storage_backend.c 51.2 KB
Newer Older
1
/*
2
 * storage_backend.c: internal storage driver backend contract
3
 *
4
 * Copyright (C) 2007-2013 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2007-2008 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24 25 26
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

#include <config.h>

#include <string.h>
27
#include <stdio.h>
28
#include <regex.h>
29
#include <sys/types.h>
E
Eric Blake 已提交
30
#include <sys/wait.h>
31
#include <unistd.h>
32 33
#include <fcntl.h>
#include <sys/stat.h>
34
#include <sys/param.h>
35
#include <dirent.h>
36
#include "dirname.h"
37 38 39 40
#ifdef __linux__
# include <sys/ioctl.h>
# include <linux/fs.h>
#endif
41

42
#if WITH_SELINUX
43
# include <selinux/selinux.h>
44
#endif
45

46
#include "datatypes.h"
47
#include "virerror.h"
48
#include "viralloc.h"
49
#include "internal.h"
50
#include "secret_conf.h"
51
#include "viruuid.h"
52
#include "virstoragefile.h"
D
Daniel P. Berrange 已提交
53
#include "storage_backend.h"
54
#include "virlog.h"
E
Eric Blake 已提交
55
#include "virfile.h"
56
#include "stat-time.h"
57
#include "virstring.h"
58

59
#if WITH_STORAGE_LVM
60
# include "storage_backend_logical.h"
61 62
#endif
#if WITH_STORAGE_ISCSI
63
# include "storage_backend_iscsi.h"
64
#endif
65
#if WITH_STORAGE_SCSI
66
# include "storage_backend_scsi.h"
67
#endif
D
Dave Allan 已提交
68
#if WITH_STORAGE_MPATH
69
# include "storage_backend_mpath.h"
D
Dave Allan 已提交
70
#endif
71
#if WITH_STORAGE_DISK
72
# include "storage_backend_disk.h"
73 74
#endif
#if WITH_STORAGE_DIR
75
# include "storage_backend_fs.h"
76
#endif
77 78 79
#if WITH_STORAGE_RBD
# include "storage_backend_rbd.h"
#endif
80 81 82
#if WITH_STORAGE_SHEEPDOG
# include "storage_backend_sheepdog.h"
#endif
83

84 85
#define VIR_FROM_THIS VIR_FROM_STORAGE

86 87 88 89 90 91 92 93 94 95 96 97 98 99
static virStorageBackendPtr backends[] = {
#if WITH_STORAGE_DIR
    &virStorageBackendDirectory,
#endif
#if WITH_STORAGE_FS
    &virStorageBackendFileSystem,
    &virStorageBackendNetFileSystem,
#endif
#if WITH_STORAGE_LVM
    &virStorageBackendLogical,
#endif
#if WITH_STORAGE_ISCSI
    &virStorageBackendISCSI,
#endif
100 101 102
#if WITH_STORAGE_SCSI
    &virStorageBackendSCSI,
#endif
D
Dave Allan 已提交
103 104 105
#if WITH_STORAGE_MPATH
    &virStorageBackendMpath,
#endif
106 107
#if WITH_STORAGE_DISK
    &virStorageBackendDisk,
108 109 110
#endif
#if WITH_STORAGE_RBD
    &virStorageBackendRBD,
111 112 113
#endif
#if WITH_STORAGE_SHEEPDOG
    &virStorageBackendSheepdog,
114 115 116
#endif
    NULL
};
117

118 119 120 121 122 123
enum {
    TOOL_QEMU_IMG,
    TOOL_KVM_IMG,
    TOOL_QCOW_CREATE,
};

124 125 126
#define READ_BLOCK_SIZE_DEFAULT  (1024 * 1024)
#define WRITE_BLOCK_SIZE_DEFAULT (4 * 1024)

127
static int ATTRIBUTE_NONNULL(2)
128
virStorageBackendCopyToFD(virStorageVolDefPtr vol,
129 130
                          virStorageVolDefPtr inputvol,
                          int fd,
131 132
                          unsigned long long *total,
                          int is_dest_file)
133 134 135
{
    int inputfd = -1;
    int amtread = -1;
136
    int ret = 0;
137 138 139
    size_t rbytes = READ_BLOCK_SIZE_DEFAULT;
    size_t wbytes = 0;
    int interval;
140
    char *zerobuf = NULL;
141
    char *buf = NULL;
142
    struct stat st;
143

144
    if ((inputfd = open(inputvol->target.path, O_RDONLY)) < 0) {
145
        ret = -errno;
146
        virReportSystemError(errno,
147 148 149
                             _("could not open input path '%s'"),
                             inputvol->target.path);
        goto cleanup;
150 151
    }

152 153 154 155 156 157 158 159 160 161 162 163 164 165 166
#ifdef __linux__
    if (ioctl(fd, BLKBSZGET, &wbytes) < 0) {
        wbytes = 0;
    }
#endif
    if ((wbytes == 0) && fstat(fd, &st) == 0)
        wbytes = st.st_blksize;
    if (wbytes < WRITE_BLOCK_SIZE_DEFAULT)
        wbytes = WRITE_BLOCK_SIZE_DEFAULT;

    if (VIR_ALLOC_N(zerobuf, wbytes) < 0) {
        ret = -errno;
        virReportOOMError();
        goto cleanup;
    }
167

168
    if (VIR_ALLOC_N(buf, rbytes) < 0) {
169
        ret = -errno;
170
        virReportOOMError();
171 172 173 174 175 176
        goto cleanup;
    }

    while (amtread != 0) {
        int amtleft;

177 178
        if (*total < rbytes)
            rbytes = *total;
179

180
        if ((amtread = saferead(inputfd, buf, rbytes)) < 0) {
181
            ret = -errno;
182
            virReportSystemError(errno,
183 184 185 186
                                 _("failed reading from file '%s'"),
                                 inputvol->target.path);
            goto cleanup;
        }
187
        *total -= amtread;
188 189 190 191 192

        /* Loop over amt read in 512 byte increments, looking for sparse
         * blocks */
        amtleft = amtread;
        do {
193
            interval = ((wbytes > amtleft) ? amtleft : wbytes);
194 195
            int offset = amtread - amtleft;

196
            if (is_dest_file && memcmp(buf+offset, zerobuf, interval) == 0) {
197
                if (lseek(fd, interval, SEEK_CUR) < 0) {
198
                    ret = -errno;
199
                    virReportSystemError(errno,
200 201 202 203 204
                                         _("cannot extend file '%s'"),
                                         vol->target.path);
                    goto cleanup;
                }
            } else if (safewrite(fd, buf+offset, interval) < 0) {
205
                ret = -errno;
206
                virReportSystemError(errno,
207 208 209 210 211
                                     _("failed writing to file '%s'"),
                                     vol->target.path);
                goto cleanup;

            }
212
        } while ((amtleft -= interval) > 0);
213 214
    }

215 216 217 218 219 220 221 222
    if (fdatasync(fd) < 0) {
        ret = -errno;
        virReportSystemError(errno, _("cannot sync data to file '%s'"),
                             vol->target.path);
        goto cleanup;
    }


223
    if (VIR_CLOSE(inputfd) < 0) {
224
        ret = -errno;
225
        virReportSystemError(errno,
226 227 228 229 230 231 232
                             _("cannot close file '%s'"),
                             inputvol->target.path);
        goto cleanup;
    }
    inputfd = -1;

cleanup:
233
    VIR_FORCE_CLOSE(inputfd);
234

235
    VIR_FREE(zerobuf);
236 237
    VIR_FREE(buf);

238 239 240
    return ret;
}

241
static int
242
virStorageBackendCreateBlockFrom(virConnectPtr conn ATTRIBUTE_UNUSED,
243
                                 virStoragePoolObjPtr pool ATTRIBUTE_UNUSED,
244 245
                                 virStorageVolDefPtr vol,
                                 virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
246
                                 unsigned int flags)
247 248 249 250
{
    int fd = -1;
    int ret = -1;
    unsigned long long remain;
251 252 253
    struct stat st;
    gid_t gid;
    uid_t uid;
254

255 256 257 258 259 260 261 262
    virCheckFlags(VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA, -1);

    if (flags & VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("metadata preallocation is not supported for block "
                         "volumes"));
        goto cleanup;
    }
E
Eric Blake 已提交
263

264
    if ((fd = open(vol->target.path, O_RDWR)) < 0) {
265
        virReportSystemError(errno,
266 267 268 269 270 271 272 273
                             _("cannot create path '%s'"),
                             vol->target.path);
        goto cleanup;
    }

    remain = vol->allocation;

    if (inputvol) {
274
        int res = virStorageBackendCopyToFD(vol, inputvol,
275
                                            fd, &remain, 0);
276 277 278 279
        if (res < 0)
            goto cleanup;
    }

280
    if (fstat(fd, &st) == -1) {
281
        virReportSystemError(errno, _("stat of '%s' failed"),
282 283 284
                             vol->target.path);
        goto cleanup;
    }
P
Philipp Hahn 已提交
285 286 287
    uid = (vol->target.perms.uid != st.st_uid) ? vol->target.perms.uid : (uid_t) -1;
    gid = (vol->target.perms.gid != st.st_gid) ? vol->target.perms.gid : (gid_t) -1;
    if (((uid != (uid_t) -1) || (gid != (gid_t) -1))
288
        && (fchown(fd, uid, gid) < 0)) {
289
        virReportSystemError(errno,
290
                             _("cannot chown '%s' to (%u, %u)"),
291 292
                             vol->target.path, (unsigned int) uid,
                             (unsigned int) gid);
293 294 295
        goto cleanup;
    }
    if (fchmod(fd, vol->target.perms.mode) < 0) {
296
        virReportSystemError(errno,
297 298 299 300
                             _("cannot set mode of '%s' to %04o"),
                             vol->target.path, vol->target.perms.mode);
        goto cleanup;
    }
301
    if (VIR_CLOSE(fd) < 0) {
302
        virReportSystemError(errno,
303 304 305 306 307 308 309 310
                             _("cannot close file '%s'"),
                             vol->target.path);
        goto cleanup;
    }
    fd = -1;

    ret = 0;
cleanup:
311
    VIR_FORCE_CLOSE(fd);
312 313 314 315

    return ret;
}

E
Eric Blake 已提交
316 317 318 319
static int
createRawFile(int fd, virStorageVolDefPtr vol,
              virStorageVolDefPtr inputvol)
{
320 321
    int ret = 0;
    unsigned long long remain;
322

323 324
    /* Seek to the final size, so the capacity is available upfront
     * for progress reporting */
E
Eric Blake 已提交
325
    if (ftruncate(fd, vol->capacity) < 0) {
326
        ret = -errno;
327
        virReportSystemError(errno,
328
                             _("cannot extend file '%s'"),
E
Eric Blake 已提交
329
                             vol->target.path);
330 331 332
        goto cleanup;
    }

E
Eric Blake 已提交
333
    remain = vol->allocation;
334

E
Eric Blake 已提交
335 336
    if (inputvol) {
        ret = virStorageBackendCopyToFD(vol, inputvol, fd, &remain, 1);
337
        if (ret < 0) {
338
            goto cleanup;
339
        }
340 341 342
    }

    if (remain) {
343 344 345 346 347
        if (safezero(fd, 0, remain) < 0) {
            ret = -errno;
            virReportSystemError(errno, _("cannot fill file '%s'"),
                                 vol->target.path);
            goto cleanup;
348
        }
349 350 351
    }

    if (fsync(fd) < 0) {
352
        ret = -errno;
353
        virReportSystemError(errno, _("cannot sync data to file '%s'"),
E
Eric Blake 已提交
354
                             vol->target.path);
355
        goto cleanup;
356 357
    }

358 359 360 361 362 363 364 365 366
cleanup:
    return ret;
}

int
virStorageBackendCreateRaw(virConnectPtr conn ATTRIBUTE_UNUSED,
                           virStoragePoolObjPtr pool,
                           virStorageVolDefPtr vol,
                           virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
367
                           unsigned int flags)
368 369
{
    int ret = -1;
E
Eric Blake 已提交
370 371 372
    int fd = -1;
    int operation_flags;

373 374 375 376 377 378 379 380
    virCheckFlags(VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA, -1);

    if (flags & VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("metadata preallocation is not supported for raw "
                         "volumes"));
        goto cleanup;
    }
381 382

    if (vol->target.encryption != NULL) {
383 384
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("storage pool does not support encrypted volumes"));
385 386
        goto cleanup;
    }
387

L
Laine Stump 已提交
388
    operation_flags = VIR_FILE_OPEN_FORCE_MODE | VIR_FILE_OPEN_FORCE_OWNER;
E
Eric Blake 已提交
389
    if (pool->def->type == VIR_STORAGE_POOL_NETFS)
L
Laine Stump 已提交
390
        operation_flags |= VIR_FILE_OPEN_FORK;
E
Eric Blake 已提交
391

392 393
    if ((fd = virFileOpenAs(vol->target.path,
                            O_RDWR | O_CREAT | O_EXCL,
L
Laine Stump 已提交
394 395 396
                            vol->target.perms.mode,
                            vol->target.perms.uid,
                            vol->target.perms.gid,
397
                            operation_flags)) < 0) {
E
Eric Blake 已提交
398
        virReportSystemError(-fd,
399
                             _("Failed to create file '%s'"),
E
Eric Blake 已提交
400 401 402 403
                             vol->target.path);
        goto cleanup;
    }

404 405
    if ((ret = createRawFile(fd, vol, inputvol)) < 0)
        /* createRawFile already reported the exact error. */
E
Eric Blake 已提交
406
        ret = -1;
407 408

cleanup:
E
Eric Blake 已提交
409
    VIR_FORCE_CLOSE(fd);
410 411 412
    return ret;
}

413 414 415 416 417 418 419 420 421
static int
virStorageGenerateSecretUUID(virConnectPtr conn,
                             unsigned char *uuid)
{
    unsigned attempt;

    for (attempt = 0; attempt < 65536; attempt++) {
        virSecretPtr tmp;
        if (virUUIDGenerate(uuid) < 0) {
422 423
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("unable to generate uuid"));
424 425
            return -1;
        }
426
        tmp = conn->secretDriver->secretLookupByUUID(conn, uuid);
427 428 429 430 431 432
        if (tmp == NULL)
            return 0;

        virSecretFree(tmp);
    }

433 434
    virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                   _("too many conflicts when generating an uuid"));
435 436 437 438

    return -1;
}

439 440 441 442 443 444 445 446 447
static int
virStorageGenerateQcowEncryption(virConnectPtr conn,
                                 virStorageVolDefPtr vol)
{
    virSecretDefPtr def = NULL;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    virStorageEncryptionPtr enc;
    virStorageEncryptionSecretPtr enc_secret = NULL;
    virSecretPtr secret = NULL;
448
    char *xml;
449 450 451
    unsigned char value[VIR_STORAGE_QCOW_PASSPHRASE_SIZE];
    int ret = -1;

452
    if (conn->secretDriver == NULL ||
453 454 455
        conn->secretDriver->secretLookupByUUID == NULL ||
        conn->secretDriver->secretDefineXML == NULL ||
        conn->secretDriver->secretSetValue == NULL) {
456 457
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("secret storage not supported"));
458 459 460 461 462
        goto cleanup;
    }

    enc = vol->target.encryption;
    if (enc->nsecrets != 0) {
463 464
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("secrets already defined"));
465 466 467 468 469
        goto cleanup;
    }

    if (VIR_ALLOC(enc_secret) < 0 || VIR_REALLOC_N(enc->secrets, 1) < 0 ||
        VIR_ALLOC(def) < 0) {
470
        virReportOOMError();
471 472 473
        goto cleanup;
    }

474
    def->ephemeral = false;
475
    def->private = false;
476
    if (virStorageGenerateSecretUUID(conn, def->uuid) < 0)
477
        goto cleanup;
478

479
    def->usage_type = VIR_SECRET_USAGE_TYPE_VOLUME;
480
    if (VIR_STRDUP(def->usage.volume, vol->target.path) < 0)
481
        goto cleanup;
482
    xml = virSecretDefFormat(def);
483 484 485 486 487
    virSecretDefFree(def);
    def = NULL;
    if (xml == NULL)
        goto cleanup;

488
    secret = conn->secretDriver->secretDefineXML(conn, xml, 0);
489 490 491 492 493 494
    if (secret == NULL) {
        VIR_FREE(xml);
        goto cleanup;
    }
    VIR_FREE(xml);

495
    if (virStorageGenerateQcowPassphrase(value) < 0)
496 497
        goto cleanup;

498
    if (conn->secretDriver->secretSetValue(secret, value, sizeof(value), 0) < 0)
499 500 501
        goto cleanup;

    enc_secret->type = VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE;
502
    memcpy(enc_secret->uuid, secret->uuid, VIR_UUID_BUFLEN);
503 504 505 506 507 508 509 510 511
    enc->format = VIR_STORAGE_ENCRYPTION_FORMAT_QCOW;
    enc->secrets[0] = enc_secret; /* Space for secrets[0] allocated above */
    enc_secret = NULL;
    enc->nsecrets = 1;

    ret = 0;

cleanup:
    if (secret != NULL) {
512
        if (ret != 0 &&
513 514
            conn->secretDriver->secretUndefine != NULL)
            conn->secretDriver->secretUndefine(secret);
515 516
        virSecretFree(secret);
    }
517
    virBufferFreeAndReset(&buf);
518 519 520 521 522
    virSecretDefFree(def);
    VIR_FREE(enc_secret);
    return ret;
}

523 524 525 526 527
static int
virStorageBackendCreateExecCommand(virStoragePoolObjPtr pool,
                                   virStorageVolDefPtr vol,
                                   virCommandPtr cmd)
{
528 529 530
    struct stat st;
    gid_t gid;
    uid_t uid;
531
    bool filecreated = false;
532 533

    if ((pool->def->type == VIR_STORAGE_POOL_NETFS)
534
        && (((getuid() == 0)
P
Philipp Hahn 已提交
535
             && (vol->target.perms.uid != (uid_t) -1)
536
             && (vol->target.perms.uid != 0))
P
Philipp Hahn 已提交
537
            || ((vol->target.perms.gid != (gid_t) -1)
538
                && (vol->target.perms.gid != getgid())))) {
539

540 541
        virCommandSetUID(cmd, vol->target.perms.uid);
        virCommandSetGID(cmd, vol->target.perms.gid);
542 543

        if (virCommandRun(cmd, NULL) == 0) {
544 545
            /* command was successfully run, check if the file was created */
            if (stat(vol->target.path, &st) >=0)
546
                filecreated = true;
547 548
        }
    }
549

550 551 552
    /* don't change uid/gid if we retry */
    virCommandSetUID(cmd, -1);
    virCommandSetGID(cmd, -1);
553

554
    if (!filecreated) {
555
        if (virCommandRun(cmd, NULL) < 0) {
556 557 558
            return -1;
        }
        if (stat(vol->target.path, &st) < 0) {
559
            virReportSystemError(errno,
560
                                 _("failed to create %s"), vol->target.path);
561 562 563 564
            return -1;
        }
    }

P
Philipp Hahn 已提交
565 566 567
    uid = (vol->target.perms.uid != st.st_uid) ? vol->target.perms.uid : (uid_t) -1;
    gid = (vol->target.perms.gid != st.st_gid) ? vol->target.perms.gid : (gid_t) -1;
    if (((uid != (uid_t) -1) || (gid != (gid_t) -1))
568
        && (chown(vol->target.path, uid, gid) < 0)) {
569
        virReportSystemError(errno,
570
                             _("cannot chown %s to (%u, %u)"),
571 572
                             vol->target.path, (unsigned int) uid,
                             (unsigned int) gid);
573 574 575
        return -1;
    }
    if (chmod(vol->target.path, vol->target.perms.mode) < 0) {
576
        virReportSystemError(errno,
577 578 579 580 581 582 583
                             _("cannot set mode of '%s' to %04o"),
                             vol->target.path, vol->target.perms.mode);
        return -1;
    }
    return 0;
}

584 585 586 587 588 589
enum {
    QEMU_IMG_BACKING_FORMAT_NONE = 0,
    QEMU_IMG_BACKING_FORMAT_FLAG,
    QEMU_IMG_BACKING_FORMAT_OPTIONS,
};

590 591
static int
virStorageBackendQEMUImgBackingFormat(const char *qemuimg)
592 593 594 595 596 597
{
    char *help = NULL;
    char *start;
    char *end;
    char *tmp;
    int ret = -1;
598
    int exitstatus;
599
    virCommandPtr cmd = virCommandNewArgList(qemuimg, "-h", NULL);
600

601 602 603
    virCommandAddEnvString(cmd, "LC_ALL=C");
    virCommandSetOutputBuffer(cmd, &help);
    virCommandClearCaps(cmd);
604

605 606 607
    /* qemuimg doesn't return zero exit status on -h,
     * therefore we need to provide pointer for storing
     * exit status, although we don't parse it any later */
608
    if (virCommandRun(cmd, &exitstatus) < 0)
609 610
        goto cleanup;

611 612
    if ((start = strstr(help, " create ")) == NULL ||
        (end = strstr(start, "\n")) == NULL) {
613 614 615
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unable to parse qemu-img output '%s'"),
                       help);
616 617
        goto cleanup;
    }
618 619
    if (((tmp = strstr(start, "-F fmt")) && tmp < end) ||
        ((tmp = strstr(start, "-F backing_fmt")) && tmp < end))
620 621 622 623 624 625 626
        ret = QEMU_IMG_BACKING_FORMAT_FLAG;
    else if ((tmp = strstr(start, "[-o options]")) && tmp < end)
        ret = QEMU_IMG_BACKING_FORMAT_OPTIONS;
    else
        ret = QEMU_IMG_BACKING_FORMAT_NONE;

cleanup:
627
    virCommandFree(cmd);
628 629 630 631
    VIR_FREE(help);
    return ret;
}

632 633 634 635 636 637 638 639
virCommandPtr
virStorageBackendCreateQemuImgCmd(virConnectPtr conn,
                                  virStoragePoolObjPtr pool,
                                  virStorageVolDefPtr vol,
                                  virStorageVolDefPtr inputvol,
                                  unsigned int flags,
                                  const char *create_tool,
                                  int imgformat)
640
{
641 642
    virCommandPtr cmd = NULL;
    bool do_encryption = (vol->target.encryption != NULL);
643
    unsigned long long int size_arg;
644 645 646 647 648
    bool preallocate = !!(flags & VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA);
    const char *type;
    const char *backingType = NULL;
    const char *inputPath = NULL;
    const char *inputType = NULL;
649

650 651
    virCheckFlags(VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA, NULL);

652 653 654 655
    /* Treat output block devices as 'raw' format */
    type = virStorageFileFormatTypeToString(vol->type == VIR_STORAGE_VOL_BLOCK ?
                                            VIR_STORAGE_FILE_RAW :
                                            vol->target.format);
656

657
    if (!type) {
658 659 660
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unknown storage vol type %d"),
                       vol->target.format);
661
        return NULL;
662
    }
663

664 665 666
    if (preallocate && vol->target.format != VIR_STORAGE_FILE_QCOW2) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("metadata preallocation only available with qcow2"));
667
        return NULL;
668
    }
669

670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689
    if (inputvol) {
        if (!(inputPath = inputvol->target.path)) {
            virReportError(VIR_ERR_INVALID_ARG, "%s",
                           _("missing input volume target path"));
            return NULL;
        }

        inputType = virStorageFileFormatTypeToString(inputvol->type == VIR_STORAGE_VOL_BLOCK ?
                                                     VIR_STORAGE_FILE_RAW :
                                                     inputvol->target.format);

        if (!inputType) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unknown storage vol type %d"),
                           inputvol->target.format);
            return NULL;
        }

    }

690
    if (vol->backingStore.path) {
691 692
        int accessRetCode = -1;
        char *absolutePath = NULL;
693

694 695
        backingType = virStorageFileFormatTypeToString(vol->backingStore.format);

696 697 698 699
        if (preallocate) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("metadata preallocation conflicts with backing"
                             " store"));
700
            return NULL;
701 702
        }

703 704 705 706 707
        /* XXX: Not strictly required: qemu-img has an option a different
         * backing store, not really sure what use it serves though, and it
         * may cause issues with lvm. Untested essentially.
         */
        if (inputvol &&
708 709 710
            STRNEQ_NULLABLE(inputvol->backingStore.path, vol->backingStore.path)) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("a different backing store cannot be specified."));
711
            return NULL;
712 713 714
        }

        if (backingType == NULL) {
715 716 717
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unknown storage vol backing store type %d"),
                           vol->backingStore.format);
718
            return NULL;
719
        }
720 721 722 723 724 725 726 727

        /* Convert relative backing store paths to absolute paths for access
         * validation.
         */
        if ('/' != *(vol->backingStore.path) &&
            virAsprintf(&absolutePath, "%s/%s", pool->def->target.path,
                        vol->backingStore.path) < 0) {
            virReportOOMError();
728
            return NULL;
729 730 731 732 733
        }
        accessRetCode = access(absolutePath ? absolutePath
                               : vol->backingStore.path, R_OK);
        VIR_FREE(absolutePath);
        if (accessRetCode != 0) {
734
            virReportSystemError(errno,
735 736
                                 _("inaccessible backing store volume %s"),
                                 vol->backingStore.path);
737
            return NULL;
738 739 740
        }
    }

741
    if (do_encryption) {
742 743
        virStorageEncryptionPtr enc;

744 745
        if (vol->target.format != VIR_STORAGE_FILE_QCOW &&
            vol->target.format != VIR_STORAGE_FILE_QCOW2) {
746 747 748
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("qcow volume encryption unsupported with "
                             "volume format %s"), type);
749
            return NULL;
750
        }
751 752 753
        enc = vol->target.encryption;
        if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW &&
            enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT) {
754 755 756
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("unsupported volume encryption format %d"),
                           vol->target.encryption->format);
757
            return NULL;
758
        }
759
        if (enc->nsecrets > 1) {
760 761
            virReportError(VIR_ERR_XML_ERROR, "%s",
                           _("too many secrets for qcow encryption"));
762
            return NULL;
763
        }
764 765 766
        if (enc->format == VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT ||
            enc->nsecrets == 0) {
            if (virStorageGenerateQcowEncryption(conn, vol) < 0)
767
                return NULL;
768
        }
769 770
    }

771
    /* Size in KB */
772
    size_arg = VIR_DIV_UP(vol->capacity, 1024);
773

774 775
    cmd = virCommandNew(create_tool);

776
    if (inputvol) {
777
        virCommandAddArgList(cmd, "convert", "-f", inputType, "-O", type, NULL);
778

779 780 781 782 783 784 785 786
        if (imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS &&
            (do_encryption || preallocate)) {
            virCommandAddArg(cmd, "-o");
            virCommandAddArgFormat(cmd, "%s%s%s", do_encryption ? "encryption=on" : "",
                                   (do_encryption && preallocate) ? "," : "",
                                   preallocate ? "preallocation=metadata" : "");
        } else if (do_encryption) {
            virCommandAddArg(cmd, "-e");
787
        }
788
        virCommandAddArgList(cmd, inputPath, vol->target.path, NULL);
789
    } else if (vol->backingStore.path) {
790 791 792
        virCommandAddArgList(cmd, "create", "-f", type,
                             "-b", vol->backingStore.path, NULL);

793 794
        switch (imgformat) {
        case QEMU_IMG_BACKING_FORMAT_FLAG:
795
            virCommandAddArgList(cmd, "-F", backingType, NULL);
796 797
            if (do_encryption)
                virCommandAddArg(cmd, "-e");
798 799
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
800 801 802
            break;

        case QEMU_IMG_BACKING_FORMAT_OPTIONS:
803 804 805
            virCommandAddArg(cmd, "-o");
            virCommandAddArgFormat(cmd, "backing_fmt=%s%s", backingType,
                                   do_encryption ? ",encryption=on" : "");
806 807
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
808
            break;
809 810

        default:
811 812
            VIR_DEBUG("Unable to set backing store format for %s with %s",
                      vol->target.path, create_tool);
813

814 815
            if (do_encryption)
                virCommandAddArg(cmd, "-e");
816 817
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
818
        }
819
    } else {
820
        virCommandAddArgList(cmd, "create", "-f", type, NULL);
821

822 823 824 825 826 827 828 829
        if (imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS &&
            (do_encryption || preallocate)) {
            virCommandAddArg(cmd, "-o");
            virCommandAddArgFormat(cmd, "%s%s%s", do_encryption ? "encryption=on" : "",
                                   (do_encryption && preallocate) ? "," : "",
                                   preallocate ? "preallocation=metadata" : "");
        } else if (do_encryption) {
            virCommandAddArg(cmd, "-e");
830
        }
831 832
        virCommandAddArg(cmd, vol->target.path);
        virCommandAddArgFormat(cmd, "%lluK", size_arg);
833
    }
834

835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871
    return cmd;
}

static int
virStorageBackendCreateQemuImg(virConnectPtr conn,
                               virStoragePoolObjPtr pool,
                               virStorageVolDefPtr vol,
                               virStorageVolDefPtr inputvol,
                               unsigned int flags)
{
    int ret = -1;
    const char *create_tool;
    int imgformat;
    virCommandPtr cmd;

    virCheckFlags(VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA, -1);

    /* KVM is usually ahead of qemu on features, so try that first */
    create_tool = virFindFileInPath("kvm-img");
    if (!create_tool)
        create_tool = virFindFileInPath("qemu-img");

    if (!create_tool) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("unable to find kvm-img or qemu-img"));
        return -1;
    }

    imgformat = virStorageBackendQEMUImgBackingFormat(create_tool);
    if (imgformat < 0)
        goto cleanup;

    cmd = virStorageBackendCreateQemuImgCmd(conn, pool, vol, inputvol, flags,
                                            create_tool, imgformat);
    if (!cmd)
        goto cleanup;

872
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
873 874

    virCommandFree(cmd);
875
cleanup:
876
    VIR_FREE(create_tool);
877
    return ret;
878 879 880 881 882 883 884
}

/*
 * Xen removed the fully-functional qemu-img, and replaced it
 * with a partially functional qcow-create. Go figure ??!?
 */
static int
885
virStorageBackendCreateQcowCreate(virConnectPtr conn ATTRIBUTE_UNUSED,
886
                                  virStoragePoolObjPtr pool,
887
                                  virStorageVolDefPtr vol,
888
                                  virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
889
                                  unsigned int flags)
890
{
891
    int ret;
E
Eric Blake 已提交
892
    char *size;
893
    virCommandPtr cmd;
894

895 896 897 898 899 900 901 902
    virCheckFlags(VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA, -1);

    if (flags & VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("metadata preallocation is not supported with "
                         "qcow-create"));
        return -1;
    }
E
Eric Blake 已提交
903

904
    if (inputvol) {
905 906
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("cannot copy from volume with qcow-create"));
907 908 909
        return -1;
    }

910
    if (vol->target.format != VIR_STORAGE_FILE_QCOW2) {
911 912 913
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unsupported storage vol type %d"),
                       vol->target.format);
914 915 916
        return -1;
    }
    if (vol->backingStore.path != NULL) {
917 918 919
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("copy-on-write image not supported with "
                         "qcow-create"));
920 921
        return -1;
    }
922
    if (vol->target.encryption != NULL) {
923 924 925
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       "%s", _("encrypted volumes not supported with "
                               "qcow-create"));
926 927
        return -1;
    }
928 929

    /* Size in MB - yes different units to qemu-img :-( */
930 931
    if (virAsprintf(&size, "%llu",
                    VIR_DIV_UP(vol->capacity, (1024 * 1024))) < 0) {
E
Eric Blake 已提交
932 933 934
        virReportOOMError();
        return -1;
    }
935

936
    cmd = virCommandNewArgList("qcow-create", size, vol->target.path, NULL);
937

938 939
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
    virCommandFree(cmd);
E
Eric Blake 已提交
940
    VIR_FREE(size);
941

942
    return ret;
943 944
}

945
virStorageBackendBuildVolFrom
946
virStorageBackendFSImageToolTypeToFunc(int tool_type)
947 948 949 950 951 952 953 954
{
    switch (tool_type) {
    case TOOL_KVM_IMG:
    case TOOL_QEMU_IMG:
        return virStorageBackendCreateQemuImg;
    case TOOL_QCOW_CREATE:
        return virStorageBackendCreateQcowCreate;
    default:
955 956 957
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Unknown file create tool type '%d'."),
                       tool_type);
958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984
    }

    return NULL;
}

int
virStorageBackendFindFSImageTool(char **tool)
{
    int tool_type = -1;
    char *tmp = NULL;

    if ((tmp = virFindFileInPath("kvm-img")) != NULL) {
        tool_type = TOOL_KVM_IMG;
    } else if ((tmp = virFindFileInPath("qemu-img")) != NULL) {
        tool_type = TOOL_QEMU_IMG;
    } else if ((tmp = virFindFileInPath("qcow-create")) != NULL) {
        tool_type = TOOL_QCOW_CREATE;
    }

    if (tool)
        *tool = tmp;
    else
        VIR_FREE(tmp);

    return tool_type;
}

985
virStorageBackendBuildVolFrom
986
virStorageBackendGetBuildVolFromFunction(virStorageVolDefPtr vol,
987 988 989 990 991 992 993 994 995 996 997
                                         virStorageVolDefPtr inputvol)
{
    int tool_type;

    if (!inputvol)
        return NULL;

    /* If either volume is a non-raw file vol, we need to use an external
     * tool for converting
     */
    if ((vol->type == VIR_STORAGE_VOL_FILE &&
998
         vol->target.format != VIR_STORAGE_FILE_RAW) ||
999
        (inputvol->type == VIR_STORAGE_VOL_FILE &&
1000
         inputvol->target.format != VIR_STORAGE_FILE_RAW)) {
1001

D
Daniel P. Berrange 已提交
1002
        if ((tool_type = virStorageBackendFindFSImageTool(NULL)) < 0) {
1003 1004 1005
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("creation of non-raw file images is "
                             "not supported without qemu-img."));
1006 1007 1008
            return NULL;
        }

1009
        return virStorageBackendFSImageToolTypeToFunc(tool_type);
1010 1011
    }

1012 1013 1014 1015
    if (vol->type == VIR_STORAGE_VOL_BLOCK)
        return virStorageBackendCreateBlockFrom;
    else
        return virStorageBackendCreateRaw;
1016
}
1017

1018

1019
virStorageBackendPtr
1020 1021
virStorageBackendForType(int type)
{
1022
    unsigned int i;
1023
    for (i = 0; backends[i]; i++)
1024 1025 1026
        if (backends[i]->type == type)
            return backends[i];

1027 1028
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("missing backend for pool type %d"), type);
1029 1030 1031 1032
    return NULL;
}


1033 1034 1035
/*
 * Allows caller to silently ignore files with improper mode
 *
1036 1037
 * Returns -1 on error, -2 if file mode is unexpected or the
 * volume is a dangling symbolic link.
1038
 */
1039
int
1040
virStorageBackendVolOpenCheckMode(const char *path, unsigned int flags)
1041
{
1042 1043
    int fd, mode = 0;
    struct stat sb;
1044
    char *base = last_component(path);
1045

1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060
    if (lstat(path, &sb) < 0) {
        virReportSystemError(errno,
                             _("cannot stat file '%s'"),
                             path);
        return -1;
    }

    if (S_ISFIFO(sb.st_mode)) {
        VIR_WARN("ignoring FIFO '%s'", path);
        return -2;
    } else if (S_ISSOCK(sb.st_mode)) {
        VIR_WARN("ignoring socket '%s'", path);
        return -2;
    }

1061
    if ((fd = open(path, O_RDONLY|O_NONBLOCK|O_NOCTTY)) < 0) {
1062
        if ((errno == ENOENT || errno == ELOOP) &&
1063
            S_ISLNK(sb.st_mode)) {
1064 1065 1066 1067
            VIR_WARN("ignoring dangling symlink '%s'", path);
            return -2;
        }

1068
        virReportSystemError(errno,
1069
                             _("cannot open volume '%s'"),
1070
                             path);
1071 1072 1073
        return -1;
    }

1074 1075 1076 1077 1078 1079 1080 1081
    if (fstat(fd, &sb) < 0) {
        virReportSystemError(errno,
                             _("cannot stat file '%s'"),
                             path);
        VIR_FORCE_CLOSE(fd);
        return -1;
    }

1082 1083 1084 1085 1086 1087
    if (S_ISREG(sb.st_mode))
        mode = VIR_STORAGE_VOL_OPEN_REG;
    else if (S_ISCHR(sb.st_mode))
        mode = VIR_STORAGE_VOL_OPEN_CHAR;
    else if (S_ISBLK(sb.st_mode))
        mode = VIR_STORAGE_VOL_OPEN_BLOCK;
1088 1089 1090 1091 1092 1093 1094 1095 1096 1097
    else if (S_ISDIR(sb.st_mode)) {
        mode = VIR_STORAGE_VOL_OPEN_DIR;

        if (STREQ(base, ".") ||
            STREQ(base, "..")) {
            VIR_FORCE_CLOSE(fd);
            VIR_INFO("Skipping special dir '%s'", base);
            return -2;
        }
    }
1098 1099

    if (!(mode & flags)) {
1100
        VIR_FORCE_CLOSE(fd);
1101
        VIR_INFO("Skipping volume '%s'", path);
1102 1103

        if (mode & VIR_STORAGE_VOL_OPEN_ERROR) {
1104 1105
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unexpected storage mode for '%s'"), path);
1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123
            return -1;
        }

        return -2;
    }

    return fd;
}

int virStorageBackendVolOpen(const char *path)
{
    return virStorageBackendVolOpenCheckMode(path,
                                             VIR_STORAGE_VOL_OPEN_DEFAULT);
}

int
virStorageBackendUpdateVolTargetInfo(virStorageVolTargetPtr target,
                                     unsigned long long *allocation,
1124 1125
                                     unsigned long long *capacity,
                                     unsigned int openflags)
1126 1127 1128
{
    int ret, fd;

1129 1130
    if ((ret = virStorageBackendVolOpenCheckMode(target->path,
                                                 openflags)) < 0)
1131 1132 1133
        return ret;

    fd = ret;
1134
    ret = virStorageBackendUpdateVolTargetInfoFD(target,
1135 1136 1137
                                                 fd,
                                                 allocation,
                                                 capacity);
1138

1139
    VIR_FORCE_CLOSE(fd);
1140 1141 1142 1143

    return ret;
}

1144
int
1145 1146 1147
virStorageBackendUpdateVolInfoFlags(virStorageVolDefPtr vol,
                                    int withCapacity,
                                    unsigned int openflags)
1148 1149 1150
{
    int ret;

1151
    if ((ret = virStorageBackendUpdateVolTargetInfo(&vol->target,
1152 1153 1154
                                    &vol->allocation,
                                    withCapacity ? &vol->capacity : NULL,
                                    openflags)) < 0)
1155 1156 1157
        return ret;

    if (vol->backingStore.path &&
1158
        (ret = virStorageBackendUpdateVolTargetInfo(&vol->backingStore,
1159 1160
                                            NULL, NULL,
                                            VIR_STORAGE_VOL_OPEN_DEFAULT)) < 0)
1161 1162 1163 1164 1165
        return ret;

    return 0;
}

1166 1167 1168 1169 1170 1171 1172
int virStorageBackendUpdateVolInfo(virStorageVolDefPtr vol,
                                   int withCapacity)
{
    return virStorageBackendUpdateVolInfoFlags(vol, withCapacity,
                                               VIR_STORAGE_VOL_OPEN_DEFAULT);
}

1173 1174 1175 1176
/*
 * virStorageBackendUpdateVolTargetInfoFD:
 * @conn: connection to report errors on
 * @target: target definition ptr of volume to update
1177
 * @fd: fd of storage volume to update, via virStorageBackendOpenVol*
1178 1179 1180
 * @allocation: If not NULL, updated allocation information will be stored
 * @capacity: If not NULL, updated capacity info will be stored
 *
1181
 * Returns 0 for success, -1 on a legitimate error condition.
1182
 */
1183
int
1184
virStorageBackendUpdateVolTargetInfoFD(virStorageVolTargetPtr target,
1185 1186 1187
                                       int fd,
                                       unsigned long long *allocation,
                                       unsigned long long *capacity)
1188 1189
{
    struct stat sb;
1190
#if WITH_SELINUX
1191 1192 1193 1194
    security_context_t filecon = NULL;
#endif

    if (fstat(fd, &sb) < 0) {
1195
        virReportSystemError(errno,
1196
                             _("cannot stat file '%s'"),
1197
                             target->path);
1198 1199 1200
        return -1;
    }

1201 1202
    if (allocation) {
        if (S_ISREG(sb.st_mode)) {
1203
#ifndef WIN32
1204
            *allocation = (unsigned long long)sb.st_blocks *
1205
                          (unsigned long long)DEV_BSIZE;
D
Daniel P. Berrange 已提交
1206
#else
1207
            *allocation = sb.st_size;
D
Daniel P. Berrange 已提交
1208
#endif
1209 1210 1211 1212 1213
            /* Regular files may be sparse, so logical size (capacity) is not same
             * as actual allocation above
             */
            if (capacity)
                *capacity = sb.st_size;
1214 1215 1216 1217 1218
        } else if (S_ISDIR(sb.st_mode)) {
            *allocation = 0;
            if (capacity)
                *capacity = 0;

1219 1220
        } else {
            off_t end;
E
Eric Blake 已提交
1221
            /* XXX this is POSIX compliant, but doesn't work for CHAR files,
1222 1223 1224 1225 1226 1227
             * only BLOCK. There is a Linux specific ioctl() for getting
             * size of both CHAR / BLOCK devices we should check for in
             * configure
             */
            end = lseek(fd, 0, SEEK_END);
            if (end == (off_t)-1) {
1228
                virReportSystemError(errno,
1229 1230 1231 1232 1233 1234 1235
                                     _("cannot seek to end of file '%s'"),
                                     target->path);
                return -1;
            }
            *allocation = end;
            if (capacity)
                *capacity = end;
1236 1237 1238
        }
    }

1239 1240 1241
    target->perms.mode = sb.st_mode & S_IRWXUGO;
    target->perms.uid = sb.st_uid;
    target->perms.gid = sb.st_gid;
1242

1243 1244 1245 1246 1247 1248 1249 1250 1251
    if (!target->timestamps && VIR_ALLOC(target->timestamps) < 0) {
        virReportOOMError();
        return -1;
    }
    target->timestamps->atime = get_stat_atime(&sb);
    target->timestamps->btime = get_stat_birthtime(&sb);
    target->timestamps->ctime = get_stat_ctime(&sb);
    target->timestamps->mtime = get_stat_mtime(&sb);

1252
    VIR_FREE(target->perms.label);
1253

1254
#if WITH_SELINUX
1255
    /* XXX: make this a security driver call */
M
Martin Kletzander 已提交
1256
    if (fgetfilecon_raw(fd, &filecon) == -1) {
1257
        if (errno != ENODATA && errno != ENOTSUP) {
1258
            virReportSystemError(errno,
1259
                                 _("cannot get file context of '%s'"),
1260
                                 target->path);
1261 1262
            return -1;
        } else {
1263
            target->perms.label = NULL;
1264 1265
        }
    } else {
1266 1267
        if (VIR_STRDUP(target->perms.label, filecon) < 0) {
            freecon(filecon);
1268 1269
            return -1;
        }
1270
        freecon(filecon);
1271 1272
    }
#else
1273
    target->perms.label = NULL;
1274 1275 1276 1277 1278
#endif

    return 0;
}

D
Dave Allan 已提交
1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313

struct diskType {
    int part_table_type;
    unsigned short offset;
    unsigned short length;
    unsigned long long magic;
};


static struct diskType const disk_types[] = {
    { VIR_STORAGE_POOL_DISK_LVM2, 0x218, 8, 0x31303020324D564CULL },
    { VIR_STORAGE_POOL_DISK_GPT,  0x200, 8, 0x5452415020494645ULL },
    { VIR_STORAGE_POOL_DISK_DVH,  0x0,   4, 0x41A9E50BULL },
    { VIR_STORAGE_POOL_DISK_MAC,  0x0,   2, 0x5245ULL },
    { VIR_STORAGE_POOL_DISK_BSD,  0x40,  4, 0x82564557ULL },
    { VIR_STORAGE_POOL_DISK_SUN,  0x1fc, 2, 0xBEDAULL },
    /*
     * NOTE: pc98 is funky; the actual signature is 0x55AA (just like dos), so
     * we can't use that.  At the moment I'm relying on the "dummy" IPL
     * bootloader data that comes from parted.  Luckily, the chances of running
     * into a pc98 machine running libvirt are approximately nil.
     */
    /*{ 0x1fe, 2, 0xAA55UL },*/
    { VIR_STORAGE_POOL_DISK_PC98, 0x0,   8, 0x314C5049000000CBULL },
    /*
     * NOTE: the order is important here; some other disk types (like GPT and
     * and PC98) also have 0x55AA at this offset.  For that reason, the DOS
     * one must be the last one.
     */
    { VIR_STORAGE_POOL_DISK_DOS,  0x1fe, 2, 0xAA55ULL },
    { -1,                         0x0,   0, 0x0ULL },
};


int
1314 1315
virStorageBackendDetectBlockVolFormatFD(virStorageVolTargetPtr target,
                                        int fd)
D
Dave Allan 已提交
1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326
{
    int i;
    off_t start;
    unsigned char buffer[1024];
    ssize_t bytes;

    /* make sure to set the target format "unknown" to begin with */
    target->format = VIR_STORAGE_POOL_DISK_UNKNOWN;

    start = lseek(fd, 0, SEEK_SET);
    if (start < 0) {
1327
        virReportSystemError(errno,
D
Dave Allan 已提交
1328 1329 1330 1331 1332 1333
                             _("cannot seek to beginning of file '%s'"),
                             target->path);
        return -1;
    }
    bytes = saferead(fd, buffer, sizeof(buffer));
    if (bytes < 0) {
1334
        virReportSystemError(errno,
D
Dave Allan 已提交
1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353
                             _("cannot read beginning of file '%s'"),
                             target->path);
        return -1;
    }

    for (i = 0; disk_types[i].part_table_type != -1; i++) {
        if (disk_types[i].offset + disk_types[i].length > bytes)
            continue;
        if (memcmp(buffer+disk_types[i].offset, &disk_types[i].magic,
            disk_types[i].length) == 0) {
            target->format = disk_types[i].part_table_type;
            break;
        }
    }

    return 0;
}


1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365
/*
 * Given a volume path directly in /dev/XXX, iterate over the
 * entries in the directory pool->def->target.path and find the
 * first symlink pointing to the volume path.
 *
 * If, the target.path is /dev/, then return the original volume
 * path.
 *
 * If no symlink is found, then return the original volume path
 *
 * Typically target.path is one of the /dev/disk/by-XXX dirs
 * with stable paths.
1366
 *
1367
 * If 'loop' is true, we use a timeout loop to give dynamic paths
1368
 * a change to appear.
1369 1370
 */
char *
1371
virStorageBackendStablePath(virStoragePoolObjPtr pool,
1372
                            const char *devpath,
1373
                            bool loop)
1374 1375 1376
{
    DIR *dh;
    struct dirent *dent;
1377
    char *stablepath;
1378
    int opentries = 0;
1379
    int retry = 0;
1380 1381 1382 1383 1384

    /* Short circuit if pool has no target, or if its /dev */
    if (pool->def->target.path == NULL ||
        STREQ(pool->def->target.path, "/dev") ||
        STREQ(pool->def->target.path, "/dev/"))
1385
        goto ret_strdup;
1386

1387
    /* Skip whole thing for a pool which isn't in /dev
1388
     * so we don't mess filesystem/dir based pools
1389 1390 1391 1392
     */
    if (!STRPREFIX(pool->def->target.path, "/dev"))
        goto ret_strdup;

1393 1394 1395 1396
    /* Logical pools are under /dev but already have stable paths */
    if (pool->def->type == VIR_STORAGE_POOL_LOGICAL)
        goto ret_strdup;

1397 1398 1399
    /* We loop here because /dev/disk/by-{id,path} may not have existed
     * before we started this operation, so we have to give it some time to
     * get created.
1400
     */
1401
 reopen:
1402
    if ((dh = opendir(pool->def->target.path)) == NULL) {
1403
        opentries++;
1404
        if (loop && errno == ENOENT && opentries < 50) {
1405 1406 1407
            usleep(100 * 1000);
            goto reopen;
        }
1408
        virReportSystemError(errno,
1409 1410
                             _("cannot read dir '%s'"),
                             pool->def->target.path);
1411 1412 1413
        return NULL;
    }

1414 1415 1416
    /* The pool is pointing somewhere like /dev/disk/by-path
     * or /dev/disk/by-id, so we need to check all symlinks in
     * the target directory and figure out which one points
1417 1418
     * to this device node.
     *
1419
     * And it might need some time till the stable path shows
1420
     * up, so add timeout to retry here.
1421
     */
1422
 retry:
1423 1424 1425 1426
    while ((dent = readdir(dh)) != NULL) {
        if (dent->d_name[0] == '.')
            continue;

1427 1428 1429
        if (virAsprintf(&stablepath, "%s/%s",
                        pool->def->target.path,
                        dent->d_name) == -1) {
1430
            virReportOOMError();
1431 1432 1433 1434 1435 1436 1437 1438 1439
            closedir(dh);
            return NULL;
        }

        if (virFileLinkPointsTo(stablepath, devpath)) {
            closedir(dh);
            return stablepath;
        }

1440
        VIR_FREE(stablepath);
1441 1442
    }

1443
    if (loop && ++retry < 100) {
1444 1445 1446 1447
        usleep(100 * 1000);
        goto retry;
    }

1448 1449
    closedir(dh);

1450
 ret_strdup:
1451 1452 1453
    /* Couldn't find any matching stable link so give back
     * the original non-stable dev path
     */
1454

1455
    ignore_value(VIR_STRDUP(stablepath, devpath));
1456 1457

    return stablepath;
1458 1459
}

D
Daniel P. Berrange 已提交
1460

1461
#ifndef WIN32
1462 1463 1464 1465
/*
 * Run an external program.
 *
 * Read its output and apply a series of regexes to each line
R
Richard W.M. Jones 已提交
1466
 * When the entire set of regexes has matched consecutively
1467 1468 1469
 * then run a callback passing in all the matches
 */
int
1470
virStorageBackendRunProgRegex(virStoragePoolObjPtr pool,
1471
                              virCommandPtr cmd,
1472 1473 1474 1475
                              int nregex,
                              const char **regex,
                              int *nvars,
                              virStorageBackendListVolRegexFunc func,
1476
                              void *data, const char *prefix)
1477
{
1478
    int fd = -1, err, ret = -1;
1479 1480 1481 1482 1483 1484 1485 1486 1487
    FILE *list = NULL;
    regex_t *reg;
    regmatch_t *vars = NULL;
    char line[1024];
    int maxReg = 0, i, j;
    int totgroups = 0, ngroup = 0, maxvars = 0;
    char **groups;

    /* Compile all regular expressions */
1488
    if (VIR_ALLOC_N(reg, nregex) < 0) {
1489
        virReportOOMError();
1490 1491 1492
        return -1;
    }

1493
    for (i = 0; i < nregex; i++) {
1494 1495 1496 1497
        err = regcomp(&reg[i], regex[i], REG_EXTENDED);
        if (err != 0) {
            char error[100];
            regerror(err, &reg[i], error, sizeof(error));
1498 1499
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Failed to compile regex %s"), error);
1500
            for (j = 0; j <= i; j++)
1501
                regfree(&reg[j]);
1502
            VIR_FREE(reg);
1503 1504 1505 1506 1507 1508 1509 1510 1511 1512
            return -1;
        }

        totgroups += nvars[i];
        if (nvars[i] > maxvars)
            maxvars = nvars[i];

    }

    /* Storage for matched variables */
1513
    if (VIR_ALLOC_N(groups, totgroups) < 0) {
1514
        virReportOOMError();
1515 1516
        goto cleanup;
    }
1517
    if (VIR_ALLOC_N(vars, maxvars+1) < 0) {
1518
        virReportOOMError();
1519 1520 1521
        goto cleanup;
    }

1522 1523
    virCommandSetOutputFD(cmd, &fd);
    if (virCommandRunAsync(cmd, NULL) < 0) {
1524 1525 1526
        goto cleanup;
    }

1527
    if ((list = VIR_FDOPEN(fd, "r")) == NULL) {
1528 1529
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("cannot read fd"));
1530 1531 1532 1533
        goto cleanup;
    }

    while (fgets(line, sizeof(line), list) != NULL) {
1534
        char *p = NULL;
1535 1536 1537 1538 1539
        /* Strip trailing newline */
        int len = strlen(line);
        if (len && line[len-1] == '\n')
            line[len-1] = '\0';

1540 1541 1542 1543 1544 1545
        /* ignore any command prefix */
        if (prefix)
            p = STRSKIP(line, prefix);
        if (!p)
            p = line;

1546
        for (i = 0; i <= maxReg && i < nregex; i++) {
1547
            if (regexec(&reg[i], p, nvars[i]+1, vars, 0) == 0) {
1548 1549 1550 1551 1552 1553
                maxReg++;

                if (i == 0)
                    ngroup = 0;

                /* NULL terminate each captured group in the line */
1554
                for (j = 0; j < nvars[i]; j++) {
1555
                    /* NB vars[0] is the full pattern, so we offset j by 1 */
1556
                    p[vars[j+1].rm_eo] = '\0';
1557
                    if (VIR_STRDUP(groups[ngroup++], p + vars[j+1].rm_so) < 0)
1558 1559 1560 1561 1562
                        goto cleanup;
                }

                /* We're matching on the last regex, so callback time */
                if (i == (nregex-1)) {
1563
                    if (((*func)(pool, groups, data)) < 0)
1564 1565 1566
                        goto cleanup;

                    /* Release matches & restart to matching the first regex */
1567
                    for (j = 0; j < totgroups; j++)
1568
                        VIR_FREE(groups[j]);
1569 1570 1571 1572 1573 1574 1575
                    maxReg = 0;
                    ngroup = 0;
                }
            }
        }
    }

1576
    ret = virCommandWait(cmd, NULL);
1577
cleanup:
1578
    if (groups) {
1579
        for (j = 0; j < totgroups; j++)
1580 1581
            VIR_FREE(groups[j]);
        VIR_FREE(groups);
1582
    }
1583
    VIR_FREE(vars);
1584

1585
    for (i = 0; i < nregex; i++)
1586 1587
        regfree(&reg[i]);

1588
    VIR_FREE(reg);
1589

1590 1591
    VIR_FORCE_FCLOSE(list);
    VIR_FORCE_CLOSE(fd);
1592

1593
    return ret;
1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607
}

/*
 * Run an external program and read from its standard output
 * a stream of tokens from IN_STREAM, applying FUNC to
 * each successive sequence of N_COLUMNS tokens.
 * If FUNC returns < 0, stop processing input and return -1.
 * Return -1 if N_COLUMNS == 0.
 * Return -1 upon memory allocation error.
 * If the number of input tokens is not a multiple of N_COLUMNS,
 * then the final FUNC call will specify a number smaller than N_COLUMNS.
 * If there are no input tokens (empty input), call FUNC with N_COLUMNS == 0.
 */
int
1608
virStorageBackendRunProgNul(virStoragePoolObjPtr pool,
1609
                            virCommandPtr cmd,
1610 1611 1612 1613 1614
                            size_t n_columns,
                            virStorageBackendListVolNulFunc func,
                            void *data)
{
    size_t n_tok = 0;
1615
    int fd = -1;
1616 1617
    FILE *fp = NULL;
    char **v;
1618
    int ret = -1;
1619 1620 1621 1622 1623
    int i;

    if (n_columns == 0)
        return -1;

1624
    if (VIR_ALLOC_N(v, n_columns) < 0) {
1625
        virReportOOMError();
1626 1627 1628 1629 1630
        return -1;
    }
    for (i = 0; i < n_columns; i++)
        v[i] = NULL;

1631 1632
    virCommandSetOutputFD(cmd, &fd);
    if (virCommandRunAsync(cmd, NULL) < 0) {
1633 1634 1635
        goto cleanup;
    }

1636
    if ((fp = VIR_FDOPEN(fd, "r")) == NULL) {
1637 1638
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("cannot open file using fd"));
1639 1640 1641 1642 1643 1644 1645 1646
        goto cleanup;
    }

    while (1) {
        char *buf = NULL;
        size_t buf_len = 0;
        /* Be careful: even when it returns -1,
           this use of getdelim allocates memory.  */
1647
        ssize_t tok_len = getdelim(&buf, &buf_len, 0, fp);
1648 1649 1650 1651
        v[n_tok] = buf;
        if (tok_len < 0) {
            /* Maybe EOF, maybe an error.
               If n_tok > 0, then we know it's an error.  */
1652
            if (n_tok && func(pool, n_tok, v, data) < 0)
1653 1654 1655 1656 1657
                goto cleanup;
            break;
        }
        ++n_tok;
        if (n_tok == n_columns) {
1658
            if (func(pool, n_tok, v, data) < 0)
1659 1660 1661
                goto cleanup;
            n_tok = 0;
            for (i = 0; i < n_columns; i++) {
1662
                VIR_FREE(v[i]);
1663 1664 1665 1666
            }
        }
    }

1667 1668 1669
    if (feof(fp) < 0) {
        virReportSystemError(errno, "%s",
                             _("read error on pipe"));
1670 1671
        goto cleanup;
    }
1672

1673
    ret = virCommandWait(cmd, NULL);
1674 1675
 cleanup:
    for (i = 0; i < n_columns; i++)
1676 1677
        VIR_FREE(v[i]);
    VIR_FREE(v);
1678

1679 1680
    VIR_FORCE_FCLOSE(fp);
    VIR_FORCE_CLOSE(fd);
1681

1682
    return ret;
1683
}
D
Daniel P. Berrange 已提交
1684

1685
#else /* WIN32 */
D
Daniel P. Berrange 已提交
1686 1687 1688 1689 1690 1691 1692 1693 1694

int
virStorageBackendRunProgRegex(virConnectPtr conn,
                              virStoragePoolObjPtr pool ATTRIBUTE_UNUSED,
                              const char *const*prog ATTRIBUTE_UNUSED,
                              int nregex ATTRIBUTE_UNUSED,
                              const char **regex ATTRIBUTE_UNUSED,
                              int *nvars ATTRIBUTE_UNUSED,
                              virStorageBackendListVolRegexFunc func ATTRIBUTE_UNUSED,
1695
                              void *data ATTRIBUTE_UNUSED)
D
Daniel P. Berrange 已提交
1696
{
1697 1698
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709
    return -1;
}

int
virStorageBackendRunProgNul(virConnectPtr conn,
                            virStoragePoolObjPtr pool ATTRIBUTE_UNUSED,
                            const char **prog ATTRIBUTE_UNUSED,
                            size_t n_columns ATTRIBUTE_UNUSED,
                            virStorageBackendListVolNulFunc func ATTRIBUTE_UNUSED,
                            void *data ATTRIBUTE_UNUSED)
{
1710 1711
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1712 1713
    return -1;
}
1714
#endif /* WIN32 */