storage_backend.c 54.9 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
#if WITH_STORAGE_GLUSTER
# include "storage_backend_gluster.h"
#endif
86

87 88
#define VIR_FROM_THIS VIR_FROM_STORAGE

89 90 91 92 93 94 95 96 97 98 99 100 101 102
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
103 104 105
#if WITH_STORAGE_SCSI
    &virStorageBackendSCSI,
#endif
D
Dave Allan 已提交
106 107 108
#if WITH_STORAGE_MPATH
    &virStorageBackendMpath,
#endif
109 110
#if WITH_STORAGE_DISK
    &virStorageBackendDisk,
111 112 113
#endif
#if WITH_STORAGE_RBD
    &virStorageBackendRBD,
114 115 116
#endif
#if WITH_STORAGE_SHEEPDOG
    &virStorageBackendSheepdog,
117 118 119
#endif
#if WITH_STORAGE_GLUSTER
    &virStorageBackendGluster,
120 121 122
#endif
    NULL
};
123

124 125 126 127 128 129
enum {
    TOOL_QEMU_IMG,
    TOOL_KVM_IMG,
    TOOL_QCOW_CREATE,
};

130 131 132
#define READ_BLOCK_SIZE_DEFAULT  (1024 * 1024)
#define WRITE_BLOCK_SIZE_DEFAULT (4 * 1024)

133
static int ATTRIBUTE_NONNULL(2)
134
virStorageBackendCopyToFD(virStorageVolDefPtr vol,
135 136
                          virStorageVolDefPtr inputvol,
                          int fd,
137
                          unsigned long long *total,
J
Ján Tomko 已提交
138
                          bool want_sparse)
139 140 141
{
    int inputfd = -1;
    int amtread = -1;
142
    int ret = 0;
143
    size_t rbytes = READ_BLOCK_SIZE_DEFAULT;
L
Li Zhang 已提交
144
    int wbytes = 0;
145
    int interval;
146
    char *zerobuf = NULL;
147
    char *buf = NULL;
148
    struct stat st;
149

150
    if ((inputfd = open(inputvol->target.path, O_RDONLY)) < 0) {
151
        ret = -errno;
152
        virReportSystemError(errno,
153 154 155
                             _("could not open input path '%s'"),
                             inputvol->target.path);
        goto cleanup;
156 157
    }

158 159 160 161 162 163 164 165 166 167 168 169 170 171
#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;
        goto cleanup;
    }
172

173
    if (VIR_ALLOC_N(buf, rbytes) < 0) {
174
        ret = -errno;
175 176 177 178 179 180
        goto cleanup;
    }

    while (amtread != 0) {
        int amtleft;

181 182
        if (*total < rbytes)
            rbytes = *total;
183

184
        if ((amtread = saferead(inputfd, buf, rbytes)) < 0) {
185
            ret = -errno;
186
            virReportSystemError(errno,
187 188 189 190
                                 _("failed reading from file '%s'"),
                                 inputvol->target.path);
            goto cleanup;
        }
191
        *total -= amtread;
192 193 194 195 196

        /* Loop over amt read in 512 byte increments, looking for sparse
         * blocks */
        amtleft = amtread;
        do {
197
            interval = ((wbytes > amtleft) ? amtleft : wbytes);
198 199
            int offset = amtread - amtleft;

200
            if (want_sparse && memcmp(buf+offset, zerobuf, interval) == 0) {
201
                if (lseek(fd, interval, SEEK_CUR) < 0) {
202
                    ret = -errno;
203
                    virReportSystemError(errno,
204 205 206 207 208
                                         _("cannot extend file '%s'"),
                                         vol->target.path);
                    goto cleanup;
                }
            } else if (safewrite(fd, buf+offset, interval) < 0) {
209
                ret = -errno;
210
                virReportSystemError(errno,
211 212 213 214 215
                                     _("failed writing to file '%s'"),
                                     vol->target.path);
                goto cleanup;

            }
216
        } while ((amtleft -= interval) > 0);
217 218
    }

219 220 221 222 223 224 225 226
    if (fdatasync(fd) < 0) {
        ret = -errno;
        virReportSystemError(errno, _("cannot sync data to file '%s'"),
                             vol->target.path);
        goto cleanup;
    }


227
    if (VIR_CLOSE(inputfd) < 0) {
228
        ret = -errno;
229
        virReportSystemError(errno,
230 231 232 233 234 235 236
                             _("cannot close file '%s'"),
                             inputvol->target.path);
        goto cleanup;
    }
    inputfd = -1;

cleanup:
237
    VIR_FORCE_CLOSE(inputfd);
238

239
    VIR_FREE(zerobuf);
240 241
    VIR_FREE(buf);

242 243 244
    return ret;
}

245
static int
246
virStorageBackendCreateBlockFrom(virConnectPtr conn ATTRIBUTE_UNUSED,
247
                                 virStoragePoolObjPtr pool ATTRIBUTE_UNUSED,
248 249
                                 virStorageVolDefPtr vol,
                                 virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
250
                                 unsigned int flags)
251 252 253 254
{
    int fd = -1;
    int ret = -1;
    unsigned long long remain;
255 256 257
    struct stat st;
    gid_t gid;
    uid_t uid;
258

259 260 261 262 263 264 265 266
    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
267

268
    if ((fd = open(vol->target.path, O_RDWR)) < 0) {
269
        virReportSystemError(errno,
270 271 272 273 274 275 276 277
                             _("cannot create path '%s'"),
                             vol->target.path);
        goto cleanup;
    }

    remain = vol->allocation;

    if (inputvol) {
278
        int res = virStorageBackendCopyToFD(vol, inputvol,
J
Ján Tomko 已提交
279
                                            fd, &remain, false);
280 281 282 283
        if (res < 0)
            goto cleanup;
    }

284
    if (fstat(fd, &st) == -1) {
285
        virReportSystemError(errno, _("stat of '%s' failed"),
286 287 288
                             vol->target.path);
        goto cleanup;
    }
P
Philipp Hahn 已提交
289 290 291
    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))
292
        && (fchown(fd, uid, gid) < 0)) {
293
        virReportSystemError(errno,
294
                             _("cannot chown '%s' to (%u, %u)"),
295 296
                             vol->target.path, (unsigned int) uid,
                             (unsigned int) gid);
297 298 299
        goto cleanup;
    }
    if (fchmod(fd, vol->target.perms.mode) < 0) {
300
        virReportSystemError(errno,
301 302 303 304
                             _("cannot set mode of '%s' to %04o"),
                             vol->target.path, vol->target.perms.mode);
        goto cleanup;
    }
305
    if (VIR_CLOSE(fd) < 0) {
306
        virReportSystemError(errno,
307 308 309 310 311 312 313 314
                             _("cannot close file '%s'"),
                             vol->target.path);
        goto cleanup;
    }
    fd = -1;

    ret = 0;
cleanup:
315
    VIR_FORCE_CLOSE(fd);
316 317 318 319

    return ret;
}

E
Eric Blake 已提交
320 321 322 323
static int
createRawFile(int fd, virStorageVolDefPtr vol,
              virStorageVolDefPtr inputvol)
{
J
Ján Tomko 已提交
324
    bool need_alloc = true;
325 326
    int ret = 0;
    unsigned long long remain;
327

328 329
    /* Seek to the final size, so the capacity is available upfront
     * for progress reporting */
E
Eric Blake 已提交
330
    if (ftruncate(fd, vol->capacity) < 0) {
331
        ret = -errno;
332
        virReportSystemError(errno,
333
                             _("cannot extend file '%s'"),
E
Eric Blake 已提交
334
                             vol->target.path);
335 336 337
        goto cleanup;
    }

E
Eric Blake 已提交
338 339
/* Avoid issues with older kernel's <linux/fs.h> namespace pollution. */
#if HAVE_FALLOCATE - 0
340
    /* Try to preallocate all requested disk space, but fall back to
341 342
     * other methods if this fails with ENOSYS or EOPNOTSUPP. If allocation
     * is 0 (or less than 0), then fallocate will fail with EINVAL.
343 344 345 346
     * NOTE: do not use posix_fallocate; posix_fallocate falls back
     * to writing zeroes block by block in case fallocate isn't
     * available, and since we're going to copy data from another
     * file it doesn't make sense to write the file twice. */
347 348 349 350 351 352 353 354 355 356
    if (vol->allocation) {
        if (fallocate(fd, 0, 0, vol->allocation) == 0) {
            need_alloc = false;
        } else if (errno != ENOSYS && errno != EOPNOTSUPP) {
            ret = -errno;
            virReportSystemError(errno,
                                 _("cannot allocate %llu bytes in file '%s'"),
                                 vol->allocation, vol->target.path);
            goto cleanup;
        }
357 358 359
    }
#endif

E
Eric Blake 已提交
360
    remain = vol->allocation;
361

E
Eric Blake 已提交
362
    if (inputvol) {
363 364 365
        /* allow zero blocks to be skipped if we've requested sparse
         * allocation (allocation < capacity) or we have already
         * been able to allocate the required space. */
J
Ján Tomko 已提交
366 367
        bool want_sparse = !need_alloc ||
                           (vol->allocation < inputvol->capacity);
368 369

        ret = virStorageBackendCopyToFD(vol, inputvol, fd, &remain, want_sparse);
370
        if (ret < 0) {
371
            goto cleanup;
372
        }
373 374
    }

375 376
    if (remain && need_alloc) {
        if (safezero(fd, vol->allocation - remain, remain) < 0) {
377 378 379 380
            ret = -errno;
            virReportSystemError(errno, _("cannot fill file '%s'"),
                                 vol->target.path);
            goto cleanup;
381
        }
382 383 384
    }

    if (fsync(fd) < 0) {
385
        ret = -errno;
386
        virReportSystemError(errno, _("cannot sync data to file '%s'"),
E
Eric Blake 已提交
387
                             vol->target.path);
388
        goto cleanup;
389 390
    }

391 392 393 394 395 396 397 398 399
cleanup:
    return ret;
}

int
virStorageBackendCreateRaw(virConnectPtr conn ATTRIBUTE_UNUSED,
                           virStoragePoolObjPtr pool,
                           virStorageVolDefPtr vol,
                           virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
400
                           unsigned int flags)
401 402
{
    int ret = -1;
E
Eric Blake 已提交
403 404 405
    int fd = -1;
    int operation_flags;

406 407 408 409 410 411 412 413
    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;
    }
414 415

    if (vol->target.encryption != NULL) {
416 417
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("storage pool does not support encrypted volumes"));
418 419
        goto cleanup;
    }
420

L
Laine Stump 已提交
421
    operation_flags = VIR_FILE_OPEN_FORCE_MODE | VIR_FILE_OPEN_FORCE_OWNER;
E
Eric Blake 已提交
422
    if (pool->def->type == VIR_STORAGE_POOL_NETFS)
L
Laine Stump 已提交
423
        operation_flags |= VIR_FILE_OPEN_FORK;
E
Eric Blake 已提交
424

425 426
    if ((fd = virFileOpenAs(vol->target.path,
                            O_RDWR | O_CREAT | O_EXCL,
L
Laine Stump 已提交
427 428 429
                            vol->target.perms.mode,
                            vol->target.perms.uid,
                            vol->target.perms.gid,
430
                            operation_flags)) < 0) {
E
Eric Blake 已提交
431
        virReportSystemError(-fd,
432
                             _("Failed to create file '%s'"),
E
Eric Blake 已提交
433 434 435 436
                             vol->target.path);
        goto cleanup;
    }

437 438
    if ((ret = createRawFile(fd, vol, inputvol)) < 0)
        /* createRawFile already reported the exact error. */
E
Eric Blake 已提交
439
        ret = -1;
440 441

cleanup:
E
Eric Blake 已提交
442
    VIR_FORCE_CLOSE(fd);
443 444 445
    return ret;
}

446 447 448 449 450 451 452 453 454
static int
virStorageGenerateSecretUUID(virConnectPtr conn,
                             unsigned char *uuid)
{
    unsigned attempt;

    for (attempt = 0; attempt < 65536; attempt++) {
        virSecretPtr tmp;
        if (virUUIDGenerate(uuid) < 0) {
455 456
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("unable to generate uuid"));
457 458
            return -1;
        }
459
        tmp = conn->secretDriver->secretLookupByUUID(conn, uuid);
460 461 462 463 464 465
        if (tmp == NULL)
            return 0;

        virSecretFree(tmp);
    }

466 467
    virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                   _("too many conflicts when generating an uuid"));
468 469 470 471

    return -1;
}

472 473 474 475 476 477 478 479 480
static int
virStorageGenerateQcowEncryption(virConnectPtr conn,
                                 virStorageVolDefPtr vol)
{
    virSecretDefPtr def = NULL;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    virStorageEncryptionPtr enc;
    virStorageEncryptionSecretPtr enc_secret = NULL;
    virSecretPtr secret = NULL;
481
    char *xml;
482 483 484
    unsigned char value[VIR_STORAGE_QCOW_PASSPHRASE_SIZE];
    int ret = -1;

485
    if (conn->secretDriver == NULL ||
486 487 488
        conn->secretDriver->secretLookupByUUID == NULL ||
        conn->secretDriver->secretDefineXML == NULL ||
        conn->secretDriver->secretSetValue == NULL) {
489 490
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("secret storage not supported"));
491 492 493 494 495
        goto cleanup;
    }

    enc = vol->target.encryption;
    if (enc->nsecrets != 0) {
496 497
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("secrets already defined"));
498 499 500 501
        goto cleanup;
    }

    if (VIR_ALLOC(enc_secret) < 0 || VIR_REALLOC_N(enc->secrets, 1) < 0 ||
502
        VIR_ALLOC(def) < 0)
503 504
        goto cleanup;

505
    def->ephemeral = false;
506
    def->private = false;
507
    if (virStorageGenerateSecretUUID(conn, def->uuid) < 0)
508
        goto cleanup;
509

510
    def->usage_type = VIR_SECRET_USAGE_TYPE_VOLUME;
511
    if (VIR_STRDUP(def->usage.volume, vol->target.path) < 0)
512
        goto cleanup;
513
    xml = virSecretDefFormat(def);
514 515 516 517 518
    virSecretDefFree(def);
    def = NULL;
    if (xml == NULL)
        goto cleanup;

519
    secret = conn->secretDriver->secretDefineXML(conn, xml, 0);
520 521 522 523 524 525
    if (secret == NULL) {
        VIR_FREE(xml);
        goto cleanup;
    }
    VIR_FREE(xml);

526
    if (virStorageGenerateQcowPassphrase(value) < 0)
527 528
        goto cleanup;

529
    if (conn->secretDriver->secretSetValue(secret, value, sizeof(value), 0) < 0)
530 531 532
        goto cleanup;

    enc_secret->type = VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE;
533
    memcpy(enc_secret->uuid, secret->uuid, VIR_UUID_BUFLEN);
534 535 536 537 538 539 540 541 542
    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) {
543
        if (ret != 0 &&
544 545
            conn->secretDriver->secretUndefine != NULL)
            conn->secretDriver->secretUndefine(secret);
546 547
        virSecretFree(secret);
    }
548
    virBufferFreeAndReset(&buf);
549 550 551 552 553
    virSecretDefFree(def);
    VIR_FREE(enc_secret);
    return ret;
}

554 555 556 557 558
static int
virStorageBackendCreateExecCommand(virStoragePoolObjPtr pool,
                                   virStorageVolDefPtr vol,
                                   virCommandPtr cmd)
{
559 560 561
    struct stat st;
    gid_t gid;
    uid_t uid;
562
    bool filecreated = false;
563 564

    if ((pool->def->type == VIR_STORAGE_POOL_NETFS)
565
        && (((geteuid() == 0)
P
Philipp Hahn 已提交
566
             && (vol->target.perms.uid != (uid_t) -1)
567
             && (vol->target.perms.uid != 0))
P
Philipp Hahn 已提交
568
            || ((vol->target.perms.gid != (gid_t) -1)
569
                && (vol->target.perms.gid != getegid())))) {
570

571 572
        virCommandSetUID(cmd, vol->target.perms.uid);
        virCommandSetGID(cmd, vol->target.perms.gid);
573 574

        if (virCommandRun(cmd, NULL) == 0) {
575 576
            /* command was successfully run, check if the file was created */
            if (stat(vol->target.path, &st) >=0)
577
                filecreated = true;
578 579
        }
    }
580

581 582 583
    /* don't change uid/gid if we retry */
    virCommandSetUID(cmd, -1);
    virCommandSetGID(cmd, -1);
584

585
    if (!filecreated) {
586
        if (virCommandRun(cmd, NULL) < 0) {
587 588 589
            return -1;
        }
        if (stat(vol->target.path, &st) < 0) {
590
            virReportSystemError(errno,
591
                                 _("failed to create %s"), vol->target.path);
592 593 594 595
            return -1;
        }
    }

P
Philipp Hahn 已提交
596 597 598
    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))
599
        && (chown(vol->target.path, uid, gid) < 0)) {
600
        virReportSystemError(errno,
601
                             _("cannot chown %s to (%u, %u)"),
602 603
                             vol->target.path, (unsigned int) uid,
                             (unsigned int) gid);
604 605 606
        return -1;
    }
    if (chmod(vol->target.path, vol->target.perms.mode) < 0) {
607
        virReportSystemError(errno,
608 609 610 611 612 613 614
                             _("cannot set mode of '%s' to %04o"),
                             vol->target.path, vol->target.perms.mode);
        return -1;
    }
    return 0;
}

615 616 617 618
enum {
    QEMU_IMG_BACKING_FORMAT_NONE = 0,
    QEMU_IMG_BACKING_FORMAT_FLAG,
    QEMU_IMG_BACKING_FORMAT_OPTIONS,
619
    QEMU_IMG_BACKING_FORMAT_OPTIONS_COMPAT,
620 621
};

622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646
static bool
virStorageBackendQemuImgSupportsCompat(const char *qemuimg)
{
    bool ret = false;
    char *output;
    virCommandPtr cmd = NULL;

    cmd = virCommandNewArgList(qemuimg, "create", "-o", "?", "-f", "qcow2",
                               "/dev/null", NULL);

    virCommandAddEnvString(cmd, "LC_ALL=C");
    virCommandSetOutputBuffer(cmd, &output);

    if (virCommandRun(cmd, NULL) < 0)
        goto cleanup;

    if (strstr(output, "\ncompat "))
        ret = true;

cleanup:
    virCommandFree(cmd);
    VIR_FREE(output);
    return ret;
}

647 648
static int
virStorageBackendQEMUImgBackingFormat(const char *qemuimg)
649 650 651 652 653 654
{
    char *help = NULL;
    char *start;
    char *end;
    char *tmp;
    int ret = -1;
655
    int exitstatus;
656
    virCommandPtr cmd = virCommandNewArgList(qemuimg, "-h", NULL);
657

658 659 660
    virCommandAddEnvString(cmd, "LC_ALL=C");
    virCommandSetOutputBuffer(cmd, &help);
    virCommandClearCaps(cmd);
661

662 663 664
    /* 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 */
665
    if (virCommandRun(cmd, &exitstatus) < 0)
666 667
        goto cleanup;

668 669
    if ((start = strstr(help, " create ")) == NULL ||
        (end = strstr(start, "\n")) == NULL) {
670 671 672
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unable to parse qemu-img output '%s'"),
                       help);
673 674
        goto cleanup;
    }
675
    if (((tmp = strstr(start, "-F fmt")) && tmp < end) ||
676
        ((tmp = strstr(start, "-F backing_fmt")) && tmp < end)) {
677
        ret = QEMU_IMG_BACKING_FORMAT_FLAG;
678 679 680 681 682 683
    } else if ((tmp = strstr(start, "[-o options]")) && tmp < end) {
        if (virStorageBackendQemuImgSupportsCompat(qemuimg))
            ret = QEMU_IMG_BACKING_FORMAT_OPTIONS_COMPAT;
        else
            ret = QEMU_IMG_BACKING_FORMAT_OPTIONS;
    } else {
684
        ret = QEMU_IMG_BACKING_FORMAT_NONE;
685
    }
686 687

cleanup:
688
    virCommandFree(cmd);
689 690 691 692
    VIR_FREE(help);
    return ret;
}

693 694 695 696
static int
virStorageBackendCreateQemuImgOpts(char **opts,
                                   const char *backingType,
                                   bool encryption,
697 698 699 700
                                   bool preallocate,
                                   int format,
                                   const char *compat,
                                   virBitmapPtr features)
701 702
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;
703
    bool b;
704
    size_t i;
705

706 707 708 709 710 711 712
    if (backingType)
        virBufferAsprintf(&buf, "backing_fmt=%s,", backingType);
    if (encryption)
        virBufferAddLit(&buf, "encryption=on,");
    if (preallocate)
        virBufferAddLit(&buf, "preallocation=metadata,");

713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729
    if (compat)
        virBufferAsprintf(&buf, "compat=%s,", compat);
    if (features && format == VIR_STORAGE_FILE_QCOW2) {
        for (i = 0; i < VIR_STORAGE_FILE_FEATURE_LAST; i++) {
            ignore_value(virBitmapGetBit(features, i, &b));
            if (b) {
                switch ((enum virStorageFileFeature) i) {
                case VIR_STORAGE_FILE_FEATURE_LAZY_REFCOUNTS:
                    if (STREQ_NULLABLE(compat, "0.10")) {
                        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                                       _("Feature %s not supported with compat"
                                         " level %s"),
                                       virStorageFileFeatureTypeToString(i),
                                       compat);
                        goto error;
                    }
                    break;
730

731
                /* coverity[dead_error_begin] */
732 733 734 735 736 737 738 739 740
                case VIR_STORAGE_FILE_FEATURE_LAST:
                    ;
                }
                virBufferAsprintf(&buf, "%s,",
                                  virStorageFileFeatureTypeToString(i));
            }
        }
    }

741 742
    virBufferTrim(&buf, ",", -1);

743 744
    if (virBufferError(&buf))
        goto no_memory;
745 746 747

    *opts = virBufferContentAndReset(&buf);
    return 0;
748 749 750 751 752 753

no_memory:
    virReportOOMError();
error:
    virBufferFreeAndReset(&buf);
    return -1;
754 755
}

756 757 758 759 760 761 762 763
virCommandPtr
virStorageBackendCreateQemuImgCmd(virConnectPtr conn,
                                  virStoragePoolObjPtr pool,
                                  virStorageVolDefPtr vol,
                                  virStorageVolDefPtr inputvol,
                                  unsigned int flags,
                                  const char *create_tool,
                                  int imgformat)
764
{
765 766
    virCommandPtr cmd = NULL;
    bool do_encryption = (vol->target.encryption != NULL);
767
    unsigned long long int size_arg;
768 769 770 771 772
    bool preallocate = !!(flags & VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA);
    const char *type;
    const char *backingType = NULL;
    const char *inputPath = NULL;
    const char *inputType = NULL;
773
    const char *compat = vol->target.compat;
774 775 776
    char *opts = NULL;
    bool convert = false;
    bool backing = false;
777

778 779
    virCheckFlags(VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA, NULL);

780 781 782 783
    /* Treat output block devices as 'raw' format */
    type = virStorageFileFormatTypeToString(vol->type == VIR_STORAGE_VOL_BLOCK ?
                                            VIR_STORAGE_FILE_RAW :
                                            vol->target.format);
784

785
    if (!type) {
786 787 788
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unknown storage vol type %d"),
                       vol->target.format);
789
        return NULL;
790
    }
791

792 793 794
    if (preallocate && vol->target.format != VIR_STORAGE_FILE_QCOW2) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("metadata preallocation only available with qcow2"));
795
        return NULL;
796
    }
797 798 799 800 801 802 803 804 805 806
    if (vol->target.compat && vol->target.format != VIR_STORAGE_FILE_QCOW2) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("compatibility option only available with qcow2"));
        return NULL;
    }
    if (vol->target.features && vol->target.format != VIR_STORAGE_FILE_QCOW2) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("format features only available with qcow2"));
        return NULL;
    }
807

808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827
    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;
        }

    }

828
    if (vol->backingStore.path) {
829 830
        int accessRetCode = -1;
        char *absolutePath = NULL;
831

832 833
        backingType = virStorageFileFormatTypeToString(vol->backingStore.format);

834 835 836 837
        if (preallocate) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("metadata preallocation conflicts with backing"
                             " store"));
838
            return NULL;
839 840
        }

841 842 843 844 845
        /* 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 &&
846 847 848
            STRNEQ_NULLABLE(inputvol->backingStore.path, vol->backingStore.path)) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("a different backing store cannot be specified."));
849
            return NULL;
850 851 852
        }

        if (backingType == NULL) {
853 854 855
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unknown storage vol backing store type %d"),
                           vol->backingStore.format);
856
            return NULL;
857
        }
858 859 860 861 862 863

        /* Convert relative backing store paths to absolute paths for access
         * validation.
         */
        if ('/' != *(vol->backingStore.path) &&
            virAsprintf(&absolutePath, "%s/%s", pool->def->target.path,
864
                        vol->backingStore.path) < 0)
865
            return NULL;
866 867 868 869
        accessRetCode = access(absolutePath ? absolutePath
                               : vol->backingStore.path, R_OK);
        VIR_FREE(absolutePath);
        if (accessRetCode != 0) {
870
            virReportSystemError(errno,
871 872
                                 _("inaccessible backing store volume %s"),
                                 vol->backingStore.path);
873
            return NULL;
874 875 876
        }
    }

877
    if (do_encryption) {
878 879
        virStorageEncryptionPtr enc;

880 881
        if (vol->target.format != VIR_STORAGE_FILE_QCOW &&
            vol->target.format != VIR_STORAGE_FILE_QCOW2) {
882 883 884
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("qcow volume encryption unsupported with "
                             "volume format %s"), type);
885
            return NULL;
886
        }
887 888 889
        enc = vol->target.encryption;
        if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW &&
            enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT) {
890 891 892
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("unsupported volume encryption format %d"),
                           vol->target.encryption->format);
893
            return NULL;
894
        }
895
        if (enc->nsecrets > 1) {
896 897
            virReportError(VIR_ERR_XML_ERROR, "%s",
                           _("too many secrets for qcow encryption"));
898
            return NULL;
899
        }
900 901 902
        if (enc->format == VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT ||
            enc->nsecrets == 0) {
            if (virStorageGenerateQcowEncryption(conn, vol) < 0)
903
                return NULL;
904
        }
905 906
    }

907
    /* Size in KB */
908
    size_arg = VIR_DIV_UP(vol->capacity, 1024);
909

910 911
    cmd = virCommandNew(create_tool);

912 913
    convert = !!inputvol;
    backing = !inputvol && vol->backingStore.path;
914

915 916 917
    if (convert)
        virCommandAddArgList(cmd, "convert", "-f", inputType, "-O", type, NULL);
    else
918
        virCommandAddArgList(cmd, "create", "-f", type, NULL);
919

920 921 922
    if (backing)
        virCommandAddArgList(cmd, "-b", vol->backingStore.path, NULL);

923 924 925 926 927
    if (imgformat >= QEMU_IMG_BACKING_FORMAT_OPTIONS) {
        if (vol->target.format == VIR_STORAGE_FILE_QCOW2 && !compat &&
            imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS_COMPAT)
            compat = "0.10";

928 929
        if (virStorageBackendCreateQemuImgOpts(&opts,
                                               backing ? backingType : NULL,
930 931
                                               do_encryption, preallocate,
                                               vol->target.format,
932
                                               compat,
933 934
                                               vol->target.features) < 0) {
            virCommandFree(cmd);
935
            return NULL;
936
        }
937 938 939 940 941 942 943 944 945 946
        if (opts)
            virCommandAddArgList(cmd, "-o", opts, NULL);
        VIR_FREE(opts);
    } else {
        if (backing) {
            if (imgformat == QEMU_IMG_BACKING_FORMAT_FLAG)
                virCommandAddArgList(cmd, "-F", backingType, NULL);
            else
                VIR_DEBUG("Unable to set backing store format for %s with %s",
                          vol->target.path, create_tool);
947
        }
948 949
        if (do_encryption)
            virCommandAddArg(cmd, "-e");
950
    }
951

952 953 954 955 956 957
    if (convert)
        virCommandAddArg(cmd, inputPath);
    virCommandAddArg(cmd, vol->target.path);
    if (!convert)
        virCommandAddArgFormat(cmd, "%lluK", size_arg);

958 959 960 961 962 963 964 965 966 967 968
    return cmd;
}

static int
virStorageBackendCreateQemuImg(virConnectPtr conn,
                               virStoragePoolObjPtr pool,
                               virStorageVolDefPtr vol,
                               virStorageVolDefPtr inputvol,
                               unsigned int flags)
{
    int ret = -1;
969
    char *create_tool;
970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994
    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;

995
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
996 997

    virCommandFree(cmd);
998
cleanup:
999
    VIR_FREE(create_tool);
1000
    return ret;
1001 1002 1003 1004 1005 1006 1007
}

/*
 * Xen removed the fully-functional qemu-img, and replaced it
 * with a partially functional qcow-create. Go figure ??!?
 */
static int
1008
virStorageBackendCreateQcowCreate(virConnectPtr conn ATTRIBUTE_UNUSED,
1009
                                  virStoragePoolObjPtr pool,
1010
                                  virStorageVolDefPtr vol,
1011
                                  virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
1012
                                  unsigned int flags)
1013
{
1014
    int ret;
E
Eric Blake 已提交
1015
    char *size;
1016
    virCommandPtr cmd;
1017

1018 1019 1020 1021 1022 1023 1024 1025
    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
1026

1027
    if (inputvol) {
1028 1029
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("cannot copy from volume with qcow-create"));
1030 1031 1032
        return -1;
    }

1033
    if (vol->target.format != VIR_STORAGE_FILE_QCOW2) {
1034 1035 1036
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unsupported storage vol type %d"),
                       vol->target.format);
1037 1038 1039
        return -1;
    }
    if (vol->backingStore.path != NULL) {
1040 1041 1042
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("copy-on-write image not supported with "
                         "qcow-create"));
1043 1044
        return -1;
    }
1045
    if (vol->target.encryption != NULL) {
1046 1047 1048
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       "%s", _("encrypted volumes not supported with "
                               "qcow-create"));
1049 1050
        return -1;
    }
1051 1052

    /* Size in MB - yes different units to qemu-img :-( */
1053
    if (virAsprintf(&size, "%llu",
1054
                    VIR_DIV_UP(vol->capacity, (1024 * 1024))) < 0)
E
Eric Blake 已提交
1055
        return -1;
1056

1057
    cmd = virCommandNewArgList("qcow-create", size, vol->target.path, NULL);
1058

1059 1060
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
    virCommandFree(cmd);
E
Eric Blake 已提交
1061
    VIR_FREE(size);
1062

1063
    return ret;
1064 1065
}

1066
virStorageBackendBuildVolFrom
1067
virStorageBackendFSImageToolTypeToFunc(int tool_type)
1068 1069 1070 1071 1072 1073 1074 1075
{
    switch (tool_type) {
    case TOOL_KVM_IMG:
    case TOOL_QEMU_IMG:
        return virStorageBackendCreateQemuImg;
    case TOOL_QCOW_CREATE:
        return virStorageBackendCreateQcowCreate;
    default:
1076 1077 1078
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Unknown file create tool type '%d'."),
                       tool_type);
1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105
    }

    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;
}

1106
virStorageBackendBuildVolFrom
1107
virStorageBackendGetBuildVolFromFunction(virStorageVolDefPtr vol,
1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118
                                         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 &&
1119
         vol->target.format != VIR_STORAGE_FILE_RAW) ||
1120
        (inputvol->type == VIR_STORAGE_VOL_FILE &&
1121
         inputvol->target.format != VIR_STORAGE_FILE_RAW)) {
1122

D
Daniel P. Berrange 已提交
1123
        if ((tool_type = virStorageBackendFindFSImageTool(NULL)) < 0) {
1124 1125 1126
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("creation of non-raw file images is "
                             "not supported without qemu-img."));
1127 1128 1129
            return NULL;
        }

1130
        return virStorageBackendFSImageToolTypeToFunc(tool_type);
1131 1132
    }

1133 1134 1135 1136
    if (vol->type == VIR_STORAGE_VOL_BLOCK)
        return virStorageBackendCreateBlockFrom;
    else
        return virStorageBackendCreateRaw;
1137
}
1138

1139

1140
virStorageBackendPtr
1141 1142
virStorageBackendForType(int type)
{
1143
    size_t i;
1144
    for (i = 0; backends[i]; i++)
1145 1146 1147
        if (backends[i]->type == type)
            return backends[i];

1148 1149
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("missing backend for pool type %d"), type);
1150 1151 1152 1153
    return NULL;
}


1154 1155 1156
/*
 * Allows caller to silently ignore files with improper mode
 *
1157 1158
 * Returns -1 on error, -2 if file mode is unexpected or the
 * volume is a dangling symbolic link.
1159
 */
1160
int
E
Eric Blake 已提交
1161 1162
virStorageBackendVolOpenCheckMode(const char *path, struct stat *sb,
                                  unsigned int flags)
1163
{
1164
    int fd, mode = 0;
1165
    char *base = last_component(path);
1166

E
Eric Blake 已提交
1167
    if (lstat(path, sb) < 0) {
1168 1169 1170 1171 1172 1173
        virReportSystemError(errno,
                             _("cannot stat file '%s'"),
                             path);
        return -1;
    }

E
Eric Blake 已提交
1174
    if (S_ISFIFO(sb->st_mode)) {
1175 1176
        VIR_WARN("ignoring FIFO '%s'", path);
        return -2;
E
Eric Blake 已提交
1177
    } else if (S_ISSOCK(sb->st_mode)) {
1178 1179 1180 1181
        VIR_WARN("ignoring socket '%s'", path);
        return -2;
    }

1182
    if ((fd = open(path, O_RDONLY|O_NONBLOCK|O_NOCTTY)) < 0) {
1183
        if ((errno == ENOENT || errno == ELOOP) &&
E
Eric Blake 已提交
1184
            S_ISLNK(sb->st_mode)) {
1185 1186 1187 1188
            VIR_WARN("ignoring dangling symlink '%s'", path);
            return -2;
        }

1189
        virReportSystemError(errno,
1190
                             _("cannot open volume '%s'"),
1191
                             path);
1192 1193 1194
        return -1;
    }

E
Eric Blake 已提交
1195
    if (fstat(fd, sb) < 0) {
1196 1197 1198 1199 1200 1201 1202
        virReportSystemError(errno,
                             _("cannot stat file '%s'"),
                             path);
        VIR_FORCE_CLOSE(fd);
        return -1;
    }

E
Eric Blake 已提交
1203
    if (S_ISREG(sb->st_mode))
1204
        mode = VIR_STORAGE_VOL_OPEN_REG;
E
Eric Blake 已提交
1205
    else if (S_ISCHR(sb->st_mode))
1206
        mode = VIR_STORAGE_VOL_OPEN_CHAR;
E
Eric Blake 已提交
1207
    else if (S_ISBLK(sb->st_mode))
1208
        mode = VIR_STORAGE_VOL_OPEN_BLOCK;
E
Eric Blake 已提交
1209
    else if (S_ISDIR(sb->st_mode)) {
1210 1211 1212 1213 1214 1215 1216 1217 1218
        mode = VIR_STORAGE_VOL_OPEN_DIR;

        if (STREQ(base, ".") ||
            STREQ(base, "..")) {
            VIR_FORCE_CLOSE(fd);
            VIR_INFO("Skipping special dir '%s'", base);
            return -2;
        }
    }
1219 1220

    if (!(mode & flags)) {
1221
        VIR_FORCE_CLOSE(fd);
1222
        VIR_INFO("Skipping volume '%s'", path);
1223 1224

        if (mode & VIR_STORAGE_VOL_OPEN_ERROR) {
1225 1226
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unexpected storage mode for '%s'"), path);
1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237
            return -1;
        }

        return -2;
    }

    return fd;
}

int virStorageBackendVolOpen(const char *path)
{
E
Eric Blake 已提交
1238 1239
    struct stat sb;
    return virStorageBackendVolOpenCheckMode(path, &sb,
1240 1241 1242 1243 1244 1245
                                             VIR_STORAGE_VOL_OPEN_DEFAULT);
}

int
virStorageBackendUpdateVolTargetInfo(virStorageVolTargetPtr target,
                                     unsigned long long *allocation,
1246 1247
                                     unsigned long long *capacity,
                                     unsigned int openflags)
1248 1249
{
    int ret, fd;
E
Eric Blake 已提交
1250
    struct stat sb;
1251

E
Eric Blake 已提交
1252
    if ((ret = virStorageBackendVolOpenCheckMode(target->path, &sb,
1253
                                                 openflags)) < 0)
1254 1255 1256
        return ret;

    fd = ret;
1257
    ret = virStorageBackendUpdateVolTargetInfoFD(target,
1258
                                                 fd,
E
Eric Blake 已提交
1259
                                                 &sb,
1260 1261
                                                 allocation,
                                                 capacity);
1262

1263
    VIR_FORCE_CLOSE(fd);
1264 1265 1266 1267

    return ret;
}

1268
int
1269 1270 1271
virStorageBackendUpdateVolInfoFlags(virStorageVolDefPtr vol,
                                    int withCapacity,
                                    unsigned int openflags)
1272 1273 1274
{
    int ret;

1275
    if ((ret = virStorageBackendUpdateVolTargetInfo(&vol->target,
1276 1277 1278
                                    &vol->allocation,
                                    withCapacity ? &vol->capacity : NULL,
                                    openflags)) < 0)
1279 1280 1281
        return ret;

    if (vol->backingStore.path &&
1282
        (ret = virStorageBackendUpdateVolTargetInfo(&vol->backingStore,
1283 1284
                                            NULL, NULL,
                                            VIR_STORAGE_VOL_OPEN_DEFAULT)) < 0)
1285 1286 1287 1288 1289
        return ret;

    return 0;
}

1290 1291 1292 1293 1294 1295 1296
int virStorageBackendUpdateVolInfo(virStorageVolDefPtr vol,
                                   int withCapacity)
{
    return virStorageBackendUpdateVolInfoFlags(vol, withCapacity,
                                               VIR_STORAGE_VOL_OPEN_DEFAULT);
}

1297 1298 1299
/*
 * virStorageBackendUpdateVolTargetInfoFD:
 * @target: target definition ptr of volume to update
1300 1301
 * @fd: fd of storage volume to update, via virStorageBackendOpenVol*, or -1
 * @sb: details about file (must match @fd, if that is provided)
1302 1303 1304
 * @allocation: If not NULL, updated allocation information will be stored
 * @capacity: If not NULL, updated capacity info will be stored
 *
1305
 * Returns 0 for success, -1 on a legitimate error condition.
1306
 */
1307
int
1308
virStorageBackendUpdateVolTargetInfoFD(virStorageVolTargetPtr target,
1309
                                       int fd,
E
Eric Blake 已提交
1310
                                       struct stat *sb,
1311 1312
                                       unsigned long long *allocation,
                                       unsigned long long *capacity)
1313
{
1314
#if WITH_SELINUX
1315 1316 1317
    security_context_t filecon = NULL;
#endif

1318
    if (allocation) {
E
Eric Blake 已提交
1319
        if (S_ISREG(sb->st_mode)) {
1320
#ifndef WIN32
E
Eric Blake 已提交
1321
            *allocation = (unsigned long long)sb->st_blocks *
1322
                          (unsigned long long)DEV_BSIZE;
D
Daniel P. Berrange 已提交
1323
#else
E
Eric Blake 已提交
1324
            *allocation = sb->st_size;
D
Daniel P. Berrange 已提交
1325
#endif
1326 1327 1328 1329
            /* Regular files may be sparse, so logical size (capacity) is not same
             * as actual allocation above
             */
            if (capacity)
E
Eric Blake 已提交
1330 1331
                *capacity = sb->st_size;
        } else if (S_ISDIR(sb->st_mode)) {
1332 1333 1334 1335
            *allocation = 0;
            if (capacity)
                *capacity = 0;

1336
        } else if (fd >= 0) {
1337
            off_t end;
E
Eric Blake 已提交
1338
            /* XXX this is POSIX compliant, but doesn't work for CHAR files,
1339 1340 1341 1342 1343 1344
             * 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) {
1345
                virReportSystemError(errno,
1346 1347 1348 1349 1350 1351 1352
                                     _("cannot seek to end of file '%s'"),
                                     target->path);
                return -1;
            }
            *allocation = end;
            if (capacity)
                *capacity = end;
1353 1354 1355
        }
    }

E
Eric Blake 已提交
1356 1357 1358
    target->perms.mode = sb->st_mode & S_IRWXUGO;
    target->perms.uid = sb->st_uid;
    target->perms.gid = sb->st_gid;
1359

1360
    if (!target->timestamps && VIR_ALLOC(target->timestamps) < 0)
1361
        return -1;
E
Eric Blake 已提交
1362 1363 1364 1365
    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);
1366

1367
    VIR_FREE(target->perms.label);
1368

1369
#if WITH_SELINUX
1370
    /* XXX: make this a security driver call */
1371
    if (fd >= 0 && fgetfilecon_raw(fd, &filecon) == -1) {
1372
        if (errno != ENODATA && errno != ENOTSUP) {
1373
            virReportSystemError(errno,
1374
                                 _("cannot get file context of '%s'"),
1375
                                 target->path);
1376 1377
            return -1;
        } else {
1378
            target->perms.label = NULL;
1379 1380
        }
    } else {
1381 1382
        if (VIR_STRDUP(target->perms.label, filecon) < 0) {
            freecon(filecon);
1383 1384
            return -1;
        }
1385
        freecon(filecon);
1386 1387
    }
#else
1388
    target->perms.label = NULL;
1389 1390 1391 1392 1393
#endif

    return 0;
}

D
Dave Allan 已提交
1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428

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
1429 1430
virStorageBackendDetectBlockVolFormatFD(virStorageVolTargetPtr target,
                                        int fd)
D
Dave Allan 已提交
1431
{
1432
    size_t i;
D
Dave Allan 已提交
1433 1434 1435 1436 1437 1438 1439 1440 1441
    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) {
1442
        virReportSystemError(errno,
D
Dave Allan 已提交
1443 1444 1445 1446 1447 1448
                             _("cannot seek to beginning of file '%s'"),
                             target->path);
        return -1;
    }
    bytes = saferead(fd, buffer, sizeof(buffer));
    if (bytes < 0) {
1449
        virReportSystemError(errno,
D
Dave Allan 已提交
1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468
                             _("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;
}


1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480
/*
 * 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.
1481
 *
1482
 * If 'loop' is true, we use a timeout loop to give dynamic paths
1483
 * a change to appear.
1484 1485
 */
char *
1486
virStorageBackendStablePath(virStoragePoolObjPtr pool,
1487
                            const char *devpath,
1488
                            bool loop)
1489 1490 1491
{
    DIR *dh;
    struct dirent *dent;
1492
    char *stablepath;
1493
    int opentries = 0;
1494
    int retry = 0;
1495 1496 1497 1498 1499

    /* 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/"))
1500
        goto ret_strdup;
1501

1502
    /* Skip whole thing for a pool which isn't in /dev
1503
     * so we don't mess filesystem/dir based pools
1504 1505 1506 1507
     */
    if (!STRPREFIX(pool->def->target.path, "/dev"))
        goto ret_strdup;

1508 1509 1510 1511
    /* Logical pools are under /dev but already have stable paths */
    if (pool->def->type == VIR_STORAGE_POOL_LOGICAL)
        goto ret_strdup;

1512 1513 1514
    /* 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.
1515
     */
1516
 reopen:
1517
    if ((dh = opendir(pool->def->target.path)) == NULL) {
1518
        opentries++;
1519
        if (loop && errno == ENOENT && opentries < 50) {
1520 1521 1522
            usleep(100 * 1000);
            goto reopen;
        }
1523
        virReportSystemError(errno,
1524 1525
                             _("cannot read dir '%s'"),
                             pool->def->target.path);
1526 1527 1528
        return NULL;
    }

1529 1530 1531
    /* 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
1532 1533
     * to this device node.
     *
1534
     * And it might need some time till the stable path shows
1535
     * up, so add timeout to retry here.
1536
     */
1537
 retry:
1538 1539 1540 1541
    while ((dent = readdir(dh)) != NULL) {
        if (dent->d_name[0] == '.')
            continue;

1542 1543 1544
        if (virAsprintf(&stablepath, "%s/%s",
                        pool->def->target.path,
                        dent->d_name) == -1) {
1545 1546 1547 1548 1549 1550 1551 1552 1553
            closedir(dh);
            return NULL;
        }

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

1554
        VIR_FREE(stablepath);
1555 1556
    }

1557
    if (loop && ++retry < 100) {
1558 1559 1560 1561
        usleep(100 * 1000);
        goto retry;
    }

1562 1563
    closedir(dh);

1564
 ret_strdup:
1565 1566 1567
    /* Couldn't find any matching stable link so give back
     * the original non-stable dev path
     */
1568

1569
    ignore_value(VIR_STRDUP(stablepath, devpath));
1570 1571

    return stablepath;
1572 1573
}

D
Daniel P. Berrange 已提交
1574

1575
#ifndef WIN32
1576 1577 1578 1579
/*
 * Run an external program.
 *
 * Read its output and apply a series of regexes to each line
R
Richard W.M. Jones 已提交
1580
 * When the entire set of regexes has matched consecutively
1581 1582 1583
 * then run a callback passing in all the matches
 */
int
1584
virStorageBackendRunProgRegex(virStoragePoolObjPtr pool,
1585
                              virCommandPtr cmd,
1586 1587 1588 1589
                              int nregex,
                              const char **regex,
                              int *nvars,
                              virStorageBackendListVolRegexFunc func,
1590
                              void *data, const char *prefix)
1591
{
1592
    int fd = -1, err, ret = -1;
1593 1594 1595 1596
    FILE *list = NULL;
    regex_t *reg;
    regmatch_t *vars = NULL;
    char line[1024];
1597 1598
    int maxReg = 0;
    size_t i, j;
1599 1600 1601 1602
    int totgroups = 0, ngroup = 0, maxvars = 0;
    char **groups;

    /* Compile all regular expressions */
1603
    if (VIR_ALLOC_N(reg, nregex) < 0)
1604 1605
        return -1;

1606
    for (i = 0; i < nregex; i++) {
1607 1608 1609 1610
        err = regcomp(&reg[i], regex[i], REG_EXTENDED);
        if (err != 0) {
            char error[100];
            regerror(err, &reg[i], error, sizeof(error));
1611 1612
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Failed to compile regex %s"), error);
1613
            for (j = 0; j < i; j++)
1614
                regfree(&reg[j]);
1615
            VIR_FREE(reg);
1616 1617 1618 1619 1620 1621 1622 1623 1624 1625
            return -1;
        }

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

    }

    /* Storage for matched variables */
1626
    if (VIR_ALLOC_N(groups, totgroups) < 0)
1627
        goto cleanup;
1628
    if (VIR_ALLOC_N(vars, maxvars+1) < 0)
1629 1630
        goto cleanup;

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

1636
    if ((list = VIR_FDOPEN(fd, "r")) == NULL) {
1637 1638
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("cannot read fd"));
1639 1640 1641 1642
        goto cleanup;
    }

    while (fgets(line, sizeof(line), list) != NULL) {
1643
        char *p = NULL;
1644 1645 1646 1647 1648
        /* Strip trailing newline */
        int len = strlen(line);
        if (len && line[len-1] == '\n')
            line[len-1] = '\0';

1649 1650 1651 1652 1653 1654
        /* ignore any command prefix */
        if (prefix)
            p = STRSKIP(line, prefix);
        if (!p)
            p = line;

1655
        for (i = 0; i <= maxReg && i < nregex; i++) {
1656
            if (regexec(&reg[i], p, nvars[i]+1, vars, 0) == 0) {
1657 1658 1659 1660 1661 1662
                maxReg++;

                if (i == 0)
                    ngroup = 0;

                /* NULL terminate each captured group in the line */
1663
                for (j = 0; j < nvars[i]; j++) {
1664
                    /* NB vars[0] is the full pattern, so we offset j by 1 */
1665
                    p[vars[j+1].rm_eo] = '\0';
1666
                    if (VIR_STRDUP(groups[ngroup++], p + vars[j+1].rm_so) < 0)
1667 1668 1669 1670 1671
                        goto cleanup;
                }

                /* We're matching on the last regex, so callback time */
                if (i == (nregex-1)) {
1672
                    if (((*func)(pool, groups, data)) < 0)
1673 1674 1675
                        goto cleanup;

                    /* Release matches & restart to matching the first regex */
1676
                    for (j = 0; j < totgroups; j++)
1677
                        VIR_FREE(groups[j]);
1678 1679 1680 1681 1682 1683 1684
                    maxReg = 0;
                    ngroup = 0;
                }
            }
        }
    }

1685
    ret = virCommandWait(cmd, NULL);
1686
cleanup:
1687
    if (groups) {
1688
        for (j = 0; j < totgroups; j++)
1689 1690
            VIR_FREE(groups[j]);
        VIR_FREE(groups);
1691
    }
1692
    VIR_FREE(vars);
1693

1694
    for (i = 0; i < nregex; i++)
1695 1696
        regfree(&reg[i]);

1697
    VIR_FREE(reg);
1698

1699 1700
    VIR_FORCE_FCLOSE(list);
    VIR_FORCE_CLOSE(fd);
1701

1702
    return ret;
1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716
}

/*
 * 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
1717
virStorageBackendRunProgNul(virStoragePoolObjPtr pool,
1718
                            virCommandPtr cmd,
1719 1720 1721 1722 1723
                            size_t n_columns,
                            virStorageBackendListVolNulFunc func,
                            void *data)
{
    size_t n_tok = 0;
1724
    int fd = -1;
1725 1726
    FILE *fp = NULL;
    char **v;
1727
    int ret = -1;
1728
    size_t i;
1729 1730 1731 1732

    if (n_columns == 0)
        return -1;

1733
    if (VIR_ALLOC_N(v, n_columns) < 0)
1734 1735 1736 1737
        return -1;
    for (i = 0; i < n_columns; i++)
        v[i] = NULL;

1738 1739
    virCommandSetOutputFD(cmd, &fd);
    if (virCommandRunAsync(cmd, NULL) < 0) {
1740 1741 1742
        goto cleanup;
    }

1743
    if ((fp = VIR_FDOPEN(fd, "r")) == NULL) {
1744 1745
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("cannot open file using fd"));
1746 1747 1748 1749 1750 1751 1752 1753
        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.  */
1754
        ssize_t tok_len = getdelim(&buf, &buf_len, 0, fp);
1755 1756 1757 1758
        v[n_tok] = buf;
        if (tok_len < 0) {
            /* Maybe EOF, maybe an error.
               If n_tok > 0, then we know it's an error.  */
1759
            if (n_tok && func(pool, n_tok, v, data) < 0)
1760 1761 1762 1763 1764
                goto cleanup;
            break;
        }
        ++n_tok;
        if (n_tok == n_columns) {
1765
            if (func(pool, n_tok, v, data) < 0)
1766 1767 1768
                goto cleanup;
            n_tok = 0;
            for (i = 0; i < n_columns; i++) {
1769
                VIR_FREE(v[i]);
1770 1771 1772 1773
            }
        }
    }

1774 1775 1776
    if (feof(fp) < 0) {
        virReportSystemError(errno, "%s",
                             _("read error on pipe"));
1777 1778
        goto cleanup;
    }
1779

1780
    ret = virCommandWait(cmd, NULL);
1781 1782
 cleanup:
    for (i = 0; i < n_columns; i++)
1783 1784
        VIR_FREE(v[i]);
    VIR_FREE(v);
1785

1786 1787
    VIR_FORCE_FCLOSE(fp);
    VIR_FORCE_CLOSE(fd);
1788

1789
    return ret;
1790
}
D
Daniel P. Berrange 已提交
1791

1792
#else /* WIN32 */
D
Daniel P. Berrange 已提交
1793 1794 1795 1796 1797 1798 1799 1800 1801

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,
1802
                              void *data ATTRIBUTE_UNUSED)
D
Daniel P. Berrange 已提交
1803
{
1804 1805
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816
    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)
{
1817 1818
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1819 1820
    return -1;
}
1821
#endif /* WIN32 */