
1174 1175 1176
    { "pci-hole64-size", QEMU_CAPS_I440FX_PCI_HOLE64_SIZE },
};

1177
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsQ35PCIHost[] = {
1178 1179 1180
    { "pci-hole64-size", QEMU_CAPS_Q35_PCI_HOLE64_SIZE },
};

1181
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsUSBStorage[] = {
1182
    { "removable", QEMU_CAPS_USB_STORAGE_REMOVABLE },
1183
    { "share-rw", QEMU_CAPS_DISK_SHARE_RW },
1184
    { "write-cache", QEMU_CAPS_DISK_WRITE_CACHE },
1185
    { "werror", QEMU_CAPS_USB_STORAGE_WERROR },
1186 1187
};

1188
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsKVMPit[] = {
1189 1190 1191
    { "lost_tick_policy", QEMU_CAPS_KVM_PIT_TICK_POLICY },
};

1192
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVGA[] = {
1193 1194 1195
    { "vgamem_mb", QEMU_CAPS_VGA_VGAMEM },
};

1196
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVmwareSvga[] = {
1197 1198 1199
    { "vgamem_mb", QEMU_CAPS_VMWARE_SVGA_VGAMEM },
};

1200
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsQxl[] = {
1201
    { "vgamem_mb", QEMU_CAPS_QXL_VGAMEM },
1202
    { "vram64_size_mb", QEMU_CAPS_QXL_VRAM64 },
1203
    { "max_outputs", QEMU_CAPS_QXL_MAX_OUTPUTS },
1204 1205
};

1206
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsVirtioGpu[] = {
1207
    { "virgl", QEMU_CAPS_VIRTIO_GPU_VIRGL },
1208
    { "max_outputs", QEMU_CAPS_VIRTIO_GPU_MAX_OUTPUTS },
1209 1210 1211
    { "disable-legacy", QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY },
    { "iommu_platform", QEMU_CAPS_VIRTIO_PCI_IOMMU_PLATFORM },
    { "ats", QEMU_CAPS_VIRTIO_PCI_ATS },
1212 1213
};

1214
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsICH9[] = {
1215 1216 1217 1218
    { "disable_s3", QEMU_CAPS_ICH9_DISABLE_S3 },
    { "disable_s4", QEMU_CAPS_ICH9_DISABLE_S4 },
};

1219
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsUSBNECXHCI[] = {
1220 1221 1222
    { "p3", QEMU_CAPS_NEC_USB_XHCI_PORTS },
};

1223
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsIntelIOMMU[] = {
1224
    { "intremap", QEMU_CAPS_INTEL_IOMMU_INTREMAP },
1225
    { "caching-mode", QEMU_CAPS_INTEL_IOMMU_CACHING_MODE },
1226
    { "eim", QEMU_CAPS_INTEL_IOMMU_EIM },
1227
    { "device-iotlb", QEMU_CAPS_INTEL_IOMMU_DEVICE_IOTLB },
1228 1229
};

1230 1231 1232 1233
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsVirtualCSSBridge[] = {
    { "cssid-unrestricted", QEMU_CAPS_CCW_CSSID_UNRESTRICTED },
};

1234 1235 1236 1237
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsMCH[] = {
    { "extended-tseg-mbytes", QEMU_CAPS_MCH_EXTENDED_TSEG_MBYTES },
};

1238 1239 1240 1241
static struct virQEMUCapsStringFlags virQEMUCapsDevicePropsNVDIMM[] = {
    { "unarmed", QEMU_CAPS_DEVICE_NVDIMM_UNARMED },
};

1242
/* see documentation for virQEMUQAPISchemaPathGet for the query format */
1243
static struct virQEMUCapsStringFlags virQEMUCapsQMPSchemaQueries[] = {
1244
    { "blockdev-add/arg-type/options/+gluster/debug-level", QEMU_CAPS_GLUSTER_DEBUG_LEVEL},
1245
    { "blockdev-add/arg-type/+gluster/debug", QEMU_CAPS_GLUSTER_DEBUG_LEVEL},
J
John Ferlan 已提交
1246
    { "blockdev-add/arg-type/+vxhs", QEMU_CAPS_VXHS},
1247
    { "blockdev-add/arg-type/+iscsi/password-secret", QEMU_CAPS_ISCSI_PASSWORD_SECRET },
1248
    { "blockdev-add/arg-type/+qcow2/encrypt/+luks/key-secret", QEMU_CAPS_QCOW2_LUKS },
1249
    { "nbd-server-start/arg-type/tls-creds", QEMU_CAPS_NBD_TLS },
1250
    { "screendump/arg-type/device", QEMU_CAPS_SCREENDUMP_DEVICE },
1251
    { "block-commit/arg-type/*top",  QEMU_CAPS_ACTIVE_COMMIT },
1252
    { "query-iothreads/ret-type/poll-max-ns", QEMU_CAPS_IOTHREAD_POLLING },
1253
    { "query-display-options/ret-type/+egl-headless/rendernode", QEMU_CAPS_EGL_HEADLESS_RENDERNODE },
1254 1255
};

1256 1257
typedef struct _virQEMUCapsObjectTypeProps virQEMUCapsObjectTypeProps;
struct _virQEMUCapsObjectTypeProps {
1258
    const char *type;
1259
    struct virQEMUCapsStringFlags *props;
1260
    size_t nprops;
1261
    int capsCondition;
1262 1263
};

1264 1265 1266 1267 1268
typedef int (*virQEMUCapsObjectTypePropsCB)(qemuMonitorPtr mon,
                                            const char *type,
                                            char ***props);

static virQEMUCapsObjectTypeProps virQEMUCapsDeviceProps[] = {
1269 1270
    { "virtio-blk-pci", virQEMUCapsDevicePropsVirtioBlk,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBlk),
1271
      -1 },
1272 1273
    { "virtio-net-pci", virQEMUCapsDevicePropsVirtioNet,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioNet),
1274
      QEMU_CAPS_DEVICE_VIRTIO_NET },
1275 1276
    { "virtio-scsi-pci", virQEMUCapsDevicePropsVirtioSCSI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioSCSI),
1277
      QEMU_CAPS_VIRTIO_SCSI },
1278 1279
    { "virtio-blk-ccw", virQEMUCapsDevicePropsVirtioBlk,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBlk),
1280
      QEMU_CAPS_VIRTIO_CCW },
1281 1282
    { "virtio-net-ccw", virQEMUCapsDevicePropsVirtioNet,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioNet),
1283
      QEMU_CAPS_DEVICE_VIRTIO_NET },
1284 1285
    { "virtio-scsi-ccw", virQEMUCapsDevicePropsVirtioSCSI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioSCSI),
1286
      QEMU_CAPS_VIRTIO_SCSI },
1287 1288
    { "virtio-blk-s390", virQEMUCapsDevicePropsVirtioBlk,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBlk),
1289
      QEMU_CAPS_VIRTIO_S390 },
1290 1291
    { "virtio-net-s390", virQEMUCapsDevicePropsVirtioNet,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioNet),
1292
      QEMU_CAPS_DEVICE_VIRTIO_NET },
1293 1294
    { "vfio-pci", virQEMUCapsDevicePropsVfioPCI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVfioPCI),
1295
      QEMU_CAPS_DEVICE_VFIO_PCI },
1296
    { "scsi-hd", virQEMUCapsDevicePropsSCSIDisk,
1297
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsSCSIDisk),
1298
      -1 },
1299
    { "ide-hd", virQEMUCapsDevicePropsIDEDrive,
1300
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsIDEDrive),
1301
      -1 },
1302 1303
    { "PIIX4_PM", virQEMUCapsDevicePropsPiix4PM,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsPiix4PM),
1304
      -1 },
1305 1306
    { "usb-redir", virQEMUCapsDevicePropsUSBRedir,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsUSBRedir),
1307
      QEMU_CAPS_USB_REDIR },
1308 1309
    { "i440FX-pcihost", virQEMUCapsDevicePropsI440FXPCIHost,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsI440FXPCIHost),
1310
      -1 },
1311 1312
    { "q35-pcihost", virQEMUCapsDevicePropsQ35PCIHost,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsQ35PCIHost),
1313
      -1 },
1314 1315
    { "usb-storage", virQEMUCapsDevicePropsUSBStorage,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsUSBStorage),
1316
      QEMU_CAPS_DEVICE_USB_STORAGE },
1317 1318
    { "kvm-pit", virQEMUCapsDevicePropsKVMPit,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsKVMPit),
1319
      -1 },
1320 1321
    { "VGA", virQEMUCapsDevicePropsVGA,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVGA),
1322
      QEMU_CAPS_DEVICE_VGA },
1323 1324
    { "vmware-svga", virQEMUCapsDevicePropsVmwareSvga,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVmwareSvga),
1325
      QEMU_CAPS_DEVICE_VMWARE_SVGA },
1326 1327
    { "qxl", virQEMUCapsDevicePropsQxl,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsQxl),
1328
      QEMU_CAPS_DEVICE_QXL },
1329 1330
    { "virtio-gpu-pci", virQEMUCapsDevicePropsVirtioGpu,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioGpu),
1331
      QEMU_CAPS_DEVICE_VIRTIO_GPU },
1332 1333
    { "virtio-gpu-device", virQEMUCapsDevicePropsVirtioGpu,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioGpu),
1334
      QEMU_CAPS_DEVICE_VIRTIO_GPU },
1335 1336
    { "ICH9-LPC", virQEMUCapsDevicePropsICH9,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsICH9),
1337
      -1 },
1338 1339
    { "virtio-balloon-pci", virQEMUCapsDevicePropsVirtioBalloon,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBalloon),
1340
      -1 },
1341 1342
    { "virtio-balloon-ccw", virQEMUCapsDevicePropsVirtioBalloon,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBalloon),
1343
      -1 },
1344 1345
    { "virtio-balloon-device", virQEMUCapsDevicePropsVirtioBalloon,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioBalloon),
1346
      -1 },
1347 1348
    { "nec-usb-xhci", virQEMUCapsDevicePropsUSBNECXHCI,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsUSBNECXHCI),
1349
      QEMU_CAPS_NEC_USB_XHCI },
1350 1351
    { "intel-iommu", virQEMUCapsDevicePropsIntelIOMMU,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsIntelIOMMU),
1352
      QEMU_CAPS_DEVICE_INTEL_IOMMU },
1353 1354
    { "spapr-pci-host-bridge", virQEMUCapsDevicePropsSpaprPCIHostBridge,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsSpaprPCIHostBridge),
1355
      QEMU_CAPS_DEVICE_SPAPR_PCI_HOST_BRIDGE },
1356 1357
    { "virtio-gpu-ccw", virQEMUCapsDevicePropsVirtioGpu,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsVirtioGpu),
1358
      QEMU_CAPS_DEVICE_VIRTIO_GPU_CCW },
1359 1360 1361
    { "virtual-css-bridge", virQEMUCapsObjectPropsVirtualCSSBridge,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsVirtualCSSBridge),
      QEMU_CAPS_CCW },
1362 1363 1364
    { "mch", virQEMUCapsDevicePropsMCH,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsMCH),
      QEMU_CAPS_DEVICE_MCH },
1365 1366 1367
    { "nvdimm", virQEMUCapsDevicePropsNVDIMM,
      ARRAY_CARDINALITY(virQEMUCapsDevicePropsNVDIMM),
      QEMU_CAPS_DEVICE_NVDIMM },
1368 1369
};

1370 1371
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsMemoryBackendFile[] = {
    { "discard-data", QEMU_CAPS_OBJECT_MEMORY_FILE_DISCARD },
1372
    { "align", QEMU_CAPS_OBJECT_MEMORY_FILE_ALIGN },
1373
    { "pmem", QEMU_CAPS_OBJECT_MEMORY_FILE_PMEM },
1374 1375
};

1376 1377 1378 1379
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsMemoryBackendMemfd[] = {
    { "hugetlb", QEMU_CAPS_OBJECT_MEMORY_MEMFD_HUGETLB },
};

1380 1381
static struct virQEMUCapsStringFlags virQEMUCapsObjectPropsSPAPRMachine[] = {
    { "cap-hpt-max-page-size", QEMU_CAPS_MACHINE_PSERIES_CAP_HPT_MAX_PAGE_SIZE },
1382
    { "cap-htm", QEMU_CAPS_MACHINE_PSERIES_CAP_HTM },
1383
    { "cap-nested-hv", QEMU_CAPS_MACHINE_PSERIES_CAP_NESTED_HV },
1384 1385
};

1386 1387 1388 1389
static virQEMUCapsObjectTypeProps virQEMUCapsObjectProps[] = {
    { "memory-backend-file", virQEMUCapsObjectPropsMemoryBackendFile,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsMemoryBackendFile),
      QEMU_CAPS_OBJECT_MEMORY_FILE },
1390 1391 1392
    { "memory-backend-memfd", virQEMUCapsObjectPropsMemoryBackendMemfd,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsMemoryBackendMemfd),
      QEMU_CAPS_OBJECT_MEMORY_MEMFD },
1393 1394 1395
    { "spapr-machine", virQEMUCapsObjectPropsSPAPRMachine,
      ARRAY_CARDINALITY(virQEMUCapsObjectPropsSPAPRMachine),
      -1 },
1396
};
1397 1398

static void
1399 1400 1401 1402 1403
virQEMUCapsProcessStringFlags(virQEMUCapsPtr qemuCaps,
                              size_t nflags,
                              struct virQEMUCapsStringFlags *flags,
                              size_t nvalues,
                              char *const*values)
1404 1405
{
    size_t i, j;
1406
    for (i = 0; i < nflags; i++) {
1407 1408 1409
        if (virQEMUCapsGet(qemuCaps, flags[i].flag))
            continue;

1410
        for (j = 0; j < nvalues; j++) {
1411
            if (STREQ(values[j], flags[i].value)) {
1412
                virQEMUCapsSet(qemuCaps, flags[i].flag);
1413 1414 1415 1416 1417 1418 1419
                break;
            }
        }
    }
}


1420
int virQEMUCapsGetDefaultVersion(virCapsPtr caps,
1421
                                 virFileCachePtr capsCache,
1422
                                 unsigned int *version)
1423
{
1424
    virQEMUCapsPtr qemucaps;
T
Tal Kain 已提交
1425
    virArch hostarch;
1426
    virCapsDomainDataPtr capsdata;
1427 1428 1429 1430

    if (*version > 0)
        return 0;

T
Tal Kain 已提交
1431
    hostarch = virArchFromHost();
1432 1433 1434
    if (!(capsdata = virCapabilitiesDomainDataLookup(caps,
            VIR_DOMAIN_OSTYPE_HVM, hostarch, VIR_DOMAIN_VIRT_QEMU,
            NULL, NULL))) {
1435
        virReportError(VIR_ERR_INTERNAL_ERROR,
1436
                       _("Cannot find suitable emulator for %s"),
T
Tal Kain 已提交
1437
                       virArchToString(hostarch));
1438 1439 1440
        return -1;
    }

1441
    qemucaps = virQEMUCapsCacheLookup(capsCache, capsdata->emulator);
1442 1443
    VIR_FREE(capsdata);
    if (!qemucaps)
1444 1445
        return -1;

1446
    *version = virQEMUCapsGetVersion(qemucaps);
1447
    virObjectUnref(qemucaps);
1448 1449
    return 0;
}
1450 1451


1452 1453


1454 1455
virQEMUCapsPtr
virQEMUCapsNew(void)
1456
{
1457
    virQEMUCapsPtr qemuCaps;
1458

1459
    if (virQEMUCapsInitialize() < 0)
1460 1461
        return NULL;

1462
    if (!(qemuCaps = virObjectNew(virQEMUCapsClass)))
1463 1464
        return NULL;

1465
    if (!(qemuCaps->flags = virBitmapNew(QEMU_CAPS_LAST)))
1466
        goto error;
1467

1468
    return qemuCaps;
1469

1470
 error:
1471
    virObjectUnref(qemuCaps);
1472
    return NULL;
1473 1474 1475
}


1476
static int
1477 1478
virQEMUCapsHostCPUDataCopy(virQEMUCapsHostCPUDataPtr dst,
                           virQEMUCapsHostCPUDataPtr src)
1479
{
1480 1481
    if (src->info &&
        !(dst->info = qemuMonitorCPUModelInfoCopy(src->info)))
1482 1483
        return -1;

1484 1485
    if (src->reported &&
        !(dst->reported = virCPUDefCopy(src->reported)))
1486 1487
        return -1;

1488 1489 1490 1491
    if (src->migratable &&
        !(dst->migratable = virCPUDefCopy(src->migratable)))
        return -1;

1492 1493 1494 1495
    if (src->full &&
        !(dst->full = virCPUDefCopy(src->full)))
        return -1;

1496 1497 1498 1499
    return 0;
}


1500
static void
1501
virQEMUCapsHostCPUDataClear(virQEMUCapsHostCPUDataPtr cpuData)
1502
{
1503
    qemuMonitorCPUModelInfoFree(cpuData->info);
1504
    virCPUDefFree(cpuData->reported);
1505
    virCPUDefFree(cpuData->migratable);
1506
    virCPUDefFree(cpuData->full);
1507 1508

    memset(cpuData, 0, sizeof(*cpuData));
1509 1510 1511
}


1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530
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;
}


1531
virQEMUCapsPtr virQEMUCapsNewCopy(virQEMUCapsPtr qemuCaps)
1532
{
1533
    virQEMUCapsPtr ret = virQEMUCapsNew();
1534 1535 1536 1537 1538
    size_t i;

    if (!ret)
        return NULL;

1539
    ret->usedQMP = qemuCaps->usedQMP;
1540
    ret->kvmSupportsNesting = qemuCaps->kvmSupportsNesting;
1541

1542 1543 1544 1545 1546
    if (VIR_STRDUP(ret->binary, qemuCaps->binary) < 0)
        goto error;

    ret->ctime = qemuCaps->ctime;

1547
    virBitmapCopy(ret->flags, qemuCaps->flags);
1548

1549 1550
    ret->version = qemuCaps->version;
    ret->kvmVersion = qemuCaps->kvmVersion;
1551
    ret->microcodeVersion = qemuCaps->microcodeVersion;
1552 1553 1554 1555

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

1556 1557 1558
    if (VIR_STRDUP(ret->kernelVersion, qemuCaps->kernelVersion) < 0)
        goto error;

1559
    ret->arch = qemuCaps->arch;
1560

1561 1562 1563 1564 1565 1566 1567 1568 1569
    if (qemuCaps->kvmCPUModels) {
        ret->kvmCPUModels = virDomainCapsCPUModelsCopy(qemuCaps->kvmCPUModels);
        if (!ret->kvmCPUModels)
            goto error;
    }

    if (qemuCaps->tcgCPUModels) {
        ret->tcgCPUModels = virDomainCapsCPUModelsCopy(qemuCaps->tcgCPUModels);
        if (!ret->tcgCPUModels)
1570
            goto error;
1571 1572
    }

1573 1574
    if (virQEMUCapsHostCPUDataCopy(&ret->kvmCPU, &qemuCaps->kvmCPU) < 0 ||
        virQEMUCapsHostCPUDataCopy(&ret->tcgCPU, &qemuCaps->tcgCPU) < 0)
1575 1576
        goto error;

1577
    if (VIR_ALLOC_N(ret->machineTypes, qemuCaps->nmachineTypes) < 0)
1578
        goto error;
1579
    ret->nmachineTypes = qemuCaps->nmachineTypes;
1580
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
1581 1582
        if (VIR_STRDUP(ret->machineTypes[i].name, qemuCaps->machineTypes[i].name) < 0 ||
            VIR_STRDUP(ret->machineTypes[i].alias, qemuCaps->machineTypes[i].alias) < 0)
1583
            goto error;
1584
        ret->machineTypes[i].maxCpus = qemuCaps->machineTypes[i].maxCpus;
1585
        ret->machineTypes[i].hotplugCpus = qemuCaps->machineTypes[i].hotplugCpus;
1586 1587
        ret->machineTypes[i].qemuDefault = qemuCaps->machineTypes[i].qemuDefault;

1588 1589
    }

1590 1591 1592 1593 1594 1595
    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];

1596 1597 1598 1599 1600
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_SEV_GUEST) &&
        virQEMUCapsSEVInfoCopy(&ret->sevCapabilities,
                               qemuCaps->sevCapabilities) < 0)
        goto error;

1601 1602
    return ret;

1603
 error:
1604 1605 1606 1607 1608
    virObjectUnref(ret);
    return NULL;
}


1609
void virQEMUCapsDispose(void *obj)
1610
{
1611
    virQEMUCapsPtr qemuCaps = obj;
1612 1613
    size_t i;

1614
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
1615 1616
        VIR_FREE(qemuCaps->machineTypes[i].name);
        VIR_FREE(qemuCaps->machineTypes[i].alias);
1617
    }
1618
    VIR_FREE(qemuCaps->machineTypes);
1619

1620 1621
    virObjectUnref(qemuCaps->kvmCPUModels);
    virObjectUnref(qemuCaps->tcgCPUModels);
1622

1623
    virBitmapFree(qemuCaps->flags);
1624

1625
    VIR_FREE(qemuCaps->package);
1626
    VIR_FREE(qemuCaps->kernelVersion);
1627
    VIR_FREE(qemuCaps->binary);
A
Andrea Bolognani 已提交
1628 1629

    VIR_FREE(qemuCaps->gicCapabilities);
1630

1631 1632
    virSEVCapabilitiesFree(qemuCaps->sevCapabilities);

1633 1634
    virQEMUCapsHostCPUDataClear(&qemuCaps->kvmCPU);
    virQEMUCapsHostCPUDataClear(&qemuCaps->tcgCPU);
1635 1636
}

1637
void
1638
virQEMUCapsSet(virQEMUCapsPtr qemuCaps,
1639
               virQEMUCapsFlags flag)
1640
{
1641
    ignore_value(virBitmapSetBit(qemuCaps->flags, flag));
1642 1643 1644 1645
}


void
1646
virQEMUCapsSetList(virQEMUCapsPtr qemuCaps, ...)
1647 1648 1649 1650
{
    va_list list;
    int flag;

1651
    va_start(list, qemuCaps);
1652
    while ((flag = va_arg(list, int)) < QEMU_CAPS_LAST)
1653
        ignore_value(virBitmapSetBit(qemuCaps->flags, flag));
1654
    va_end(list);
1655 1656 1657 1658
}


void
1659
virQEMUCapsClear(virQEMUCapsPtr qemuCaps,
1660
                 virQEMUCapsFlags flag)
1661
{
1662
    ignore_value(virBitmapClearBit(qemuCaps->flags, flag));
1663 1664 1665
}


1666
char *virQEMUCapsFlagsString(virQEMUCapsPtr qemuCaps)
1667
{
1668
    return virBitmapToString(qemuCaps->flags, true, false);
1669 1670 1671 1672
}


bool
1673
virQEMUCapsGet(virQEMUCapsPtr qemuCaps,
1674
               virQEMUCapsFlags flag)
1675
{
J
Ján Tomko 已提交
1676
    return qemuCaps && virBitmapIsBitSet(qemuCaps->flags, flag);
1677
}
1678 1679


D
Daniel P. Berrange 已提交
1680
bool virQEMUCapsHasPCIMultiBus(virQEMUCapsPtr qemuCaps,
1681
                               const virDomainDef *def)
D
Daniel P. Berrange 已提交
1682
{
1683 1684
    /* x86_64 and i686 support PCI-multibus on all machine types
     * since forever */
1685
    if (ARCH_IS_X86(def->os.arch))
D
Daniel P. Berrange 已提交
1686 1687 1688
        return true;

    if (def->os.arch == VIR_ARCH_PPC ||
1689
        ARCH_IS_PPC64(def->os.arch)) {
D
Daniel P. Berrange 已提交
1690 1691 1692 1693 1694
        /*
         * Usage of pci.0 naming:
         *
         *    ref405ep: no pci
         *       taihu: no pci
1695
         *      bamboo: 1.1.0 (<= 1.5.0, so basically forever)
D
Daniel P. Berrange 已提交
1696 1697
         *       mac99: 2.0.0
         *     g3beige: 2.0.0
1698
         *        prep: 1.4.0 (<= 1.5.0, so basically forever)
D
Daniel P. Berrange 已提交
1699 1700 1701 1702 1703 1704
         *     pseries: 2.0.0
         *   mpc8544ds: forever
         * virtex-m507: no pci
         *     ppce500: 1.6.0
         */

1705 1706 1707 1708 1709 1710 1711 1712
        /* 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
1713 1714 1715 1716 1717 1718 1719
        if (qemuCaps->version >= 2000000)
            return true;

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

1720 1721 1722
        if (STREQ(def->os.machine, "bamboo") ||
            STREQ(def->os.machine, "mpc8544ds") ||
            STREQ(def->os.machine, "prep")) {
D
Daniel P. Berrange 已提交
1723
            return true;
1724
        }
D
Daniel P. Berrange 已提交
1725 1726 1727 1728

        return false;
    }

1729 1730 1731 1732
    /* S390 supports PCI-multibus. */
    if (ARCH_IS_S390(def->os.arch))
        return true;

1733 1734 1735 1736
    /* If the virt machine, both on ARM and RISC-V, supports PCI,
     * then it also supports multibus */
    if (qemuDomainIsARMVirt(def) ||
        qemuDomainIsRISCVVirt(def)) {
1737
        return true;
1738
    }
1739

D
Daniel P. Berrange 已提交
1740 1741 1742 1743
    return false;
}


1744
const char *virQEMUCapsGetBinary(virQEMUCapsPtr qemuCaps)
1745
{
1746
    return qemuCaps->binary;
1747 1748
}

1749 1750 1751 1752 1753 1754 1755 1756 1757

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


1758
virArch virQEMUCapsGetArch(virQEMUCapsPtr qemuCaps)
1759
{
1760
    return qemuCaps->arch;
1761 1762 1763
}


1764
unsigned int virQEMUCapsGetVersion(virQEMUCapsPtr qemuCaps)
1765
{
1766
    return qemuCaps->version;
1767 1768 1769
}


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


1776 1777 1778 1779 1780 1781
const char *virQEMUCapsGetPackage(virQEMUCapsPtr qemuCaps)
{
    return qemuCaps->package;
}


1782 1783
int
virQEMUCapsAddCPUDefinitions(virQEMUCapsPtr qemuCaps,
1784
                             virDomainVirtType type,
1785
                             const char **name,
1786 1787
                             size_t count,
                             virDomainCapsCPUUsable usable)
1788
{
1789
    size_t i;
1790
    virDomainCapsCPUModelsPtr cpus = NULL;
1791

1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805
    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;
    }
1806 1807

    for (i = 0; i < count; i++) {
1808
        if (virDomainCapsCPUModelsAdd(cpus, name[i], -1, usable, NULL) < 0)
1809
            return -1;
1810
    }
1811

1812 1813 1814 1815
    return 0;
}


1816
virDomainCapsCPUModelsPtr
1817
virQEMUCapsGetCPUDefinitions(virQEMUCapsPtr qemuCaps,
1818
                             virDomainVirtType type)
1819
{
1820
    if (type == VIR_DOMAIN_VIRT_KVM)
1821
        return qemuCaps->kvmCPUModels;
1822
    else
1823
        return qemuCaps->tcgCPUModels;
1824 1825 1826
}


1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837
static virQEMUCapsHostCPUDataPtr
virQEMUCapsGetHostCPUData(virQEMUCapsPtr qemuCaps,
                          virDomainVirtType type)
{
    if (type == VIR_DOMAIN_VIRT_KVM)
        return &qemuCaps->kvmCPU;
    else
        return &qemuCaps->tcgCPU;
}


1838
virCPUDefPtr
1839
virQEMUCapsGetHostModel(virQEMUCapsPtr qemuCaps,
1840 1841
                        virDomainVirtType type,
                        virQEMUCapsHostCPUType cpuType)
1842
{
1843 1844
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);

1845 1846 1847
    switch (cpuType) {
    case VIR_QEMU_CAPS_HOST_CPU_REPORTED:
        return cpuData->reported;
1848 1849 1850

    case VIR_QEMU_CAPS_HOST_CPU_MIGRATABLE:
        return cpuData->migratable;
1851 1852 1853 1854 1855

    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;
1856 1857 1858
    }

    return NULL;
1859 1860 1861
}


1862 1863 1864
static void
virQEMUCapsSetHostModel(virQEMUCapsPtr qemuCaps,
                        virDomainVirtType type,
1865
                        virCPUDefPtr reported,
1866 1867
                        virCPUDefPtr migratable,
                        virCPUDefPtr full)
1868
{
1869 1870
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);

1871 1872
    cpuData->reported = reported;
    cpuData->migratable = migratable;
1873
    cpuData->full = full;
1874 1875 1876
}


1877 1878 1879 1880 1881 1882
bool
virQEMUCapsIsCPUModeSupported(virQEMUCapsPtr qemuCaps,
                              virCapsPtr caps,
                              virDomainVirtType type,
                              virCPUMode mode)
{
1883 1884
    virDomainCapsCPUModelsPtr cpus;

1885 1886 1887 1888 1889 1890
    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:
1891 1892
        return !!virQEMUCapsGetHostModel(qemuCaps, type,
                                         VIR_QEMU_CAPS_HOST_CPU_REPORTED);
1893 1894

    case VIR_CPU_MODE_CUSTOM:
1895 1896 1897 1898 1899
        if (type == VIR_DOMAIN_VIRT_KVM)
            cpus = qemuCaps->kvmCPUModels;
        else
            cpus = qemuCaps->tcgCPUModels;
        return cpus && cpus->nmodels > 0;
1900 1901 1902 1903 1904 1905 1906 1907 1908

    case VIR_CPU_MODE_LAST:
        break;
    }

    return false;
}


1909 1910 1911
int virQEMUCapsGetMachineTypesCaps(virQEMUCapsPtr qemuCaps,
                                   size_t *nmachines,
                                   virCapsGuestMachinePtr **machines)
1912 1913 1914 1915
{
    size_t i;

    *machines = NULL;
1916
    *nmachines = qemuCaps->nmachineTypes;
1917

1918 1919 1920 1921
    if (*nmachines &&
        VIR_ALLOC_N(*machines, qemuCaps->nmachineTypes) < 0)
        goto error;

1922
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
1923 1924
        virCapsGuestMachinePtr mach;
        if (VIR_ALLOC(mach) < 0)
1925
            goto error;
1926
        (*machines)[i] = mach;
1927 1928 1929
        if (qemuCaps->machineTypes[i].alias) {
            if (VIR_STRDUP(mach->name, qemuCaps->machineTypes[i].alias) < 0 ||
                VIR_STRDUP(mach->canonical, qemuCaps->machineTypes[i].name) < 0)
1930
                goto error;
1931
        } else {
1932
            if (VIR_STRDUP(mach->name, qemuCaps->machineTypes[i].name) < 0)
1933
                goto error;
1934
        }
1935
        mach->maxCpus = qemuCaps->machineTypes[i].maxCpus;
1936 1937
    }

1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975
    /* 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++;
    }

1976 1977
    return 0;

1978
 error:
1979 1980 1981 1982 1983 1984 1985
    virCapabilitiesFreeMachines(*machines, *nmachines);
    *nmachines = 0;
    *machines = NULL;
    return -1;
}


1986 1987 1988 1989 1990 1991 1992 1993
/**
 * 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.
 */
1994 1995
const char *virQEMUCapsGetCanonicalMachine(virQEMUCapsPtr qemuCaps,
                                           const char *name)
1996 1997 1998
{
    size_t i;

1999 2000
    if (!name || !qemuCaps)
        return name;
2001

2002
    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
2003
        if (!qemuCaps->machineTypes[i].alias)
2004
            continue;
2005 2006
        if (STREQ(qemuCaps->machineTypes[i].alias, name))
            return qemuCaps->machineTypes[i].name;
2007 2008 2009 2010
    }

    return name;
}
2011

2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023
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;
}
2024

2025 2026 2027 2028 2029 2030 2031 2032 2033
int virQEMUCapsGetMachineMaxCpus(virQEMUCapsPtr qemuCaps,
                                 const char *name)
{
    size_t i;

    if (!name)
        return 0;

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
2034
        if (!qemuCaps->machineTypes[i].maxCpus)
2035
            continue;
2036 2037
        if (STREQ(qemuCaps->machineTypes[i].name, name))
            return qemuCaps->machineTypes[i].maxCpus;
2038 2039 2040 2041 2042 2043
    }

    return 0;
}


2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057
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;
}


2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081
/**
 * 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;
}


2082 2083 2084 2085 2086 2087 2088
virSEVCapabilityPtr
virQEMUCapsGetSEVCapabilities(virQEMUCapsPtr qemuCaps)
{
    return qemuCaps->sevCapabilities;
}


2089
static int
2090 2091
virQEMUCapsProbeQMPCommands(virQEMUCapsPtr qemuCaps,
                            qemuMonitorPtr mon)
2092 2093 2094 2095 2096 2097 2098
{
    char **commands = NULL;
    int ncommands;

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

2099 2100 2101 2102
    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsCommands),
                                  virQEMUCapsCommands,
                                  ncommands, commands);
2103
    virStringListFreeCount(commands, ncommands);
2104

2105 2106 2107 2108
    /* 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) &&
2109 2110 2111
        qemuMonitorSupportsActiveCommit(mon))
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_ACTIVE_COMMIT);

2112 2113 2114 2115 2116
    return 0;
}


static int
2117 2118
virQEMUCapsProbeQMPEvents(virQEMUCapsPtr qemuCaps,
                          qemuMonitorPtr mon)
2119 2120 2121 2122
{
    char **events = NULL;
    int nevents;

2123
    /* we can probe events also from the QMP schema so we can skip this here */
2124
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_QMP_SCHEMA))
2125
        return 0;
2126 2127 2128

    if ((nevents = qemuMonitorGetEvents(mon, &events)) < 0)
        return -1;
2129

2130 2131 2132 2133
    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsEvents),
                                  virQEMUCapsEvents,
                                  nevents, events);
2134
    virStringListFreeCount(events, nevents);
2135 2136 2137 2138

    return 0;
}

2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167
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;
}
2168

2169
static int
2170
virQEMUCapsProbeQMPDevices(virQEMUCapsPtr qemuCaps,
2171
                           qemuMonitorPtr mon)
2172 2173 2174 2175 2176 2177
{
    int nvalues;
    char **values;

    if ((nvalues = qemuMonitorGetObjectTypes(mon, &values)) < 0)
        return -1;
2178 2179 2180 2181
    virQEMUCapsProcessStringFlags(qemuCaps,
                                  ARRAY_CARDINALITY(virQEMUCapsObjectTypes),
                                  virQEMUCapsObjectTypes,
                                  nvalues, values);
2182
    virStringListFreeCount(values, nvalues);
2183

2184 2185 2186 2187 2188 2189
    if (virQEMUCapsProbeQMPGenericProps(qemuCaps,
                                        mon,
                                        virQEMUCapsDeviceProps,
                                        ARRAY_CARDINALITY(virQEMUCapsDeviceProps),
                                        qemuMonitorGetDeviceProps) < 0)
        return -1;
2190

2191 2192 2193 2194 2195 2196 2197 2198
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_QOM_LIST_PROPERTIES) &&
        virQEMUCapsProbeQMPGenericProps(qemuCaps,
                                        mon,
                                        virQEMUCapsObjectProps,
                                        ARRAY_CARDINALITY(virQEMUCapsObjectProps),
                                        qemuMonitorGetObjectProps) < 0)
        return -1;

2199 2200 2201 2202
    return 0;
}


2203 2204 2205 2206 2207 2208 2209 2210 2211 2212
/* 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.
 */
2213
static const char *preferredMachines[] =
2214
{
2215 2216
    NULL, /* VIR_ARCH_NONE (not a real arch :) */
    "clipper", /* VIR_ARCH_ALPHA */
S
Stefan Schallenberg 已提交
2217
    "integratorcp", /* VIR_ARCH_ARMV6L */
2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244
    "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
2245 2246
    "spike_v1.10", /* VIR_ARCH_RISCV32 */
    "spike_v1.10", /* VIR_ARCH_RISCV64 */
2247 2248 2249
    NULL, /* VIR_ARCH_S390 (no QEMU impl) */
    "s390-ccw-virtio", /* VIR_ARCH_S390X */
    "shix", /* VIR_ARCH_SH4 */
L
Lubomir Rintel 已提交
2250

2251 2252 2253 2254 2255
    "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
2256

2257 2258
    "sim", /* VIR_ARCH_XTENSA */
    "sim", /* VIR_ARCH_XTENSAEB */
2259
};
2260
verify(ARRAY_CARDINALITY(preferredMachines) == VIR_ARCH_LAST);
2261 2262


2263
static int
2264 2265
virQEMUCapsProbeQMPMachineTypes(virQEMUCapsPtr qemuCaps,
                                qemuMonitorPtr mon)
2266 2267 2268 2269 2270
{
    qemuMonitorMachineInfoPtr *machines = NULL;
    int nmachines = 0;
    int ret = -1;
    size_t i;
2271 2272 2273
    ssize_t defIdx = -1;
    ssize_t preferredIdx = -1;
    const char *preferredMachine = preferredMachines[qemuCaps->arch];
2274 2275

    if ((nmachines = qemuMonitorGetMachines(mon, &machines)) < 0)
2276
        return -1;
2277

2278
    if (VIR_ALLOC_N(qemuCaps->machineTypes, nmachines) < 0)
2279 2280
        goto cleanup;

2281
    for (i = 0; i < nmachines; i++) {
2282
        struct virQEMUCapsMachineType *mach;
2283 2284
        if (STREQ(machines[i]->name, "none"))
            continue;
2285 2286 2287 2288 2289

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

        if (VIR_STRDUP(mach->alias, machines[i]->alias) < 0 ||
            VIR_STRDUP(mach->name, machines[i]->name) < 0)
2290
            goto cleanup;
2291 2292

        mach->maxCpus = machines[i]->maxCpus;
2293
        mach->hotplugCpus = machines[i]->hotplugCpus;
2294

2295 2296 2297 2298 2299 2300
        if (preferredMachine &&
            (STREQ_NULLABLE(mach->alias, preferredMachine) ||
             STREQ(mach->name, preferredMachine))) {
            preferredIdx = qemuCaps->nmachineTypes - 1;
        }

2301 2302
        if (machines[i]->isDefault) {
            mach->qemuDefault = true;
2303
            defIdx = qemuCaps->nmachineTypes - 1;
2304
        }
2305
    }
2306

2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319
    /*
     * 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);
2320 2321 2322

    ret = 0;

2323
 cleanup:
2324
    for (i = 0; i < nmachines; i++)
2325 2326 2327 2328 2329 2330
        qemuMonitorMachineInfoFree(machines[i]);
    VIR_FREE(machines);
    return ret;
}


2331 2332
virDomainCapsCPUModelsPtr
virQEMUCapsFetchCPUDefinitions(qemuMonitorPtr mon)
2333
{
2334 2335 2336
    virDomainCapsCPUModelsPtr models = NULL;
    qemuMonitorCPUDefInfoPtr *cpus = NULL;
    int ncpus = 0;
2337
    size_t i;
2338

2339
    if ((ncpus = qemuMonitorGetCPUDefinitions(mon, &cpus)) < 0)
2340
        return NULL;
2341

2342
    if (!(models = virDomainCapsCPUModelsNew(ncpus)))
2343
        goto error;
2344

2345
    for (i = 0; i < ncpus; i++) {
2346 2347 2348 2349 2350 2351 2352
        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;

2353
        if (virDomainCapsCPUModelsAddSteal(models, &cpus[i]->name, usable,
2354
                                           &cpus[i]->blockers) < 0)
2355
            goto error;
2356 2357 2358 2359 2360 2361
    }

 cleanup:
    for (i = 0; i < ncpus; i++)
        qemuMonitorCPUDefInfoFree(cpus[i]);
    VIR_FREE(cpus);
2362 2363 2364 2365 2366 2367
    return models;

 error:
    virObjectUnref(models);
    models = NULL;
    goto cleanup;
2368 2369
}

2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392

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;
}


2393 2394
static int
virQEMUCapsProbeQMPHostCPU(virQEMUCapsPtr qemuCaps,
2395 2396
                           qemuMonitorPtr mon,
                           bool tcg)
2397
{
2398
    qemuMonitorCPUModelInfoPtr modelInfo = NULL;
2399 2400
    qemuMonitorCPUModelInfoPtr nonMigratable = NULL;
    virHashTablePtr hash = NULL;
2401
    const char *model;
2402
    qemuMonitorCPUModelExpansionType type;
2403 2404
    virDomainVirtType virtType;
    virQEMUCapsHostCPUDataPtr cpuData;
2405
    int ret = -1;
2406 2407

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_CPU_MODEL_EXPANSION))
2408 2409
        return 0;

2410
    if (tcg || !virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
2411
        virtType = VIR_DOMAIN_VIRT_QEMU;
2412 2413
        model = "max";
    } else {
2414
        virtType = VIR_DOMAIN_VIRT_KVM;
2415 2416 2417
        model = "host";
    }

2418 2419
    cpuData = virQEMUCapsGetHostCPUData(qemuCaps, virtType);

2420 2421 2422 2423 2424 2425 2426 2427 2428 2429
    /* 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;

2430 2431
    if (qemuMonitorGetCPUModelExpansion(mon, type, model, true, &modelInfo) < 0)
        goto cleanup;
2432 2433

    /* Try to check migratability of each feature. */
2434
    if (modelInfo &&
2435 2436
        qemuMonitorGetCPUModelExpansion(mon, type, model, false,
                                        &nonMigratable) < 0)
2437
        goto cleanup;
2438 2439 2440 2441 2442 2443 2444

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

        if (!(hash = virHashCreate(0, NULL)))
2445
            goto cleanup;
2446

2447 2448
        for (i = 0; i < modelInfo->nprops; i++) {
            prop = modelInfo->props + i;
2449
            if (virHashAddEntry(hash, prop->name, prop) < 0)
2450
                goto cleanup;
2451 2452 2453 2454 2455 2456 2457 2458 2459
        }

        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;

2460
            if (prop->value.boolean) {
2461
                prop->migratable = VIR_TRISTATE_BOOL_YES;
2462 2463 2464 2465
            } else if (nmProp->value.boolean) {
                prop->value.boolean = true;
                prop->migratable = VIR_TRISTATE_BOOL_NO;
            }
2466 2467
        }

2468
        modelInfo->migratability = true;
2469 2470
    }

2471
    VIR_STEAL_PTR(cpuData->info, modelInfo);
2472 2473 2474 2475 2476
    ret = 0;

 cleanup:
    virHashFree(hash);
    qemuMonitorCPUModelInfoFree(nonMigratable);
2477
    qemuMonitorCPUModelInfoFree(modelInfo);
2478 2479

    return ret;
2480 2481
}

2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533

/**
 * 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;
}


2534 2535
struct tpmTypeToCaps {
    int type;
2536
    virQEMUCapsFlags caps;
2537 2538 2539 2540 2541 2542 2543
};

static const struct tpmTypeToCaps virQEMUCapsTPMTypesToCaps[] = {
    {
        .type = VIR_DOMAIN_TPM_TYPE_PASSTHROUGH,
        .caps = QEMU_CAPS_DEVICE_TPM_PASSTHROUGH,
    },
2544 2545 2546 2547
    {
        .type = VIR_DOMAIN_TPM_TYPE_EMULATOR,
        .caps = QEMU_CAPS_DEVICE_TPM_EMULATOR,
    },
2548 2549 2550 2551 2552 2553 2554
};

const struct tpmTypeToCaps virQEMUCapsTPMModelsToCaps[] = {
    {
        .type = VIR_DOMAIN_TPM_MODEL_TIS,
        .caps = QEMU_CAPS_DEVICE_TPM_TIS,
    },
2555 2556 2557 2558
    {
        .type = VIR_DOMAIN_TPM_MODEL_CRB,
        .caps = QEMU_CAPS_DEVICE_TPM_CRB,
    },
2559 2560 2561 2562 2563 2564
};

static int
virQEMUCapsProbeQMPTPM(virQEMUCapsPtr qemuCaps,
                       qemuMonitorPtr mon)
{
2565 2566
    int nentries;
    size_t i;
2567
    char **entries = NULL;
S
Stefan Berger 已提交
2568

2569 2570 2571 2572 2573 2574 2575
    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);
2576
            if (virStringListHasString((const char **)entries, needle))
2577 2578 2579 2580
                virQEMUCapsSet(qemuCaps,
                               virQEMUCapsTPMModelsToCaps[i].caps);
        }
    }
2581
    virStringListFree(entries);
2582 2583 2584 2585 2586 2587 2588 2589

    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);
2590
            if (virStringListHasString((const char **)entries, needle))
2591 2592 2593
                virQEMUCapsSet(qemuCaps, virQEMUCapsTPMTypesToCaps[i].caps);
        }
    }
2594
    virStringListFree(entries);
2595 2596 2597 2598

    return 0;
}

2599

2600
static int
2601 2602
virQEMUCapsProbeQMPKVMState(virQEMUCapsPtr qemuCaps,
                            qemuMonitorPtr mon)
2603 2604 2605 2606 2607 2608 2609
{
    bool enabled = false;
    bool present = false;

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

A
Andrea Bolognani 已提交
2610 2611
    if (present && enabled)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_KVM);
2612 2613 2614 2615

    return 0;
}

2616 2617 2618 2619 2620 2621 2622 2623
struct virQEMUCapsCommandLineProps {
    const char *option;
    const char *param;
    int flag;
};

static struct virQEMUCapsCommandLineProps virQEMUCapsCommandLine[] = {
    { "machine", "mem-merge", QEMU_CAPS_MEM_MERGE },
2624
    { "machine", "vmport", QEMU_CAPS_MACHINE_VMPORT_OPT },
O
Osier Yang 已提交
2625
    { "drive", "discard", QEMU_CAPS_DRIVE_DISCARD },
2626
    { "drive", "detect-zeroes", QEMU_CAPS_DRIVE_DETECT_ZEROES },
2627
    { "realtime", "mlock", QEMU_CAPS_REALTIME_MLOCK },
2628
    { "boot-opts", "strict", QEMU_CAPS_BOOT_STRICT },
2629
    { "boot-opts", "reboot-timeout", QEMU_CAPS_REBOOT_TIMEOUT },
2630
    { "boot-opts", "splash-time", QEMU_CAPS_SPLASH_TIMEOUT },
2631
    { "spice", "disable-agent-file-xfer", QEMU_CAPS_SPICE_FILE_XFER_DISABLE },
2632
    { "msg", "timestamp", QEMU_CAPS_MSG_TIMESTAMP },
2633
    { "numa", NULL, QEMU_CAPS_NUMA },
2634
    { "drive", "throttling.bps-total-max", QEMU_CAPS_DRIVE_IOTUNE_MAX},
2635 2636
    { "machine", "aes-key-wrap", QEMU_CAPS_AES_KEY_WRAP },
    { "machine", "dea-key-wrap", QEMU_CAPS_DEA_KEY_WRAP },
2637
    { "chardev", "append", QEMU_CAPS_CHARDEV_FILE_APPEND },
2638
    { "spice", "gl", QEMU_CAPS_SPICE_GL },
2639
    { "chardev", "logfile", QEMU_CAPS_CHARDEV_LOGFILE },
2640
    { "name", "debug-threads", QEMU_CAPS_NAME_DEBUG_THREADS },
2641
    { "name", "guest", QEMU_CAPS_NAME_GUEST },
2642
    { "spice", "unix", QEMU_CAPS_SPICE_UNIX },
2643
    { "drive", "throttling.bps-total-max-length", QEMU_CAPS_DRIVE_IOTUNE_MAX_LENGTH },
2644
    { "drive", "throttling.group", QEMU_CAPS_DRIVE_IOTUNE_GROUP },
2645
    { "spice", "rendernode", QEMU_CAPS_SPICE_RENDERNODE },
2646
    { "machine", "kernel_irqchip", QEMU_CAPS_MACHINE_KERNEL_IRQCHIP },
2647
    { "machine", "loadparm", QEMU_CAPS_LOADPARM },
2648
    { "vnc", "vnc", QEMU_CAPS_VNC_MULTI_SERVERS },
2649
    { "chardev", "reconnect", QEMU_CAPS_CHARDEV_RECONNECT },
2650
    { "sandbox", "enable", QEMU_CAPS_SECCOMP_SANDBOX },
2651
    { "sandbox", "elevateprivileges", QEMU_CAPS_SECCOMP_BLACKLIST },
2652
    { "chardev", "fd", QEMU_CAPS_CHARDEV_FD_PASS },
2653 2654 2655 2656 2657 2658
};

static int
virQEMUCapsProbeQMPCommandLine(virQEMUCapsPtr qemuCaps,
                               qemuMonitorPtr mon)
{
2659
    bool found = false;
2660 2661 2662 2663 2664 2665 2666
    int nvalues;
    char **values;
    size_t i, j;

    for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsCommandLine); i++) {
        if ((nvalues = qemuMonitorGetCommandLineOptionParameters(mon,
                                                                 virQEMUCapsCommandLine[i].option,
2667 2668
                                                                 &values,
                                                                 &found)) < 0)
2669
            return -1;
2670 2671 2672 2673

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

2674
        for (j = 0; j < nvalues; j++) {
2675
            if (STREQ_NULLABLE(virQEMUCapsCommandLine[i].param, values[j])) {
2676 2677 2678 2679
                virQEMUCapsSet(qemuCaps, virQEMUCapsCommandLine[i].flag);
                break;
            }
        }
2680
        virStringListFree(values);
2681 2682 2683 2684
    }

    return 0;
}
2685

2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699
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);
2700
    virStringListFreeCount(caps, ncaps);
2701 2702 2703 2704

    return 0;
}

A
Andrea Bolognani 已提交
2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724
/**
 * 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;

2725
    virQEMUCapsSetGICCapabilities(qemuCaps, caps, ncaps);
A
Andrea Bolognani 已提交
2726 2727 2728 2729

    return 0;
}

2730

2731
/* Returns -1 on error, 0 if SEV is not supported, 1 if SEV is supported */
2732 2733 2734 2735
static int
virQEMUCapsProbeQMPSEVCapabilities(virQEMUCapsPtr qemuCaps,
                                   qemuMonitorPtr mon)
{
2736
    int rc = -1;
2737 2738
    virSEVCapability *caps = NULL;

2739 2740
    if ((rc = qemuMonitorGetSEVCapabilities(mon, &caps)) <= 0)
        return rc;
2741

2742 2743
    virSEVCapabilitiesFree(qemuCaps->sevCapabilities);
    qemuCaps->sevCapabilities = caps;
2744
    return rc;
2745 2746 2747
}


2748
bool
2749
virQEMUCapsCPUFilterFeatures(const char *name,
2750
                             void *opaque)
2751
{
2752
    virArch *arch = opaque;
2753

2754
    if (!ARCH_IS_X86(*arch))
2755 2756
        return true;

2757 2758 2759 2760 2761 2762 2763 2764 2765
    if (STREQ(name, "cmt") ||
        STREQ(name, "mbm_total") ||
        STREQ(name, "mbm_local"))
        return false;

    return true;
}


2766 2767 2768
/**
 * 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,
2769
 *          2 when cpu model info is not supported for this configuration,
2770 2771 2772 2773
 *         -1 on error.
 */
static int
virQEMUCapsInitCPUModelS390(virQEMUCapsPtr qemuCaps,
2774
                            virDomainVirtType type,
2775
                            qemuMonitorCPUModelInfoPtr modelInfo,
2776 2777
                            virCPUDefPtr cpu,
                            bool migratable)
2778
{
2779
    size_t i;
2780

2781
    if (!modelInfo) {
2782 2783 2784 2785 2786 2787 2788 2789
        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;
2790
    }
J
Jiri Denemark 已提交
2791

2792 2793
    if (VIR_STRDUP(cpu->model, modelInfo->name) < 0 ||
        VIR_ALLOC_N(cpu->features, modelInfo->nprops) < 0)
2794
        return -1;
2795 2796 2797 2798 2799

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

    for (i = 0; i < modelInfo->nprops; i++) {
2800 2801
        virCPUFeatureDefPtr feature = cpu->features + cpu->nfeatures;
        qemuMonitorCPUPropertyPtr prop = modelInfo->props + i;
2802

2803 2804
        if (prop->type != QEMU_MONITOR_CPU_PROPERTY_BOOLEAN)
            continue;
2805

2806 2807
        if (VIR_STRDUP(feature->name, prop->name) < 0)
            return -1;
2808 2809 2810 2811 2812 2813

        if (!prop->value.boolean ||
            (migratable && prop->migratable == VIR_TRISTATE_BOOL_NO))
            feature->policy = VIR_CPU_FEATURE_DISABLE;
        else
            feature->policy = VIR_CPU_FEATURE_REQUIRE;
2814 2815 2816
        cpu->nfeatures++;
    }

2817 2818
    return 0;
}
2819

2820

2821 2822 2823 2824 2825 2826 2827 2828
/**
 * 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,
2829
                           qemuMonitorCPUModelInfoPtr model,
2830 2831
                           virCPUDefPtr cpu,
                           bool migratable)
2832 2833 2834 2835
{
    virCPUDataPtr data = NULL;
    unsigned long long sigFamily = 0;
    unsigned long long sigModel = 0;
2836
    unsigned long long sigStepping = 0;
2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850
    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:
2851 2852 2853 2854 2855
            if (!prop->value.boolean ||
                (migratable && prop->migratable == VIR_TRISTATE_BOOL_NO))
                continue;

            if (virCPUx86DataAddFeature(data, prop->name) < 0)
2856
                goto cleanup;
2857

2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870
            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;
2871 2872
            else if (STREQ(prop->name, "stepping"))
                sigStepping = prop->value.number;
2873 2874 2875 2876 2877 2878 2879
            break;

        case QEMU_MONITOR_CPU_PROPERTY_LAST:
            break;
        }
    }

2880
    if (virCPUx86DataSetSignature(data, sigFamily, sigModel, sigStepping) < 0)
2881 2882
        goto cleanup;

2883
    if (cpuDecode(cpu, data, virQEMUCapsGetCPUDefinitions(qemuCaps, type)) < 0)
2884 2885 2886 2887 2888 2889 2890 2891 2892 2893
        goto cleanup;

    ret = 0;

 cleanup:
    virCPUDataFree(data);
    return ret;
}


2894 2895
/**
 * Returns  0 when host CPU model provided by QEMU was filled in qemuCaps,
2896 2897
 *          1 when the caller should fall back to other methods,
 *          2 when cpu model info is not supported for this configuration,
2898 2899
 *         -1 on error.
 */
2900
int
2901
virQEMUCapsInitCPUModel(virQEMUCapsPtr qemuCaps,
2902
                        virDomainVirtType type,
2903 2904
                        virCPUDefPtr cpu,
                        bool migratable)
2905
{
2906
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);
2907 2908
    int ret = 1;

2909
    if (migratable && cpuData->info && !cpuData->info->migratability)
2910 2911
        return 1;

2912
    if (ARCH_IS_S390(qemuCaps->arch)) {
2913
        ret = virQEMUCapsInitCPUModelS390(qemuCaps, type, cpuData->info,
2914 2915 2916 2917 2918
                                          cpu, migratable);
    } else if (ARCH_IS_X86(qemuCaps->arch)) {
        ret = virQEMUCapsInitCPUModelX86(qemuCaps, type, cpuData->info,
                                         cpu, migratable);
    }
2919

2920 2921 2922
    if (ret == 0)
        cpu->fallback = VIR_CPU_FALLBACK_FORBID;

2923
    return ret;
2924 2925 2926
}


2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943
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;
}


2944 2945
void
virQEMUCapsInitHostCPUModel(virQEMUCapsPtr qemuCaps,
2946
                            virArch hostArch,
2947
                            virDomainVirtType type)
2948 2949
{
    virCPUDefPtr cpu = NULL;
2950
    virCPUDefPtr cpuExpanded = NULL;
2951
    virCPUDefPtr migCPU = NULL;
2952
    virCPUDefPtr hostCPU = NULL;
2953 2954
    virCPUDefPtr fullCPU = NULL;
    size_t i;
2955
    int rc;
2956

2957
    if (!virQEMUCapsGuestIsNative(hostArch, qemuCaps->arch))
2958 2959
        return;

2960
    if (!(cpu = virQEMUCapsNewHostCPUModel()))
2961 2962
        goto error;

2963
    if ((rc = virQEMUCapsInitCPUModel(qemuCaps, type, cpu, false)) < 0) {
2964 2965
        goto error;
    } else if (rc == 1) {
2966
        VIR_DEBUG("No host CPU model info from QEMU; probing host CPU directly");
2967

2968
        hostCPU = virQEMUCapsProbeHostCPUForEmulator(hostArch, qemuCaps, type);
2969 2970
        if (!hostCPU ||
            virCPUDefCopyModelFilter(cpu, hostCPU, true,
2971
                                     virQEMUCapsCPUFilterFeatures,
2972
                                     &qemuCaps->arch) < 0)
2973
            goto error;
2974 2975 2976 2977 2978
    } else if (rc == 2) {
        VIR_DEBUG("QEMU does not provide CPU model for arch=%s virttype=%s",
                  virArchToString(qemuCaps->arch),
                  virDomainVirtTypeToString(type));
        goto error;
2979 2980 2981
    } else if (type == VIR_DOMAIN_VIRT_KVM &&
               virCPUGetHostIsSupported(qemuCaps->arch)) {
        if (!(fullCPU = virCPUGetHost(qemuCaps->arch, VIR_CPU_TYPE_GUEST,
2982
                                      NULL, NULL)))
2983 2984
            goto error;

2985 2986 2987 2988 2989 2990 2991
        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,
2992 2993 2994
                                       VIR_CPU_FEATURE_REQUIRE) < 0)
                goto error;
        }
2995 2996
    }

2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009
    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;
    }

3010
    virQEMUCapsSetHostModel(qemuCaps, type, cpu, migCPU, fullCPU);
3011

3012
 cleanup:
3013
    virCPUDefFree(cpuExpanded);
3014
    virCPUDefFree(hostCPU);
3015 3016 3017 3018
    return;

 error:
    virCPUDefFree(cpu);
3019
    virCPUDefFree(migCPU);
3020
    virCPUDefFree(fullCPU);
3021
    virResetLastError();
3022
    goto cleanup;
3023 3024 3025
}


3026 3027 3028 3029 3030
void
virQEMUCapsSetCPUModelInfo(virQEMUCapsPtr qemuCaps,
                           virDomainVirtType type,
                           qemuMonitorCPUModelInfoPtr modelInfo)
{
3031 3032 3033
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);

    cpuData->info = modelInfo;
3034 3035 3036
}


3037 3038
static int
virQEMUCapsLoadHostCPUModelInfo(virQEMUCapsPtr qemuCaps,
3039 3040
                                xmlXPathContextPtr ctxt,
                                virDomainVirtType virtType)
3041 3042 3043
{
    char *str = NULL;
    xmlNodePtr hostCPUNode;
3044
    xmlNodePtr *nodes = NULL;
3045
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
3046 3047 3048 3049
    qemuMonitorCPUModelInfoPtr hostCPU = NULL;
    int ret = -1;
    size_t i;
    int n;
3050
    int val;
3051

3052 3053 3054 3055 3056 3057
    if (virtType == VIR_DOMAIN_VIRT_KVM)
        hostCPUNode = virXPathNode("./hostCPU[@type='kvm']", ctxt);
    else
        hostCPUNode = virXPathNode("./hostCPU[@type='tcg']", ctxt);

    if (!hostCPUNode) {
3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071
        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;
    }

3072 3073 3074 3075 3076 3077 3078 3079 3080
    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);

3081 3082
    ctxt->node = hostCPUNode;

3083
    if ((n = virXPathNodeSet("./property", ctxt, &nodes)) > 0) {
3084 3085 3086 3087 3088 3089
        if (VIR_ALLOC_N(hostCPU->props, n) < 0)
            goto cleanup;

        hostCPU->nprops = n;

        for (i = 0; i < n; i++) {
3090 3091 3092 3093 3094
            qemuMonitorCPUPropertyPtr prop = hostCPU->props + i;

            ctxt->node = nodes[i];

            if (!(prop->name = virXMLPropString(ctxt->node, "name"))) {
3095 3096
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing 'name' attribute for a host CPU"
3097
                                 " model property in QEMU capabilities cache"));
3098 3099 3100
                goto cleanup;
            }

3101
            if (!(str = virXMLPropString(ctxt->node, "type")) ||
3102
                (val = qemuMonitorCPUPropertyTypeFromString(str)) < 0) {
3103
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
3104 3105
                               _("missing or invalid CPU model property type "
                                 "in QEMU capabilities cache"));
3106 3107 3108
                goto cleanup;
            }
            VIR_FREE(str);
3109

3110
            prop->type = val;
3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141
            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;
            }
3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154

            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);
            }
3155 3156 3157
        }
    }

3158
    virQEMUCapsSetCPUModelInfo(qemuCaps, virtType, hostCPU);
3159 3160 3161 3162 3163
    hostCPU = NULL;
    ret = 0;

 cleanup:
    VIR_FREE(str);
3164
    VIR_FREE(nodes);
3165 3166 3167 3168 3169
    qemuMonitorCPUModelInfoFree(hostCPU);
    return ret;
}


3170 3171
static int
virQEMUCapsLoadCPUModels(virQEMUCapsPtr qemuCaps,
3172 3173
                         xmlXPathContextPtr ctxt,
                         virDomainVirtType type)
3174
{
3175
    virDomainCapsCPUModelsPtr cpus = NULL;
3176 3177 3178 3179 3180
    xmlNodePtr *nodes = NULL;
    char *str = NULL;
    size_t i;
    int n;
    int ret = -1;
3181 3182 3183 3184
    xmlNodePtr node;
    xmlNodePtr *blockerNodes = NULL;
    char **blockers = NULL;
    int nblockers;
3185

3186 3187 3188 3189 3190 3191
    if (type == VIR_DOMAIN_VIRT_KVM)
        n = virXPathNodeSet("./cpu[@type='kvm']", ctxt, &nodes);
    else
        n = virXPathNodeSet("./cpu[@type='tcg']", ctxt, &nodes);

    if (n < 0) {
3192 3193 3194 3195 3196 3197 3198 3199 3200 3201
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to parse qemu capabilities cpus"));
        goto cleanup;
    }

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

3202
    if (!(cpus = virDomainCapsCPUModelsNew(n)))
3203 3204
        goto cleanup;

3205 3206 3207 3208 3209
    if (type == VIR_DOMAIN_VIRT_KVM)
        qemuCaps->kvmCPUModels = cpus;
    else
        qemuCaps->tcgCPUModels = cpus;

3210
    for (i = 0; i < n; i++) {
3211 3212 3213 3214 3215 3216 3217 3218 3219 3220
        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);

3221 3222 3223 3224 3225 3226
        if (!(str = virXMLPropString(nodes[i], "name"))) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("missing cpu name in QEMU capabilities cache"));
            goto cleanup;
        }

3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251
        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;
                }
            }
3252
            VIR_FREE(blockerNodes);
3253 3254 3255
        }

        if (virDomainCapsCPUModelsAddSteal(cpus, &str, usable, &blockers) < 0)
3256 3257 3258 3259 3260 3261 3262 3263
            goto cleanup;
    }

    ret = 0;

 cleanup:
    VIR_FREE(nodes);
    VIR_FREE(str);
3264 3265
    VIR_FREE(blockerNodes);
    virStringListFree(blockers);
3266 3267 3268 3269
    return ret;
}


3270 3271 3272 3273 3274
struct _virQEMUCapsCachePriv {
    char *libDir;
    uid_t runUid;
    gid_t runGid;
    virArch hostArch;
3275
    unsigned int microcodeVersion;
3276
    char *kernelVersion;
3277 3278 3279 3280

    /* cache whether /dev/kvm is usable as runUid:runGuid */
    virTristateBool kvmUsable;
    time_t kvmCtime;
3281 3282 3283 3284 3285
};
typedef struct _virQEMUCapsCachePriv virQEMUCapsCachePriv;
typedef virQEMUCapsCachePriv *virQEMUCapsCachePrivPtr;


3286
static void
3287
virQEMUCapsCachePrivFree(void *privData)
3288
{
3289 3290
    virQEMUCapsCachePrivPtr priv = privData;

3291
    VIR_FREE(priv->libDir);
3292
    VIR_FREE(priv->kernelVersion);
3293 3294 3295 3296
    VIR_FREE(priv);
}


3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348
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;
}


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

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

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

    ctxt->node = xmlDocGetRootElement(doc);

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

    if (virXPathLongLong("string(./qemuctime)", ctxt, &l) < 0) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing qemuctime in QEMU capabilities XML"));
        goto cleanup;
    }
3403
    qemuCaps->ctime = (time_t)l;
3404 3405 3406 3407 3408 3409

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

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

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

3452 3453 3454 3455 3456 3457 3458
    if (virXPathUInt("string(./microcodeVersion)", ctxt,
                     &qemuCaps->microcodeVersion) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing microcode version in QEMU capabilities cache"));
        goto cleanup;
    }

3459 3460 3461 3462 3463 3464
    if (virXPathBoolean("boolean(./package)", ctxt) > 0) {
        qemuCaps->package = virXPathString("string(./package)", ctxt);
        if (!qemuCaps->package &&
            VIR_STRDUP(qemuCaps->package, "") < 0)
            goto cleanup;
    }
3465

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

3472 3473 3474 3475 3476 3477 3478 3479 3480 3481
    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
3482
    VIR_FREE(str);
3483

3484 3485
    if (virQEMUCapsLoadHostCPUModelInfo(qemuCaps, ctxt, VIR_DOMAIN_VIRT_KVM) < 0 ||
        virQEMUCapsLoadHostCPUModelInfo(qemuCaps, ctxt, VIR_DOMAIN_VIRT_QEMU) < 0)
3486 3487
        goto cleanup;

3488 3489
    if (virQEMUCapsLoadCPUModels(qemuCaps, ctxt, VIR_DOMAIN_VIRT_KVM) < 0 ||
        virQEMUCapsLoadCPUModels(qemuCaps, ctxt, VIR_DOMAIN_VIRT_QEMU) < 0)
3490 3491 3492 3493 3494 3495 3496 3497 3498
        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;
3499
        if (VIR_ALLOC_N(qemuCaps->machineTypes, qemuCaps->nmachineTypes) < 0)
3500 3501 3502
            goto cleanup;

        for (i = 0; i < n; i++) {
3503
            if (!(qemuCaps->machineTypes[i].name = virXMLPropString(nodes[i], "name"))) {
3504 3505 3506 3507
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing machine name in QEMU capabilities cache"));
                goto cleanup;
            }
3508
            qemuCaps->machineTypes[i].alias = virXMLPropString(nodes[i], "alias");
3509 3510 3511

            str = virXMLPropString(nodes[i], "maxCpus");
            if (str &&
3512
                virStrToLong_ui(str, NULL, 10, &(qemuCaps->machineTypes[i].maxCpus)) < 0) {
3513 3514 3515 3516
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("malformed machine cpu count in QEMU capabilities cache"));
                goto cleanup;
            }
J
Ján Tomko 已提交
3517
            VIR_FREE(str);
3518 3519 3520 3521 3522

            str = virXMLPropString(nodes[i], "hotplugCpus");
            if (STREQ_NULLABLE(str, "yes"))
                qemuCaps->machineTypes[i].hotplugCpus = true;
            VIR_FREE(str);
3523 3524 3525 3526 3527

            str = virXMLPropString(nodes[i], "default");
            if (STREQ_NULLABLE(str, "yes"))
                qemuCaps->machineTypes[i].qemuDefault = true;
            VIR_FREE(str);
3528 3529 3530 3531
        }
    }
    VIR_FREE(nodes);

A
Andrea Bolognani 已提交
3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597
    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);

3598 3599 3600
    if (virQEMUCapsParseSEVInfo(qemuCaps, ctxt) < 0)
        goto cleanup;

3601 3602
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_QEMU);
3603

3604 3605 3606
    if (virXPathBoolean("boolean(./kvmSupportsNesting)", ctxt) > 0)
        qemuCaps->kvmSupportsNesting = true;

3607
    ret = 0;
3608
 cleanup:
J
Ján Tomko 已提交
3609
    VIR_FREE(str);
3610 3611 3612 3613 3614 3615 3616
    VIR_FREE(nodes);
    xmlXPathFreeContext(ctxt);
    xmlFreeDoc(doc);
    return ret;
}


3617 3618
static void
virQEMUCapsFormatHostCPUModelInfo(virQEMUCapsPtr qemuCaps,
3619 3620
                                  virBufferPtr buf,
                                  virDomainVirtType type)
3621
{
3622 3623 3624
    virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type);
    qemuMonitorCPUModelInfoPtr model = cpuData->info;
    const char *typeStr = type == VIR_DOMAIN_VIRT_KVM ? "kvm" : "tcg";
3625 3626
    size_t i;

3627 3628 3629
    if (!model)
        return;

3630 3631 3632 3633
    virBufferAsprintf(buf,
                      "<hostCPU type='%s' model='%s' migratability='%s'>\n",
                      typeStr, model->name,
                      model->migratability ? "yes" : "no");
3634 3635 3636
    virBufferAdjustIndent(buf, 2);

    for (i = 0; i < model->nprops; i++) {
3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659
        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;
        }
3660 3661 3662 3663 3664

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

3665
        virBufferAddLit(buf, "/>\n");
3666 3667 3668 3669 3670 3671 3672
    }

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


3673 3674
static void
virQEMUCapsFormatCPUModels(virQEMUCapsPtr qemuCaps,
3675 3676
                           virBufferPtr buf,
                           virDomainVirtType type)
3677
{
3678 3679
    virDomainCapsCPUModelsPtr cpus;
    const char *typeStr;
3680 3681
    size_t i;

3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693
    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++) {
3694 3695
        virDomainCapsCPUModelPtr cpu = cpus->models + i;

3696
        virBufferAsprintf(buf, "<cpu type='%s' ", typeStr);
3697 3698 3699 3700 3701
        virBufferEscapeString(buf, "name='%s'", cpu->name);
        if (cpu->usable) {
            virBufferAsprintf(buf, " usable='%s'",
                              virDomainCapsCPUUsableTypeToString(cpu->usable));
        }
3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716

        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");
        }
3717 3718 3719 3720
    }
}


3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738
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");
}


3739
char *
3740
virQEMUCapsFormatCache(virQEMUCapsPtr qemuCaps)
3741 3742
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;
3743
    char *ret = NULL;
3744 3745 3746
    size_t i;

    virBufferAddLit(&buf, "<qemuCaps>\n");
3747
    virBufferAdjustIndent(&buf, 2);
3748

3749
    virBufferAsprintf(&buf, "<qemuctime>%llu</qemuctime>\n",
3750
                      (long long)qemuCaps->ctime);
3751
    virBufferAsprintf(&buf, "<selfctime>%llu</selfctime>\n",
3752
                      (long long)qemuCaps->libvirtCtime);
3753
    virBufferAsprintf(&buf, "<selfvers>%lu</selfvers>\n",
3754
                      (unsigned long)qemuCaps->libvirtVersion);
3755 3756 3757

    for (i = 0; i < QEMU_CAPS_LAST; i++) {
        if (virQEMUCapsGet(qemuCaps, i)) {
3758
            virBufferAsprintf(&buf, "<flag name='%s'/>\n",
3759 3760 3761 3762
                              virQEMUCapsTypeToString(i));
        }
    }

3763
    virBufferAsprintf(&buf, "<version>%d</version>\n",
3764 3765
                      qemuCaps->version);

3766
    virBufferAsprintf(&buf, "<kvmVersion>%d</kvmVersion>\n",
3767 3768
                      qemuCaps->kvmVersion);

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

3772 3773 3774 3775
    if (qemuCaps->package)
        virBufferAsprintf(&buf, "<package>%s</package>\n",
                          qemuCaps->package);

3776 3777 3778 3779
    if (qemuCaps->kernelVersion)
        virBufferAsprintf(&buf, "<kernelVersion>%s</kernelVersion>\n",
                          qemuCaps->kernelVersion);

3780
    virBufferAsprintf(&buf, "<arch>%s</arch>\n",
3781 3782
                      virArchToString(qemuCaps->arch));

3783 3784
    virQEMUCapsFormatHostCPUModelInfo(qemuCaps, &buf, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsFormatHostCPUModelInfo(qemuCaps, &buf, VIR_DOMAIN_VIRT_QEMU);
3785

3786 3787
    virQEMUCapsFormatCPUModels(qemuCaps, &buf, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsFormatCPUModels(qemuCaps, &buf, VIR_DOMAIN_VIRT_QEMU);
3788 3789

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
3790
        virBufferEscapeString(&buf, "<machine name='%s'",
3791 3792
                              qemuCaps->machineTypes[i].name);
        if (qemuCaps->machineTypes[i].alias)
3793
            virBufferEscapeString(&buf, " alias='%s'",
3794
                              qemuCaps->machineTypes[i].alias);
3795 3796
        if (qemuCaps->machineTypes[i].hotplugCpus)
            virBufferAddLit(&buf, " hotplugCpus='yes'");
3797
        virBufferAsprintf(&buf, " maxCpus='%u'",
3798
                          qemuCaps->machineTypes[i].maxCpus);
3799 3800 3801
        if (qemuCaps->machineTypes[i].qemuDefault)
            virBufferAddLit(&buf, " default='yes'");
        virBufferAddLit(&buf, "/>\n");
3802 3803
    }

A
Andrea Bolognani 已提交
3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819
    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");
    }

3820 3821 3822
    if (qemuCaps->sevCapabilities)
        virQEMUCapsFormatSEVInfo(qemuCaps, &buf);

3823 3824 3825
    if (qemuCaps->kvmSupportsNesting)
        virBufferAddLit(&buf, "<kvmSupportsNesting/>\n");

3826
    virBufferAdjustIndent(&buf, -2);
3827 3828
    virBufferAddLit(&buf, "</qemuCaps>\n");

3829 3830 3831 3832 3833 3834 3835 3836
    if (virBufferCheckError(&buf) == 0)
        ret = virBufferContentAndReset(&buf);

    return ret;
}


static int
3837 3838 3839
virQEMUCapsSaveFile(void *data,
                    const char *filename,
                    void *privData ATTRIBUTE_UNUSED)
3840
{
3841
    virQEMUCapsPtr qemuCaps = data;
3842 3843
    char *xml = NULL;
    int ret = -1;
3844

3845
    xml = virQEMUCapsFormatCache(qemuCaps);
3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856

    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,
3857
              (long long)qemuCaps->libvirtCtime);
3858 3859 3860 3861 3862 3863 3864 3865

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


3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900
/* 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;
}


3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948
/* 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;
}


3949
static bool
3950 3951
virQEMUCapsIsValid(void *data,
                   void *privData)
3952
{
3953 3954
    virQEMUCapsPtr qemuCaps = data;
    virQEMUCapsCachePrivPtr priv = privData;
3955
    bool kvmUsable;
3956
    struct stat sb;
3957
    bool kvmSupportsNesting;
3958 3959 3960 3961

    if (!qemuCaps->binary)
        return true;

3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973
    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;
    }

3974 3975 3976 3977 3978 3979
    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;
3980 3981
    }

3982
    if (sb.st_ctime != qemuCaps->ctime) {
3983 3984 3985
        VIR_DEBUG("Outdated capabilities for '%s': QEMU binary changed "
                  "(%lld vs %lld)",
                  qemuCaps->binary,
3986
                  (long long)sb.st_ctime, (long long)qemuCaps->ctime);
3987 3988 3989
        return false;
    }

3990 3991 3992 3993 3994 3995 3996 3997
    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;
    }

3998
    kvmUsable = virQEMUCapsKVMUsable(priv);
3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015

    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;
    }

4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033
    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;
        }
4034 4035 4036 4037 4038 4039 4040 4041

        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;
        }
4042 4043
    }

4044 4045 4046 4047
    return true;
}


4048 4049 4050 4051 4052 4053 4054 4055
/**
 * virQEMUCapsInitQMPArch:
 * @qemuCaps: QEMU capabilities
 * @mon: QEMU monitor
 *
 * Initialize the architecture for @qemuCaps by asking @mon.
 *
 * Returns: 0 on success, <0 on failure
4056 4057
 */
static int
4058
virQEMUCapsInitQMPArch(virQEMUCapsPtr qemuCaps,
4059 4060 4061 4062 4063 4064
                            qemuMonitorPtr mon)
{
    char *archstr = NULL;
    int ret = -1;

    if (!(archstr = qemuMonitorGetTargetArch(mon)))
4065
        goto cleanup;
4066 4067 4068 4069 4070 4071 4072

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

4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086
    ret = 0;

 cleanup:
    VIR_FREE(archstr);
    return ret;
}


/**
 * virQEMUCapsInitQMPBasicArch:
 * @qemuCaps: QEMU capabilities
 *
 * Initialize @qemuCaps with basic architecture-dependent capabilities.
 */
4087
void
4088 4089
virQEMUCapsInitQMPBasicArch(virQEMUCapsPtr qemuCaps)
{
4090 4091 4092
    /* ACPI only works on x86 and aarch64 */
    if (ARCH_IS_X86(qemuCaps->arch) ||
        qemuCaps->arch == VIR_ARCH_AARCH64) {
4093
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_NO_ACPI);
4094 4095
    }

J
Ján Tomko 已提交
4096 4097
    /* HPET is x86 specific */
    if (ARCH_IS_X86(qemuCaps->arch))
J
Ján Tomko 已提交
4098
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_NO_HPET);
4099
}
4100

4101 4102 4103 4104 4105 4106

static int
virQEMUCapsProbeQMPSchemaCapabilities(virQEMUCapsPtr qemuCaps,
                                      qemuMonitorPtr mon)
{
    struct virQEMUCapsStringFlags *entry;
4107 4108
    virJSONValuePtr schemareply;
    virHashTablePtr schema = NULL;
4109 4110
    size_t i;

4111
    if (!(schemareply = qemuMonitorQueryQMPSchema(mon)))
4112 4113
        return -1;

4114 4115 4116 4117
    if (!(schema = virQEMUQAPISchemaConvert(schemareply)))
        return -1;
    schemareply = NULL;

4118 4119 4120
    for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsQMPSchemaQueries); i++) {
        entry = virQEMUCapsQMPSchemaQueries + i;

4121
        if (virQEMUQAPISchemaPathExists(entry->value, schema))
4122 4123 4124
            virQEMUCapsSet(qemuCaps, entry->flag);
    }

4125 4126 4127 4128 4129 4130 4131 4132
    /* probe also for basic event support */
    for (i = 0; i < ARRAY_CARDINALITY(virQEMUCapsEvents); i++) {
        entry = virQEMUCapsEvents + i;

        if (virQEMUQAPISchemaPathExists(entry->value, schema))
            virQEMUCapsSet(qemuCaps, entry->flag);
    }

4133 4134 4135 4136
    virHashFree(schema);
    return 0;
}

J
Ján Tomko 已提交
4137
#define QEMU_MIN_MAJOR 1
J
Ján Tomko 已提交
4138
#define QEMU_MIN_MINOR 5
J
Ján Tomko 已提交
4139
#define QEMU_MIN_MICRO 0
4140

4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154
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 (qemuMonitorGetVersion(mon,
                              &major, &minor, &micro,
                              &package) < 0) {
        VIR_DEBUG("Failed to query monitor version %s",
4155
                  virGetLastErrorMessage());
4156 4157 4158 4159 4160 4161
        goto cleanup;
    }

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

J
Ján Tomko 已提交
4162 4163 4164 4165 4166 4167
    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);
4168 4169 4170 4171
        goto cleanup;
    }

    qemuCaps->version = major * 1000000 + minor * 1000 + micro;
4172
    qemuCaps->package = package;
4173 4174
    qemuCaps->usedQMP = true;

4175
    if (virQEMUCapsInitQMPArch(qemuCaps, mon) < 0)
4176 4177
        goto cleanup;

4178 4179
    virQEMUCapsInitQMPBasicArch(qemuCaps);

4180 4181 4182
    if (qemuCaps->version >= 1006000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_DEVICE_VIDEO_PRIMARY);

4183 4184 4185 4186
    /* vmport option is supported v2.2.0 onwards */
    if (qemuCaps->version >= 2002000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_VMPORT_OPT);

4187 4188 4189 4190 4191 4192
    /* -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);

4193 4194 4195
    /* 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
4196
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_VHOSTUSER_MULTIQUEUE);
4197

M
Michal Privoznik 已提交
4198 4199 4200 4201
    /* smm option is supported from v2.4.0 */
    if (qemuCaps->version >= 2004000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_SMM_OPT);

4202 4203 4204 4205
    /* sdl -gl option is supported from v2.4.0 (qemu commit id 0b71a5d5) */
    if (qemuCaps->version >= 2004000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_SDL_GL);

4206 4207 4208 4209
    /* Since 2.4.50 ARM virt machine supports gic-version option */
    if (qemuCaps->version >= 2004050)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACH_VIRT_GIC_VERSION);

4210 4211 4212 4213
    /* no way to query if -machine kernel_irqchip supports split */
    if (qemuCaps->version >= 2006000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_MACHINE_KERNEL_IRQCHIP_SPLIT);

4214 4215 4216 4217 4218 4219 4220
    /* 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);
    }

4221 4222 4223 4224 4225
    /* '-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);

4226 4227 4228 4229
    /* no way to query for -numa dist */
    if (qemuCaps->version >= 2010000)
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_NUMA_DIST);

4230 4231 4232 4233 4234 4235
    /* 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);
    }

4236 4237
    if (virQEMUCapsProbeQMPCommands(qemuCaps, mon) < 0)
        goto cleanup;
J
Jiri Denemark 已提交
4238 4239 4240 4241 4242

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

4243 4244
    if (virQEMUCapsProbeQMPEvents(qemuCaps, mon) < 0)
        goto cleanup;
4245
    if (virQEMUCapsProbeQMPDevices(qemuCaps, mon) < 0)
4246 4247 4248
        goto cleanup;
    if (virQEMUCapsProbeQMPMachineTypes(qemuCaps, mon) < 0)
        goto cleanup;
4249
    if (virQEMUCapsProbeQMPCPUDefinitions(qemuCaps, mon, false) < 0)
4250 4251 4252 4253 4254
        goto cleanup;
    if (virQEMUCapsProbeQMPTPM(qemuCaps, mon) < 0)
        goto cleanup;
    if (virQEMUCapsProbeQMPCommandLine(qemuCaps, mon) < 0)
        goto cleanup;
4255 4256
    if (virQEMUCapsProbeQMPMigrationCapabilities(qemuCaps, mon) < 0)
        goto cleanup;
4257 4258 4259
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_QMP_SCHEMA) &&
        virQEMUCapsProbeQMPSchemaCapabilities(qemuCaps, mon) < 0)
        goto cleanup;
4260
    if (virQEMUCapsProbeQMPHostCPU(qemuCaps, mon, false) < 0)
4261
        goto cleanup;
4262

4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273
    /* '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
4274 4275
    /* GIC capabilities, eg. available GIC versions */
    if ((qemuCaps->arch == VIR_ARCH_AARCH64 ||
S
Stefan Schallenberg 已提交
4276
         qemuCaps->arch == VIR_ARCH_ARMV6L ||
A
Andrea Bolognani 已提交
4277 4278 4279 4280
         qemuCaps->arch == VIR_ARCH_ARMV7L) &&
        virQEMUCapsProbeQMPGICCapabilities(qemuCaps, mon) < 0)
        goto cleanup;

4281 4282 4283 4284 4285 4286 4287
    /* 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);

4288 4289 4290 4291
    if (ARCH_IS_X86(qemuCaps->arch) &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_CPU_MODEL_EXPANSION))
        virQEMUCapsSet(qemuCaps, QEMU_CAPS_CPU_CACHE);

4292 4293 4294 4295 4296
    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);
4297 4298
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CCW_CSSID_UNRESTRICTED))
            virQEMUCapsClear(qemuCaps, QEMU_CAPS_DEVICE_VFIO_CCW);
4299 4300
    }

4301 4302
    /* Probe for SEV capabilities */
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_SEV_GUEST)) {
4303 4304 4305 4306 4307 4308
        int rc = virQEMUCapsProbeQMPSEVCapabilities(qemuCaps, mon);

        if (rc < 0)
            goto cleanup;

        if (rc == 0)
4309 4310 4311
            virQEMUCapsClear(qemuCaps, QEMU_CAPS_SEV_GUEST);
    }

4312 4313 4314 4315 4316
    /* To avoid guest ABI regression, blockdev shall be enabled only when
     * we are able to pass the custom 'device_id' for SCSI disks and cdroms. */
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SCSI_DISK_DEVICE_ID))
        virQEMUCapsClear(qemuCaps, QEMU_CAPS_BLOCKDEV);

4317
    ret = 0;
4318
 cleanup:
4319 4320 4321
    return ret;
}

4322

4323 4324 4325 4326 4327 4328
int
virQEMUCapsInitQMPMonitorTCG(virQEMUCapsPtr qemuCaps ATTRIBUTE_UNUSED,
                             qemuMonitorPtr mon)
{
    int ret = -1;

4329 4330 4331
    if (virQEMUCapsProbeQMPCPUDefinitions(qemuCaps, mon, true) < 0)
        goto cleanup;

4332 4333 4334
    if (virQEMUCapsProbeQMPHostCPU(qemuCaps, mon, true) < 0)
        goto cleanup;

4335 4336 4337 4338 4339 4340
    ret = 0;
 cleanup:
    return ret;
}


4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360
#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 },
    };

    virLogMessage(&virLogSelf,
                  VIR_LOG_WARN,
                  __FILE__, __LINE__, __func__,
                  meta,
                  _("Failed to probe capabilities for %s: %s"),
                  binary, virGetLastErrorMessage());
}


4361
static int
4362 4363 4364 4365 4366
virQEMUCapsInitQMPSingle(virQEMUCapsPtr qemuCaps,
                         const char *libDir,
                         uid_t runUid,
                         gid_t runGid,
                         bool onlyTCG)
4367
{
4368
    qemuProcessQMPPtr proc = NULL;
4369 4370
    int ret = -1;

4371
    if (!(proc = qemuProcessQMPNew(qemuCaps->binary, libDir,
4372
                                   runUid, runGid, onlyTCG)))
4373 4374
        goto cleanup;

4375
    if (qemuProcessQMPStart(proc) < 0)
4376 4377
        goto cleanup;

4378 4379 4380 4381
    if (onlyTCG)
        ret = virQEMUCapsInitQMPMonitorTCG(qemuCaps, proc->mon);
    else
        ret = virQEMUCapsInitQMPMonitor(qemuCaps, proc->mon);
4382 4383

 cleanup:
4384 4385 4386
    if (ret < 0)
        virQEMUCapsLogProbeFailure(qemuCaps->binary);

4387
    qemuProcessQMPFree(proc);
4388 4389 4390 4391
    return ret;
}


4392 4393 4394 4395
static int
virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps,
                   const char *libDir,
                   uid_t runUid,
4396
                   gid_t runGid)
4397
{
4398
    if (virQEMUCapsInitQMPSingle(qemuCaps, libDir, runUid, runGid, false) < 0)
4399 4400 4401 4402 4403 4404 4405 4406
        return -1;

    /*
     * If KVM was enabled during the first probe, we need to explicitly probe
     * for TCG capabilities by asking the same binary again and turning KVM
     * off.
     */
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM) &&
4407
        virQEMUCapsInitQMPSingle(qemuCaps, libDir, runUid, runGid, true) < 0)
4408 4409 4410 4411 4412 4413
        return -1;

    return 0;
}


4414
virQEMUCapsPtr
4415
virQEMUCapsNewForBinaryInternal(virArch hostArch,
4416
                                const char *binary,
4417 4418 4419
                                const char *libDir,
                                uid_t runUid,
                                gid_t runGid,
4420
                                unsigned int microcodeVersion,
J
Ján Tomko 已提交
4421
                                const char *kernelVersion)
4422
{
4423
    virQEMUCapsPtr qemuCaps;
4424 4425
    struct stat sb;

4426 4427 4428
    if (!(qemuCaps = virQEMUCapsNew()))
        goto error;

4429 4430
    if (VIR_STRDUP(qemuCaps->binary, binary) < 0)
        goto error;
4431 4432 4433 4434 4435 4436 4437 4438

    /* 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;
    }
4439
    qemuCaps->ctime = sb.st_ctime;
4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450

    /* 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;
    }

4451
    if (virQEMUCapsInitQMP(qemuCaps, libDir, runUid, runGid) < 0)
4452 4453
        goto error;

4454 4455
    qemuCaps->libvirtCtime = virGetSelfLastChanged();
    qemuCaps->libvirtVersion = LIBVIR_VERSION_NUMBER;
4456

4457 4458
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_KVM);
    virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_QEMU);
4459

4460
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) {
4461 4462
        qemuCaps->microcodeVersion = microcodeVersion;

4463 4464
        if (VIR_STRDUP(qemuCaps->kernelVersion, kernelVersion) < 0)
            goto error;
4465 4466

        qemuCaps->kvmSupportsNesting = virQEMUCapsKVMSupportsNesting();
4467 4468
    }

4469
 cleanup:
4470
    return qemuCaps;
4471

4472
 error:
4473 4474
    virObjectUnref(qemuCaps);
    qemuCaps = NULL;
4475
    goto cleanup;
4476 4477
}

4478 4479 4480
static void *
virQEMUCapsNewData(const char *binary,
                   void *privData)
4481
{
4482 4483 4484 4485 4486 4487 4488
    virQEMUCapsCachePrivPtr priv = privData;

    return virQEMUCapsNewForBinaryInternal(priv->hostArch,
                                           binary,
                                           priv->libDir,
                                           priv->runUid,
                                           priv->runGid,
4489
                                           priv->microcodeVersion,
J
Ján Tomko 已提交
4490
                                           priv->kernelVersion);
4491
}
4492 4493


4494 4495 4496 4497 4498 4499 4500
static void *
virQEMUCapsLoadFile(const char *filename,
                    const char *binary,
                    void *privData)
{
    virQEMUCapsPtr qemuCaps = virQEMUCapsNew();
    virQEMUCapsCachePrivPtr priv = privData;
4501

4502 4503
    if (!qemuCaps)
        return NULL;
4504

4505 4506 4507 4508 4509 4510 4511
    if (VIR_STRDUP(qemuCaps->binary, binary) < 0)
        goto error;

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

 cleanup:
4512 4513 4514 4515
    return qemuCaps;

 error:
    virObjectUnref(qemuCaps);
4516 4517
    qemuCaps = NULL;
    goto cleanup;
4518 4519
}

4520

4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553
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]);
    }

4554 4555
    if (!virQEMUCapsGetMachineHotplugCpus(qemuCaps, machineType))
        virQEMUCapsClear(qemuCaps, QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS);
4556 4557 4558
}


4559 4560 4561 4562 4563 4564 4565 4566 4567 4568
virFileCacheHandlers qemuCapsCacheHandlers = {
    .isValid = virQEMUCapsIsValid,
    .newData = virQEMUCapsNewData,
    .loadFile = virQEMUCapsLoadFile,
    .saveFile = virQEMUCapsSaveFile,
    .privFree = virQEMUCapsCachePrivFree,
};


virFileCachePtr
4569
virQEMUCapsCacheNew(const char *libDir,
4570
                    const char *cacheDir,
4571
                    uid_t runUid,
4572 4573
                    gid_t runGid,
                    unsigned int microcodeVersion)
4574
{
4575 4576 4577
    char *capsCacheDir = NULL;
    virFileCachePtr cache = NULL;
    virQEMUCapsCachePrivPtr priv = NULL;
4578
    struct utsname uts;
4579

4580
    if (virAsprintf(&capsCacheDir, "%s/capabilities", cacheDir) < 0)
4581
        goto error;
4582 4583

    if (!(cache = virFileCacheNew(capsCacheDir, "xml", &qemuCapsCacheHandlers)))
4584
        goto error;
4585

4586
    if (VIR_ALLOC(priv) < 0)
4587
        goto error;
4588
    virFileCacheSetPriv(cache, priv);
4589

4590
    if (VIR_STRDUP(priv->libDir, libDir) < 0)
4591 4592
        goto error;

4593
    priv->hostArch = virArchFromHost();
4594

4595 4596
    priv->runUid = runUid;
    priv->runGid = runGid;
4597
    priv->microcodeVersion = microcodeVersion;
4598
    priv->kvmUsable = VIR_TRISTATE_BOOL_ABSENT;
4599

4600 4601 4602 4603
    if (uname(&uts) == 0 &&
        virAsprintf(&priv->kernelVersion, "%s %s", uts.release, uts.version) < 0)
        goto error;

4604 4605
 cleanup:
    VIR_FREE(capsCacheDir);
4606 4607
    return cache;

4608
 error:
4609 4610 4611
    virObjectUnref(cache);
    cache = NULL;
    goto cleanup;
4612 4613 4614
}


4615
virQEMUCapsPtr
4616
virQEMUCapsCacheLookup(virFileCachePtr cache,
4617
                       const char *binary)
4618
{
4619
    virQEMUCapsPtr ret = NULL;
4620

4621
    ret = virFileCacheLookup(cache, binary);
4622 4623

    VIR_DEBUG("Returning caps %p for %s", ret, binary);
4624 4625 4626 4627
    return ret;
}


4628
virQEMUCapsPtr
4629
virQEMUCapsCacheLookupCopy(virFileCachePtr cache,
4630
                           const char *binary,
4631
                           const char *machineType)
4632
{
4633
    virQEMUCapsPtr qemuCaps = virQEMUCapsCacheLookup(cache, binary);
4634
    virQEMUCapsPtr ret;
4635

4636
    if (!qemuCaps)
4637 4638
        return NULL;

4639 4640
    ret = virQEMUCapsNewCopy(qemuCaps);
    virObjectUnref(qemuCaps);
4641 4642 4643 4644

    if (!ret)
        return NULL;

4645
    virQEMUCapsFilterByMachineType(ret, machineType);
4646 4647 4648 4649
    return ret;
}


4650 4651 4652 4653 4654
static int
virQEMUCapsCompareArch(const void *payload,
                       const void *name ATTRIBUTE_UNUSED,
                       const void *opaque)
{
4655
    struct virQEMUCapsSearchData *data = (struct virQEMUCapsSearchData *)opaque;
4656 4657
    const virQEMUCaps *qemuCaps = payload;

4658 4659 4660 4661 4662 4663 4664 4665 4666
    if (qemuCaps->arch != data->arch)
        return false;

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

    return true;
4667 4668 4669 4670
}


virQEMUCapsPtr
4671
virQEMUCapsCacheLookupByArch(virFileCachePtr cache,
4672 4673 4674
                             virArch arch)
{
    virQEMUCapsPtr ret = NULL;
4675 4676 4677 4678
    const char *binaryFilters[] = {
        "qemu-system-",
        NULL,
    };
4679 4680 4681 4682
    virArch archs[] = {
        arch,
        virQEMUCapsFindTarget(virArchFromHost(), arch),
    };
4683
    size_t i;
4684 4685
    size_t j;

4686 4687 4688 4689 4690 4691
    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],
            };
4692

4693 4694 4695 4696
            ret = virFileCacheLookupByFunc(cache, virQEMUCapsCompareArch, &data);
            if (ret)
                goto done;
        }
4697 4698
    }

4699 4700 4701 4702 4703
    virReportError(VIR_ERR_INVALID_ARG,
                   _("unable to find any emulator to serve '%s' "
                     "architecture"), virArchToString(arch));

 done:
4704 4705
    VIR_DEBUG("Returning caps %p for arch %s", ret, virArchToString(arch));

4706 4707 4708 4709
    return ret;
}


4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795
/**
 * 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 {
4796
        machine = virQEMUCapsGetPreferredMachine(qemuCaps);
4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827
    }

    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;
}

4828 4829 4830 4831 4832 4833 4834
bool
virQEMUCapsSupportsVmport(virQEMUCapsPtr qemuCaps,
                          const virDomainDef *def)
{
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_VMPORT_OPT))
        return false;

4835 4836
    return qemuDomainIsI440FX(def) ||
        qemuDomainIsQ35(def) ||
4837 4838 4839 4840
        STREQ(def->os.machine, "isapc");
}


4841 4842 4843 4844 4845 4846 4847
bool
virQEMUCapsIsMachineSupported(virQEMUCapsPtr qemuCaps,
                              const char *canonical_machine)
{
    size_t i;

    for (i = 0; i < qemuCaps->nmachineTypes; i++) {
4848
        if (STREQ(canonical_machine, qemuCaps->machineTypes[i].name))
4849 4850 4851 4852
            return true;
    }
    return false;
}
4853 4854


4855 4856 4857 4858
/*
 * The preferred machine to use if none is listed explicitly
 * Note that this may differ from QEMU's own default machine
 */
4859
const char *
4860
virQEMUCapsGetPreferredMachine(virQEMUCapsPtr qemuCaps)
4861 4862 4863
{
    if (!qemuCaps->nmachineTypes)
        return NULL;
4864
    return qemuCaps->machineTypes[0].name;
4865
}
4866 4867


4868
static int
4869
virQEMUCapsFillDomainLoaderCaps(virDomainCapsLoaderPtr capsLoader,
4870 4871
                                virFirmwarePtr *firmwares,
                                size_t nfirmwares)
4872
{
4873 4874
    size_t i;

4875
    capsLoader->supported = true;
4876

4877
    if (VIR_ALLOC_N(capsLoader->values.values, nfirmwares) < 0)
4878 4879
        return -1;

4880 4881
    for (i = 0; i < nfirmwares; i++) {
        const char *filename = firmwares[i]->name;
4882 4883 4884 4885 4886 4887

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

4888
        if (VIR_STRDUP(capsLoader->values.values[capsLoader->values.nvalues],
4889 4890
                       filename) < 0)
            return -1;
4891
        capsLoader->values.nvalues++;
4892 4893
    }

4894
    VIR_DOMAIN_CAPS_ENUM_SET(capsLoader->type,
4895 4896
                             VIR_DOMAIN_LOADER_TYPE_ROM);

4897 4898
    VIR_DOMAIN_CAPS_ENUM_SET(capsLoader->type,
                             VIR_DOMAIN_LOADER_TYPE_PFLASH);
4899 4900


4901 4902 4903
    VIR_DOMAIN_CAPS_ENUM_SET(capsLoader->readonly,
                             VIR_TRISTATE_BOOL_YES,
                             VIR_TRISTATE_BOOL_NO);
4904
    return 0;
4905 4906 4907
}


4908
static int
4909
virQEMUCapsFillDomainOSCaps(virDomainCapsOSPtr os,
4910 4911
                            virFirmwarePtr *firmwares,
                            size_t nfirmwares)
4912
{
4913
    virDomainCapsLoaderPtr capsLoader = &os->loader;
4914

4915
    os->supported = true;
4916
    if (virQEMUCapsFillDomainLoaderCaps(capsLoader, firmwares, nfirmwares) < 0)
4917 4918
        return -1;
    return 0;
4919 4920 4921
}


4922 4923 4924 4925 4926
static int
virQEMUCapsFillDomainCPUCaps(virCapsPtr caps,
                             virQEMUCapsPtr qemuCaps,
                             virDomainCapsPtr domCaps)
{
4927 4928
    if (virQEMUCapsIsCPUModeSupported(qemuCaps, caps, domCaps->virttype,
                                      VIR_CPU_MODE_HOST_PASSTHROUGH))
4929 4930
        domCaps->cpu.hostPassthrough = true;

4931
    if (virQEMUCapsIsCPUModeSupported(qemuCaps, caps, domCaps->virttype,
4932
                                      VIR_CPU_MODE_HOST_MODEL)) {
4933 4934
        virCPUDefPtr cpu = virQEMUCapsGetHostModel(qemuCaps, domCaps->virttype,
                                                   VIR_QEMU_CAPS_HOST_CPU_REPORTED);
4935 4936
        domCaps->cpu.hostModel = virCPUDefCopy(cpu);
    }
4937 4938 4939 4940 4941

    if (virQEMUCapsIsCPUModeSupported(qemuCaps, caps, domCaps->virttype,
                                      VIR_CPU_MODE_CUSTOM)) {
        virDomainCapsCPUModelsPtr filtered = NULL;
        char **models = NULL;
4942
        const char *blacklist[] = { "host", NULL };
4943

J
Jiri Denemark 已提交
4944
        if (virCPUGetModels(domCaps->arch, &models) >= 0) {
4945 4946 4947 4948 4949 4950 4951 4952
            virDomainCapsCPUModelsPtr cpus;

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

            filtered = virDomainCapsCPUModelsFilter(cpus,
4953
                                                    (const char **)models,
4954
                                                    blacklist);
4955
            virStringListFree(models);
4956 4957
        }
        domCaps->cpu.custom = filtered;
4958
    }
4959 4960 4961 4962 4963

    return 0;
}


4964 4965 4966 4967 4968 4969 4970 4971 4972 4973
static int
virQEMUCapsFillDomainIOThreadCaps(virQEMUCapsPtr qemuCaps,
                                  virDomainCapsPtr domCaps)
{
    domCaps->iothreads = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_IOTHREAD);

    return 0;
}


4974
static int
4975
virQEMUCapsFillDomainDeviceDiskCaps(virQEMUCapsPtr qemuCaps,
4976
                                    const char *machine,
4977 4978
                                    virDomainCapsDeviceDiskPtr disk)
{
4979
    disk->supported = true;
4980 4981 4982
    /* QEMU supports all of these */
    VIR_DOMAIN_CAPS_ENUM_SET(disk->diskDevice,
                             VIR_DOMAIN_DISK_DEVICE_DISK,
4983 4984
                             VIR_DOMAIN_DISK_DEVICE_CDROM,
                             VIR_DOMAIN_DISK_DEVICE_LUN);
4985 4986

    /* PowerPC pseries based VMs do not support floppy device */
4987
    if (!qemuDomainMachineIsPSeries(machine, qemuCaps->arch)) {
4988
        VIR_DOMAIN_CAPS_ENUM_SET(disk->diskDevice, VIR_DOMAIN_DISK_DEVICE_FLOPPY);
4989 4990
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_FDC);
    }
4991

4992
    if (qemuDomainMachineHasBuiltinIDE(machine, qemuCaps->arch))
4993 4994
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_IDE);

4995 4996 4997 4998 4999 5000 5001
    VIR_DOMAIN_CAPS_ENUM_SET(disk->bus,
                             VIR_DOMAIN_DISK_BUS_SCSI,
                             VIR_DOMAIN_DISK_BUS_VIRTIO,
                             /* VIR_DOMAIN_DISK_BUS_SD */);

    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_USB_STORAGE))
        VIR_DOMAIN_CAPS_ENUM_SET(disk->bus, VIR_DOMAIN_DISK_BUS_USB);
5002 5003 5004 5005

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

5006
    return 0;
5007 5008 5009
}


5010 5011 5012 5013 5014 5015
static int
virQEMUCapsFillDomainDeviceGraphicsCaps(virQEMUCapsPtr qemuCaps,
                                        virDomainCapsDeviceGraphicsPtr dev)
{
    dev->supported = true;

J
Ján Tomko 已提交
5016
    VIR_DOMAIN_CAPS_ENUM_SET(dev->type, VIR_DOMAIN_GRAPHICS_TYPE_SDL);
5017 5018 5019 5020 5021 5022 5023 5024 5025
    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;
}


5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037
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);
5038
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_QXL))
5039 5040 5041 5042 5043 5044 5045 5046
        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;
}


5047
static int
5048 5049 5050 5051 5052 5053
virQEMUCapsFillDomainDeviceHostdevCaps(virQEMUCapsPtr qemuCaps,
                                       virDomainCapsDeviceHostdevPtr hostdev)
{
    bool supportsPassthroughKVM = qemuHostdevHostSupportsPassthroughLegacy();
    bool supportsPassthroughVFIO = qemuHostdevHostSupportsPassthroughVFIO();

5054
    hostdev->supported = true;
5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066
    /* 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,
5067 5068
                             VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI,
                             VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI);
5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080

    /* 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
5081
    if (supportsPassthroughKVM) {
5082 5083 5084 5085
        VIR_DOMAIN_CAPS_ENUM_SET(hostdev->pciBackend,
                                 VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT,
                                 VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM);
    }
5086
    return 0;
5087 5088 5089
}


5090 5091 5092 5093 5094 5095 5096
/**
 * virQEMUCapsSupportsGICVersion:
 * @qemuCaps: QEMU capabilities
 * @virtType: domain type
 * @version: GIC version
 *
 * Checks the QEMU binary with capabilities @qemuCaps supports a specific
5097 5098
 * GIC version for a domain of type @virtType. If @qemuCaps is NULL, the GIC
 * @version is considered unsupported.
5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109
 *
 * Returns: true if the binary supports the requested GIC version, false
 *          otherwise
 */
bool
virQEMUCapsSupportsGICVersion(virQEMUCapsPtr qemuCaps,
                              virDomainVirtType virtType,
                              virGICVersion version)
{
    size_t i;

5110 5111 5112
    if (!qemuCaps)
        return false;

5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131
    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;
}


5132 5133 5134 5135 5136 5137 5138 5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155
/**
 * 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;
5156
    virGICVersion version;
5157

5158
    if (!qemuDomainMachineIsARMVirt(domCaps->machine, domCaps->arch))
5159 5160
        return 0;

5161 5162 5163 5164 5165 5166
    for (version = VIR_GIC_VERSION_LAST - 1;
         version > VIR_GIC_VERSION_NONE;
         version--) {
        if (!virQEMUCapsSupportsGICVersion(qemuCaps,
                                           domCaps->virttype,
                                           version))
5167 5168 5169 5170
            continue;

        gic->supported = true;
        VIR_DOMAIN_CAPS_ENUM_SET(gic->version,
5171
                                 version);
5172 5173 5174 5175 5176 5177
    }

    return 0;
}


5178 5179 5180 5181 5182 5183 5184 5185 5186 5187 5188 5189 5190 5191 5192 5193
/**
 * 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;
5194
    VIR_AUTOPTR(virSEVCapability) sev = NULL;
5195 5196 5197 5198 5199 5200 5201 5202

    if (!cap)
        return 0;

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

    if (VIR_STRDUP(sev->pdh, cap->pdh) < 0)
5203
        return -1;
5204 5205

    if (VIR_STRDUP(sev->cert_chain, cap->cert_chain) < 0)
5206
        return -1;
5207 5208 5209 5210 5211

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

5212
    return 0;
5213 5214 5215
}


5216
int
5217 5218
virQEMUCapsFillDomainCaps(virCapsPtr caps,
                          virDomainCapsPtr domCaps,
5219
                          virQEMUCapsPtr qemuCaps,
5220
                          virFirmwarePtr *firmwares,
5221
                          size_t nfirmwares)
5222
{
5223
    virDomainCapsOSPtr os = &domCaps->os;
5224 5225
    virDomainCapsDeviceDiskPtr disk = &domCaps->disk;
    virDomainCapsDeviceHostdevPtr hostdev = &domCaps->hostdev;
5226
    virDomainCapsDeviceGraphicsPtr graphics = &domCaps->graphics;
5227
    virDomainCapsDeviceVideoPtr video = &domCaps->video;
5228

5229 5230
    domCaps->maxvcpus = virQEMUCapsGetMachineMaxCpus(qemuCaps,
                                                     domCaps->machine);
5231
    if (domCaps->virttype == VIR_DOMAIN_VIRT_KVM) {
5232 5233 5234 5235 5236 5237
        int hostmaxvcpus;

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

        domCaps->maxvcpus = MIN(domCaps->maxvcpus, hostmaxvcpus);
5238
    }
5239

5240 5241 5242
    domCaps->vmcoreinfo = virQEMUCapsGet(qemuCaps,
                                         QEMU_CAPS_DEVICE_VMCOREINFO);

5243 5244
    domCaps->genid = virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VMGENID);

5245
    if (virQEMUCapsFillDomainOSCaps(os, firmwares, nfirmwares) < 0 ||
5246
        virQEMUCapsFillDomainCPUCaps(caps, qemuCaps, domCaps) < 0 ||
5247
        virQEMUCapsFillDomainIOThreadCaps(qemuCaps, domCaps) < 0 ||
5248 5249 5250
        virQEMUCapsFillDomainDeviceDiskCaps(qemuCaps,
                                            domCaps->machine, disk) < 0 ||
        virQEMUCapsFillDomainDeviceGraphicsCaps(qemuCaps, graphics) < 0 ||
5251
        virQEMUCapsFillDomainDeviceVideoCaps(qemuCaps, video) < 0 ||
5252
        virQEMUCapsFillDomainDeviceHostdevCaps(qemuCaps, hostdev) < 0 ||
5253 5254
        virQEMUCapsFillDomainFeatureGICCaps(qemuCaps, domCaps) < 0 ||
        virQEMUCapsFillDomainFeatureSEVCaps(qemuCaps, domCaps) < 0)
5255
        return -1;
5256

5257
    return 0;
5258
}
5259 5260 5261 5262 5263 5264 5265 5266


void
virQEMUCapsSetMicrocodeVersion(virQEMUCapsPtr qemuCaps,
                               unsigned int microcodeVersion)
{
    qemuCaps->microcodeVersion = microcodeVersion;
}
5267 5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283


/**
 * 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);
}