xen_hypervisor.c 109.9 KB
Newer Older
1 2 3
/*
 * xen_internal.c: direct access to Xen hypervisor level
 *
4
 * Copyright (C) 2005-2013 Red Hat, Inc.
5
 *
O
Osier Yang 已提交
6 7 8 9 10 11 12 13 14 15 16
 * 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
17
 * License along with this library.  If not, see
O
Osier Yang 已提交
18
 * <http://www.gnu.org/licenses/>.
19 20 21 22
 *
 * Daniel Veillard <veillard@redhat.com>
 */

23
#include <config.h>
24

25 26
#include <stdio.h>
#include <string.h>
27
/* required for uint8_t, uint32_t, etc ... */
28 29 30 31 32 33 34
#include <stdint.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <unistd.h>
#include <fcntl.h>
#include <sys/mman.h>
#include <sys/ioctl.h>
35
#include <limits.h>
36
#include <stdint.h>
37 38
#include <regex.h>
#include <errno.h>
39

J
John Levon 已提交
40
#ifdef __sun
41
# include <sys/systeminfo.h>
J
John Levon 已提交
42

43
# include <priv.h>
J
John Levon 已提交
44

45 46 47
# ifndef PRIV_XVM_CONTROL
#  define PRIV_XVM_CONTROL ((const char *)"xvm_control")
# endif
J
John Levon 已提交
48 49 50

#endif /* __sun */

51
/* required for dom0_getdomaininfo_t */
52
#include <xen/dom0_ops.h>
53
#include <xen/version.h>
54
#ifdef HAVE_XEN_LINUX_PRIVCMD_H
55
# include <xen/linux/privcmd.h>
56
#else
57 58 59
# ifdef HAVE_XEN_SYS_PRIVCMD_H
#  include <xen/sys/privcmd.h>
# endif
60
#endif
61

62 63 64
/* required for shutdown flags */
#include <xen/sched.h>

65
#include "virerror.h"
66
#include "virlog.h"
67
#include "datatypes.h"
68
#include "driver.h"
69
#include "virutil.h"
70 71
#include "xen_driver.h"
#include "xen_hypervisor.h"
72
#include "xs_internal.h"
73
#include "virstatslinux.h"
74
#include "block_stats.h"
75
#include "xend_internal.h"
76
#include "virbuffer.h"
77
#include "capabilities.h"
78
#include "viralloc.h"
79
#include "virthread.h"
E
Eric Blake 已提交
80
#include "virfile.h"
81
#include "virnodesuspend.h"
82
#include "virtypedparam.h"
83

84 85
#define VIR_FROM_THIS VIR_FROM_XEN

86
/*
87
 * so far there is 2 versions of the structures usable for doing
88 89 90 91
 * hypervisor calls.
 */
/* the old one */
typedef struct v0_hypercall_struct {
92 93
    unsigned long op;
    unsigned long arg[5];
94
} v0_hypercall_t;
95 96

#ifdef __linux__
97
# define XEN_V0_IOCTL_HYPERCALL_CMD \
98 99 100 101 102 103 104
        _IOC(_IOC_NONE, 'P', 0, sizeof(v0_hypercall_t))
/* the new one */
typedef struct v1_hypercall_struct
{
    uint64_t op;
    uint64_t arg[5];
} v1_hypercall_t;
105
# define XEN_V1_IOCTL_HYPERCALL_CMD                  \
106
    _IOC(_IOC_NONE, 'P', 0, sizeof(v1_hypercall_t))
107
typedef v1_hypercall_t hypercall_t;
108
#elif defined(__sun)
109 110
typedef privcmd_hypercall_t hypercall_t;
#else
111
# error "unsupported platform"
112
#endif
113 114

#ifndef __HYPERVISOR_sysctl
115
# define __HYPERVISOR_sysctl 35
116 117
#endif
#ifndef __HYPERVISOR_domctl
118
# define __HYPERVISOR_domctl 36
119
#endif
120

121
#ifdef WITH_RHEL5_API
122
# define SYS_IFACE_MIN_VERS_NUMA 3
123
#else
124
# define SYS_IFACE_MIN_VERS_NUMA 4
125 126
#endif

127
static int xen_ioctl_hypercall_cmd = 0;
P
Philipp Hahn 已提交
128 129 130 131 132 133 134
static struct xenHypervisorVersions hv_versions = {
    .hv = 0,
    .hypervisor = 2,
    .sys_interface = -1,
    .dom_interface = -1,
};

135
static int kb_per_pages = 0;
136

137 138 139 140 141 142 143 144
/* Regular expressions used by xenHypervisorGetCapabilities, and
 * compiled once by xenHypervisorInit.  Note that these are POSIX.2
 * extended regular expressions (regex(7)).
 */
static const char *flags_hvm_re = "^flags[[:blank:]]+:.* (vmx|svm)[[:space:]]";
static regex_t flags_hvm_rec;
static const char *flags_pae_re = "^flags[[:blank:]]+:.* pae[[:space:]]";
static regex_t flags_pae_rec;
145
static const char *xen_cap_re = "(xen|hvm)-[[:digit:]]+\\.[[:digit:]]+-(x86_32|x86_64|ia64|powerpc64)(p|be)?";
146 147
static regex_t xen_cap_rec;

148 149 150 151
/*
 * The content of the structures for a getdomaininfolist system hypercall
 */
#ifndef DOMFLAGS_DYING
152 153 154 155 156 157 158 159 160 161
# define DOMFLAGS_DYING     (1<<0) /* Domain is scheduled to die.             */
# define DOMFLAGS_HVM       (1<<1) /* Domain is HVM                           */
# define DOMFLAGS_SHUTDOWN  (1<<2) /* The guest OS has shut down.             */
# define DOMFLAGS_PAUSED    (1<<3) /* Currently paused by control software.   */
# define DOMFLAGS_BLOCKED   (1<<4) /* Currently blocked pending an event.     */
# define DOMFLAGS_RUNNING   (1<<5) /* Domain is currently running.            */
# define DOMFLAGS_CPUMASK      255 /* CPU to which this domain is bound.      */
# define DOMFLAGS_CPUSHIFT       8
# define DOMFLAGS_SHUTDOWNMASK 255 /* DOMFLAGS_SHUTDOWN guest-supplied code.  */
# define DOMFLAGS_SHUTDOWNSHIFT 16
162 163
#endif

164 165 166 167 168
/*
 * These flags explain why a system is in the state of "shutdown".  Normally,
 * They are defined in xen/sched.h
 */
#ifndef SHUTDOWN_poweroff
169 170 171 172
# define SHUTDOWN_poweroff   0  /* Domain exited normally. Clean up and kill. */
# define SHUTDOWN_reboot     1  /* Clean up, kill, and then restart.          */
# define SHUTDOWN_suspend    2  /* Clean up, save suspend info, kill.         */
# define SHUTDOWN_crash      3  /* Tell controller we've crashed.             */
173 174
#endif

175 176 177 178 179 180
#define XEN_V0_OP_GETDOMAININFOLIST	38
#define XEN_V1_OP_GETDOMAININFOLIST	38
#define XEN_V2_OP_GETDOMAININFOLIST	6

struct xen_v0_getdomaininfo {
    domid_t  domain;	/* the domain number */
R
Richard W.M. Jones 已提交
181
    uint32_t flags;	/* flags, see before */
182 183
    uint64_t tot_pages;	/* total number of pages used */
    uint64_t max_pages;	/* maximum number of pages allowed */
184
    unsigned long shared_info_frame; /* MFN of shared_info struct */
185 186 187 188 189 190 191 192
    uint64_t cpu_time;  /* CPU time used */
    uint32_t nr_online_vcpus;  /* Number of VCPUs currently online. */
    uint32_t max_vcpu_id; /* Maximum VCPUID in use by this domain. */
    uint32_t ssidref;
    xen_domain_handle_t handle;
};
typedef struct xen_v0_getdomaininfo xen_v0_getdomaininfo;

193 194
struct xen_v2_getdomaininfo {
    domid_t  domain;	/* the domain number */
R
Richard W.M. Jones 已提交
195
    uint32_t flags;	/* flags, see before */
196 197 198 199 200 201 202 203 204 205 206
    uint64_t tot_pages;	/* total number of pages used */
    uint64_t max_pages;	/* maximum number of pages allowed */
    uint64_t shared_info_frame; /* MFN of shared_info struct */
    uint64_t cpu_time;  /* CPU time used */
    uint32_t nr_online_vcpus;  /* Number of VCPUs currently online. */
    uint32_t max_vcpu_id; /* Maximum VCPUID in use by this domain. */
    uint32_t ssidref;
    xen_domain_handle_t handle;
};
typedef struct xen_v2_getdomaininfo xen_v2_getdomaininfo;

207 208 209 210 211 212 213

/* As of Hypervisor Call v2,  DomCtl v5 we are now 8-byte aligned
   even on 32-bit archs when dealing with uint64_t */
#define ALIGN_64 __attribute__((aligned(8)))

struct xen_v2d5_getdomaininfo {
    domid_t  domain;	/* the domain number */
R
Richard W.M. Jones 已提交
214
    uint32_t flags;	/* flags, see before */
215 216 217 218 219 220 221 222 223 224 225
    uint64_t tot_pages ALIGN_64;	/* total number of pages used */
    uint64_t max_pages ALIGN_64;	/* maximum number of pages allowed */
    uint64_t shared_info_frame ALIGN_64; /* MFN of shared_info struct */
    uint64_t cpu_time ALIGN_64;  /* CPU time used */
    uint32_t nr_online_vcpus;  /* Number of VCPUs currently online. */
    uint32_t max_vcpu_id; /* Maximum VCPUID in use by this domain. */
    uint32_t ssidref;
    xen_domain_handle_t handle;
};
typedef struct xen_v2d5_getdomaininfo xen_v2d5_getdomaininfo;

226 227 228 229 230 231 232 233 234 235 236 237 238 239 240
struct xen_v2d6_getdomaininfo {
    domid_t  domain;	/* the domain number */
    uint32_t flags;	/* flags, see before */
    uint64_t tot_pages ALIGN_64;	/* total number of pages used */
    uint64_t max_pages ALIGN_64;	/* maximum number of pages allowed */
    uint64_t shr_pages ALIGN_64;    /* number of shared pages */
    uint64_t shared_info_frame ALIGN_64; /* MFN of shared_info struct */
    uint64_t cpu_time ALIGN_64;  /* CPU time used */
    uint32_t nr_online_vcpus;  /* Number of VCPUs currently online. */
    uint32_t max_vcpu_id; /* Maximum VCPUID in use by this domain. */
    uint32_t ssidref;
    xen_domain_handle_t handle;
};
typedef struct xen_v2d6_getdomaininfo xen_v2d6_getdomaininfo;

J
Jim Fehlig 已提交
241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256
struct xen_v2d7_getdomaininfo {
    domid_t  domain;	/* the domain number */
    uint32_t flags;	/* flags, see before */
    uint64_t tot_pages ALIGN_64;	/* total number of pages used */
    uint64_t max_pages ALIGN_64;	/* maximum number of pages allowed */
    uint64_t shr_pages ALIGN_64;    /* number of shared pages */
    uint64_t shared_info_frame ALIGN_64; /* MFN of shared_info struct */
    uint64_t cpu_time ALIGN_64;  /* CPU time used */
    uint32_t nr_online_vcpus;  /* Number of VCPUs currently online. */
    uint32_t max_vcpu_id; /* Maximum VCPUID in use by this domain. */
    uint32_t ssidref;
    xen_domain_handle_t handle;
    uint32_t cpupool;
};
typedef struct xen_v2d7_getdomaininfo xen_v2d7_getdomaininfo;

J
Jim Fehlig 已提交
257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273
struct xen_v2d8_getdomaininfo {
    domid_t  domain;	/* the domain number */
    uint32_t flags;	/* flags, see before */
    uint64_t tot_pages ALIGN_64;	/* total number of pages used */
    uint64_t max_pages ALIGN_64;	/* maximum number of pages allowed */
    uint64_t shr_pages ALIGN_64;    /* number of shared pages */
    uint64_t paged_pages ALIGN_64;    /* number of paged pages */
    uint64_t shared_info_frame ALIGN_64; /* MFN of shared_info struct */
    uint64_t cpu_time ALIGN_64;  /* CPU time used */
    uint32_t nr_online_vcpus;  /* Number of VCPUs currently online. */
    uint32_t max_vcpu_id; /* Maximum VCPUID in use by this domain. */
    uint32_t ssidref;
    xen_domain_handle_t handle;
    uint32_t cpupool;
};
typedef struct xen_v2d8_getdomaininfo xen_v2d8_getdomaininfo;

274 275 276
union xen_getdomaininfo {
    struct xen_v0_getdomaininfo v0;
    struct xen_v2_getdomaininfo v2;
277
    struct xen_v2d5_getdomaininfo v2d5;
278
    struct xen_v2d6_getdomaininfo v2d6;
J
Jim Fehlig 已提交
279
    struct xen_v2d7_getdomaininfo v2d7;
J
Jim Fehlig 已提交
280
    struct xen_v2d8_getdomaininfo v2d8;
281 282 283 284 285 286
};
typedef union xen_getdomaininfo xen_getdomaininfo;

union xen_getdomaininfolist {
    struct xen_v0_getdomaininfo *v0;
    struct xen_v2_getdomaininfo *v2;
287
    struct xen_v2d5_getdomaininfo *v2d5;
288
    struct xen_v2d6_getdomaininfo *v2d6;
J
Jim Fehlig 已提交
289
    struct xen_v2d7_getdomaininfo *v2d7;
J
Jim Fehlig 已提交
290
    struct xen_v2d8_getdomaininfo *v2d8;
291 292 293
};
typedef union xen_getdomaininfolist xen_getdomaininfolist;

294 295 296 297 298 299 300 301 302 303 304 305

struct xen_v2_getschedulerid {
    uint32_t sched_id; /* Get Scheduler ID from Xen */
};
typedef struct xen_v2_getschedulerid xen_v2_getschedulerid;


union xen_getschedulerid {
    struct xen_v2_getschedulerid *v2;
};
typedef union xen_getschedulerid xen_getschedulerid;

306 307 308 309 310 311 312 313 314
struct xen_v2s4_availheap {
    uint32_t min_bitwidth;  /* Smallest address width (zero if don't care). */
    uint32_t max_bitwidth;  /* Largest address width (zero if don't care). */
    int32_t  node;          /* NUMA node (-1 for sum across all nodes). */
    uint64_t avail_bytes;   /* Bytes available in the specified region. */
};

typedef struct xen_v2s4_availheap  xen_v2s4_availheap;

315 316 317 318 319 320 321 322 323
struct xen_v2s5_availheap {
    uint32_t min_bitwidth;  /* Smallest address width (zero if don't care). */
    uint32_t max_bitwidth;  /* Largest address width (zero if don't care). */
    int32_t  node;          /* NUMA node (-1 for sum across all nodes). */
    uint64_t avail_bytes ALIGN_64;   /* Bytes available in the specified region. */
};

typedef struct xen_v2s5_availheap  xen_v2s5_availheap;

324

325
#define XEN_GETDOMAININFOLIST_ALLOC(domlist, size)                      \
P
Philipp Hahn 已提交
326
    (hv_versions.hypervisor < 2 ?                                       \
327
     (VIR_ALLOC_N(domlist.v0, (size)) == 0) :                           \
J
Jim Fehlig 已提交
328 329 330
     (hv_versions.dom_interface >= 8 ?                                  \
      (VIR_ALLOC_N(domlist.v2d8, (size)) == 0) :                        \
     (hv_versions.dom_interface == 7 ?                                  \
J
Jim Fehlig 已提交
331
      (VIR_ALLOC_N(domlist.v2d7, (size)) == 0) :                        \
P
Philipp Hahn 已提交
332
     (hv_versions.dom_interface == 6 ?                                  \
333
      (VIR_ALLOC_N(domlist.v2d6, (size)) == 0) :                        \
P
Philipp Hahn 已提交
334
     (hv_versions.dom_interface == 5 ?                                  \
335
      (VIR_ALLOC_N(domlist.v2d5, (size)) == 0) :                        \
J
Jim Fehlig 已提交
336
      (VIR_ALLOC_N(domlist.v2, (size)) == 0))))))
337

338
#define XEN_GETDOMAININFOLIST_FREE(domlist)            \
P
Philipp Hahn 已提交
339
    (hv_versions.hypervisor < 2 ?                      \
340
     VIR_FREE(domlist.v0) :                            \
J
Jim Fehlig 已提交
341 342 343
     (hv_versions.dom_interface >= 8 ?                 \
      VIR_FREE(domlist.v2d8) :                         \
     (hv_versions.dom_interface == 7 ?                 \
J
Jim Fehlig 已提交
344
      VIR_FREE(domlist.v2d7) :                         \
P
Philipp Hahn 已提交
345
     (hv_versions.dom_interface == 6 ?                 \
346
      VIR_FREE(domlist.v2d6) :                         \
P
Philipp Hahn 已提交
347
     (hv_versions.dom_interface == 5 ?                 \
348
      VIR_FREE(domlist.v2d5) :                         \
J
Jim Fehlig 已提交
349
      VIR_FREE(domlist.v2))))))
350

351
#define XEN_GETDOMAININFOLIST_CLEAR(domlist, size)            \
P
Philipp Hahn 已提交
352
    (hv_versions.hypervisor < 2 ?                             \
353
     memset(domlist.v0, 0, sizeof(*domlist.v0) * size) :      \
J
Jim Fehlig 已提交
354 355 356
     (hv_versions.dom_interface >= 8 ?                        \
      memset(domlist.v2d8, 0, sizeof(*domlist.v2d8) * size) : \
     (hv_versions.dom_interface == 7 ?                        \
J
Jim Fehlig 已提交
357
      memset(domlist.v2d7, 0, sizeof(*domlist.v2d7) * size) : \
P
Philipp Hahn 已提交
358
     (hv_versions.dom_interface == 6 ?                        \
359
      memset(domlist.v2d6, 0, sizeof(*domlist.v2d6) * size) : \
P
Philipp Hahn 已提交
360
     (hv_versions.dom_interface == 5 ?                        \
361
      memset(domlist.v2d5, 0, sizeof(*domlist.v2d5) * size) : \
J
Jim Fehlig 已提交
362
      memset(domlist.v2, 0, sizeof(*domlist.v2) * size))))))
363 364

#define XEN_GETDOMAININFOLIST_DOMAIN(domlist, n)    \
P
Philipp Hahn 已提交
365
    (hv_versions.hypervisor < 2 ?                   \
366
     domlist.v0[n].domain :                         \
J
Jim Fehlig 已提交
367 368 369
     (hv_versions.dom_interface >= 8 ?              \
      domlist.v2d8[n].domain :                      \
     (hv_versions.dom_interface == 7 ?              \
J
Jim Fehlig 已提交
370
      domlist.v2d7[n].domain :                      \
P
Philipp Hahn 已提交
371
     (hv_versions.dom_interface == 6 ?              \
372
      domlist.v2d6[n].domain :                      \
P
Philipp Hahn 已提交
373
     (hv_versions.dom_interface == 5 ?              \
374
      domlist.v2d5[n].domain :                      \
J
Jim Fehlig 已提交
375
      domlist.v2[n].domain)))))
376

377
#define XEN_GETDOMAININFOLIST_UUID(domlist, n)      \
P
Philipp Hahn 已提交
378
    (hv_versions.hypervisor < 2 ?                   \
379
     domlist.v0[n].handle :                         \
J
Jim Fehlig 已提交
380 381 382
     (hv_versions.dom_interface >= 8 ?              \
      domlist.v2d8[n].handle :                      \
     (hv_versions.dom_interface == 7 ?              \
J
Jim Fehlig 已提交
383
      domlist.v2d7[n].handle :                      \
P
Philipp Hahn 已提交
384
     (hv_versions.dom_interface == 6 ?              \
385
      domlist.v2d6[n].handle :                      \
P
Philipp Hahn 已提交
386
     (hv_versions.dom_interface == 5 ?              \
387
      domlist.v2d5[n].handle :                      \
J
Jim Fehlig 已提交
388
      domlist.v2[n].handle)))))
389

390
#define XEN_GETDOMAININFOLIST_DATA(domlist)        \
P
Philipp Hahn 已提交
391
    (hv_versions.hypervisor < 2 ?                  \
392
     (void*)(domlist->v0) :                        \
J
Jim Fehlig 已提交
393 394 395
     (hv_versions.dom_interface >= 8 ?             \
      (void*)(domlist->v2d8) :                     \
     (hv_versions.dom_interface == 7 ?             \
J
Jim Fehlig 已提交
396
      (void*)(domlist->v2d7) :                     \
P
Philipp Hahn 已提交
397
     (hv_versions.dom_interface == 6 ?             \
398
      (void*)(domlist->v2d6) :                     \
P
Philipp Hahn 已提交
399
     (hv_versions.dom_interface == 5 ?             \
400
      (void*)(domlist->v2d5) :                     \
J
Jim Fehlig 已提交
401
      (void*)(domlist->v2))))))
402 403

#define XEN_GETDOMAININFO_SIZE                     \
P
Philipp Hahn 已提交
404
    (hv_versions.hypervisor < 2 ?                  \
405
     sizeof(xen_v0_getdomaininfo) :                \
J
Jim Fehlig 已提交
406 407 408
     (hv_versions.dom_interface >= 8 ?             \
      sizeof(xen_v2d8_getdomaininfo) :             \
     (hv_versions.dom_interface == 7 ?             \
J
Jim Fehlig 已提交
409
      sizeof(xen_v2d7_getdomaininfo) :             \
P
Philipp Hahn 已提交
410
     (hv_versions.dom_interface == 6 ?             \
411
      sizeof(xen_v2d6_getdomaininfo) :             \
P
Philipp Hahn 已提交
412
     (hv_versions.dom_interface == 5 ?             \
413
      sizeof(xen_v2d5_getdomaininfo) :             \
J
Jim Fehlig 已提交
414
      sizeof(xen_v2_getdomaininfo))))))
415 416

#define XEN_GETDOMAININFO_CLEAR(dominfo)                           \
P
Philipp Hahn 已提交
417
    (hv_versions.hypervisor < 2 ?                                  \
418
     memset(&(dominfo.v0), 0, sizeof(xen_v0_getdomaininfo)) :      \
J
Jim Fehlig 已提交
419 420 421
     (hv_versions.dom_interface >= 8 ?                             \
      memset(&(dominfo.v2d8), 0, sizeof(xen_v2d8_getdomaininfo)) : \
     (hv_versions.dom_interface == 7 ?                             \
J
Jim Fehlig 已提交
422
      memset(&(dominfo.v2d7), 0, sizeof(xen_v2d7_getdomaininfo)) : \
P
Philipp Hahn 已提交
423
     (hv_versions.dom_interface == 6 ?                             \
424
      memset(&(dominfo.v2d6), 0, sizeof(xen_v2d6_getdomaininfo)) : \
P
Philipp Hahn 已提交
425
     (hv_versions.dom_interface == 5 ?                             \
426
      memset(&(dominfo.v2d5), 0, sizeof(xen_v2d5_getdomaininfo)) : \
J
Jim Fehlig 已提交
427
      memset(&(dominfo.v2), 0, sizeof(xen_v2_getdomaininfo)))))))
428 429

#define XEN_GETDOMAININFO_DOMAIN(dominfo)       \
P
Philipp Hahn 已提交
430
    (hv_versions.hypervisor < 2 ?               \
431
     dominfo.v0.domain :                        \
J
Jim Fehlig 已提交
432 433 434
     (hv_versions.dom_interface >= 8 ?          \
      dominfo.v2d8.domain :                     \
     (hv_versions.dom_interface == 7 ?          \
J
Jim Fehlig 已提交
435
      dominfo.v2d7.domain :                     \
P
Philipp Hahn 已提交
436
     (hv_versions.dom_interface == 6 ?          \
437
      dominfo.v2d6.domain :                     \
P
Philipp Hahn 已提交
438
     (hv_versions.dom_interface == 5 ?          \
439
      dominfo.v2d5.domain :                     \
J
Jim Fehlig 已提交
440
      dominfo.v2.domain)))))
441 442

#define XEN_GETDOMAININFO_CPUTIME(dominfo)      \
P
Philipp Hahn 已提交
443
    (hv_versions.hypervisor < 2 ?               \
444
     dominfo.v0.cpu_time :                      \
J
Jim Fehlig 已提交
445 446 447
     (hv_versions.dom_interface >= 8 ?          \
      dominfo.v2d8.cpu_time :                   \
     (hv_versions.dom_interface == 7 ?          \
J
Jim Fehlig 已提交
448
      dominfo.v2d7.cpu_time :                   \
P
Philipp Hahn 已提交
449
     (hv_versions.dom_interface == 6 ?          \
450
      dominfo.v2d6.cpu_time :                   \
P
Philipp Hahn 已提交
451
     (hv_versions.dom_interface == 5 ?          \
452
      dominfo.v2d5.cpu_time :                   \
J
Jim Fehlig 已提交
453
      dominfo.v2.cpu_time)))))
454

455 456

#define XEN_GETDOMAININFO_CPUCOUNT(dominfo)     \
P
Philipp Hahn 已提交
457
    (hv_versions.hypervisor < 2 ?               \
458
     dominfo.v0.nr_online_vcpus :               \
J
Jim Fehlig 已提交
459 460 461
     (hv_versions.dom_interface >= 8 ?          \
      dominfo.v2d8.nr_online_vcpus :            \
     (hv_versions.dom_interface == 7 ?          \
J
Jim Fehlig 已提交
462
      dominfo.v2d7.nr_online_vcpus :            \
P
Philipp Hahn 已提交
463
     (hv_versions.dom_interface == 6 ?          \
464
      dominfo.v2d6.nr_online_vcpus :            \
P
Philipp Hahn 已提交
465
     (hv_versions.dom_interface == 5 ?          \
466
      dominfo.v2d5.nr_online_vcpus :            \
J
Jim Fehlig 已提交
467
      dominfo.v2.nr_online_vcpus)))))
468

J
Jim Fehlig 已提交
469
#define XEN_GETDOMAININFO_MAXCPUID(dominfo)     \
P
Philipp Hahn 已提交
470
    (hv_versions.hypervisor < 2 ?               \
471
     dominfo.v0.max_vcpu_id :                   \
J
Jim Fehlig 已提交
472 473 474
     (hv_versions.dom_interface >= 8 ?          \
      dominfo.v2d8.max_vcpu_id :                \
     (hv_versions.dom_interface == 7 ?          \
J
Jim Fehlig 已提交
475
      dominfo.v2d7.max_vcpu_id :                \
P
Philipp Hahn 已提交
476
     (hv_versions.dom_interface == 6 ?          \
477
      dominfo.v2d6.max_vcpu_id :                \
P
Philipp Hahn 已提交
478
     (hv_versions.dom_interface == 5 ?          \
479
      dominfo.v2d5.max_vcpu_id :                \
J
Jim Fehlig 已提交
480
      dominfo.v2.max_vcpu_id)))))
481

482
#define XEN_GETDOMAININFO_FLAGS(dominfo)        \
P
Philipp Hahn 已提交
483
    (hv_versions.hypervisor < 2 ?               \
484
     dominfo.v0.flags :                         \
J
Jim Fehlig 已提交
485 486 487
     (hv_versions.dom_interface >= 8 ?          \
      dominfo.v2d8.flags :                      \
     (hv_versions.dom_interface == 7 ?          \
J
Jim Fehlig 已提交
488
      dominfo.v2d7.flags :                      \
P
Philipp Hahn 已提交
489
     (hv_versions.dom_interface == 6 ?          \
490
      dominfo.v2d6.flags :                      \
P
Philipp Hahn 已提交
491
     (hv_versions.dom_interface == 5 ?          \
492
      dominfo.v2d5.flags :                      \
J
Jim Fehlig 已提交
493
      dominfo.v2.flags)))))
494 495

#define XEN_GETDOMAININFO_TOT_PAGES(dominfo)    \
P
Philipp Hahn 已提交
496
    (hv_versions.hypervisor < 2 ?               \
497
     dominfo.v0.tot_pages :                     \
J
Jim Fehlig 已提交
498 499 500
     (hv_versions.dom_interface >= 8 ?          \
      dominfo.v2d8.tot_pages :                  \
     (hv_versions.dom_interface == 7 ?          \
J
Jim Fehlig 已提交
501
      dominfo.v2d7.tot_pages :                  \
P
Philipp Hahn 已提交
502
     (hv_versions.dom_interface == 6 ?          \
503
      dominfo.v2d6.tot_pages :                  \
P
Philipp Hahn 已提交
504
     (hv_versions.dom_interface == 5 ?          \
505
      dominfo.v2d5.tot_pages :                  \
J
Jim Fehlig 已提交
506
      dominfo.v2.tot_pages)))))
507 508

#define XEN_GETDOMAININFO_MAX_PAGES(dominfo)    \
P
Philipp Hahn 已提交
509
    (hv_versions.hypervisor < 2 ?               \
510
     dominfo.v0.max_pages :                     \
J
Jim Fehlig 已提交
511 512 513
     (hv_versions.dom_interface >= 8 ?          \
      dominfo.v2d8.max_pages :                  \
     (hv_versions.dom_interface == 7 ?          \
J
Jim Fehlig 已提交
514
      dominfo.v2d7.max_pages :                  \
P
Philipp Hahn 已提交
515
     (hv_versions.dom_interface == 6 ?          \
516
      dominfo.v2d6.max_pages :                  \
P
Philipp Hahn 已提交
517
     (hv_versions.dom_interface == 5 ?          \
518
      dominfo.v2d5.max_pages :                  \
J
Jim Fehlig 已提交
519
      dominfo.v2.max_pages)))))
520

521
#define XEN_GETDOMAININFO_UUID(dominfo)         \
P
Philipp Hahn 已提交
522
    (hv_versions.hypervisor < 2 ?               \
523
     dominfo.v0.handle :                        \
J
Jim Fehlig 已提交
524 525 526
     (hv_versions.dom_interface >= 8 ?          \
      dominfo.v2d8.handle :                     \
     (hv_versions.dom_interface == 7 ?          \
J
Jim Fehlig 已提交
527
      dominfo.v2d7.handle :                     \
P
Philipp Hahn 已提交
528
     (hv_versions.dom_interface == 6 ?          \
529
      dominfo.v2d6.handle :                     \
P
Philipp Hahn 已提交
530
     (hv_versions.dom_interface == 5 ?          \
531
      dominfo.v2d5.handle :                     \
J
Jim Fehlig 已提交
532
      dominfo.v2.handle)))))
533

534

535 536 537 538
static int
lock_pages(void *addr, size_t len)
{
#ifdef __linux__
539 540 541 542 543 544 545
    if (mlock(addr, len) < 0) {
        virReportSystemError(errno,
                             _("Unable to lock %zu bytes of memory"),
                             len);
        return -1;
    }
    return 0;
546
#elif defined(__sun)
547
    return 0;
548 549 550 551 552 553 554
#endif
}

static int
unlock_pages(void *addr, size_t len)
{
#ifdef __linux__
555 556 557 558 559 560 561
    if (munlock(addr, len) < 0) {
        virReportSystemError(errno,
                             _("Unable to unlock %zu bytes of memory"),
                             len);
        return -1;
    }
    return 0;
562
#elif defined(__sun)
563
    return 0;
564 565 566
#endif
}

567 568

struct xen_v0_getdomaininfolistop {
569 570 571 572 573
    domid_t   first_domain;
    uint32_t  max_domains;
    struct xen_v0_getdomaininfo *buffer;
    uint32_t  num_domains;
};
574 575 576 577 578 579 580 581 582 583 584
typedef struct xen_v0_getdomaininfolistop xen_v0_getdomaininfolistop;


struct xen_v2_getdomaininfolistop {
    domid_t   first_domain;
    uint32_t  max_domains;
    struct xen_v2_getdomaininfo *buffer;
    uint32_t  num_domains;
};
typedef struct xen_v2_getdomaininfolistop xen_v2_getdomaininfolistop;

585 586 587 588
/* As of HV version 2, sysctl version 3 the *buffer pointer is 64-bit aligned */
struct xen_v2s3_getdomaininfolistop {
    domid_t   first_domain;
    uint32_t  max_domains;
589 590
#ifdef __BIG_ENDIAN__
    struct {
591
        int __pad[(sizeof(long long) - sizeof(struct xen_v2d5_getdomaininfo *)) / sizeof(int)];
592 593 594
        struct xen_v2d5_getdomaininfo *v;
    } buffer;
#else
595 596 597 598
    union {
        struct xen_v2d5_getdomaininfo *v;
        uint64_t pad ALIGN_64;
    } buffer;
599
#endif
600 601 602 603
    uint32_t  num_domains;
};
typedef struct xen_v2s3_getdomaininfolistop xen_v2s3_getdomaininfolistop;

604

605 606 607 608 609 610 611

struct xen_v0_domainop {
    domid_t   domain;
};
typedef struct xen_v0_domainop xen_v0_domainop;

/*
612
 * The information for a destroydomain system hypercall
613 614 615 616 617 618
 */
#define XEN_V0_OP_DESTROYDOMAIN	9
#define XEN_V1_OP_DESTROYDOMAIN	9
#define XEN_V2_OP_DESTROYDOMAIN	2

/*
619
 * The information for a pausedomain system hypercall
620 621 622 623 624 625
 */
#define XEN_V0_OP_PAUSEDOMAIN	10
#define XEN_V1_OP_PAUSEDOMAIN	10
#define XEN_V2_OP_PAUSEDOMAIN	3

/*
626
 * The information for an unpausedomain system hypercall
627 628 629 630 631 632
 */
#define XEN_V0_OP_UNPAUSEDOMAIN	11
#define XEN_V1_OP_UNPAUSEDOMAIN	11
#define XEN_V2_OP_UNPAUSEDOMAIN	4

/*
E
Eric Blake 已提交
633
 * The information for a setmaxmem system hypercall
634 635 636
 */
#define XEN_V0_OP_SETMAXMEM	28
#define XEN_V1_OP_SETMAXMEM	28
637
#define XEN_V2_OP_SETMAXMEM	11
638 639 640 641 642 643 644 645 646 647 648 649 650

struct xen_v0_setmaxmem {
    domid_t	domain;
    uint64_t	maxmem;
};
typedef struct xen_v0_setmaxmem xen_v0_setmaxmem;
typedef struct xen_v0_setmaxmem xen_v1_setmaxmem;

struct xen_v2_setmaxmem {
    uint64_t	maxmem;
};
typedef struct xen_v2_setmaxmem xen_v2_setmaxmem;

651 652 653 654 655
struct xen_v2d5_setmaxmem {
    uint64_t	maxmem ALIGN_64;
};
typedef struct xen_v2d5_setmaxmem xen_v2d5_setmaxmem;

656
/*
E
Eric Blake 已提交
657
 * The information for a setmaxvcpu system hypercall
658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675
 */
#define XEN_V0_OP_SETMAXVCPU	41
#define XEN_V1_OP_SETMAXVCPU	41
#define XEN_V2_OP_SETMAXVCPU	15

struct xen_v0_setmaxvcpu {
    domid_t	domain;
    uint32_t	maxvcpu;
};
typedef struct xen_v0_setmaxvcpu xen_v0_setmaxvcpu;
typedef struct xen_v0_setmaxvcpu xen_v1_setmaxvcpu;

struct xen_v2_setmaxvcpu {
    uint32_t	maxvcpu;
};
typedef struct xen_v2_setmaxvcpu xen_v2_setmaxvcpu;

/*
E
Eric Blake 已提交
676
 * The information for a setvcpumap system hypercall
677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701
 * Note that between 1 and 2 the limitation to 64 physical CPU was lifted
 * hence the difference in structures
 */
#define XEN_V0_OP_SETVCPUMAP	20
#define XEN_V1_OP_SETVCPUMAP	20
#define XEN_V2_OP_SETVCPUMAP	9

struct xen_v0_setvcpumap {
    domid_t	domain;
    uint32_t	vcpu;
    cpumap_t    cpumap;
};
typedef struct xen_v0_setvcpumap xen_v0_setvcpumap;
typedef struct xen_v0_setvcpumap xen_v1_setvcpumap;

struct xen_v2_cpumap {
    uint8_t    *bitmap;
    uint32_t    nr_cpus;
};
struct xen_v2_setvcpumap {
    uint32_t	vcpu;
    struct xen_v2_cpumap cpumap;
};
typedef struct xen_v2_setvcpumap xen_v2_setvcpumap;

702 703
/* HV version 2, Dom version 5 requires 64-bit alignment */
struct xen_v2d5_cpumap {
704 705
#ifdef __BIG_ENDIAN__
    struct {
706
        int __pad[(sizeof(long long) - sizeof(uint8_t *)) / sizeof(int)];
707 708 709
        uint8_t *v;
    } bitmap;
#else
710 711 712 713
    union {
        uint8_t    *v;
        uint64_t   pad ALIGN_64;
    } bitmap;
714
#endif
715 716 717 718 719 720 721 722
    uint32_t    nr_cpus;
};
struct xen_v2d5_setvcpumap {
    uint32_t	vcpu;
    struct xen_v2d5_cpumap cpumap;
};
typedef struct xen_v2d5_setvcpumap xen_v2d5_setvcpumap;

723
/*
E
Eric Blake 已提交
724
 * The information for a vcpuinfo system hypercall
725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752
 */
#define XEN_V0_OP_GETVCPUINFO   43
#define XEN_V1_OP_GETVCPUINFO	43
#define XEN_V2_OP_GETVCPUINFO   14

struct xen_v0_vcpuinfo {
    domid_t	domain;		/* owner's domain */
    uint32_t	vcpu;		/* the vcpu number */
    uint8_t	online;		/* seen as on line */
    uint8_t	blocked;	/* blocked on event */
    uint8_t	running;	/* scheduled on CPU */
    uint64_t    cpu_time;	/* nanosecond of CPU used */
    uint32_t	cpu;		/* current mapping */
    cpumap_t	cpumap;		/* deprecated in V2 */
};
typedef struct xen_v0_vcpuinfo xen_v0_vcpuinfo;
typedef struct xen_v0_vcpuinfo xen_v1_vcpuinfo;

struct xen_v2_vcpuinfo {
    uint32_t	vcpu;		/* the vcpu number */
    uint8_t	online;		/* seen as on line */
    uint8_t	blocked;	/* blocked on event */
    uint8_t	running;	/* scheduled on CPU */
    uint64_t    cpu_time;	/* nanosecond of CPU used */
    uint32_t	cpu;		/* current mapping */
};
typedef struct xen_v2_vcpuinfo xen_v2_vcpuinfo;

753 754 755 756 757 758 759 760 761 762
struct xen_v2d5_vcpuinfo {
    uint32_t	vcpu;		/* the vcpu number */
    uint8_t	online;		/* seen as on line */
    uint8_t	blocked;	/* blocked on event */
    uint8_t	running;	/* scheduled on CPU */
    uint64_t    cpu_time ALIGN_64; /* nanosecond of CPU used */
    uint32_t	cpu;		/* current mapping */
};
typedef struct xen_v2d5_vcpuinfo xen_v2d5_vcpuinfo;

763 764 765 766 767
/*
 * from V2 the pinning of a vcpu is read with a separate call
 */
#define XEN_V2_OP_GETVCPUMAP	25
typedef struct xen_v2_setvcpumap xen_v2_getvcpumap;
768
typedef struct xen_v2d5_setvcpumap xen_v2d5_getvcpumap;
769

770 771 772 773 774
/*
 * from V2 we get the scheduler information
 */
#define XEN_V2_OP_GETSCHEDULERID	4

775 776 777
/*
 * from V2 we get the available heap information
 */
E
Eric Blake 已提交
778
#define XEN_V2_OP_GETAVAILHEAP		9
779

780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811
/*
 * from V2 we get the scheduler parameter
 */
#define XEN_V2_OP_SCHEDULER		16
/* Scheduler types. */
#define XEN_SCHEDULER_SEDF       4
#define XEN_SCHEDULER_CREDIT     5
/* get/set scheduler parameters */
#define XEN_DOMCTL_SCHEDOP_putinfo 0
#define XEN_DOMCTL_SCHEDOP_getinfo 1

struct xen_v2_setschedinfo {
    uint32_t sched_id;
    uint32_t cmd;
    union {
        struct xen_domctl_sched_sedf {
            uint64_t period ALIGN_64;
            uint64_t slice  ALIGN_64;
            uint64_t latency ALIGN_64;
            uint32_t extratime;
            uint32_t weight;
        } sedf;
        struct xen_domctl_sched_credit {
            uint16_t weight;
            uint16_t cap;
        } credit;
    } u;
};
typedef struct xen_v2_setschedinfo xen_v2_setschedinfo;
typedef struct xen_v2_setschedinfo xen_v2_getschedinfo;


812 813 814 815 816 817 818 819 820
/*
 * The hypercall operation structures also have changed on
 * changeset 86d26e6ec89b
 */
/* the old structure */
struct xen_op_v0 {
    uint32_t cmd;
    uint32_t interface_version;
    union {
821 822 823 824 825 826 827
        xen_v0_getdomaininfolistop getdomaininfolist;
        xen_v0_domainop          domain;
        xen_v0_setmaxmem         setmaxmem;
        xen_v0_setmaxvcpu        setmaxvcpu;
        xen_v0_setvcpumap        setvcpumap;
        xen_v0_vcpuinfo          getvcpuinfo;
        uint8_t padding[128];
828 829 830 831 832 833 834 835 836 837
    } u;
};
typedef struct xen_op_v0 xen_op_v0;
typedef struct xen_op_v0 xen_op_v1;

/* the new structure for systems operations */
struct xen_op_v2_sys {
    uint32_t cmd;
    uint32_t interface_version;
    union {
838 839
        xen_v2_getdomaininfolistop   getdomaininfolist;
        xen_v2s3_getdomaininfolistop getdomaininfolists3;
840
        xen_v2_getschedulerid        getschedulerid;
841
        xen_v2s4_availheap           availheap;
842
        xen_v2s5_availheap           availheap5;
843
        uint8_t padding[128];
844 845 846 847 848 849 850 851 852 853
    } u;
};
typedef struct xen_op_v2_sys xen_op_v2_sys;

/* the new structure for domains operation */
struct xen_op_v2_dom {
    uint32_t cmd;
    uint32_t interface_version;
    domid_t  domain;
    union {
854
        xen_v2_setmaxmem         setmaxmem;
855
        xen_v2d5_setmaxmem       setmaxmemd5;
856 857
        xen_v2_setmaxvcpu        setmaxvcpu;
        xen_v2_setvcpumap        setvcpumap;
858
        xen_v2d5_setvcpumap      setvcpumapd5;
859
        xen_v2_vcpuinfo          getvcpuinfo;
860
        xen_v2d5_vcpuinfo        getvcpuinfod5;
861
        xen_v2_getvcpumap        getvcpumap;
862
        xen_v2d5_getvcpumap      getvcpumapd5;
863 864
        xen_v2_setschedinfo      setschedinfo;
        xen_v2_getschedinfo      getschedinfo;
865
        uint8_t padding[128];
866 867 868
    } u;
};
typedef struct xen_op_v2_dom xen_op_v2_dom;
869 870


871
#ifdef __linux__
872 873
# define XEN_HYPERVISOR_SOCKET	"/proc/xen/privcmd"
# define HYPERVISOR_CAPABILITIES	"/sys/hypervisor/properties/capabilities"
874
#elif defined(__sun)
875
# define XEN_HYPERVISOR_SOCKET	"/dev/xen/privcmd"
876
#else
877
# error "unsupported platform"
878
#endif
879

880
static unsigned long long xenHypervisorGetMaxMemory(virDomainPtr domain);
881

882
struct xenUnifiedDriver xenHypervisorDriver = {
E
Eric Blake 已提交
883 884 885 886 887 888 889 890 891 892 893 894 895 896
    .xenClose = xenHypervisorClose,
    .xenVersion = xenHypervisorGetVersion,
    .xenDomainSuspend = xenHypervisorPauseDomain,
    .xenDomainResume = xenHypervisorResumeDomain,
    .xenDomainDestroyFlags = xenHypervisorDestroyDomainFlags,
    .xenDomainGetOSType = xenHypervisorDomainGetOSType,
    .xenDomainGetMaxMemory = xenHypervisorGetMaxMemory,
    .xenDomainSetMaxMemory = xenHypervisorSetMaxMemory,
    .xenDomainGetInfo = xenHypervisorGetDomainInfo,
    .xenDomainPinVcpu = xenHypervisorPinVcpu,
    .xenDomainGetVcpus = xenHypervisorGetVcpus,
    .xenDomainGetSchedulerType = xenHypervisorGetSchedulerType,
    .xenDomainGetSchedulerParameters = xenHypervisorGetSchedulerParameters,
    .xenDomainSetSchedulerParameters = xenHypervisorSetSchedulerParameters,
897 898
};

899 900 901
/**
 * xenHypervisorDoV0Op:
 * @handle: the handle to the Xen hypervisor
R
Richard W.M. Jones 已提交
902
 * @op: pointer to the hypervisor operation structure
903
 *
E
Eric Blake 已提交
904 905
 * Do a hypervisor operation though the old interface,
 * this leads to a hypervisor call through ioctl.
906 907 908 909 910 911 912 913 914 915
 *
 * Returns 0 in case of success and -1 in case of error.
 */
static int
xenHypervisorDoV0Op(int handle, xen_op_v0 * op)
{
    int ret;
    v0_hypercall_t hc;

    memset(&hc, 0, sizeof(hc));
P
Philipp Hahn 已提交
916
    op->interface_version = hv_versions.hv << 8;
917 918 919
    hc.op = __HYPERVISOR_dom0_op;
    hc.arg[0] = (unsigned long) op;

920
    if (lock_pages(op, sizeof(dom0_op_t)) < 0)
921
        return -1;
922 923 924

    ret = ioctl(handle, xen_ioctl_hypercall_cmd, (unsigned long) &hc);
    if (ret < 0) {
925 926 927
        virReportSystemError(errno,
                             _("Unable to issue hypervisor ioctl %d"),
                             xen_ioctl_hypercall_cmd);
928 929
    }

930
    if (unlock_pages(op, sizeof(dom0_op_t)) < 0)
931 932 933
        ret = -1;

    if (ret < 0)
934
        return -1;
935

936
    return 0;
937 938 939 940
}
/**
 * xenHypervisorDoV1Op:
 * @handle: the handle to the Xen hypervisor
R
Richard W.M. Jones 已提交
941
 * @op: pointer to the hypervisor operation structure
942
 *
E
Eric Blake 已提交
943
 * Do a hypervisor v1 operation, this leads to a hypervisor call through
944 945 946 947 948 949 950 951 952 953 954 955 956 957 958
 * ioctl.
 *
 * Returns 0 in case of success and -1 in case of error.
 */
static int
xenHypervisorDoV1Op(int handle, xen_op_v1* op)
{
    int ret;
    hypercall_t hc;

    memset(&hc, 0, sizeof(hc));
    op->interface_version = DOM0_INTERFACE_VERSION;
    hc.op = __HYPERVISOR_dom0_op;
    hc.arg[0] = (unsigned long) op;

959
    if (lock_pages(op, sizeof(dom0_op_t)) < 0)
960
        return -1;
961 962 963

    ret = ioctl(handle, xen_ioctl_hypercall_cmd, (unsigned long) &hc);
    if (ret < 0) {
964 965 966
        virReportSystemError(errno,
                             _("Unable to issue hypervisor ioctl %d"),
                             xen_ioctl_hypercall_cmd);
967 968
    }

969
    if (unlock_pages(op, sizeof(dom0_op_t)) < 0)
970 971 972
        ret = -1;

    if (ret < 0)
973
        return -1;
974

975
    return 0;
976 977 978 979 980 981 982
}

/**
 * xenHypervisorDoV2Sys:
 * @handle: the handle to the Xen hypervisor
 * @op: pointer to the hypervisor operation structure
 *
E
Eric Blake 已提交
983
 * Do a hypervisor v2 system operation, this leads to a hypervisor
984 985 986 987 988 989 990 991 992 993 994
 * call through ioctl.
 *
 * Returns 0 in case of success and -1 in case of error.
 */
static int
xenHypervisorDoV2Sys(int handle, xen_op_v2_sys* op)
{
    int ret;
    hypercall_t hc;

    memset(&hc, 0, sizeof(hc));
P
Philipp Hahn 已提交
995
    op->interface_version = hv_versions.sys_interface;
996 997 998
    hc.op = __HYPERVISOR_sysctl;
    hc.arg[0] = (unsigned long) op;

999
    if (lock_pages(op, sizeof(dom0_op_t)) < 0)
1000
        return -1;
1001 1002 1003

    ret = ioctl(handle, xen_ioctl_hypercall_cmd, (unsigned long) &hc);
    if (ret < 0) {
1004 1005 1006
        virReportSystemError(errno,
                             _("Unable to issue hypervisor ioctl %d"),
                             xen_ioctl_hypercall_cmd);
1007 1008
    }

1009
    if (unlock_pages(op, sizeof(dom0_op_t)) < 0)
1010 1011 1012
        ret = -1;

    if (ret < 0)
1013
        return -1;
1014

1015
    return 0;
1016 1017 1018 1019 1020 1021 1022
}

/**
 * xenHypervisorDoV2Dom:
 * @handle: the handle to the Xen hypervisor
 * @op: pointer to the hypervisor domain operation structure
 *
E
Eric Blake 已提交
1023
 * Do a hypervisor v2 domain operation, this leads to a hypervisor
1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034
 * call through ioctl.
 *
 * Returns 0 in case of success and -1 in case of error.
 */
static int
xenHypervisorDoV2Dom(int handle, xen_op_v2_dom* op)
{
    int ret;
    hypercall_t hc;

    memset(&hc, 0, sizeof(hc));
P
Philipp Hahn 已提交
1035
    op->interface_version = hv_versions.dom_interface;
1036 1037 1038
    hc.op = __HYPERVISOR_domctl;
    hc.arg[0] = (unsigned long) op;

1039
    if (lock_pages(op, sizeof(dom0_op_t)) < 0)
1040
        return -1;
1041 1042 1043

    ret = ioctl(handle, xen_ioctl_hypercall_cmd, (unsigned long) &hc);
    if (ret < 0) {
1044 1045 1046
        virReportSystemError(errno,
                             _("Unable to issue hypervisor ioctl %d"),
                             xen_ioctl_hypercall_cmd);
1047 1048
    }

1049
    if (unlock_pages(op, sizeof(dom0_op_t)) < 0)
1050 1051 1052
        ret = -1;

    if (ret < 0)
1053
        return -1;
1054

1055
    return 0;
1056 1057 1058 1059 1060 1061 1062 1063 1064
}

/**
 * virXen_getdomaininfolist:
 * @handle: the hypervisor handle
 * @first_domain: first domain in the range
 * @maxids: maximum number of domains to list
 * @dominfos: output structures
 *
1065
 * Do a low level hypercall to list existing domains information
1066 1067 1068 1069
 *
 * Returns the number of domains or -1 in case of failure
 */
static int
1070 1071 1072
virXen_getdomaininfolist(int handle,
                         int first_domain,
                         int maxids,
1073
                         xen_getdomaininfolist *dominfos)
1074 1075 1076
{
    int ret = -1;

1077
    if (lock_pages(XEN_GETDOMAININFOLIST_DATA(dominfos),
1078
                   XEN_GETDOMAININFO_SIZE * maxids) < 0)
1079
        return -1;
1080

P
Philipp Hahn 已提交
1081
    if (hv_versions.hypervisor > 1) {
1082 1083 1084
        xen_op_v2_sys op;

        memset(&op, 0, sizeof(op));
1085
        op.cmd = XEN_V2_OP_GETDOMAININFOLIST;
1086

P
Philipp Hahn 已提交
1087
        if (hv_versions.sys_interface < 3) {
1088 1089 1090 1091 1092 1093 1094 1095 1096 1097
            op.u.getdomaininfolist.first_domain = (domid_t) first_domain;
            op.u.getdomaininfolist.max_domains = maxids;
            op.u.getdomaininfolist.buffer = dominfos->v2;
            op.u.getdomaininfolist.num_domains = maxids;
        } else {
            op.u.getdomaininfolists3.first_domain = (domid_t) first_domain;
            op.u.getdomaininfolists3.max_domains = maxids;
            op.u.getdomaininfolists3.buffer.v = dominfos->v2d5;
            op.u.getdomaininfolists3.num_domains = maxids;
        }
1098
        ret = xenHypervisorDoV2Sys(handle, &op);
1099 1100

        if (ret == 0) {
P
Philipp Hahn 已提交
1101
            if (hv_versions.sys_interface < 3)
1102 1103 1104 1105
                ret = op.u.getdomaininfolist.num_domains;
            else
                ret = op.u.getdomaininfolists3.num_domains;
        }
P
Philipp Hahn 已提交
1106
    } else if (hv_versions.hypervisor == 1) {
1107 1108 1109
        xen_op_v1 op;

        memset(&op, 0, sizeof(op));
1110 1111 1112 1113 1114 1115 1116 1117
        op.cmd = XEN_V1_OP_GETDOMAININFOLIST;
        op.u.getdomaininfolist.first_domain = (domid_t) first_domain;
        op.u.getdomaininfolist.max_domains = maxids;
        op.u.getdomaininfolist.buffer = dominfos->v0;
        op.u.getdomaininfolist.num_domains = maxids;
        ret = xenHypervisorDoV1Op(handle, &op);
        if (ret == 0)
            ret = op.u.getdomaininfolist.num_domains;
P
Philipp Hahn 已提交
1118
    } else if (hv_versions.hypervisor == 0) {
1119 1120 1121
        xen_op_v0 op;

        memset(&op, 0, sizeof(op));
1122 1123 1124 1125 1126 1127 1128 1129
        op.cmd = XEN_V0_OP_GETDOMAININFOLIST;
        op.u.getdomaininfolist.first_domain = (domid_t) first_domain;
        op.u.getdomaininfolist.max_domains = maxids;
        op.u.getdomaininfolist.buffer = dominfos->v0;
        op.u.getdomaininfolist.num_domains = maxids;
        ret = xenHypervisorDoV0Op(handle, &op);
        if (ret == 0)
            ret = op.u.getdomaininfolist.num_domains;
1130
    }
1131
    if (unlock_pages(XEN_GETDOMAININFOLIST_DATA(dominfos),
1132
                     XEN_GETDOMAININFO_SIZE * maxids) < 0)
1133
        ret = -1;
1134

1135
    return ret;
1136 1137
}

1138
static int
1139 1140
virXen_getdomaininfo(int handle, int first_domain, xen_getdomaininfo *dominfo)
{
1141 1142
    xen_getdomaininfolist dominfos;

P
Philipp Hahn 已提交
1143
    if (hv_versions.hypervisor < 2) {
1144 1145 1146 1147 1148 1149 1150 1151 1152
        dominfos.v0 = &(dominfo->v0);
    } else {
        dominfos.v2 = &(dominfo->v2);
    }

    return virXen_getdomaininfolist(handle, first_domain, 1, &dominfos);
}


1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165
/**
 * xenHypervisorGetSchedulerType:
 * @domain: pointer to the Xen Hypervisor block
 * @nparams:give a number of scheduler parameters.
 *
 * Do a low level hypercall to get scheduler type
 *
 * Returns scheduler name or NULL in case of failure
 */
char *
xenHypervisorGetSchedulerType(virDomainPtr domain, int *nparams)
{
    char *schedulertype = NULL;
1166
    xenUnifiedPrivatePtr priv = domain->conn->privateData;
1167

1168
    if (priv->handle < 0) {
1169 1170
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("priv->handle invalid"));
1171 1172 1173
        return NULL;
    }
    if (domain->id < 0) {
1174 1175
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("domain is not running"));
1176 1177 1178 1179
        return NULL;
    }

    /*
P
Philipp Hahn 已提交
1180
     * Support only hv_versions.dom_interface >=5
1181
     * (Xen3.1.0 or later)
1182
     * TODO: check on Xen 3.0.3
1183
     */
P
Philipp Hahn 已提交
1184
    if (hv_versions.dom_interface < 5) {
1185 1186
        virReportError(VIR_ERR_NO_XEN, "%s",
                       _("unsupported in dom interface < 5"));
1187 1188 1189
        return NULL;
    }

P
Philipp Hahn 已提交
1190
    if (hv_versions.hypervisor > 1) {
1191 1192 1193 1194 1195 1196 1197
        xen_op_v2_sys op;
        int ret;

        memset(&op, 0, sizeof(op));
        op.cmd = XEN_V2_OP_GETSCHEDULERID;
        ret = xenHypervisorDoV2Sys(priv->handle, &op);
        if (ret < 0)
1198
            return NULL;
1199 1200

        switch (op.u.getschedulerid.sched_id){
1201 1202
            case XEN_SCHEDULER_SEDF:
                schedulertype = strdup("sedf");
1203
                if (schedulertype == NULL)
1204
                    virReportOOMError();
1205
                if (nparams)
1206
                    *nparams = XEN_SCHED_SEDF_NPARAM;
1207 1208 1209
                break;
            case XEN_SCHEDULER_CREDIT:
                schedulertype = strdup("credit");
1210
                if (schedulertype == NULL)
1211
                    virReportOOMError();
1212
                if (nparams)
1213
                    *nparams = XEN_SCHED_CRED_NPARAM;
1214 1215 1216
                break;
            default:
                break;
1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227
        }
    }

    return schedulertype;
}

/**
 * xenHypervisorGetSchedulerParameters:
 * @domain: pointer to the Xen Hypervisor block
 * @params: pointer to scheduler parameters.
 *     This memory area should be allocated before calling.
1228 1229
 * @nparams: this parameter must be at least as large as
 *     the given number of scheduler parameters.
1230 1231 1232 1233 1234 1235 1236 1237
 *     from xenHypervisorGetSchedulerType().
 *
 * Do a low level hypercall to get scheduler parameters
 *
 * Returns 0 or -1 in case of failure
 */
int
xenHypervisorGetSchedulerParameters(virDomainPtr domain,
1238 1239
                                    virTypedParameterPtr params,
                                    int *nparams)
1240
{
1241
    xenUnifiedPrivatePtr priv = domain->conn->privateData;
1242

1243
    if (priv->handle < 0) {
1244 1245
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("priv->handle invalid"));
1246 1247 1248
        return -1;
    }
    if (domain->id < 0) {
1249 1250
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("domain is not running"));
1251 1252 1253 1254
        return -1;
    }

    /*
P
Philipp Hahn 已提交
1255
     * Support only hv_versions.dom_interface >=5
1256 1257 1258
     * (Xen3.1.0 or later)
     * TODO: check on Xen 3.0.3
     */
P
Philipp Hahn 已提交
1259
    if (hv_versions.dom_interface < 5) {
1260 1261
        virReportError(VIR_ERR_NO_XEN, "%s",
                       _("unsupported in dom interface < 5"));
1262 1263 1264
        return -1;
    }

P
Philipp Hahn 已提交
1265
    if (hv_versions.hypervisor > 1) {
1266 1267 1268 1269 1270 1271 1272 1273
        xen_op_v2_sys op_sys;
        xen_op_v2_dom op_dom;
        int ret;

        memset(&op_sys, 0, sizeof(op_sys));
        op_sys.cmd = XEN_V2_OP_GETSCHEDULERID;
        ret = xenHypervisorDoV2Sys(priv->handle, &op_sys);
        if (ret < 0)
1274
            return -1;
1275 1276

        switch (op_sys.u.getschedulerid.sched_id){
1277
            case XEN_SCHEDULER_SEDF:
1278
                if (*nparams < XEN_SCHED_SEDF_NPARAM) {
1279 1280
                    virReportError(VIR_ERR_INVALID_ARG,
                                   "%s", _("Invalid parameter count"));
1281 1282 1283
                    return -1;
                }

1284 1285
                /* TODO: Implement for Xen/SEDF */
                TODO
1286
                return -1;
1287 1288 1289 1290 1291 1292 1293 1294
            case XEN_SCHEDULER_CREDIT:
                memset(&op_dom, 0, sizeof(op_dom));
                op_dom.cmd = XEN_V2_OP_SCHEDULER;
                op_dom.domain = (domid_t) domain->id;
                op_dom.u.getschedinfo.sched_id = XEN_SCHEDULER_CREDIT;
                op_dom.u.getschedinfo.cmd = XEN_DOMCTL_SCHEDOP_getinfo;
                ret = xenHypervisorDoV2Dom(priv->handle, &op_dom);
                if (ret < 0)
1295
                    return -1;
1296

1297 1298 1299 1300
                if (virTypedParameterAssign(&params[0],
                                            VIR_DOMAIN_SCHEDULER_WEIGHT,
                                            VIR_TYPED_PARAM_UINT,
                                            op_dom.u.getschedinfo.u.credit.weight) < 0)
C
Chris Lalancette 已提交
1301
                    return -1;
1302 1303 1304 1305 1306 1307

                if (*nparams > 1 &&
                    virTypedParameterAssign(&params[1],
                                            VIR_DOMAIN_SCHEDULER_CAP,
                                            VIR_TYPED_PARAM_UINT,
                                            op_dom.u.getschedinfo.u.credit.cap) < 0)
1308
                        return -1;
1309

1310 1311
                if (*nparams > XEN_SCHED_CRED_NPARAM)
                    *nparams = XEN_SCHED_CRED_NPARAM;
1312 1313
                break;
            default:
1314 1315 1316
                virReportError(VIR_ERR_INVALID_ARG,
                               _("Unknown scheduler %d"),
                               op_sys.u.getschedulerid.sched_id);
1317
                return -1;
1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334
        }
    }

    return 0;
}

/**
 * xenHypervisorSetSchedulerParameters:
 * @domain: pointer to the Xen Hypervisor block
 * @nparams:give a number of scheduler setting parameters .
 *
 * Do a low level hypercall to set scheduler parameters
 *
 * Returns 0 or -1 in case of failure
 */
int
xenHypervisorSetSchedulerParameters(virDomainPtr domain,
1335 1336
                                    virTypedParameterPtr params,
                                    int nparams)
1337 1338
{
    int i;
1339
    unsigned int val;
1340
    xenUnifiedPrivatePtr priv = domain->conn->privateData;
1341
    char buf[256];
1342

1343 1344 1345
    if (nparams == 0) {
        /* nothing to do, exit early */
        return 0;
1346 1347
    }

1348 1349 1350 1351 1352 1353 1354 1355
    if (virTypedParameterArrayValidate(params, nparams,
                                       VIR_DOMAIN_SCHEDULER_WEIGHT,
                                       VIR_TYPED_PARAM_UINT,
                                       VIR_DOMAIN_SCHEDULER_CAP,
                                       VIR_TYPED_PARAM_UINT,
                                       NULL) < 0)
        return -1;

1356
    if (priv->handle < 0) {
1357 1358
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("priv->handle invalid"));
1359 1360 1361
        return -1;
    }
    if (domain->id < 0) {
1362 1363
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("domain is not running"));
1364 1365 1366 1367
        return -1;
    }

    /*
P
Philipp Hahn 已提交
1368
     * Support only hv_versions.dom_interface >=5
1369 1370 1371
     * (Xen3.1.0 or later)
     * TODO: check on Xen 3.0.3
     */
P
Philipp Hahn 已提交
1372
    if (hv_versions.dom_interface < 5) {
1373 1374
        virReportError(VIR_ERR_NO_XEN, "%s",
                       _("unsupported in dom interface < 5"));
1375 1376 1377
        return -1;
    }

P
Philipp Hahn 已提交
1378
    if (hv_versions.hypervisor > 1) {
1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391
        xen_op_v2_sys op_sys;
        xen_op_v2_dom op_dom;
        int ret;

        memset(&op_sys, 0, sizeof(op_sys));
        op_sys.cmd = XEN_V2_OP_GETSCHEDULERID;
        ret = xenHypervisorDoV2Sys(priv->handle, &op_sys);
        if (ret == -1) return -1;

        switch (op_sys.u.getschedulerid.sched_id){
        case XEN_SCHEDULER_SEDF:
            /* TODO: Implement for Xen/SEDF */
            TODO
1392
            return -1;
1393 1394 1395 1396 1397 1398 1399 1400
        case XEN_SCHEDULER_CREDIT: {
            memset(&op_dom, 0, sizeof(op_dom));
            op_dom.cmd = XEN_V2_OP_SCHEDULER;
            op_dom.domain = (domid_t) domain->id;
            op_dom.u.getschedinfo.sched_id = XEN_SCHEDULER_CREDIT;
            op_dom.u.getschedinfo.cmd = XEN_DOMCTL_SCHEDOP_putinfo;

            /*
1401 1402
             * credit scheduler parameters
             * following values do not change the parameters
1403 1404 1405 1406 1407
             */
            op_dom.u.getschedinfo.u.credit.weight = 0;
            op_dom.u.getschedinfo.u.credit.cap    = (uint16_t)~0U;

            for (i = 0; i < nparams; i++) {
1408
                memset(&buf, 0, sizeof(buf));
1409
                if (STREQ(params[i].field, VIR_DOMAIN_SCHEDULER_WEIGHT)) {
1410 1411
                    val = params[i].value.ui;
                    if ((val < 1) || (val > USHRT_MAX)) {
1412 1413 1414
                        virReportError(VIR_ERR_INVALID_ARG,
                                       _("Credit scheduler weight parameter (%d) "
                                         "is out of range (1-65535)"), val);
1415
                        return -1;
1416
                    }
1417
                    op_dom.u.getschedinfo.u.credit.weight = val;
1418
                } else if (STREQ(params[i].field, VIR_DOMAIN_SCHEDULER_CAP)) {
1419
                    val = params[i].value.ui;
1420
                    if (val >= USHRT_MAX) {
1421 1422 1423
                        virReportError(VIR_ERR_INVALID_ARG,
                                       _("Credit scheduler cap parameter (%d) is "
                                         "out of range (0-65534)"), val);
1424
                        return -1;
1425
                    }
1426
                    op_dom.u.getschedinfo.u.credit.cap = val;
1427
                }
1428 1429 1430 1431
            }

            ret = xenHypervisorDoV2Dom(priv->handle, &op_dom);
            if (ret < 0)
1432
                return -1;
1433
            break;
1434
        }
1435
        default:
1436 1437 1438
            virReportError(VIR_ERR_INVALID_ARG,
                           _("Unknown scheduler %d"),
                           op_sys.u.getschedulerid.sched_id);
1439 1440 1441
            return -1;
        }
    }
1442

1443 1444 1445
    return 0;
}

1446 1447

int
1448 1449 1450
xenHypervisorDomainBlockStats(virDomainPtr dom,
                              const char *path,
                              struct _virDomainBlockStats *stats)
1451
{
1452
#ifdef __linux__
1453
    xenUnifiedPrivatePtr priv = dom->conn->privateData;
D
Daniel P. Berrange 已提交
1454
    int ret;
1455

D
Daniel P. Berrange 已提交
1456 1457
    xenUnifiedLock(priv);
    /* Need to lock because it hits the xenstore handle :-( */
1458
    ret = xenLinuxDomainBlockStats(priv, dom, path, stats);
D
Daniel P. Berrange 已提交
1459 1460
    xenUnifiedUnlock(priv);
    return ret;
1461
#else
1462 1463
    virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                   _("block statistics not supported on this platform"));
1464
    return -1;
1465
#endif
1466 1467 1468 1469 1470 1471 1472 1473 1474 1475
}

/* Paths have the form vif<domid>.<n> (this interface checks that
 * <domid> is the real domain ID and returns an error if not).
 *
 * In future we may allow you to query bridge stats (virbrX or
 * xenbrX), but that will probably be through a separate
 * virNetwork interface, as yet not decided.
 */
int
1476 1477 1478
xenHypervisorDomainInterfaceStats(virDomainPtr dom,
                                  const char *path,
                                  struct _virDomainInterfaceStats *stats)
1479
{
1480
#ifdef __linux__
1481 1482
    int rqdomid, device;

1483 1484 1485
    /* Verify that the vif requested is one belonging to the current
     * domain.
     */
1486
    if (sscanf(path, "vif%d.%d", &rqdomid, &device) != 2) {
1487 1488
        virReportError(VIR_ERR_INVALID_ARG, "%s",
                       _("invalid path, should be vif<domid>.<n>."));
1489 1490 1491
        return -1;
    }
    if (rqdomid != dom->id) {
1492 1493
        virReportError(VIR_ERR_INVALID_ARG, "%s",
                       _("invalid path, vif<domid> should match this domain ID"));
1494 1495 1496
        return -1;
    }

1497
    return linuxDomainInterfaceStats(path, stats);
1498
#else
1499 1500
    virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                   _("/proc/net/dev: Interface not found"));
1501
    return -1;
1502
#endif
1503 1504
}

1505 1506 1507 1508 1509 1510 1511 1512 1513 1514
/**
 * virXen_pausedomain:
 * @handle: the hypervisor handle
 * @id: the domain id
 *
 * Do a low level hypercall to pause the domain
 *
 * Returns 0 or -1 in case of failure
 */
static int
1515
virXen_pausedomain(int handle, int id)
1516 1517 1518
{
    int ret = -1;

P
Philipp Hahn 已提交
1519
    if (hv_versions.hypervisor > 1) {
1520 1521 1522
        xen_op_v2_dom op;

        memset(&op, 0, sizeof(op));
1523 1524 1525
        op.cmd = XEN_V2_OP_PAUSEDOMAIN;
        op.domain = (domid_t) id;
        ret = xenHypervisorDoV2Dom(handle, &op);
P
Philipp Hahn 已提交
1526
    } else if (hv_versions.hypervisor == 1) {
1527 1528 1529
        xen_op_v1 op;

        memset(&op, 0, sizeof(op));
1530 1531 1532
        op.cmd = XEN_V1_OP_PAUSEDOMAIN;
        op.u.domain.domain = (domid_t) id;
        ret = xenHypervisorDoV1Op(handle, &op);
P
Philipp Hahn 已提交
1533
    } else if (hv_versions.hypervisor == 0) {
1534 1535 1536
        xen_op_v0 op;

        memset(&op, 0, sizeof(op));
1537 1538 1539
        op.cmd = XEN_V0_OP_PAUSEDOMAIN;
        op.u.domain.domain = (domid_t) id;
        ret = xenHypervisorDoV0Op(handle, &op);
1540
    }
1541
    return ret;
1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553
}

/**
 * virXen_unpausedomain:
 * @handle: the hypervisor handle
 * @id: the domain id
 *
 * Do a low level hypercall to unpause the domain
 *
 * Returns 0 or -1 in case of failure
 */
static int
1554
virXen_unpausedomain(int handle, int id)
1555 1556 1557
{
    int ret = -1;

P
Philipp Hahn 已提交
1558
    if (hv_versions.hypervisor > 1) {
1559 1560 1561
        xen_op_v2_dom op;

        memset(&op, 0, sizeof(op));
1562 1563 1564
        op.cmd = XEN_V2_OP_UNPAUSEDOMAIN;
        op.domain = (domid_t) id;
        ret = xenHypervisorDoV2Dom(handle, &op);
P
Philipp Hahn 已提交
1565
    } else if (hv_versions.hypervisor == 1) {
1566 1567 1568
        xen_op_v1 op;

        memset(&op, 0, sizeof(op));
1569 1570 1571
        op.cmd = XEN_V1_OP_UNPAUSEDOMAIN;
        op.u.domain.domain = (domid_t) id;
        ret = xenHypervisorDoV1Op(handle, &op);
P
Philipp Hahn 已提交
1572
    } else if (hv_versions.hypervisor == 0) {
1573 1574 1575
        xen_op_v0 op;

        memset(&op, 0, sizeof(op));
1576 1577 1578
        op.cmd = XEN_V0_OP_UNPAUSEDOMAIN;
        op.u.domain.domain = (domid_t) id;
        ret = xenHypervisorDoV0Op(handle, &op);
1579
    }
1580
    return ret;
1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592
}

/**
 * virXen_destroydomain:
 * @handle: the hypervisor handle
 * @id: the domain id
 *
 * Do a low level hypercall to destroy the domain
 *
 * Returns 0 or -1 in case of failure
 */
static int
1593
virXen_destroydomain(int handle, int id)
1594 1595 1596
{
    int ret = -1;

P
Philipp Hahn 已提交
1597
    if (hv_versions.hypervisor > 1) {
1598 1599 1600
        xen_op_v2_dom op;

        memset(&op, 0, sizeof(op));
1601 1602 1603
        op.cmd = XEN_V2_OP_DESTROYDOMAIN;
        op.domain = (domid_t) id;
        ret = xenHypervisorDoV2Dom(handle, &op);
P
Philipp Hahn 已提交
1604
    } else if (hv_versions.hypervisor == 1) {
1605 1606 1607
        xen_op_v1 op;

        memset(&op, 0, sizeof(op));
1608 1609 1610
        op.cmd = XEN_V1_OP_DESTROYDOMAIN;
        op.u.domain.domain = (domid_t) id;
        ret = xenHypervisorDoV1Op(handle, &op);
P
Philipp Hahn 已提交
1611
    } else if (hv_versions.hypervisor == 0) {
1612 1613 1614
        xen_op_v0 op;

        memset(&op, 0, sizeof(op));
1615 1616 1617
        op.cmd = XEN_V0_OP_DESTROYDOMAIN;
        op.u.domain.domain = (domid_t) id;
        ret = xenHypervisorDoV0Op(handle, &op);
1618
    }
1619
    return ret;
1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632
}

/**
 * virXen_setmaxmem:
 * @handle: the hypervisor handle
 * @id: the domain id
 * @memory: the amount of memory in kilobytes
 *
 * Do a low level hypercall to change the max memory amount
 *
 * Returns 0 or -1 in case of failure
 */
static int
1633
virXen_setmaxmem(int handle, int id, unsigned long memory)
1634 1635 1636
{
    int ret = -1;

P
Philipp Hahn 已提交
1637
    if (hv_versions.hypervisor > 1) {
1638 1639 1640
        xen_op_v2_dom op;

        memset(&op, 0, sizeof(op));
1641 1642
        op.cmd = XEN_V2_OP_SETMAXMEM;
        op.domain = (domid_t) id;
P
Philipp Hahn 已提交
1643
        if (hv_versions.dom_interface < 5)
1644 1645 1646
            op.u.setmaxmem.maxmem = memory;
        else
            op.u.setmaxmemd5.maxmem = memory;
1647
        ret = xenHypervisorDoV2Dom(handle, &op);
P
Philipp Hahn 已提交
1648
    } else if (hv_versions.hypervisor == 1) {
1649 1650 1651
        xen_op_v1 op;

        memset(&op, 0, sizeof(op));
1652 1653 1654 1655
        op.cmd = XEN_V1_OP_SETMAXMEM;
        op.u.setmaxmem.domain = (domid_t) id;
        op.u.setmaxmem.maxmem = memory;
        ret = xenHypervisorDoV1Op(handle, &op);
P
Philipp Hahn 已提交
1656
    } else if (hv_versions.hypervisor == 0) {
1657
        xen_op_v0 op;
1658 1659

        memset(&op, 0, sizeof(op));
1660 1661 1662 1663
        op.cmd = XEN_V0_OP_SETMAXMEM;
        op.u.setmaxmem.domain = (domid_t) id;
        op.u.setmaxmem.maxmem = memory;
        ret = xenHypervisorDoV0Op(handle, &op);
1664
    }
1665
    return ret;
1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678
}

/**
 * virXen_setmaxvcpus:
 * @handle: the hypervisor handle
 * @id: the domain id
 * @vcpus: the numbers of vcpus
 *
 * Do a low level hypercall to change the max vcpus amount
 *
 * Returns 0 or -1 in case of failure
 */
static int
1679
virXen_setmaxvcpus(int handle, int id, unsigned int vcpus)
1680 1681 1682
{
    int ret = -1;

P
Philipp Hahn 已提交
1683
    if (hv_versions.hypervisor > 1) {
1684 1685 1686
        xen_op_v2_dom op;

        memset(&op, 0, sizeof(op));
1687 1688 1689 1690
        op.cmd = XEN_V2_OP_SETMAXVCPU;
        op.domain = (domid_t) id;
        op.u.setmaxvcpu.maxvcpu = vcpus;
        ret = xenHypervisorDoV2Dom(handle, &op);
P
Philipp Hahn 已提交
1691
    } else if (hv_versions.hypervisor == 1) {
1692 1693 1694
        xen_op_v1 op;

        memset(&op, 0, sizeof(op));
1695 1696 1697 1698
        op.cmd = XEN_V1_OP_SETMAXVCPU;
        op.u.setmaxvcpu.domain = (domid_t) id;
        op.u.setmaxvcpu.maxvcpu = vcpus;
        ret = xenHypervisorDoV1Op(handle, &op);
P
Philipp Hahn 已提交
1699
    } else if (hv_versions.hypervisor == 0) {
1700
        xen_op_v0 op;
1701 1702

        memset(&op, 0, sizeof(op));
1703 1704 1705 1706
        op.cmd = XEN_V0_OP_SETMAXVCPU;
        op.u.setmaxvcpu.domain = (domid_t) id;
        op.u.setmaxvcpu.maxvcpu = vcpus;
        ret = xenHypervisorDoV0Op(handle, &op);
1707
    }
1708
    return ret;
1709 1710 1711 1712 1713 1714 1715 1716
}

/**
 * virXen_setvcpumap:
 * @handle: the hypervisor handle
 * @id: the domain id
 * @vcpu: the vcpu to map
 * @cpumap: the bitmap for this vcpu
1717
 * @maplen: the size of the bitmap in bytes
1718 1719 1720 1721 1722 1723
 *
 * Do a low level hypercall to change the pinning for vcpu
 *
 * Returns 0 or -1 in case of failure
 */
static int
1724 1725 1726 1727 1728
virXen_setvcpumap(int handle,
                  int id,
                  unsigned int vcpu,
                  unsigned char * cpumap,
                  int maplen)
1729 1730
{
    int ret = -1;
1731 1732 1733
    unsigned char *new = NULL;
    unsigned char *bitmap = NULL;
    uint32_t nr_cpus;
1734

P
Philipp Hahn 已提交
1735
    if (hv_versions.hypervisor > 1) {
1736 1737
        xen_op_v2_dom op;

1738
        if (lock_pages(cpumap, maplen) < 0)
1739
            return -1;
1740

1741
        memset(&op, 0, sizeof(op));
1742 1743
        op.cmd = XEN_V2_OP_SETVCPUMAP;
        op.domain = (domid_t) id;
1744 1745 1746 1747

        /* The allocated memory to cpumap must be 'sizeof(uint64_t)' byte *
         * for Xen, and also nr_cpus must be 'sizeof(uint64_t) * 8'       */
        if (maplen < 8) {
1748
            if (VIR_ALLOC_N(new, sizeof(uint64_t)) < 0) {
1749
                virReportOOMError();
1750
                return -1;
1751 1752 1753 1754 1755 1756 1757 1758 1759
            }
            memcpy(new, cpumap, maplen);
            bitmap = new;
            nr_cpus = sizeof(uint64_t) * 8;
        } else {
            bitmap = cpumap;
            nr_cpus = maplen * 8;
        }

P
Philipp Hahn 已提交
1760
        if (hv_versions.dom_interface < 5) {
1761
            op.u.setvcpumap.vcpu = vcpu;
1762 1763
            op.u.setvcpumap.cpumap.bitmap = bitmap;
            op.u.setvcpumap.cpumap.nr_cpus = nr_cpus;
1764 1765
        } else {
            op.u.setvcpumapd5.vcpu = vcpu;
1766 1767
            op.u.setvcpumapd5.cpumap.bitmap.v = bitmap;
            op.u.setvcpumapd5.cpumap.nr_cpus = nr_cpus;
1768
        }
1769
        ret = xenHypervisorDoV2Dom(handle, &op);
1770
        VIR_FREE(new);
1771

1772
        if (unlock_pages(cpumap, maplen) < 0)
1773
            ret = -1;
1774
    } else {
1775
        cpumap_t xen_cpumap; /* limited to 64 CPUs in old hypervisors */
1776
        uint64_t *pm = &xen_cpumap;
1777
        int j;
1778

1779
        if ((maplen > (int)sizeof(cpumap_t)) || (sizeof(cpumap_t) & 7))
1780
            return -1;
1781

1782
        memset(&xen_cpumap, 0, sizeof(cpumap_t));
1783
        for (j = 0; j < maplen; j++) {
1784 1785 1786
            if ((j & 7) == 0)
                pm = (uint64_t *)((uint64_t)&xen_cpumap + (j & ~0x7UL));
            *pm |= (uint64_t)cpumap[j] << (8 * (j & 7));
1787
        }
1788

P
Philipp Hahn 已提交
1789
        if (hv_versions.hypervisor == 1) {
1790 1791 1792 1793 1794 1795 1796 1797
            xen_op_v1 op;

            memset(&op, 0, sizeof(op));
            op.cmd = XEN_V1_OP_SETVCPUMAP;
            op.u.setvcpumap.domain = (domid_t) id;
            op.u.setvcpumap.vcpu = vcpu;
            op.u.setvcpumap.cpumap = xen_cpumap;
            ret = xenHypervisorDoV1Op(handle, &op);
P
Philipp Hahn 已提交
1798
        } else if (hv_versions.hypervisor == 0) {
1799 1800 1801 1802 1803 1804 1805 1806 1807
            xen_op_v0 op;

            memset(&op, 0, sizeof(op));
            op.cmd = XEN_V0_OP_SETVCPUMAP;
            op.u.setvcpumap.domain = (domid_t) id;
            op.u.setvcpumap.vcpu = vcpu;
            op.u.setvcpumap.cpumap = xen_cpumap;
            ret = xenHypervisorDoV0Op(handle, &op);
        }
1808
    }
1809
    return ret;
1810
}
1811

1812

1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825
/**
 * virXen_getvcpusinfo:
 * @handle: the hypervisor handle
 * @id: the domain id
 * @vcpu: the vcpu to map
 * @cpumap: the bitmap for this vcpu
 * @maplen: the size of the bitmap in bytes
 *
 * Do a low level hypercall to change the pinning for vcpu
 *
 * Returns 0 or -1 in case of failure
 */
static int
1826 1827 1828 1829 1830 1831
virXen_getvcpusinfo(int handle,
                    int id,
                    unsigned int vcpu,
                    virVcpuInfoPtr ipt,
                    unsigned char *cpumap,
                    int maplen)
1832 1833 1834
{
    int ret = -1;

P
Philipp Hahn 已提交
1835
    if (hv_versions.hypervisor > 1) {
1836 1837 1838
        xen_op_v2_dom op;

        memset(&op, 0, sizeof(op));
1839 1840
        op.cmd = XEN_V2_OP_GETVCPUINFO;
        op.domain = (domid_t) id;
P
Philipp Hahn 已提交
1841
        if (hv_versions.dom_interface < 5)
1842 1843 1844
            op.u.getvcpuinfo.vcpu = (uint16_t) vcpu;
        else
            op.u.getvcpuinfod5.vcpu = (uint16_t) vcpu;
1845
        ret = xenHypervisorDoV2Dom(handle, &op);
1846

1847
        if (ret < 0)
1848
            return -1;
1849
        ipt->number = vcpu;
P
Philipp Hahn 已提交
1850
        if (hv_versions.dom_interface < 5) {
1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871
            if (op.u.getvcpuinfo.online) {
                if (op.u.getvcpuinfo.running)
                    ipt->state = VIR_VCPU_RUNNING;
                if (op.u.getvcpuinfo.blocked)
                    ipt->state = VIR_VCPU_BLOCKED;
            } else
                ipt->state = VIR_VCPU_OFFLINE;

            ipt->cpuTime = op.u.getvcpuinfo.cpu_time;
            ipt->cpu = op.u.getvcpuinfo.online ? (int)op.u.getvcpuinfo.cpu : -1;
        } else {
            if (op.u.getvcpuinfod5.online) {
                if (op.u.getvcpuinfod5.running)
                    ipt->state = VIR_VCPU_RUNNING;
                if (op.u.getvcpuinfod5.blocked)
                    ipt->state = VIR_VCPU_BLOCKED;
            } else
                ipt->state = VIR_VCPU_OFFLINE;

            ipt->cpuTime = op.u.getvcpuinfod5.cpu_time;
            ipt->cpu = op.u.getvcpuinfod5.online ? (int)op.u.getvcpuinfod5.cpu : -1;
1872 1873
        }
        if ((cpumap != NULL) && (maplen > 0)) {
1874
            if (lock_pages(cpumap, maplen) < 0)
1875
                return -1;
1876

1877
            memset(cpumap, 0, maplen);
1878 1879 1880
            memset(&op, 0, sizeof(op));
            op.cmd = XEN_V2_OP_GETVCPUMAP;
            op.domain = (domid_t) id;
P
Philipp Hahn 已提交
1881
            if (hv_versions.dom_interface < 5) {
1882 1883 1884 1885 1886 1887 1888 1889
                op.u.getvcpumap.vcpu = vcpu;
                op.u.getvcpumap.cpumap.bitmap = cpumap;
                op.u.getvcpumap.cpumap.nr_cpus = maplen * 8;
            } else {
                op.u.getvcpumapd5.vcpu = vcpu;
                op.u.getvcpumapd5.cpumap.bitmap.v = cpumap;
                op.u.getvcpumapd5.cpumap.nr_cpus = maplen * 8;
            }
1890
            ret = xenHypervisorDoV2Dom(handle, &op);
1891
            if (unlock_pages(cpumap, maplen) < 0)
1892 1893
                ret = -1;
        }
1894
    } else {
1895 1896 1897 1898 1899
        int mapl = maplen;
        int cpu;

        if (maplen > (int)sizeof(cpumap_t))
            mapl = (int)sizeof(cpumap_t);
1900

P
Philipp Hahn 已提交
1901
        if (hv_versions.hypervisor == 1) {
1902 1903 1904 1905 1906 1907 1908 1909
            xen_op_v1 op;

            memset(&op, 0, sizeof(op));
            op.cmd = XEN_V1_OP_GETVCPUINFO;
            op.u.getvcpuinfo.domain = (domid_t) id;
            op.u.getvcpuinfo.vcpu = vcpu;
            ret = xenHypervisorDoV1Op(handle, &op);
            if (ret < 0)
1910
                return -1;
1911
            ipt->number = vcpu;
1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924
            if (op.u.getvcpuinfo.online) {
                if (op.u.getvcpuinfo.running) ipt->state = VIR_VCPU_RUNNING;
                if (op.u.getvcpuinfo.blocked) ipt->state = VIR_VCPU_BLOCKED;
            }
            else ipt->state = VIR_VCPU_OFFLINE;
            ipt->cpuTime = op.u.getvcpuinfo.cpu_time;
            ipt->cpu = op.u.getvcpuinfo.online ? (int)op.u.getvcpuinfo.cpu : -1;
            if ((cpumap != NULL) && (maplen > 0)) {
                for (cpu = 0; cpu < (mapl * 8); cpu++) {
                    if (op.u.getvcpuinfo.cpumap & ((uint64_t)1<<cpu))
                        VIR_USE_CPU(cpumap, cpu);
                }
            }
P
Philipp Hahn 已提交
1925
        } else if (hv_versions.hypervisor == 0) {
1926 1927 1928 1929 1930 1931 1932 1933
            xen_op_v1 op;

            memset(&op, 0, sizeof(op));
            op.cmd = XEN_V0_OP_GETVCPUINFO;
            op.u.getvcpuinfo.domain = (domid_t) id;
            op.u.getvcpuinfo.vcpu = vcpu;
            ret = xenHypervisorDoV0Op(handle, &op);
            if (ret < 0)
1934
                return -1;
1935
            ipt->number = vcpu;
1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949
            if (op.u.getvcpuinfo.online) {
                if (op.u.getvcpuinfo.running) ipt->state = VIR_VCPU_RUNNING;
                if (op.u.getvcpuinfo.blocked) ipt->state = VIR_VCPU_BLOCKED;
            }
            else ipt->state = VIR_VCPU_OFFLINE;
            ipt->cpuTime = op.u.getvcpuinfo.cpu_time;
            ipt->cpu = op.u.getvcpuinfo.online ? (int)op.u.getvcpuinfo.cpu : -1;
            if ((cpumap != NULL) && (maplen > 0)) {
                for (cpu = 0; cpu < (mapl * 8); cpu++) {
                    if (op.u.getvcpuinfo.cpumap & ((uint64_t)1<<cpu))
                        VIR_USE_CPU(cpumap, cpu);
                }
            }
        }
1950
    }
1951
    return ret;
1952
}
1953

1954 1955
/**
 * xenHypervisorInit:
P
Philipp Hahn 已提交
1956 1957
 * @override_versions: pointer to optional struct xenHypervisorVersions with
 *     version information used instead of automatic version detection.
1958 1959 1960
 *
 * Initialize the hypervisor layer. Try to detect the kind of interface
 * used i.e. pre or post changeset 10277
P
Philipp Hahn 已提交
1961 1962
 *
 * Returns 0 or -1 in case of failure
1963
 */
1964
int
P
Philipp Hahn 已提交
1965
xenHypervisorInit(struct xenHypervisorVersions *override_versions)
1966
{
1967
    int fd, ret, cmd, errcode;
1968
    hypercall_t hc;
1969
    v0_hypercall_t v0_hc;
1970
    xen_getdomaininfo info;
D
Daniel Veillard 已提交
1971
    virVcpuInfoPtr ipt = NULL;
1972

1973 1974 1975 1976
    /* Compile regular expressions used by xenHypervisorGetCapabilities.
     * Note that errors here are really internal errors since these
     * regexps should never fail to compile.
     */
1977
    errcode = regcomp(&flags_hvm_rec, flags_hvm_re, REG_EXTENDED);
1978 1979
    if (errcode != 0) {
        char error[100];
1980 1981
        regerror(errcode, &flags_hvm_rec, error, sizeof(error));
        regfree(&flags_hvm_rec);
1982
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s", error);
1983 1984
        return -1;
    }
1985
    errcode = regcomp(&flags_pae_rec, flags_pae_re, REG_EXTENDED);
1986 1987
    if (errcode != 0) {
        char error[100];
1988 1989 1990
        regerror(errcode, &flags_pae_rec, error, sizeof(error));
        regfree(&flags_pae_rec);
        regfree(&flags_hvm_rec);
1991
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s", error);
1992 1993
        return -1;
    }
1994
    errcode = regcomp(&xen_cap_rec, xen_cap_re, REG_EXTENDED);
1995 1996
    if (errcode != 0) {
        char error[100];
1997 1998 1999 2000
        regerror(errcode, &xen_cap_rec, error, sizeof(error));
        regfree(&xen_cap_rec);
        regfree(&flags_pae_rec);
        regfree(&flags_hvm_rec);
2001
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s", error);
2002 2003 2004
        return -1;
    }

P
Philipp Hahn 已提交
2005 2006 2007 2008 2009
    if (override_versions) {
      hv_versions = *override_versions;
      return 0;
    }

2010
    /* Xen hypervisor version detection begins. */
2011 2012
    ret = open(XEN_HYPERVISOR_SOCKET, O_RDWR);
    if (ret < 0) {
P
Philipp Hahn 已提交
2013
        hv_versions.hypervisor = -1;
2014
        return -1;
2015 2016 2017
    }
    fd = ret;

2018 2019 2020 2021
    /*
     * The size of the hypervisor call block changed July 2006
     * this detect if we are using the new or old hypercall_t structure
     */
2022 2023 2024 2025 2026 2027 2028 2029
    hc.op = __HYPERVISOR_xen_version;
    hc.arg[0] = (unsigned long) XENVER_version;
    hc.arg[1] = 0;

    cmd = IOCTL_PRIVCMD_HYPERCALL;
    ret = ioctl(fd, cmd, (unsigned long) &hc);

    if ((ret != -1) && (ret != 0)) {
2030
        VIR_DEBUG("Using new hypervisor call: %X", ret);
P
Philipp Hahn 已提交
2031
        hv_versions.hv = ret;
2032 2033
        xen_ioctl_hypercall_cmd = cmd;
        goto detect_v2;
2034
    }
2035

2036
#ifndef __sun
2037 2038 2039 2040 2041 2042 2043 2044
    /*
     * check if the old hypercall are actually working
     */
    v0_hc.op = __HYPERVISOR_xen_version;
    v0_hc.arg[0] = (unsigned long) XENVER_version;
    v0_hc.arg[1] = 0;
    cmd = _IOC(_IOC_NONE, 'P', 0, sizeof(v0_hypercall_t));
    ret = ioctl(fd, cmd, (unsigned long) &v0_hc);
2045
    if ((ret != -1) && (ret != 0)) {
2046
        VIR_DEBUG("Using old hypervisor call: %X", ret);
P
Philipp Hahn 已提交
2047
        hv_versions.hv = ret;
2048
        xen_ioctl_hypercall_cmd = cmd;
P
Philipp Hahn 已提交
2049
        hv_versions.hypervisor = 0;
2050
        goto done;
2051
    }
2052
#endif
2053

2054
    /*
R
Richard W.M. Jones 已提交
2055
     * we failed to make any hypercall
2056 2057
     */

P
Philipp Hahn 已提交
2058
    hv_versions.hypervisor = -1;
2059 2060 2061
    virReportSystemError(errno,
                         _("Unable to issue hypervisor ioctl %lu"),
                         (unsigned long)IOCTL_PRIVCMD_HYPERCALL);
2062
    VIR_FORCE_CLOSE(fd);
2063
    return -1;
2064

2065
 detect_v2:
2066 2067 2068 2069 2070
    /*
     * The hypercalls were refactored into 3 different section in August 2006
     * Try to detect if we are running a version post 3.0.2 with the new ones
     * or the old ones
     */
P
Philipp Hahn 已提交
2071
    hv_versions.hypervisor = 2;
2072

2073
    if (VIR_ALLOC(ipt) < 0) {
2074
        virReportOOMError();
2075
        return -1;
2076
    }
2077
    /* Currently consider RHEL5.0 Fedora7, xen-3.1, and xen-unstable */
P
Philipp Hahn 已提交
2078
    hv_versions.sys_interface = 2; /* XEN_SYSCTL_INTERFACE_VERSION */
2079
    if (virXen_getdomaininfo(fd, 0, &info) == 1) {
2080
        /* RHEL 5.0 */
P
Philipp Hahn 已提交
2081
        hv_versions.dom_interface = 3; /* XEN_DOMCTL_INTERFACE_VERSION */
2082
        if (virXen_getvcpusinfo(fd, 0, 0, ipt, NULL, 0) == 0){
2083
            VIR_DEBUG("Using hypervisor call v2, sys ver2 dom ver3");
2084 2085 2086
            goto done;
        }
        /* Fedora 7 */
P
Philipp Hahn 已提交
2087
        hv_versions.dom_interface = 4; /* XEN_DOMCTL_INTERFACE_VERSION */
2088
        if (virXen_getvcpusinfo(fd, 0, 0, ipt, NULL, 0) == 0){
2089
            VIR_DEBUG("Using hypervisor call v2, sys ver2 dom ver4");
2090 2091 2092 2093
            goto done;
        }
    }

P
Philipp Hahn 已提交
2094
    hv_versions.sys_interface = 3; /* XEN_SYSCTL_INTERFACE_VERSION */
2095
    if (virXen_getdomaininfo(fd, 0, &info) == 1) {
2096
        /* xen-3.1 */
P
Philipp Hahn 已提交
2097
        hv_versions.dom_interface = 5; /* XEN_DOMCTL_INTERFACE_VERSION */
2098
        if (virXen_getvcpusinfo(fd, 0, 0, ipt, NULL, 0) == 0){
2099
            VIR_DEBUG("Using hypervisor call v2, sys ver3 dom ver5");
2100 2101
            goto done;
        }
2102
    }
2103

P
Philipp Hahn 已提交
2104
    hv_versions.sys_interface = 4; /* XEN_SYSCTL_INTERFACE_VERSION */
2105
    if (virXen_getdomaininfo(fd, 0, &info) == 1) {
2106
        /* Fedora 8 */
P
Philipp Hahn 已提交
2107
        hv_versions.dom_interface = 5; /* XEN_DOMCTL_INTERFACE_VERSION */
2108
        if (virXen_getvcpusinfo(fd, 0, 0, ipt, NULL, 0) == 0){
2109
            VIR_DEBUG("Using hypervisor call v2, sys ver4 dom ver5");
2110 2111 2112 2113
            goto done;
        }
    }

P
Philipp Hahn 已提交
2114
    hv_versions.sys_interface = 6; /* XEN_SYSCTL_INTERFACE_VERSION */
2115 2116
    if (virXen_getdomaininfo(fd, 0, &info) == 1) {
        /* Xen 3.2, Fedora 9 */
P
Philipp Hahn 已提交
2117
        hv_versions.dom_interface = 5; /* XEN_DOMCTL_INTERFACE_VERSION */
2118
        if (virXen_getvcpusinfo(fd, 0, 0, ipt, NULL, 0) == 0){
2119
            VIR_DEBUG("Using hypervisor call v2, sys ver6 dom ver5");
2120 2121
            goto done;
        }
J
Jim Fehlig 已提交
2122 2123 2124
    }

    /* Xen 4.0 */
P
Philipp Hahn 已提交
2125
    hv_versions.sys_interface = 7; /* XEN_SYSCTL_INTERFACE_VERSION */
J
Jim Fehlig 已提交
2126
    if (virXen_getdomaininfo(fd, 0, &info) == 1) {
P
Philipp Hahn 已提交
2127
        hv_versions.dom_interface = 6; /* XEN_DOMCTL_INTERFACE_VERSION */
2128
        VIR_DEBUG("Using hypervisor call v2, sys ver7 dom ver6");
J
Jim Fehlig 已提交
2129
        goto done;
2130 2131
    }

J
Jim Fehlig 已提交
2132 2133 2134
    /* Xen 4.1
     * sysctl version 8 -> xen-unstable c/s 21118:28e5409e3fb3
     * domctl version 7 -> xen-unstable c/s 21212:de94884a669c
J
Jim Fehlig 已提交
2135
     * domctl version 8 -> xen-unstable c/s 23874:651aed73b39c
J
Jim Fehlig 已提交
2136
     */
P
Philipp Hahn 已提交
2137
    hv_versions.sys_interface = 8; /* XEN_SYSCTL_INTERFACE_VERSION */
J
Jim Fehlig 已提交
2138
    if (virXen_getdomaininfo(fd, 0, &info) == 1) {
P
Philipp Hahn 已提交
2139
        hv_versions.dom_interface = 7; /* XEN_DOMCTL_INTERFACE_VERSION */
J
Jim Fehlig 已提交
2140 2141 2142 2143 2144 2145 2146 2147 2148
        if (virXen_getvcpusinfo(fd, 0, 0, ipt, NULL, 0) == 0){
            VIR_DEBUG("Using hypervisor call v2, sys ver8 dom ver7");
            goto done;
        }
        hv_versions.dom_interface = 8; /* XEN_DOMCTL_INTERFACE_VERSION */
        if (virXen_getvcpusinfo(fd, 0, 0, ipt, NULL, 0) == 0){
            VIR_DEBUG("Using hypervisor call v2, sys ver8 dom ver8");
            goto done;
        }
J
Jim Fehlig 已提交
2149 2150
    }

2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162
    /* Xen 4.2
     * sysctl version 9 -> xen-unstable c/s 24102:dc8e55c90604
     * domctl version 8 -> unchanged from Xen 4.1
     */
    hv_versions.sys_interface = 9; /* XEN_SYSCTL_INTERFACE_VERSION */
    if (virXen_getdomaininfo(fd, 0, &info) == 1) {
        hv_versions.dom_interface = 8; /* XEN_DOMCTL_INTERFACE_VERSION */
        if (virXen_getvcpusinfo(fd, 0, 0, ipt, NULL, 0) == 0){
            VIR_DEBUG("Using hypervisor call v2, sys ver9 dom ver8");
            goto done;
        }
    }
2163

J
Jim Fehlig 已提交
2164 2165 2166 2167 2168 2169 2170
    hv_versions.hypervisor = 1;
    hv_versions.sys_interface = -1;
    if (virXen_getdomaininfo(fd, 0, &info) == 1) {
        VIR_DEBUG("Using hypervisor call v1");
        goto done;
    }

2171
    /*
R
Richard W.M. Jones 已提交
2172
     * we failed to make the getdomaininfolist hypercall
2173
     */
P
Philipp Hahn 已提交
2174
    hv_versions.hypervisor = -1;
2175 2176 2177 2178
    virReportSystemError(errno,
                         _("Unable to issue hypervisor ioctl %lu"),
                         (unsigned long)IOCTL_PRIVCMD_HYPERCALL);
    VIR_DEBUG("Failed to find any Xen hypervisor method");
2179
    VIR_FORCE_CLOSE(fd);
2180
    VIR_FREE(ipt);
2181
    return -1;
2182

2183
 done:
2184
    VIR_FORCE_CLOSE(fd);
2185
    VIR_FREE(ipt);
2186
    return 0;
2187 2188
}

2189 2190 2191 2192 2193 2194 2195

static int xenHypervisorOnceInit(void) {
    return xenHypervisorInit(NULL);
}

VIR_ONCE_GLOBAL_INIT(xenHypervisor)

2196 2197
/**
 * xenHypervisorOpen:
2198 2199 2200
 * @conn: pointer to the connection block
 * @name: URL for the target, NULL for local
 * @flags: combination of virDrvOpenFlag(s)
2201 2202 2203
 *
 * Connects to the Xen hypervisor.
 *
2204
 * Returns 0 or -1 in case of error.
2205
 */
2206
virDrvOpenStatus
2207
xenHypervisorOpen(virConnectPtr conn,
2208
                  virConnectAuthPtr auth ATTRIBUTE_UNUSED,
E
Eric Blake 已提交
2209
                  unsigned int flags)
2210
{
2211
    int ret;
2212
    xenUnifiedPrivatePtr priv = conn->privateData;
2213

E
Eric Blake 已提交
2214 2215
    virCheckFlags(VIR_CONNECT_RO, VIR_DRV_OPEN_ERROR);

2216 2217
    if (xenHypervisorInitialize() < 0)
        return -1;
2218

2219
    priv->handle = -1;
2220

2221
    ret = open(XEN_HYPERVISOR_SOCKET, O_RDWR);
2222
    if (ret < 0) {
2223
        virReportError(VIR_ERR_NO_XEN, "%s", XEN_HYPERVISOR_SOCKET);
2224
        return -1;
2225
    }
2226 2227

    priv->handle = ret;
2228

2229
    return 0;
2230 2231 2232 2233
}

/**
 * xenHypervisorClose:
2234
 * @conn: pointer to the connection block
2235 2236 2237 2238 2239
 *
 * Close the connection to the Xen hypervisor.
 *
 * Returns 0 in case of success or -1 in case of error.
 */
2240
int
2241
xenHypervisorClose(virConnectPtr conn)
2242
{
2243
    int ret;
2244
    xenUnifiedPrivatePtr priv = conn->privateData;
2245 2246 2247 2248

    if (priv->handle < 0)
        return -1;

2249
    ret = VIR_CLOSE(priv->handle);
2250
    if (ret < 0)
2251
        return -1;
2252

2253
    return 0;
2254 2255 2256
}


2257 2258
/**
 * xenHypervisorGetVersion:
2259 2260
 * @conn: pointer to the connection block
 * @hvVer: where to store the version
2261 2262 2263
 *
 * Call the hypervisor to extracts his own internal API version
 *
2264
 * Returns 0 in case of success, -1 in case of error
2265
 */
2266 2267
int
xenHypervisorGetVersion(virConnectPtr conn, unsigned long *hvVer)
2268
{
2269
    xenUnifiedPrivatePtr priv = conn->privateData;
2270

2271
    if (priv->handle < 0)
2272
        return -1;
P
Philipp Hahn 已提交
2273
    *hvVer = (hv_versions.hv >> 16) * 1000000 + (hv_versions.hv & 0xFFFF) * 1000;
2274
    return 0;
2275 2276
}

2277
struct guest_arch {
2278
    virArch arch;
2279 2280 2281 2282 2283 2284 2285
    int hvm;
    int pae;
    int nonpae;
    int ia64_be;
};


2286
static int xenDefaultConsoleType(const char *ostype,
2287
                                 virArch arch ATTRIBUTE_UNUSED)
2288 2289 2290 2291 2292 2293 2294
{
    if (STREQ(ostype, "hvm"))
        return VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_SERIAL;
    else
        return VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_XEN;
}

2295
static virCapsPtr
2296
xenHypervisorBuildCapabilities(virConnectPtr conn, virArch hostarch,
2297
                               int host_pae,
2298
                               const char *hvm_type,
2299
                               struct guest_arch *guest_archs,
2300 2301
                               int nr_guest_archs)
{
2302 2303
    virCapsPtr caps;
    int i;
P
Philipp Hahn 已提交
2304 2305
    int hv_major = hv_versions.hv >> 16;
    int hv_minor = hv_versions.hv & 0xFFFF;
2306

2307
    if ((caps = virCapabilitiesNew(hostarch, 1, 1)) == NULL)
2308
        goto no_memory;
2309 2310 2311

    virCapabilitiesSetMacPrefix(caps, (unsigned char[]){ 0x00, 0x16, 0x3e });

2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324
    if (hvm_type && STRNEQ(hvm_type, "") &&
        virCapabilitiesAddHostFeature(caps, hvm_type) < 0)
        goto no_memory;
    if (host_pae &&
        virCapabilitiesAddHostFeature(caps, "pae") < 0)
        goto no_memory;


    if (virCapabilitiesAddHostMigrateTransport(caps,
                                               "xenmigr") < 0)
        goto no_memory;


P
Philipp Hahn 已提交
2325
    if (hv_versions.sys_interface >= SYS_IFACE_MIN_VERS_NUMA && conn != NULL) {
2326
        if (xenDaemonNodeGetTopology(conn, caps) != 0) {
2327
            virObjectUnref(caps);
2328 2329 2330 2331 2332 2333
            return NULL;
        }
    }

    for (i = 0; i < nr_guest_archs; ++i) {
        virCapsGuestPtr guest;
2334 2335 2336
        char const *const xen_machines[] = {guest_archs[i].hvm ? "xenfv" : "xenpv"};
        virCapsGuestMachinePtr *machines;

2337
        if ((machines = virCapabilitiesAllocMachines(xen_machines, 1)) == NULL)
2338
            goto no_memory;
2339 2340 2341

        if ((guest = virCapabilitiesAddGuest(caps,
                                             guest_archs[i].hvm ? "hvm" : "xen",
2342 2343
                                             guest_archs[i].arch,
                                             (hostarch == VIR_ARCH_X86_64 ?
2344 2345 2346 2347 2348 2349
                                              "/usr/lib64/xen/bin/qemu-dm" :
                                              "/usr/lib/xen/bin/qemu-dm"),
                                             (guest_archs[i].hvm ?
                                              "/usr/lib/xen/boot/hvmloader" :
                                              NULL),
                                             1,
2350 2351
                                             machines)) == NULL) {
            virCapabilitiesFreeMachines(machines, 1);
2352
            goto no_memory;
2353
        }
2354
        machines = NULL;
2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390

        if (virCapabilitiesAddGuestDomain(guest,
                                          "xen",
                                          NULL,
                                          NULL,
                                          0,
                                          NULL) == NULL)
            goto no_memory;

        if (guest_archs[i].pae &&
            virCapabilitiesAddGuestFeature(guest,
                                           "pae",
                                           1,
                                           0) == NULL)
            goto no_memory;

        if (guest_archs[i].nonpae &&
            virCapabilitiesAddGuestFeature(guest,
                                           "nonpae",
                                           1,
                                           0) == NULL)
            goto no_memory;

        if (guest_archs[i].ia64_be &&
            virCapabilitiesAddGuestFeature(guest,
                                           "ia64_be",
                                           1,
                                           0) == NULL)
            goto no_memory;

        if (guest_archs[i].hvm) {
            if (virCapabilitiesAddGuestFeature(guest,
                                               "acpi",
                                               1, 1) == NULL)
                goto no_memory;

2391
            /* In Xen 3.1.0, APIC is always on and can't be toggled */
2392 2393 2394 2395 2396 2397 2398
            if (virCapabilitiesAddGuestFeature(guest,
                                               "apic",
                                               1,
                                               (hv_major > 3 &&
                                                hv_minor > 0 ?
                                                0 : 1)) == NULL)
                goto no_memory;
2399 2400 2401 2402 2403 2404 2405 2406 2407 2408

            /* Xen 3.3.x and beyond supports enabling/disabling
             * hardware assisted paging.  Default is off.
             */
            if ((hv_major == 3 && hv_minor >= 3) || (hv_major > 3))
                if (virCapabilitiesAddGuestFeature(guest,
                                                   "hap",
                                                   0,
                                                   1) == NULL)
                    goto no_memory;
2409 2410 2411 2412 2413 2414 2415 2416 2417 2418

            /* Xen 3.4.x and beyond supports the Viridian (Hyper-V)
             * enlightenment interface.  Default is off.
             */
            if ((hv_major == 3 && hv_minor >= 4) || (hv_major > 3))
                if (virCapabilitiesAddGuestFeature(guest,
                                                   "viridian",
                                                   0,
                                                   1) == NULL)
                    goto no_memory;
2419
        }
2420

2421 2422
    }

2423
    caps->defaultConsoleTargetType = xenDefaultConsoleType;
2424

2425 2426 2427
    return caps;

 no_memory:
2428
    virObjectUnref(caps);
2429 2430 2431
    return NULL;
}

2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449
#ifdef __sun

static int
get_cpu_flags(virConnectPtr conn, const char **hvm, int *pae, int *longmode)
{
    struct {
        uint32_t r_eax, r_ebx, r_ecx, r_edx;
    } regs;

    char tmpbuf[20];
    int ret = 0;
    int fd;

    /* returns -1, errno 22 if in 32-bit mode */
    *longmode = (sysinfo(SI_ARCHITECTURE_64, tmpbuf, sizeof(tmpbuf)) != -1);

    if ((fd = open("/dev/cpu/self/cpuid", O_RDONLY)) == -1 ||
        pread(fd, &regs, sizeof(regs), 0) != sizeof(regs)) {
2450
        virReportSystemError(errno, "%s", _("could not read CPU flags"));
2451 2452 2453 2454 2455 2456 2457
        goto out;
    }

    *pae = 0;
    *hvm = "";

    if (STREQLEN((const char *)&regs.r_ebx, "AuthcAMDenti", 12)) {
2458
        if (pread(fd, &regs, sizeof(regs), 0x80000001) == sizeof(regs)) {
2459 2460 2461 2462 2463 2464 2465 2466
            /* Read secure virtual machine bit (bit 2 of ECX feature ID) */
            if ((regs.r_ecx >> 2) & 1) {
                *hvm = "svm";
            }
            if ((regs.r_edx >> 6) & 1)
                *pae = 1;
        }
    } else if (STREQLEN((const char *)&regs.r_ebx, "GenuntelineI", 12)) {
2467
        if (pread(fd, &regs, sizeof(regs), 0x00000001) == sizeof(regs)) {
2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478
            /* Read VMXE feature bit (bit 5 of ECX feature ID) */
            if ((regs.r_ecx >> 5) & 1)
                *hvm = "vmx";
            if ((regs.r_edx >> 6) & 1)
                *pae = 1;
        }
    }

    ret = 1;

out:
2479
    VIR_FORCE_CLOSE(fd);
2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494
    return ret;
}

static virCapsPtr
xenHypervisorMakeCapabilitiesSunOS(virConnectPtr conn)
{
    struct guest_arch guest_arches[32];
    int i = 0;
    virCapsPtr caps = NULL;
    int pae, longmode;
    const char *hvm;

    if (!get_cpu_flags(conn, &hvm, &pae, &longmode))
        return NULL;

2495
    guest_arches[i].arch = VIR_ARCH_I686;
2496 2497 2498 2499 2500 2501 2502
    guest_arches[i].hvm = 0;
    guest_arches[i].pae = pae;
    guest_arches[i].nonpae = !pae;
    guest_arches[i].ia64_be = 0;
    i++;

    if (longmode) {
2503
        guest_arches[i].arch = VIR_ARCH_X86_64;
2504 2505 2506 2507 2508 2509 2510 2511
        guest_arches[i].hvm = 0;
        guest_arches[i].pae = 0;
        guest_arches[i].nonpae = 0;
        guest_arches[i].ia64_be = 0;
        i++;
    }

    if (hvm[0] != '\0') {
2512
        guest_arches[i].arch = VIR_ARCH_I686;
2513 2514 2515 2516 2517 2518 2519
        guest_arches[i].hvm = 1;
        guest_arches[i].pae = pae;
        guest_arches[i].nonpae = 1;
        guest_arches[i].ia64_be = 0;
        i++;

        if (longmode) {
2520
            guest_arches[i].arch = VIR_ARCH_X86_64;
2521 2522 2523 2524 2525 2526 2527 2528 2529
            guest_arches[i].hvm = 1;
            guest_arches[i].pae = 0;
            guest_arches[i].nonpae = 0;
            guest_arches[i].ia64_be = 0;
            i++;
        }
    }

    if ((caps = xenHypervisorBuildCapabilities(conn,
2530
                                               virArchFromHost(),
2531 2532
                                               pae, hvm,
                                               guest_arches, i)) == NULL)
2533
        virReportOOMError();
2534 2535 2536 2537 2538 2539

    return caps;
}

#endif /* __sun */

2540
/**
2541
 * xenHypervisorMakeCapabilitiesInternal:
2542
 * @conn: pointer to the connection block
2543 2544
 * @cpuinfo: file handle containing /proc/cpuinfo data, or NULL
 * @capabilities: file handle containing /sys/hypervisor/properties/capabilities data, or NULL
2545 2546 2547
 *
 * Return the capabilities of this hypervisor.
 */
2548
virCapsPtr
2549
xenHypervisorMakeCapabilitiesInternal(virConnectPtr conn,
2550
                                      virArch hostarch,
2551 2552
                                      FILE *cpuinfo,
                                      FILE *capabilities)
2553 2554
{
    char line[1024], *str, *token;
2555
    regmatch_t subs[4];
2556
    char *saveptr = NULL;
2557
    int i;
2558 2559 2560

    char hvm_type[4] = ""; /* "vmx" or "svm" (or "" if not in CPU). */
    int host_pae = 0;
2561
    struct guest_arch guest_archs[32];
2562
    int nr_guest_archs = 0;
2563
    virCapsPtr caps = NULL;
2564

2565 2566
    memset(guest_archs, 0, sizeof(guest_archs));

2567 2568 2569 2570
    /* /proc/cpuinfo: flags: Intel calls HVM "vmx", AMD calls it "svm".
     * It's not clear if this will work on IA64, let alone other
     * architectures and non-Linux. (XXX)
     */
2571
    if (cpuinfo) {
2572 2573
        while (fgets(line, sizeof(line), cpuinfo)) {
            if (regexec(&flags_hvm_rec, line, sizeof(subs)/sizeof(regmatch_t), subs, 0) == 0
2574
                && subs[0].rm_so != -1) {
C
Chris Lalancette 已提交
2575 2576 2577 2578
                if (virStrncpy(hvm_type,
                               &line[subs[1].rm_so],
                               subs[1].rm_eo-subs[1].rm_so,
                               sizeof(hvm_type)) == NULL)
2579
                    goto no_memory;
2580
            } else if (regexec(&flags_pae_rec, line, 0, NULL, 0) == 0)
2581 2582
                host_pae = 1;
        }
2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609
    }

    /* Most of the useful info is in /sys/hypervisor/properties/capabilities
     * which is documented in the code in xen-unstable.hg/xen/arch/.../setup.c.
     *
     * It is a space-separated list of supported guest architectures.
     *
     * For x86:
     *    TYP-VER-ARCH[p]
     *    ^   ^   ^    ^
     *    |   |   |    +-- PAE supported
     *    |   |   +------- x86_32 or x86_64
     *    |   +----------- the version of Xen, eg. "3.0"
     *    +--------------- "xen" or "hvm" for para or full virt respectively
     *
     * For PPC this file appears to be always empty (?)
     *
     * For IA64:
     *    TYP-VER-ARCH[be]
     *    ^   ^   ^    ^
     *    |   |   |    +-- Big-endian supported
     *    |   |   +------- always "ia64"
     *    |   +----------- the version of Xen, eg. "3.0"
     *    +--------------- "xen" or "hvm" for para or full virt respectively
     */

    /* Expecting one line in this file - ignore any more. */
2610
    if ((capabilities) && (fgets(line, sizeof(line), capabilities))) {
2611 2612 2613 2614
        /* Split the line into tokens.  strtok_r is OK here because we "own"
         * this buffer.  Parse out the features from each token.
         */
        for (str = line, nr_guest_archs = 0;
2615
             nr_guest_archs < sizeof(guest_archs) / sizeof(guest_archs[0])
2616
                 && (token = strtok_r(str, " ", &saveptr)) != NULL;
2617 2618
             str = NULL) {

2619 2620
            if (regexec(&xen_cap_rec, token, sizeof(subs) / sizeof(subs[0]),
                        subs, 0) == 0) {
2621
                int hvm = STRPREFIX(&token[subs[1].rm_so], "hvm");
2622 2623
                int pae = 0, nonpae = 0, ia64_be = 0;
                virArch arch;
2624 2625

                if (STRPREFIX(&token[subs[2].rm_so], "x86_32")) {
2626
                    arch = VIR_ARCH_I686;
2627 2628
                    if (subs[3].rm_so != -1 &&
                        STRPREFIX(&token[subs[3].rm_so], "p"))
2629 2630 2631 2632
                        pae = 1;
                    else
                        nonpae = 1;
                }
2633
                else if (STRPREFIX(&token[subs[2].rm_so], "x86_64")) {
2634
                    arch = VIR_ARCH_X86_64;
2635
                }
2636
                else if (STRPREFIX(&token[subs[2].rm_so], "ia64")) {
2637
                    arch = VIR_ARCH_ITANIUM;
2638 2639
                    if (subs[3].rm_so != -1 &&
                        STRPREFIX(&token[subs[3].rm_so], "be"))
2640 2641
                        ia64_be = 1;
                }
2642
                else if (STRPREFIX(&token[subs[2].rm_so], "powerpc64")) {
2643
                    arch = VIR_ARCH_PPC64;
2644
                } else {
2645
                    /* XXX surely no other Xen archs exist. Arrrrrrrrrm  */
2646 2647
                    continue;
                }
2648

2649 2650
                /* Search for existing matching (model,hvm) tuple */
                for (i = 0 ; i < nr_guest_archs ; i++) {
2651
                    if (guest_archs[i].arch == arch &&
2652 2653 2654 2655
                        guest_archs[i].hvm == hvm) {
                        break;
                    }
                }
2656

2657
                /* Too many arch flavours - highly unlikely ! */
J
Jim Meyering 已提交
2658
                if (i >= ARRAY_CARDINALITY(guest_archs))
2659 2660 2661 2662 2663
                    continue;
                /* Didn't find a match, so create a new one */
                if (i == nr_guest_archs)
                    nr_guest_archs++;

2664
                guest_archs[i].arch = arch;
2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676
                guest_archs[i].hvm = hvm;

                /* Careful not to overwrite a previous positive
                   setting with a negative one here - some archs
                   can do both pae & non-pae, but Xen reports
                   separately capabilities so we're merging archs */
                if (pae)
                    guest_archs[i].pae = pae;
                if (nonpae)
                    guest_archs[i].nonpae = nonpae;
                if (ia64_be)
                    guest_archs[i].ia64_be = ia64_be;
2677 2678 2679 2680
            }
        }
    }

2681
    if ((caps = xenHypervisorBuildCapabilities(conn,
2682
                                               hostarch,
2683 2684 2685 2686 2687
                                               host_pae,
                                               hvm_type,
                                               guest_archs,
                                               nr_guest_archs)) == NULL)
        goto no_memory;
2688

2689
    return caps;
2690

2691
 no_memory:
2692
    virReportOOMError();
2693
    virObjectUnref(caps);
2694 2695 2696 2697
    return NULL;
}

/**
2698
 * xenHypervisorMakeCapabilities:
2699 2700 2701
 *
 * Return the capabilities of this hypervisor.
 */
2702
virCapsPtr
2703
xenHypervisorMakeCapabilities(virConnectPtr conn)
2704
{
2705 2706 2707
#ifdef __sun
    return xenHypervisorMakeCapabilitiesSunOS(conn);
#else
2708
    virCapsPtr caps = NULL;
2709 2710
    FILE *cpuinfo, *capabilities;

2711
    cpuinfo = fopen("/proc/cpuinfo", "r");
2712 2713
    if (cpuinfo == NULL) {
        if (errno != ENOENT) {
2714
            virReportSystemError(errno,
2715 2716
                                 _("cannot read file %s"),
                                 "/proc/cpuinfo");
2717 2718 2719 2720
            return NULL;
        }
    }

2721
    capabilities = fopen("/sys/hypervisor/properties/capabilities", "r");
2722 2723
    if (capabilities == NULL) {
        if (errno != ENOENT) {
2724
            VIR_FORCE_FCLOSE(cpuinfo);
2725
            virReportSystemError(errno,
2726 2727
                                 _("cannot read file %s"),
                                 "/sys/hypervisor/properties/capabilities");
2728 2729 2730 2731
            return NULL;
        }
    }

2732
    caps = xenHypervisorMakeCapabilitiesInternal(conn,
2733
                                                 virArchFromHost(),
2734 2735
                                                 cpuinfo,
                                                 capabilities);
2736
    if (caps == NULL)
2737
        goto cleanup;
2738

2739 2740 2741
    if (virNodeSuspendGetTargetMask(&caps->host.powerMgmt) < 0)
        VIR_WARN("Failed to get host power management capabilities");

2742
cleanup:
2743 2744
    VIR_FORCE_FCLOSE(cpuinfo);
    VIR_FORCE_FCLOSE(capabilities);
2745

2746
    return caps;
2747
#endif /* __sun */
2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758
}



/**
 * xenHypervisorGetCapabilities:
 * @conn: pointer to the connection block
 *
 * Return the capabilities of this hypervisor.
 */
char *
2759
xenHypervisorGetCapabilities(virConnectPtr conn)
2760
{
2761
    xenUnifiedPrivatePtr priv = conn->privateData;
2762 2763 2764
    char *xml;

    if (!(xml = virCapabilitiesFormatXML(priv->caps))) {
2765
        virReportOOMError();
2766 2767 2768
        return NULL;
    }

2769
    return xml;
2770 2771
}

2772

2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783
/**
 * xenHypervisorNumOfDomains:
 * @conn: pointer to the connection block
 *
 * Provides the number of active domains.
 *
 * Returns the number of domain found or -1 in case of error
 */
int
xenHypervisorNumOfDomains(virConnectPtr conn)
{
2784
    xen_getdomaininfolist dominfos;
2785 2786 2787
    int ret, nbids;
    static int last_maxids = 2;
    int maxids = last_maxids;
2788
    xenUnifiedPrivatePtr priv = conn->privateData;
2789

2790
    if (priv->handle < 0)
2791
        return -1;
2792

2793 2794
 retry:
    if (!(XEN_GETDOMAININFOLIST_ALLOC(dominfos, maxids))) {
2795
        virReportOOMError();
2796
        return -1;
2797 2798
    }

2799 2800
    XEN_GETDOMAININFOLIST_CLEAR(dominfos, maxids);

2801
    ret = virXen_getdomaininfolist(priv->handle, 0, maxids, &dominfos);
2802

2803
    XEN_GETDOMAININFOLIST_FREE(dominfos);
2804 2805

    if (ret < 0)
2806
        return -1;
2807

2808
    nbids = ret;
2809 2810 2811 2812 2813
    /* Can't possibly have more than 65,000 concurrent guests
     * so limit how many times we try, to avoid increasing
     * without bound & thus allocating all of system memory !
     * XXX I'll regret this comment in a few years time ;-)
     */
2814
    if (nbids == maxids) {
2815 2816 2817 2818 2819 2820
        if (maxids < 65000) {
            last_maxids *= 2;
            maxids *= 2;
            goto retry;
        }
        nbids = -1;
2821 2822
    }
    if ((nbids < 0) || (nbids > maxids))
2823 2824
        return -1;
    return nbids;
2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839
}

/**
 * xenHypervisorListDomains:
 * @conn: pointer to the connection block
 * @ids: array to collect the list of IDs of active domains
 * @maxids: size of @ids
 *
 * Collect the list of active domains, and store their ID in @maxids
 *
 * Returns the number of domain found or -1 in case of error
 */
int
xenHypervisorListDomains(virConnectPtr conn, int *ids, int maxids)
{
2840
    xen_getdomaininfolist dominfos;
2841
    int ret, nbids, i;
2842
    xenUnifiedPrivatePtr priv = conn->privateData;
2843

2844
    if (priv->handle < 0)
2845
        return -1;
2846

2847
    if (maxids == 0)
2848
        return 0;
2849

2850
    if (!(XEN_GETDOMAININFOLIST_ALLOC(dominfos, maxids))) {
2851
        virReportOOMError();
2852
        return -1;
2853
    }
2854 2855

    XEN_GETDOMAININFOLIST_CLEAR(dominfos, maxids);
2856 2857
    memset(ids, 0, maxids * sizeof(int));

2858
    ret = virXen_getdomaininfolist(priv->handle, 0, maxids, &dominfos);
2859 2860

    if (ret < 0) {
2861
        XEN_GETDOMAININFOLIST_FREE(dominfos);
2862
        return -1;
2863 2864
    }

2865
    nbids = ret;
2866
    if ((nbids < 0) || (nbids > maxids)) {
2867
        XEN_GETDOMAININFOLIST_FREE(dominfos);
2868
        return -1;
2869 2870 2871
    }

    for (i = 0;i < nbids;i++) {
2872
        ids[i] = XEN_GETDOMAININFOLIST_DOMAIN(dominfos, i);
2873 2874
    }

2875
    XEN_GETDOMAININFOLIST_FREE(dominfos);
2876
    return nbids;
2877 2878
}

2879 2880

char *
2881
xenHypervisorDomainGetOSType(virDomainPtr dom)
2882
{
2883
    xenUnifiedPrivatePtr priv = dom->conn->privateData;
2884
    xen_getdomaininfo dominfo;
2885
    char *ostype = NULL;
2886

2887
    if (priv->handle < 0) {
2888 2889
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("domain shut off or invalid"));
2890
        return NULL;
2891
    }
2892 2893

    /* HV's earlier than 3.1.0 don't include the HVM flags in guests status*/
P
Philipp Hahn 已提交
2894 2895
    if (hv_versions.hypervisor < 2 ||
        hv_versions.dom_interface < 4) {
2896 2897
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("unsupported in dom interface < 4"));
2898
        return NULL;
2899
    }
2900 2901 2902

    XEN_GETDOMAININFO_CLEAR(dominfo);

2903
    if (virXen_getdomaininfo(priv->handle, dom->id, &dominfo) < 0) {
2904 2905
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("cannot get domain details"));
2906
        return NULL;
2907
    }
2908

2909
    if (XEN_GETDOMAININFO_DOMAIN(dominfo) != dom->id) {
2910 2911
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("cannot get domain details"));
2912
        return NULL;
2913
    }
2914 2915

    if (XEN_GETDOMAININFO_FLAGS(dominfo) & DOMFLAGS_HVM)
2916 2917 2918 2919 2920
        ostype = strdup("hvm");
    else
        ostype = strdup("linux");

    if (ostype == NULL)
2921
        virReportOOMError();
2922 2923

    return ostype;
2924 2925
}

2926
int
2927
xenHypervisorHasDomain(virConnectPtr conn, int id)
2928
{
2929
    xenUnifiedPrivatePtr priv = conn->privateData;
2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945
    xen_getdomaininfo dominfo;

    if (priv->handle < 0)
        return 0;

    XEN_GETDOMAININFO_CLEAR(dominfo);

    if (virXen_getdomaininfo(priv->handle, id, &dominfo) < 0)
        return 0;

    if (XEN_GETDOMAININFO_DOMAIN(dominfo) != id)
        return 0;

    return 1;
}

2946
virDomainPtr
2947
xenHypervisorLookupDomainByID(virConnectPtr conn, int id)
2948
{
2949
    xenUnifiedPrivatePtr priv = conn->privateData;
2950 2951 2952 2953 2954
    xen_getdomaininfo dominfo;
    virDomainPtr ret;
    char *name;

    if (priv->handle < 0)
2955
        return NULL;
2956 2957 2958 2959

    XEN_GETDOMAININFO_CLEAR(dominfo);

    if (virXen_getdomaininfo(priv->handle, id, &dominfo) < 0)
2960
        return NULL;
2961 2962

    if (XEN_GETDOMAININFO_DOMAIN(dominfo) != id)
2963
        return NULL;
2964

D
Daniel P. Berrange 已提交
2965 2966 2967 2968
    xenUnifiedLock(priv);
    name = xenStoreDomainGetName(conn, id);
    xenUnifiedUnlock(priv);
    if (!name)
2969
        return NULL;
2970 2971 2972 2973

    ret = virGetDomain(conn, name, XEN_GETDOMAININFO_UUID(dominfo));
    if (ret)
        ret->id = id;
2974
    VIR_FREE(name);
2975 2976 2977 2978 2979
    return ret;
}


virDomainPtr
2980
xenHypervisorLookupDomainByUUID(virConnectPtr conn, const unsigned char *uuid)
2981 2982
{
    xen_getdomaininfolist dominfos;
2983
    xenUnifiedPrivatePtr priv = conn->privateData;
2984 2985 2986 2987 2988
    virDomainPtr ret;
    char *name;
    int maxids = 100, nids, i, id;

    if (priv->handle < 0)
2989
        return NULL;
2990 2991 2992

 retry:
    if (!(XEN_GETDOMAININFOLIST_ALLOC(dominfos, maxids))) {
2993
        virReportOOMError();
2994
        return NULL;
2995 2996 2997 2998 2999 3000 3001 3002
    }

    XEN_GETDOMAININFOLIST_CLEAR(dominfos, maxids);

    nids = virXen_getdomaininfolist(priv->handle, 0, maxids, &dominfos);

    if (nids < 0) {
        XEN_GETDOMAININFOLIST_FREE(dominfos);
3003
        return NULL;
3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016
    }

    /* Can't possibly have more than 65,000 concurrent guests
     * so limit how many times we try, to avoid increasing
     * without bound & thus allocating all of system memory !
     * XXX I'll regret this comment in a few years time ;-)
     */
    if (nids == maxids) {
        XEN_GETDOMAININFOLIST_FREE(dominfos);
        if (maxids < 65000) {
            maxids *= 2;
            goto retry;
        }
3017
        return NULL;
3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029
    }

    id = -1;
    for (i = 0 ; i < nids ; i++) {
        if (memcmp(XEN_GETDOMAININFOLIST_UUID(dominfos, i), uuid, VIR_UUID_BUFLEN) == 0) {
            id = XEN_GETDOMAININFOLIST_DOMAIN(dominfos, i);
            break;
        }
    }
    XEN_GETDOMAININFOLIST_FREE(dominfos);

    if (id == -1)
3030
        return NULL;
3031

D
Daniel P. Berrange 已提交
3032 3033 3034 3035
    xenUnifiedLock(priv);
    name = xenStoreDomainGetName(conn, id);
    xenUnifiedUnlock(priv);
    if (!name)
3036
        return NULL;
3037 3038 3039 3040

    ret = virGetDomain(conn, name, uuid);
    if (ret)
        ret->id = id;
3041
    VIR_FREE(name);
3042 3043 3044
    return ret;
}

3045
/**
3046
 * xenHypervisorGetMaxVcpus:
3047 3048 3049 3050
 *
 * Returns the maximum of CPU defined by Xen.
 */
int
3051
xenHypervisorGetMaxVcpus(virConnectPtr conn, const char *type ATTRIBUTE_UNUSED)
3052
{
3053
    xenUnifiedPrivatePtr priv = conn->privateData;
3054 3055

    if (priv->handle < 0)
3056
        return -1;
3057 3058 3059 3060

    return MAX_VIRT_CPUS;
}

3061
/**
3062 3063 3064
 * xenHypervisorGetDomMaxMemory:
 * @conn: connection data
 * @id: domain id
3065
 *
3066
 * Retrieve the maximum amount of physical memory allocated to a
3067
 * domain.
3068 3069 3070
 *
 * Returns the memory size in kilobytes or 0 in case of error.
 */
3071 3072
unsigned long
xenHypervisorGetDomMaxMemory(virConnectPtr conn, int id)
3073
{
3074
    xenUnifiedPrivatePtr priv = conn->privateData;
3075
    xen_getdomaininfo dominfo;
3076 3077
    int ret;

3078 3079
    if (priv->handle < 0)
        return 0;
3080

3081 3082
    if (kb_per_pages == 0) {
        kb_per_pages = sysconf(_SC_PAGESIZE) / 1024;
3083 3084
        if (kb_per_pages <= 0)
            kb_per_pages = 4;
3085 3086
    }

3087
    XEN_GETDOMAININFO_CLEAR(dominfo);
3088

3089
    ret = virXen_getdomaininfo(priv->handle, id, &dominfo);
3090

3091
    if ((ret < 0) || (XEN_GETDOMAININFO_DOMAIN(dominfo) != id))
3092
        return 0;
3093

3094
    return (unsigned long) XEN_GETDOMAININFO_MAX_PAGES(dominfo) * kb_per_pages;
3095 3096
}

3097 3098 3099
/**
 * xenHypervisorGetMaxMemory:
 * @domain: a domain object or NULL
3100
 *
3101 3102 3103 3104 3105 3106
 * Retrieve the maximum amount of physical memory allocated to a
 * domain. If domain is NULL, then this get the amount of memory reserved
 * to Domain0 i.e. the domain where the application runs.
 *
 * Returns the memory size in kilobytes or 0 in case of error.
 */
3107
static unsigned long long ATTRIBUTE_NONNULL(1)
3108 3109
xenHypervisorGetMaxMemory(virDomainPtr domain)
{
3110
    xenUnifiedPrivatePtr priv = domain->conn->privateData;
3111 3112

    if (priv->handle < 0 || domain->id < 0)
3113
        return 0;
3114

3115
    return xenHypervisorGetDomMaxMemory(domain->conn, domain->id);
3116 3117
}

3118
/**
3119 3120 3121
 * xenHypervisorGetDomInfo:
 * @conn: connection data
 * @id: the domain ID
3122
 * @info: the place where information should be stored
3123
 *
E
Eric Blake 已提交
3124
 * Do a hypervisor call to get the related set of domain information.
3125 3126 3127 3128
 *
 * Returns 0 in case of success, -1 in case of error.
 */
int
3129
xenHypervisorGetDomInfo(virConnectPtr conn, int id, virDomainInfoPtr info)
3130
{
3131
    xenUnifiedPrivatePtr priv = conn->privateData;
3132
    xen_getdomaininfo dominfo;
3133
    int ret;
3134
    uint32_t domain_flags, domain_state, domain_shutdown_cause;
3135 3136 3137

    if (kb_per_pages == 0) {
        kb_per_pages = sysconf(_SC_PAGESIZE) / 1024;
3138 3139
        if (kb_per_pages <= 0)
            kb_per_pages = 4;
3140
    }
3141

3142
    if (priv->handle < 0)
3143
        return -1;
3144

3145
    memset(info, 0, sizeof(virDomainInfo));
3146
    XEN_GETDOMAININFO_CLEAR(dominfo);
3147

3148
    ret = virXen_getdomaininfo(priv->handle, id, &dominfo);
3149

3150
    if ((ret < 0) || (XEN_GETDOMAININFO_DOMAIN(dominfo) != id))
3151
        return -1;
3152

3153
    domain_flags = XEN_GETDOMAININFO_FLAGS(dominfo);
3154 3155
    domain_flags &= ~DOMFLAGS_HVM; /* Mask out HVM flags */
    domain_state = domain_flags & 0xFF; /* Mask out high bits */
3156
    switch (domain_state) {
3157 3158 3159 3160
        case DOMFLAGS_DYING:
            info->state = VIR_DOMAIN_SHUTDOWN;
            break;
        case DOMFLAGS_SHUTDOWN:
3161 3162 3163 3164 3165 3166 3167 3168 3169
            /* The domain is shutdown.  Determine the cause. */
            domain_shutdown_cause = domain_flags >> DOMFLAGS_SHUTDOWNSHIFT;
            switch (domain_shutdown_cause) {
                case SHUTDOWN_crash:
                    info->state = VIR_DOMAIN_CRASHED;
                    break;
                default:
                    info->state = VIR_DOMAIN_SHUTOFF;
            }
3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181
            break;
        case DOMFLAGS_PAUSED:
            info->state = VIR_DOMAIN_PAUSED;
            break;
        case DOMFLAGS_BLOCKED:
            info->state = VIR_DOMAIN_BLOCKED;
            break;
        case DOMFLAGS_RUNNING:
            info->state = VIR_DOMAIN_RUNNING;
            break;
        default:
            info->state = VIR_DOMAIN_NOSTATE;
3182 3183 3184 3185 3186 3187 3188
    }

    /*
     * the API brings back the cpu time in nanoseconds,
     * convert to microseconds, same thing convert to
     * kilobytes from page counts
     */
3189
    info->cpuTime = XEN_GETDOMAININFO_CPUTIME(dominfo);
3190
    info->memory = XEN_GETDOMAININFO_TOT_PAGES(dominfo) * kb_per_pages;
3191
    info->maxMem = XEN_GETDOMAININFO_MAX_PAGES(dominfo);
3192
    if (info->maxMem != UINT_MAX)
3193
        info->maxMem *= kb_per_pages;
3194
    info->nrVirtCpu = XEN_GETDOMAININFO_CPUCOUNT(dominfo);
3195
    return 0;
3196 3197
}

3198 3199 3200
/**
 * xenHypervisorGetDomainInfo:
 * @domain: pointer to the domain block
3201
 * @info: the place where information should be stored
3202
 *
E
Eric Blake 已提交
3203
 * Do a hypervisor call to get the related set of domain information.
3204 3205 3206 3207 3208 3209
 *
 * Returns 0 in case of success, -1 in case of error.
 */
int
xenHypervisorGetDomainInfo(virDomainPtr domain, virDomainInfoPtr info)
{
3210
    xenUnifiedPrivatePtr priv = domain->conn->privateData;
3211

3212
    if (priv->handle < 0 || domain->id < 0)
3213
        return -1;
3214

3215
    return xenHypervisorGetDomInfo(domain->conn, domain->id, info);
3216 3217 3218

}

3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233
/**
 * xenHypervisorGetDomainState:
 * @domain: pointer to the domain block
 * @state: returned state of the domain
 * @reason: returned reason for the state
 * @flags: additional flags, 0 for now
 *
 * Do a hypervisor call to get the related set of domain information.
 *
 * Returns 0 in case of success, -1 in case of error.
 */
int
xenHypervisorGetDomainState(virDomainPtr domain,
                            int *state,
                            int *reason,
E
Eric Blake 已提交
3234
                            unsigned int flags)
3235 3236 3237 3238
{
    xenUnifiedPrivatePtr priv = domain->conn->privateData;
    virDomainInfo info;

E
Eric Blake 已提交
3239 3240
    virCheckFlags(0, -1);

3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253
    if (priv->handle < 0 || domain->id < 0)
        return -1;

    if (xenHypervisorGetDomInfo(domain->conn, domain->id, &info) < 0)
        return -1;

    *state = info.state;
    if (reason)
        *reason = 0;

    return 0;
}

3254 3255 3256 3257
/**
 * xenHypervisorNodeGetCellsFreeMemory:
 * @conn: pointer to the hypervisor connection
 * @freeMems: pointer to the array of unsigned long long
3258 3259
 * @startCell: index of first cell to return freeMems info on.
 * @maxCells: Maximum number of cells for which freeMems information can
3260 3261 3262 3263
 *            be returned.
 *
 * This call returns the amount of free memory in one or more NUMA cells.
 * The @freeMems array must be allocated by the caller and will be filled
3264 3265 3266
 * with the amount of free memory in kilobytes for each cell requested,
 * starting with startCell (in freeMems[0]), up to either
 * (startCell + maxCells), or the number of additional cells in the node,
3267 3268 3269 3270 3271
 * whichever is smaller.
 *
 * Returns the number of entries filled in freeMems, or -1 in case of error.
 */
int
3272 3273 3274 3275
xenHypervisorNodeGetCellsFreeMemory(virConnectPtr conn,
                                    unsigned long long *freeMems,
                                    int startCell,
                                    int maxCells)
3276 3277 3278
{
    xen_op_v2_sys op_sys;
    int i, j, ret;
3279
    xenUnifiedPrivatePtr priv = conn->privateData;
D
Daniel P. Berrange 已提交
3280 3281

    if (priv->nbNodeCells < 0) {
3282 3283
        virReportError(VIR_ERR_XEN_CALL, "%s",
                       _("cannot determine actual number of cells"));
3284
        return -1;
3285 3286
    }

D
Daniel P. Berrange 已提交
3287
    if ((maxCells < 1) || (startCell >= priv->nbNodeCells)) {
3288 3289
        virReportError(VIR_ERR_INVALID_ARG, "%s",
                       _("invalid argument"));
3290 3291
        return -1;
    }
3292

3293
    /*
P
Philipp Hahn 已提交
3294
     * Support only hv_versions.sys_interface >=4
3295
     */
P
Philipp Hahn 已提交
3296
    if (hv_versions.sys_interface < SYS_IFACE_MIN_VERS_NUMA) {
3297 3298
        virReportError(VIR_ERR_XEN_CALL, "%s",
                       _("unsupported in sys interface < 4"));
3299 3300 3301 3302
        return -1;
    }

    if (priv->handle < 0) {
3303 3304
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("priv->handle invalid"));
3305 3306 3307 3308 3309 3310
        return -1;
    }

    memset(&op_sys, 0, sizeof(op_sys));
    op_sys.cmd = XEN_V2_OP_GETAVAILHEAP;

D
Daniel P. Berrange 已提交
3311
    for (i = startCell, j = 0;(i < priv->nbNodeCells) && (j < maxCells);i++,j++) {
P
Philipp Hahn 已提交
3312
        if (hv_versions.sys_interface >= 5)
3313 3314 3315
            op_sys.u.availheap5.node = i;
        else
            op_sys.u.availheap.node = i;
3316 3317
        ret = xenHypervisorDoV2Sys(priv->handle, &op_sys);
        if (ret < 0) {
3318
            return -1;
3319
        }
P
Philipp Hahn 已提交
3320
        if (hv_versions.sys_interface >= 5)
3321 3322 3323
            freeMems[j] = op_sys.u.availheap5.avail_bytes;
        else
            freeMems[j] = op_sys.u.availheap.avail_bytes;
3324
    }
3325
    return j;
3326 3327 3328
}


3329 3330
/**
 * xenHypervisorPauseDomain:
3331
 * @domain: pointer to the domain block
3332
 *
E
Eric Blake 已提交
3333
 * Do a hypervisor call to pause the given domain
3334 3335 3336 3337
 *
 * Returns 0 in case of success, -1 in case of error.
 */
int
3338
xenHypervisorPauseDomain(virDomainPtr domain)
3339
{
3340
    int ret;
3341
    xenUnifiedPrivatePtr priv = domain->conn->privateData;
3342 3343

    if (priv->handle < 0 || domain->id < 0)
3344
        return -1;
3345

3346
    ret = virXen_pausedomain(priv->handle, domain->id);
3347
    if (ret < 0)
3348 3349
        return -1;
    return 0;
3350 3351 3352 3353
}

/**
 * xenHypervisorResumeDomain:
3354
 * @domain: pointer to the domain block
3355
 *
E
Eric Blake 已提交
3356
 * Do a hypervisor call to resume the given domain
3357 3358 3359 3360
 *
 * Returns 0 in case of success, -1 in case of error.
 */
int
3361
xenHypervisorResumeDomain(virDomainPtr domain)
3362
{
3363
    int ret;
3364
    xenUnifiedPrivatePtr priv = domain->conn->privateData;
3365

3366
    if (priv->handle < 0 || domain->id < 0)
3367
        return -1;
3368

3369
    ret = virXen_unpausedomain(priv->handle, domain->id);
3370
    if (ret < 0)
3371 3372
        return -1;
    return 0;
3373 3374 3375
}

/**
3376
 * xenHypervisorDestroyDomainFlags:
3377
 * @domain: pointer to the domain block
3378
 * @flags: an OR'ed set of virDomainDestroyFlagsValues
3379
 *
E
Eric Blake 已提交
3380
 * Do a hypervisor call to destroy the given domain
3381
 *
3382 3383 3384
 * Calling this function with no @flags set (equal to zero)
 * is equivalent to calling xenHypervisorDestroyDomain.
 *
3385 3386 3387
 * Returns 0 in case of success, -1 in case of error.
 */
int
3388
xenHypervisorDestroyDomainFlags(virDomainPtr domain, unsigned int flags)
3389
{
3390
    int ret;
3391
    xenUnifiedPrivatePtr priv = domain->conn->privateData;
3392

3393 3394
    virCheckFlags(0, -1);

3395
    if (priv->handle < 0 || domain->id < 0)
3396
        return -1;
3397

3398
    ret = virXen_destroydomain(priv->handle, domain->id);
3399
    if (ret < 0)
3400 3401
        return -1;
    return 0;
3402 3403
}

3404 3405
/**
 * xenHypervisorSetMaxMemory:
3406
 * @domain: pointer to the domain block
3407 3408
 * @memory: the max memory size in kilobytes.
 *
E
Eric Blake 已提交
3409
 * Do a hypervisor call to change the maximum amount of memory used
3410 3411 3412 3413
 *
 * Returns 0 in case of success, -1 in case of error.
 */
int
3414
xenHypervisorSetMaxMemory(virDomainPtr domain, unsigned long memory)
3415
{
3416
    int ret;
3417
    xenUnifiedPrivatePtr priv = domain->conn->privateData;
3418 3419

    if (priv->handle < 0 || domain->id < 0)
3420
        return -1;
3421

3422
    ret = virXen_setmaxmem(priv->handle, domain->id, memory);
3423
    if (ret < 0)
3424 3425
        return -1;
    return 0;
3426
}
3427

3428

3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441
/**
 * xenHypervisorSetVcpus:
 * @domain: pointer to domain object
 * @nvcpus: the new number of virtual CPUs for this domain
 *
 * Dynamically change the number of virtual CPUs used by the domain.
 *
 * Returns 0 in case of success, -1 in case of failure.
 */

int
xenHypervisorSetVcpus(virDomainPtr domain, unsigned int nvcpus)
{
3442
    int ret;
3443
    xenUnifiedPrivatePtr priv = domain->conn->privateData;
3444

3445
    if (priv->handle < 0 || domain->id < 0 || nvcpus < 1)
3446
        return -1;
3447

3448
    ret = virXen_setmaxvcpus(priv->handle, domain->id, nvcpus);
3449
    if (ret < 0)
3450 3451
        return -1;
    return 0;
3452 3453 3454 3455 3456 3457 3458 3459
}

/**
 * xenHypervisorPinVcpu:
 * @domain: pointer to domain object
 * @vcpu: virtual CPU number
 * @cpumap: pointer to a bit map of real CPUs (in 8-bit bytes)
 * @maplen: length of cpumap in bytes
3460
 *
3461 3462 3463 3464 3465 3466 3467 3468 3469
 * Dynamically change the real CPUs which can be allocated to a virtual CPU.
 *
 * Returns 0 in case of success, -1 in case of failure.
 */

int
xenHypervisorPinVcpu(virDomainPtr domain, unsigned int vcpu,
                     unsigned char *cpumap, int maplen)
{
3470
    int ret;
3471
    xenUnifiedPrivatePtr priv = domain->conn->privateData;
3472

3473
    if (priv->handle < 0 || domain->id < 0)
3474
        return -1;
3475

3476
    ret = virXen_setvcpumap(priv->handle, domain->id, vcpu,
3477 3478
                            cpumap, maplen);
    if (ret < 0)
3479 3480
        return -1;
    return 0;
3481 3482 3483 3484 3485 3486 3487
}

/**
 * virDomainGetVcpus:
 * @domain: pointer to domain object, or NULL for Domain0
 * @info: pointer to an array of virVcpuInfo structures (OUT)
 * @maxinfo: number of structures in info array
E
Eric Blake 已提交
3488
 * @cpumaps: pointer to a bit map of real CPUs for all vcpus of this domain (in 8-bit bytes) (OUT)
D
Daniel Veillard 已提交
3489
 *	If cpumaps is NULL, then no cpumap information is returned by the API.
3490 3491 3492 3493 3494 3495
 *	It's assumed there is <maxinfo> cpumap in cpumaps array.
 *	The memory allocated to cpumaps must be (maxinfo * maplen) bytes
 *	(ie: calloc(maxinfo, maplen)).
 *	One cpumap inside cpumaps has the format described in virDomainPinVcpu() API.
 * @maplen: number of bytes in one cpumap, from 1 up to size of CPU map in
 *	underlying virtualization system (Xen...).
3496
 *
3497
 * Extract information about virtual CPUs of domain, store it in info array
R
Richard W.M. Jones 已提交
3498
 * and also in cpumaps if this pointer isn't NULL.
3499 3500 3501 3502
 *
 * Returns the number of info filled in case of success, -1 in case of failure.
 */
int
3503 3504 3505 3506 3507
xenHypervisorGetVcpus(virDomainPtr domain,
                      virVcpuInfoPtr info,
                      int maxinfo,
                      unsigned char *cpumaps,
                      int maplen)
3508
{
3509
    xen_getdomaininfo dominfo;
3510
    int ret;
3511
    xenUnifiedPrivatePtr priv = domain->conn->privateData;
3512
    virVcpuInfoPtr ipt;
3513
    int nbinfo, i;
3514

3515
    if (priv->handle < 0 || domain->id < 0 || sizeof(cpumap_t) & 7) {
3516 3517
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("domain shut off or invalid"));
3518
        return -1;
3519
    }
3520

3521
    /* first get the number of virtual CPUs in this domain */
3522
    XEN_GETDOMAININFO_CLEAR(dominfo);
3523
    ret = virXen_getdomaininfo(priv->handle, domain->id,
3524
                               &dominfo);
3525

3526
    if ((ret < 0) || (XEN_GETDOMAININFO_DOMAIN(dominfo) != domain->id)) {
3527 3528
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("cannot get domain details"));
3529
        return -1;
3530
    }
3531
    nbinfo = XEN_GETDOMAININFO_CPUCOUNT(dominfo) + 1;
3532 3533 3534
    if (nbinfo > maxinfo) nbinfo = maxinfo;

    if (cpumaps != NULL)
3535
        memset(cpumaps, 0, maxinfo * maplen);
3536

3537 3538
    for (i = 0, ipt = info; i < nbinfo; i++, ipt++) {
        if ((cpumaps != NULL) && (i < maxinfo)) {
3539
            ret = virXen_getvcpusinfo(priv->handle, domain->id, i,
3540 3541 3542
                                      ipt,
                                      (unsigned char *)VIR_GET_CPUMAP(cpumaps, maplen, i),
                                      maplen);
3543
            if (ret < 0) {
3544 3545
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("cannot get VCPUs info"));
3546
                return -1;
3547
            }
3548
        } else {
3549
            ret = virXen_getvcpusinfo(priv->handle, domain->id, i,
3550
                                      ipt, NULL, 0);
3551
            if (ret < 0) {
3552 3553
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("cannot get VCPUs info"));
3554
                return -1;
3555
            }
3556
        }
3557 3558 3559
    }
    return nbinfo;
}
3560

3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574
/**
 * xenHypervisorGetVcpuMax:
 *
 *  Returns the maximum number of virtual CPUs supported for
 *  the guest VM. If the guest is inactive, this is the maximum
 *  of CPU defined by Xen. If the guest is running this reflect
 *  the maximum number of virtual CPUs the guest was booted with.
 */
int
xenHypervisorGetVcpuMax(virDomainPtr domain)
{
    xen_getdomaininfo dominfo;
    int ret;
    int maxcpu;
3575
    xenUnifiedPrivatePtr priv = domain->conn->privateData;
3576

3577
    if (priv->handle < 0)
3578
        return -1;
3579 3580 3581 3582 3583 3584

    /* inactive domain */
    if (domain->id < 0) {
        maxcpu = MAX_VIRT_CPUS;
    } else {
        XEN_GETDOMAININFO_CLEAR(dominfo);
3585
        ret = virXen_getdomaininfo(priv->handle, domain->id,
3586 3587 3588
                                   &dominfo);

        if ((ret < 0) || (XEN_GETDOMAININFO_DOMAIN(dominfo) != domain->id))
3589
            return -1;
3590 3591 3592 3593 3594 3595
        maxcpu = XEN_GETDOMAININFO_MAXCPUID(dominfo) + 1;
    }

    return maxcpu;
}

J
John Levon 已提交
3596 3597 3598 3599 3600 3601
/**
 * xenHavePrivilege()
 *
 * Return true if the current process should be able to connect to Xen.
 */
int
3602
xenHavePrivilege(void)
J
John Levon 已提交
3603 3604
{
#ifdef __sun
3605
    return priv_ineffect(PRIV_XVM_CONTROL);
J
John Levon 已提交
3606
#else
3607
    return access(XEN_HYPERVISOR_SOCKET, R_OK) == 0;
J
John Levon 已提交
3608 3609
#endif
}