storage_backend.c 64.2 KB
Newer Older
1
/*
2
 * storage_backend.c: internal storage driver backend contract
3
 *
4
 * Copyright (C) 2007-2014 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
#ifdef __linux__
# include <sys/ioctl.h>
# include <linux/fs.h>
C
Chunyan Liu 已提交
40 41 42
# ifndef FS_NOCOW_FL
#  define FS_NOCOW_FL                     0x00800000 /* Do not cow file */
# endif
43
#endif
44

45
#if WITH_SELINUX
46
# include <selinux/selinux.h>
47
#endif
48

49 50 51 52
#if HAVE_LINUX_BTRFS_H
# include <linux/btrfs.h>
#endif

53
#include "datatypes.h"
54
#include "virerror.h"
55
#include "viralloc.h"
56
#include "internal.h"
57
#include "secret_conf.h"
58
#include "viruuid.h"
59
#include "virstoragefile.h"
D
Daniel P. Berrange 已提交
60
#include "storage_backend.h"
61
#include "virlog.h"
E
Eric Blake 已提交
62
#include "virfile.h"
63
#include "stat-time.h"
64
#include "virstring.h"
65
#include "virxml.h"
66
#include "fdstream.h"
67

68
#if WITH_STORAGE_LVM
69
# include "storage_backend_logical.h"
70 71
#endif
#if WITH_STORAGE_ISCSI
72
# include "storage_backend_iscsi.h"
73
#endif
74
#if WITH_STORAGE_SCSI
75
# include "storage_backend_scsi.h"
76
#endif
D
Dave Allan 已提交
77
#if WITH_STORAGE_MPATH
78
# include "storage_backend_mpath.h"
D
Dave Allan 已提交
79
#endif
80
#if WITH_STORAGE_DISK
81
# include "storage_backend_disk.h"
82 83
#endif
#if WITH_STORAGE_DIR
84
# include "storage_backend_fs.h"
85
#endif
86 87 88
#if WITH_STORAGE_RBD
# include "storage_backend_rbd.h"
#endif
89 90 91
#if WITH_STORAGE_SHEEPDOG
# include "storage_backend_sheepdog.h"
#endif
92 93 94
#if WITH_STORAGE_GLUSTER
# include "storage_backend_gluster.h"
#endif
R
Roman Bogorodskiy 已提交
95 96 97
#if WITH_STORAGE_ZFS
# include "storage_backend_zfs.h"
#endif
98

99 100
#define VIR_FROM_THIS VIR_FROM_STORAGE

101 102
VIR_LOG_INIT("storage.storage_backend");

103 104 105 106 107 108 109 110 111 112 113 114 115 116
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
117 118 119
#if WITH_STORAGE_SCSI
    &virStorageBackendSCSI,
#endif
D
Dave Allan 已提交
120 121 122
#if WITH_STORAGE_MPATH
    &virStorageBackendMpath,
#endif
123 124
#if WITH_STORAGE_DISK
    &virStorageBackendDisk,
125 126 127
#endif
#if WITH_STORAGE_RBD
    &virStorageBackendRBD,
128 129 130
#endif
#if WITH_STORAGE_SHEEPDOG
    &virStorageBackendSheepdog,
131 132 133
#endif
#if WITH_STORAGE_GLUSTER
    &virStorageBackendGluster,
R
Roman Bogorodskiy 已提交
134 135 136
#endif
#if WITH_STORAGE_ZFS
    &virStorageBackendZFS,
137 138 139
#endif
    NULL
};
140

141 142

static virStorageFileBackendPtr fileBackends[] = {
143 144 145
#if WITH_STORAGE_FS
    &virStorageFileBackendFile,
    &virStorageFileBackendBlock,
146 147 148
#endif
#if WITH_STORAGE_GLUSTER
    &virStorageFileBackendGluster,
149
#endif
150 151 152 153
    NULL
};


154 155 156 157 158 159
enum {
    TOOL_QEMU_IMG,
    TOOL_KVM_IMG,
    TOOL_QCOW_CREATE,
};

160 161 162
#define READ_BLOCK_SIZE_DEFAULT  (1024 * 1024)
#define WRITE_BLOCK_SIZE_DEFAULT (4 * 1024)

163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182
/*
 * Perform the O(1) btrfs clone operation, if possible.
 * Upon success, return 0.  Otherwise, return -1 and set errno.
 */
#if HAVE_LINUX_BTRFS_H
static inline int
btrfsCloneFile(int dest_fd, int src_fd)
{
    return ioctl(dest_fd, BTRFS_IOC_CLONE, src_fd);
}
#else
static inline int
btrfsCloneFile(int dest_fd ATTRIBUTE_UNUSED,
               int src_fd ATTRIBUTE_UNUSED)
{
    errno = ENOTSUP;
    return -1;
}
#endif

183
static int ATTRIBUTE_NONNULL(2)
184
virStorageBackendCopyToFD(virStorageVolDefPtr vol,
185 186
                          virStorageVolDefPtr inputvol,
                          int fd,
187
                          unsigned long long *total,
188 189
                          bool want_sparse,
                          bool reflink_copy)
190 191 192
{
    int inputfd = -1;
    int amtread = -1;
193
    int ret = 0;
194
    size_t rbytes = READ_BLOCK_SIZE_DEFAULT;
L
Li Zhang 已提交
195
    int wbytes = 0;
196
    int interval;
197
    char *zerobuf = NULL;
198
    char *buf = NULL;
199
    struct stat st;
200

201
    if ((inputfd = open(inputvol->target.path, O_RDONLY)) < 0) {
202
        ret = -errno;
203
        virReportSystemError(errno,
204 205 206
                             _("could not open input path '%s'"),
                             inputvol->target.path);
        goto cleanup;
207 208
    }

209
#ifdef __linux__
210
    if (ioctl(fd, BLKBSZGET, &wbytes) < 0)
211 212 213 214 215 216 217 218 219 220 221
        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;
    }
222

223
    if (VIR_ALLOC_N(buf, rbytes) < 0) {
224
        ret = -errno;
225 226 227
        goto cleanup;
    }

228 229 230 231 232 233 234 235 236 237 238 239 240
    if (reflink_copy) {
        if (btrfsCloneFile(fd, inputfd) < 0) {
            ret = -errno;
            virReportSystemError(errno,
                                 _("failed to clone files from '%s'"),
                                 inputvol->target.path);
            goto cleanup;
        } else {
            VIR_DEBUG("btrfs clone finished.");
            goto cleanup;
        }
    }

241 242 243
    while (amtread != 0) {
        int amtleft;

244 245
        if (*total < rbytes)
            rbytes = *total;
246

247
        if ((amtread = saferead(inputfd, buf, rbytes)) < 0) {
248
            ret = -errno;
249
            virReportSystemError(errno,
250 251 252 253
                                 _("failed reading from file '%s'"),
                                 inputvol->target.path);
            goto cleanup;
        }
254
        *total -= amtread;
255 256 257 258 259

        /* Loop over amt read in 512 byte increments, looking for sparse
         * blocks */
        amtleft = amtread;
        do {
260
            interval = ((wbytes > amtleft) ? amtleft : wbytes);
261 262
            int offset = amtread - amtleft;

263
            if (want_sparse && memcmp(buf+offset, zerobuf, interval) == 0) {
264
                if (lseek(fd, interval, SEEK_CUR) < 0) {
265
                    ret = -errno;
266
                    virReportSystemError(errno,
267 268 269 270 271
                                         _("cannot extend file '%s'"),
                                         vol->target.path);
                    goto cleanup;
                }
            } else if (safewrite(fd, buf+offset, interval) < 0) {
272
                ret = -errno;
273
                virReportSystemError(errno,
274 275 276 277 278
                                     _("failed writing to file '%s'"),
                                     vol->target.path);
                goto cleanup;

            }
279
        } while ((amtleft -= interval) > 0);
280 281
    }

282 283 284 285 286 287 288 289
    if (fdatasync(fd) < 0) {
        ret = -errno;
        virReportSystemError(errno, _("cannot sync data to file '%s'"),
                             vol->target.path);
        goto cleanup;
    }


290
    if (VIR_CLOSE(inputfd) < 0) {
291
        ret = -errno;
292
        virReportSystemError(errno,
293 294 295 296 297 298
                             _("cannot close file '%s'"),
                             inputvol->target.path);
        goto cleanup;
    }
    inputfd = -1;

299
 cleanup:
300
    VIR_FORCE_CLOSE(inputfd);
301

302
    VIR_FREE(zerobuf);
303 304
    VIR_FREE(buf);

305 306 307
    return ret;
}

308
static int
309
virStorageBackendCreateBlockFrom(virConnectPtr conn ATTRIBUTE_UNUSED,
310
                                 virStoragePoolObjPtr pool ATTRIBUTE_UNUSED,
311 312
                                 virStorageVolDefPtr vol,
                                 virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
313
                                 unsigned int flags)
314 315 316 317
{
    int fd = -1;
    int ret = -1;
    unsigned long long remain;
318 319 320
    struct stat st;
    gid_t gid;
    uid_t uid;
321
    mode_t mode;
322
    bool reflink_copy = false;
323

324 325 326
    virCheckFlags(VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA |
                  VIR_STORAGE_VOL_CREATE_REFLINK,
                  -1);
327 328 329 330 331 332 333

    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
334

335 336 337
    if (flags & VIR_STORAGE_VOL_CREATE_REFLINK)
        reflink_copy = true;

338
    if ((fd = open(vol->target.path, O_RDWR)) < 0) {
339
        virReportSystemError(errno,
340 341 342 343 344
                             _("cannot create path '%s'"),
                             vol->target.path);
        goto cleanup;
    }

345
    remain = vol->target.capacity;
346 347

    if (inputvol) {
348
        int res = virStorageBackendCopyToFD(vol, inputvol,
349
                                            fd, &remain, false, reflink_copy);
350 351 352 353
        if (res < 0)
            goto cleanup;
    }

354
    if (fstat(fd, &st) == -1) {
355
        virReportSystemError(errno, _("stat of '%s' failed"),
356 357 358
                             vol->target.path);
        goto cleanup;
    }
359 360 361 362
    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;
P
Philipp Hahn 已提交
363
    if (((uid != (uid_t) -1) || (gid != (gid_t) -1))
364
        && (fchown(fd, uid, gid) < 0)) {
365
        virReportSystemError(errno,
366
                             _("cannot chown '%s' to (%u, %u)"),
367 368
                             vol->target.path, (unsigned int) uid,
                             (unsigned int) gid);
369 370
        goto cleanup;
    }
371 372 373 374

    mode = (vol->target.perms->mode == (mode_t) -1 ?
            VIR_STORAGE_DEFAULT_VOL_PERM_MODE : vol->target.perms->mode);
    if (fchmod(fd, mode) < 0) {
375
        virReportSystemError(errno,
376
                             _("cannot set mode of '%s' to %04o"),
377
                             vol->target.path, mode);
378 379
        goto cleanup;
    }
380
    if (VIR_CLOSE(fd) < 0) {
381
        virReportSystemError(errno,
382 383 384 385 386 387 388
                             _("cannot close file '%s'"),
                             vol->target.path);
        goto cleanup;
    }
    fd = -1;

    ret = 0;
389
 cleanup:
390
    VIR_FORCE_CLOSE(fd);
391 392 393 394

    return ret;
}

E
Eric Blake 已提交
395 396
static int
createRawFile(int fd, virStorageVolDefPtr vol,
397 398
              virStorageVolDefPtr inputvol,
              bool reflink_copy)
E
Eric Blake 已提交
399
{
J
Ján Tomko 已提交
400
    bool need_alloc = true;
401
    int ret = 0;
402
    unsigned long long pos = 0;
403

404 405 406 407 408 409
    /* If the new allocation is lower than the capacity of the original file,
     * the cloned volume will be sparse */
    if (inputvol &&
        vol->target.allocation < inputvol->target.capacity)
        need_alloc = false;

410 411
    /* Seek to the final size, so the capacity is available upfront
     * for progress reporting */
412
    if (ftruncate(fd, vol->target.capacity) < 0) {
413
        ret = -errno;
414
        virReportSystemError(errno,
415
                             _("cannot extend file '%s'"),
E
Eric Blake 已提交
416
                             vol->target.path);
417 418 419
        goto cleanup;
    }

E
Eric Blake 已提交
420 421
/* Avoid issues with older kernel's <linux/fs.h> namespace pollution. */
#if HAVE_FALLOCATE - 0
422
    /* Try to preallocate all requested disk space, but fall back to
423 424
     * other methods if this fails with ENOSYS or EOPNOTSUPP. If allocation
     * is 0 (or less than 0), then fallocate will fail with EINVAL.
425 426 427 428
     * 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. */
429
    if (vol->target.allocation && need_alloc) {
430
        if (fallocate(fd, 0, 0, vol->target.allocation) == 0) {
431 432 433 434 435
            need_alloc = false;
        } else if (errno != ENOSYS && errno != EOPNOTSUPP) {
            ret = -errno;
            virReportSystemError(errno,
                                 _("cannot allocate %llu bytes in file '%s'"),
436
                                 vol->target.allocation, vol->target.path);
437 438
            goto cleanup;
        }
439 440 441
    }
#endif

E
Eric Blake 已提交
442
    if (inputvol) {
443
        unsigned long long remain = inputvol->target.capacity;
444 445 446
        /* 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. */
447
        ret = virStorageBackendCopyToFD(vol, inputvol, fd, &remain,
448
                                        !need_alloc, reflink_copy);
449
        if (ret < 0)
450
            goto cleanup;
451

452 453 454 455
        /* If the new allocation is greater than the original capacity,
         * but fallocate failed, fill the rest with zeroes.
         */
        pos = inputvol->target.capacity - remain;
456 457
    }

458
    if (need_alloc && (vol->target.allocation - pos > 0)) {
459
        if (safezero(fd, pos, vol->target.allocation - pos) < 0) {
460 461 462 463
            ret = -errno;
            virReportSystemError(errno, _("cannot fill file '%s'"),
                                 vol->target.path);
            goto cleanup;
464
        }
465 466 467
    }

    if (fsync(fd) < 0) {
468
        ret = -errno;
469
        virReportSystemError(errno, _("cannot sync data to file '%s'"),
E
Eric Blake 已提交
470
                             vol->target.path);
471
        goto cleanup;
472 473
    }

474
 cleanup:
475 476 477 478 479 480 481 482
    return ret;
}

int
virStorageBackendCreateRaw(virConnectPtr conn ATTRIBUTE_UNUSED,
                           virStoragePoolObjPtr pool,
                           virStorageVolDefPtr vol,
                           virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
483
                           unsigned int flags)
484 485
{
    int ret = -1;
E
Eric Blake 已提交
486 487
    int fd = -1;
    int operation_flags;
488
    bool reflink_copy = false;
E
Eric Blake 已提交
489

490 491 492
    virCheckFlags(VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA |
                  VIR_STORAGE_VOL_CREATE_REFLINK,
                  -1);
493 494 495 496 497 498 499

    if (flags & VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("metadata preallocation is not supported for raw "
                         "volumes"));
        goto cleanup;
    }
500

501 502 503 504 505 506
    if (vol->target.backingStore) {
        virReportError(VIR_ERR_NO_SUPPORT, "%s",
                       _("backing storage not supported for raw volumes"));
        goto cleanup;
    }

507 508 509 510
    if (flags & VIR_STORAGE_VOL_CREATE_REFLINK)
        reflink_copy = true;


511
    if (vol->target.encryption != NULL) {
512 513
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("storage pool does not support encrypted volumes"));
514 515
        goto cleanup;
    }
516

L
Laine Stump 已提交
517
    operation_flags = VIR_FILE_OPEN_FORCE_MODE | VIR_FILE_OPEN_FORCE_OWNER;
E
Eric Blake 已提交
518
    if (pool->def->type == VIR_STORAGE_POOL_NETFS)
L
Laine Stump 已提交
519
        operation_flags |= VIR_FILE_OPEN_FORK;
E
Eric Blake 已提交
520

521 522
    if ((fd = virFileOpenAs(vol->target.path,
                            O_RDWR | O_CREAT | O_EXCL,
523 524 525
                            (vol->target.perms->mode ?
                             VIR_STORAGE_DEFAULT_VOL_PERM_MODE :
                             vol->target.perms->mode),
526 527
                            vol->target.perms->uid,
                            vol->target.perms->gid,
528
                            operation_flags)) < 0) {
E
Eric Blake 已提交
529
        virReportSystemError(-fd,
530
                             _("Failed to create file '%s'"),
E
Eric Blake 已提交
531 532 533 534
                             vol->target.path);
        goto cleanup;
    }

C
Chunyan Liu 已提交
535 536 537 538 539 540
    if (vol->target.nocow) {
#ifdef __linux__
        int attr;

        /* Set NOCOW flag. This is an optimisation for btrfs.
         * The FS_IOC_SETFLAGS ioctl return value will be ignored since any
541
         * failure of this operation should not block the volume creation.
C
Chunyan Liu 已提交
542
         */
543 544 545
        if (ioctl(fd, FS_IOC_GETFLAGS, &attr) < 0) {
            virReportSystemError(errno, "%s", _("Failed to get fs flags"));
        } else {
C
Chunyan Liu 已提交
546
            attr |= FS_NOCOW_FL;
547 548 549 550
            if (ioctl(fd, FS_IOC_SETFLAGS, &attr) < 0) {
                virReportSystemError(errno, "%s",
                                     _("Failed to set NOCOW flag"));
            }
C
Chunyan Liu 已提交
551 552 553 554
        }
#endif
    }

555
    if ((ret = createRawFile(fd, vol, inputvol, reflink_copy)) < 0)
556
        /* createRawFile already reported the exact error. */
E
Eric Blake 已提交
557
        ret = -1;
558

559
 cleanup:
E
Eric Blake 已提交
560
    VIR_FORCE_CLOSE(fd);
561 562 563
    return ret;
}

564 565 566 567 568 569 570 571 572
static int
virStorageGenerateSecretUUID(virConnectPtr conn,
                             unsigned char *uuid)
{
    unsigned attempt;

    for (attempt = 0; attempt < 65536; attempt++) {
        virSecretPtr tmp;
        if (virUUIDGenerate(uuid) < 0) {
573 574
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("unable to generate uuid"));
575 576
            return -1;
        }
577
        tmp = conn->secretDriver->secretLookupByUUID(conn, uuid);
578 579 580
        if (tmp == NULL)
            return 0;

581
        virObjectUnref(tmp);
582 583
    }

584
    virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
N
Nehal J Wani 已提交
585
                   _("too many conflicts when generating a uuid"));
586 587 588 589

    return -1;
}

590 591 592 593 594 595 596 597 598
static int
virStorageGenerateQcowEncryption(virConnectPtr conn,
                                 virStorageVolDefPtr vol)
{
    virSecretDefPtr def = NULL;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    virStorageEncryptionPtr enc;
    virStorageEncryptionSecretPtr enc_secret = NULL;
    virSecretPtr secret = NULL;
599
    char *xml;
600 601 602
    unsigned char value[VIR_STORAGE_QCOW_PASSPHRASE_SIZE];
    int ret = -1;

603
    if (conn->secretDriver == NULL ||
604 605 606
        conn->secretDriver->secretLookupByUUID == NULL ||
        conn->secretDriver->secretDefineXML == NULL ||
        conn->secretDriver->secretSetValue == NULL) {
607 608
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("secret storage not supported"));
609 610 611 612 613
        goto cleanup;
    }

    enc = vol->target.encryption;
    if (enc->nsecrets != 0) {
614 615
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("secrets already defined"));
616 617 618 619
        goto cleanup;
    }

    if (VIR_ALLOC(enc_secret) < 0 || VIR_REALLOC_N(enc->secrets, 1) < 0 ||
620
        VIR_ALLOC(def) < 0)
621 622
        goto cleanup;

623
    def->ephemeral = false;
624
    def->private = false;
625
    if (virStorageGenerateSecretUUID(conn, def->uuid) < 0)
626
        goto cleanup;
627

628
    def->usage_type = VIR_SECRET_USAGE_TYPE_VOLUME;
629
    if (VIR_STRDUP(def->usage.volume, vol->target.path) < 0)
630
        goto cleanup;
631
    xml = virSecretDefFormat(def);
632 633 634 635 636
    virSecretDefFree(def);
    def = NULL;
    if (xml == NULL)
        goto cleanup;

637
    secret = conn->secretDriver->secretDefineXML(conn, xml, 0);
638 639 640 641 642 643
    if (secret == NULL) {
        VIR_FREE(xml);
        goto cleanup;
    }
    VIR_FREE(xml);

644
    if (virStorageGenerateQcowPassphrase(value) < 0)
645 646
        goto cleanup;

647
    if (conn->secretDriver->secretSetValue(secret, value, sizeof(value), 0) < 0)
648 649 650
        goto cleanup;

    enc_secret->type = VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE;
651
    memcpy(enc_secret->uuid, secret->uuid, VIR_UUID_BUFLEN);
652 653 654 655 656 657 658
    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;

659
 cleanup:
660
    if (secret != NULL) {
661
        if (ret != 0 &&
662 663
            conn->secretDriver->secretUndefine != NULL)
            conn->secretDriver->secretUndefine(secret);
664
        virObjectUnref(secret);
665
    }
666
    virBufferFreeAndReset(&buf);
667 668 669 670 671
    virSecretDefFree(def);
    VIR_FREE(enc_secret);
    return ret;
}

672 673 674 675 676
static int
virStorageBackendCreateExecCommand(virStoragePoolObjPtr pool,
                                   virStorageVolDefPtr vol,
                                   virCommandPtr cmd)
{
677 678 679
    struct stat st;
    gid_t gid;
    uid_t uid;
680
    mode_t mode;
681
    bool filecreated = false;
682 683

    if ((pool->def->type == VIR_STORAGE_POOL_NETFS)
684
        && (((geteuid() == 0)
685 686 687 688
             && (vol->target.perms->uid != (uid_t) -1)
             && (vol->target.perms->uid != 0))
            || ((vol->target.perms->gid != (gid_t) -1)
                && (vol->target.perms->gid != getegid())))) {
689

690 691
        virCommandSetUID(cmd, vol->target.perms->uid);
        virCommandSetGID(cmd, vol->target.perms->gid);
692 693

        if (virCommandRun(cmd, NULL) == 0) {
694
            /* command was successfully run, check if the file was created */
695
            if (stat(vol->target.path, &st) >= 0)
696
                filecreated = true;
697 698
        }
    }
699

700 701 702
    /* don't change uid/gid if we retry */
    virCommandSetUID(cmd, -1);
    virCommandSetGID(cmd, -1);
703

704
    if (!filecreated) {
705
        if (virCommandRun(cmd, NULL) < 0)
706 707
            return -1;
        if (stat(vol->target.path, &st) < 0) {
708
            virReportSystemError(errno,
709
                                 _("failed to create %s"), vol->target.path);
710 711 712 713
            return -1;
        }
    }

714 715 716 717
    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;
P
Philipp Hahn 已提交
718
    if (((uid != (uid_t) -1) || (gid != (gid_t) -1))
719
        && (chown(vol->target.path, uid, gid) < 0)) {
720
        virReportSystemError(errno,
721
                             _("cannot chown %s to (%u, %u)"),
722 723
                             vol->target.path, (unsigned int) uid,
                             (unsigned int) gid);
724 725
        return -1;
    }
726 727 728 729

    mode = (vol->target.perms->mode == (mode_t) -1 ?
            VIR_STORAGE_DEFAULT_VOL_PERM_MODE : vol->target.perms->mode);
    if (chmod(vol->target.path, mode) < 0) {
730
        virReportSystemError(errno,
731
                             _("cannot set mode of '%s' to %04o"),
732
                             vol->target.path, mode);
733 734 735 736 737
        return -1;
    }
    return 0;
}

738 739 740 741
enum {
    QEMU_IMG_BACKING_FORMAT_NONE = 0,
    QEMU_IMG_BACKING_FORMAT_FLAG,
    QEMU_IMG_BACKING_FORMAT_OPTIONS,
742
    QEMU_IMG_BACKING_FORMAT_OPTIONS_COMPAT,
743 744
};

745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763
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;

764
 cleanup:
765 766 767 768 769
    virCommandFree(cmd);
    VIR_FREE(output);
    return ret;
}

770 771
static int
virStorageBackendQEMUImgBackingFormat(const char *qemuimg)
772 773 774 775 776 777
{
    char *help = NULL;
    char *start;
    char *end;
    char *tmp;
    int ret = -1;
778
    int exitstatus;
779
    virCommandPtr cmd = virCommandNewArgList(qemuimg, "-h", NULL);
780

781 782 783
    virCommandAddEnvString(cmd, "LC_ALL=C");
    virCommandSetOutputBuffer(cmd, &help);
    virCommandClearCaps(cmd);
784

785 786 787
    /* 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 */
788
    if (virCommandRun(cmd, &exitstatus) < 0)
789 790
        goto cleanup;

791 792
    if ((start = strstr(help, " create ")) == NULL ||
        (end = strstr(start, "\n")) == NULL) {
793 794 795
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unable to parse qemu-img output '%s'"),
                       help);
796 797
        goto cleanup;
    }
798
    if (((tmp = strstr(start, "-F fmt")) && tmp < end) ||
799
        ((tmp = strstr(start, "-F backing_fmt")) && tmp < end)) {
800
        ret = QEMU_IMG_BACKING_FORMAT_FLAG;
801 802 803 804 805 806
    } 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 {
807
        ret = QEMU_IMG_BACKING_FORMAT_NONE;
808
    }
809

810
 cleanup:
811
    virCommandFree(cmd);
812 813 814 815
    VIR_FREE(help);
    return ret;
}

816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832
struct _virStorageBackendQemuImgInfo {
    int format;
    const char *path;
    unsigned long long size_arg;
    bool encryption;
    bool preallocate;
    const char *compat;
    virBitmapPtr features;
    bool nocow;

    const char *backingPath;
    int backingFormat;

    const char *inputPath;
    int inputFormat;
};

833 834
static int
virStorageBackendCreateQemuImgOpts(char **opts,
835
                                   struct _virStorageBackendQemuImgInfo info)
836 837
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;
838

839 840 841 842
    if (info.backingPath)
        virBufferAsprintf(&buf, "backing_fmt=%s,",
                          virStorageFileFormatTypeToString(info.backingFormat));
    if (info.encryption)
843
        virBufferAddLit(&buf, "encryption=on,");
844
    if (info.preallocate)
845
        virBufferAddLit(&buf, "preallocation=metadata,");
846
    if (info.nocow)
C
Chunyan Liu 已提交
847
        virBufferAddLit(&buf, "nocow=on,");
848

849 850
    if (info.compat)
        virBufferAsprintf(&buf, "compat=%s,", info.compat);
J
Ján Tomko 已提交
851

852
    if (info.features && info.format == VIR_STORAGE_FILE_QCOW2) {
J
Ján Tomko 已提交
853 854 855 856 857 858 859 860
        if (virBitmapIsBitSet(info.features,
                              VIR_STORAGE_FILE_FEATURE_LAZY_REFCOUNTS)) {
            if (STREQ_NULLABLE(info.compat, "0.10")) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("lazy_refcounts not supported with compat"
                                 " level %s"),
                               info.compat);
                goto error;
861
            }
J
Ján Tomko 已提交
862
            virBufferAddLit(&buf, "lazy_refcounts,");
863 864 865
        }
    }

866 867
    virBufferTrim(&buf, ",", -1);

868 869
    if (virBufferCheckError(&buf) < 0)
        goto error;
870 871 872

    *opts = virBufferContentAndReset(&buf);
    return 0;
873

874
 error:
875 876
    virBufferFreeAndReset(&buf);
    return -1;
877 878
}

879 880 881
/* Create a qemu-img virCommand from the supplied binary path,
 * volume definitions and imgformat
 */
882
virCommandPtr
883 884 885 886 887 888 889
virStorageBackendCreateQemuImgCmdFromVol(virConnectPtr conn,
                                         virStoragePoolObjPtr pool,
                                         virStorageVolDefPtr vol,
                                         virStorageVolDefPtr inputvol,
                                         unsigned int flags,
                                         const char *create_tool,
                                         int imgformat)
890
{
891
    virCommandPtr cmd = NULL;
892 893 894
    const char *type;
    const char *backingType = NULL;
    const char *inputType = NULL;
895
    char *opts = NULL;
896 897 898 899 900 901 902 903 904
    struct _virStorageBackendQemuImgInfo info = {
        .format = vol->target.format,
        .path = vol->target.path,
        .encryption = vol->target.encryption != NULL,
        .preallocate = !!(flags & VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA),
        .compat = vol->target.compat,
        .features = vol->target.features,
        .nocow = vol->target.nocow,
    };
905

906 907
    virCheckFlags(VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA, NULL);

908
    /* Treat output block devices as 'raw' format */
909 910
    if (vol->type == VIR_STORAGE_VOL_BLOCK)
        info.format = VIR_STORAGE_FILE_RAW;
911

912
    if (!(type = virStorageFileFormatTypeToString(info.format))) {
913 914
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unknown storage vol type %d"),
915
                       info.format);
916
        return NULL;
917
    }
918

919
    if (info.preallocate && info.format != VIR_STORAGE_FILE_QCOW2) {
920 921
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("metadata preallocation only available with qcow2"));
922
        return NULL;
923
    }
924
    if (info.compat && info.format != VIR_STORAGE_FILE_QCOW2) {
925 926 927 928
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("compatibility option only available with qcow2"));
        return NULL;
    }
929
    if (info.features && info.format != VIR_STORAGE_FILE_QCOW2) {
930 931 932 933
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("format features only available with qcow2"));
        return NULL;
    }
934

935
    if (inputvol) {
936
        if (!(info.inputPath = inputvol->target.path)) {
937 938 939 940 941
            virReportError(VIR_ERR_INVALID_ARG, "%s",
                           _("missing input volume target path"));
            return NULL;
        }

942 943 944 945
        info.inputFormat = inputvol->target.format;
        if (inputvol->type == VIR_STORAGE_VOL_BLOCK)
            info.inputFormat = VIR_STORAGE_FILE_RAW;
        if (!(inputType = virStorageFileFormatTypeToString(info.inputFormat))) {
946 947
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unknown storage vol type %d"),
948
                           info.inputFormat);
949 950 951 952
            return NULL;
        }
    }

953
    if (vol->target.backingStore) {
954 955
        int accessRetCode = -1;
        char *absolutePath = NULL;
956

957 958
        info.backingFormat = vol->target.backingStore->format;
        info.backingPath = vol->target.backingStore->path;
959

960
        if (info.preallocate) {
961 962 963
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("metadata preallocation conflicts with backing"
                             " store"));
964
            return NULL;
965 966
        }

967 968 969 970
        /* 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.
         */
971
        if (inputvol && inputvol->target.backingStore &&
972
            STRNEQ_NULLABLE(inputvol->target.backingStore->path, info.backingPath)) {
973 974
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("a different backing store cannot be specified."));
975
            return NULL;
976 977
        }

978
        if (!(backingType = virStorageFileFormatTypeToString(info.backingFormat))) {
979 980
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unknown storage vol backing store type %d"),
981
                           info.backingFormat);
982
            return NULL;
983
        }
984 985 986 987

        /* Convert relative backing store paths to absolute paths for access
         * validation.
         */
988
        if ('/' != *(info.backingPath) &&
989
            virAsprintf(&absolutePath, "%s/%s", pool->def->target.path,
990
                        info.backingPath) < 0)
991
            return NULL;
992
        accessRetCode = access(absolutePath ? absolutePath : info.backingPath, R_OK);
993 994
        VIR_FREE(absolutePath);
        if (accessRetCode != 0) {
995
            virReportSystemError(errno,
996
                                 _("inaccessible backing store volume %s"),
997
                                 info.backingPath);
998
            return NULL;
999 1000 1001
        }
    }

1002
    if (info.encryption) {
1003 1004
        virStorageEncryptionPtr enc;

1005 1006
        if (info.format != VIR_STORAGE_FILE_QCOW &&
            info.format != VIR_STORAGE_FILE_QCOW2) {
1007 1008 1009
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("qcow volume encryption unsupported with "
                             "volume format %s"), type);
1010
            return NULL;
1011
        }
1012 1013 1014
        enc = vol->target.encryption;
        if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW &&
            enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT) {
1015 1016 1017
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("unsupported volume encryption format %d"),
                           vol->target.encryption->format);
1018
            return NULL;
1019
        }
1020
        if (enc->nsecrets > 1) {
1021 1022
            virReportError(VIR_ERR_XML_ERROR, "%s",
                           _("too many secrets for qcow encryption"));
1023
            return NULL;
1024
        }
1025 1026 1027
        if (enc->format == VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT ||
            enc->nsecrets == 0) {
            if (virStorageGenerateQcowEncryption(conn, vol) < 0)
1028
                return NULL;
1029
        }
1030 1031
    }

1032
    /* Size in KB */
1033
    info.size_arg = VIR_DIV_UP(vol->target.capacity, 1024);
1034

1035 1036
    cmd = virCommandNew(create_tool);

1037 1038 1039 1040 1041
    /* ignore the backing volume when we're converting a volume */
    if (info.inputPath) {
        info.backingPath = NULL;
        backingType = NULL;
    }
1042

1043
    if (info.inputPath)
1044 1045
        virCommandAddArgList(cmd, "convert", "-f", inputType, "-O", type, NULL);
    else
1046
        virCommandAddArgList(cmd, "create", "-f", type, NULL);
1047

1048 1049
    if (info.backingPath)
        virCommandAddArgList(cmd, "-b", info.backingPath, NULL);
1050

1051
    if (imgformat >= QEMU_IMG_BACKING_FORMAT_OPTIONS) {
1052
        if (info.format == VIR_STORAGE_FILE_QCOW2 && !info.compat &&
1053
            imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS_COMPAT)
1054 1055 1056
            info.compat = "0.10";

        if (virStorageBackendCreateQemuImgOpts(&opts, info) < 0) {
1057
            virCommandFree(cmd);
1058
            return NULL;
1059
        }
1060 1061 1062 1063
        if (opts)
            virCommandAddArgList(cmd, "-o", opts, NULL);
        VIR_FREE(opts);
    } else {
1064
        if (info.backingPath) {
1065 1066 1067 1068
            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",
1069
                          info.path, create_tool);
1070
        }
1071
        if (info.encryption)
1072
            virCommandAddArg(cmd, "-e");
1073
    }
1074

1075 1076 1077
    if (info.inputPath)
        virCommandAddArg(cmd, info.inputPath);
    virCommandAddArg(cmd, info.path);
1078
    if (!info.inputPath && (info.size_arg || !info.backingPath))
1079
        virCommandAddArgFormat(cmd, "%lluK", info.size_arg);
1080

1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091
    return cmd;
}

static int
virStorageBackendCreateQemuImg(virConnectPtr conn,
                               virStoragePoolObjPtr pool,
                               virStorageVolDefPtr vol,
                               virStorageVolDefPtr inputvol,
                               unsigned int flags)
{
    int ret = -1;
1092
    char *create_tool;
1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112
    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;

1113 1114
    cmd = virStorageBackendCreateQemuImgCmdFromVol(conn, pool, vol, inputvol,
                                                   flags, create_tool, imgformat);
1115 1116 1117
    if (!cmd)
        goto cleanup;

1118
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
1119 1120

    virCommandFree(cmd);
1121
 cleanup:
1122
    VIR_FREE(create_tool);
1123
    return ret;
1124 1125 1126 1127 1128 1129 1130
}

/*
 * Xen removed the fully-functional qemu-img, and replaced it
 * with a partially functional qcow-create. Go figure ??!?
 */
static int
1131
virStorageBackendCreateQcowCreate(virConnectPtr conn ATTRIBUTE_UNUSED,
1132
                                  virStoragePoolObjPtr pool,
1133
                                  virStorageVolDefPtr vol,
1134
                                  virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
1135
                                  unsigned int flags)
1136
{
1137
    int ret;
E
Eric Blake 已提交
1138
    char *size;
1139
    virCommandPtr cmd;
1140

1141 1142 1143 1144 1145 1146 1147 1148
    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
1149

1150
    if (inputvol) {
1151 1152
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("cannot copy from volume with qcow-create"));
1153 1154 1155
        return -1;
    }

1156
    if (vol->target.format != VIR_STORAGE_FILE_QCOW2) {
1157 1158 1159
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unsupported storage vol type %d"),
                       vol->target.format);
1160 1161
        return -1;
    }
1162
    if (vol->target.backingStore != NULL) {
1163 1164 1165
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("copy-on-write image not supported with "
                         "qcow-create"));
1166 1167
        return -1;
    }
1168
    if (vol->target.encryption != NULL) {
1169 1170 1171
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       "%s", _("encrypted volumes not supported with "
                               "qcow-create"));
1172 1173
        return -1;
    }
1174 1175

    /* Size in MB - yes different units to qemu-img :-( */
1176
    if (virAsprintf(&size, "%llu",
1177
                    VIR_DIV_UP(vol->target.capacity, (1024 * 1024))) < 0)
E
Eric Blake 已提交
1178
        return -1;
1179

1180
    cmd = virCommandNewArgList("qcow-create", size, vol->target.path, NULL);
1181

1182 1183
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
    virCommandFree(cmd);
E
Eric Blake 已提交
1184
    VIR_FREE(size);
1185

1186
    return ret;
1187 1188
}

1189
virStorageBackendBuildVolFrom
1190
virStorageBackendFSImageToolTypeToFunc(int tool_type)
1191 1192 1193 1194 1195 1196 1197 1198
{
    switch (tool_type) {
    case TOOL_KVM_IMG:
    case TOOL_QEMU_IMG:
        return virStorageBackendCreateQemuImg;
    case TOOL_QCOW_CREATE:
        return virStorageBackendCreateQcowCreate;
    default:
1199 1200 1201
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Unknown file create tool type '%d'."),
                       tool_type);
1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228
    }

    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;
}

1229
virStorageBackendBuildVolFrom
1230
virStorageBackendGetBuildVolFromFunction(virStorageVolDefPtr vol,
1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241
                                         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 &&
1242
         vol->target.format != VIR_STORAGE_FILE_RAW) ||
1243
        (inputvol->type == VIR_STORAGE_VOL_FILE &&
1244
         inputvol->target.format != VIR_STORAGE_FILE_RAW)) {
1245

D
Daniel P. Berrange 已提交
1246
        if ((tool_type = virStorageBackendFindFSImageTool(NULL)) < 0) {
1247 1248 1249
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("creation of non-raw file images is "
                             "not supported without qemu-img."));
1250 1251 1252
            return NULL;
        }

1253
        return virStorageBackendFSImageToolTypeToFunc(tool_type);
1254 1255
    }

1256 1257 1258 1259
    if (vol->type == VIR_STORAGE_VOL_BLOCK)
        return virStorageBackendCreateBlockFrom;
    else
        return virStorageBackendCreateRaw;
1260
}
1261

1262

1263
virStorageBackendPtr
1264 1265
virStorageBackendForType(int type)
{
1266
    size_t i;
1267
    for (i = 0; backends[i]; i++)
1268 1269 1270
        if (backends[i]->type == type)
            return backends[i];

1271
    virReportError(VIR_ERR_INTERNAL_ERROR,
1272 1273
                   _("missing backend for pool type %d (%s)"),
                   type, NULLSTR(virStoragePoolTypeToString(type)));
1274 1275 1276 1277
    return NULL;
}


1278
virStorageFileBackendPtr
1279 1280 1281
virStorageFileBackendForTypeInternal(int type,
                                     int protocol,
                                     bool report)
1282 1283 1284 1285 1286
{
    size_t i;

    for (i = 0; fileBackends[i]; i++) {
        if (fileBackends[i]->type == type) {
E
Eric Blake 已提交
1287
            if (type == VIR_STORAGE_TYPE_NETWORK &&
1288 1289 1290 1291 1292 1293 1294
                fileBackends[i]->protocol != protocol)
                continue;

            return fileBackends[i];
        }
    }

1295 1296 1297
    if (!report)
        return NULL;

E
Eric Blake 已提交
1298
    if (type == VIR_STORAGE_TYPE_NETWORK) {
1299 1300 1301
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("missing storage backend for network files "
                         "using %s protocol"),
1302
                       virStorageNetProtocolTypeToString(protocol));
1303 1304 1305
    } else {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("missing storage backend for '%s' storage"),
E
Eric Blake 已提交
1306
                       virStorageTypeToString(type));
1307 1308 1309 1310 1311 1312
    }

    return NULL;
}


1313 1314 1315 1316 1317 1318 1319 1320
virStorageFileBackendPtr
virStorageFileBackendForType(int type,
                             int protocol)
{
    return virStorageFileBackendForTypeInternal(type, protocol, true);
}


1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354
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 },
};


static int
1355
virStorageBackendDetectBlockVolFormatFD(virStorageSourcePtr target,
1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394
                                        int fd)
{
    size_t 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) {
        virReportSystemError(errno,
                             _("cannot seek to beginning of file '%s'"),
                             target->path);
        return -1;
    }
    bytes = saferead(fd, buffer, sizeof(buffer));
    if (bytes < 0) {
        virReportSystemError(errno,
                             _("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;
}


1395 1396 1397
/*
 * Allows caller to silently ignore files with improper mode
 *
1398 1399 1400
 * Returns -1 on error. If VIR_STORAGE_VOL_OPEN_NOERROR is passed, we
 * return -2 if file mode is unexpected or the volume is a dangling
 * symbolic link.
1401
 */
1402
int
1403 1404
virStorageBackendVolOpen(const char *path, struct stat *sb,
                         unsigned int flags)
1405
{
1406
    int fd, mode = 0;
1407
    char *base = last_component(path);
1408
    bool noerror = (flags & VIR_STORAGE_VOL_OPEN_NOERROR);
1409

E
Eric Blake 已提交
1410
    if (lstat(path, sb) < 0) {
1411 1412 1413 1414 1415 1416 1417 1418 1419
        if (errno == ENOENT) {
            if (noerror) {
                VIR_WARN("ignoring missing file '%s'", path);
                return -2;
            }
            virReportError(VIR_ERR_NO_STORAGE_VOL,
                           _("no storage vol with matching path '%s'"),
                           path);
            return -1;
1420
        }
1421 1422 1423 1424 1425 1426
        virReportSystemError(errno,
                             _("cannot stat file '%s'"),
                             path);
        return -1;
    }

E
Eric Blake 已提交
1427
    if (S_ISFIFO(sb->st_mode)) {
1428 1429 1430 1431 1432 1433 1434
        if (noerror) {
            VIR_WARN("ignoring FIFO '%s'", path);
            return -2;
        }
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Volume path '%s' is a FIFO"), path);
        return -1;
E
Eric Blake 已提交
1435
    } else if (S_ISSOCK(sb->st_mode)) {
1436 1437 1438 1439 1440 1441 1442
        if (noerror) {
            VIR_WARN("ignoring socket '%s'", path);
            return -2;
        }
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Volume path '%s' is a socket"), path);
        return -1;
1443 1444
    }

1445 1446 1447 1448 1449 1450
    /* O_NONBLOCK should only matter during open() for fifos and
     * sockets, which we already filtered; but using it prevents a
     * TOCTTOU race.  However, later on we will want to read() the
     * header from this fd, and virFileRead* routines require a
     * blocking fd, so fix it up after verifying we avoided a
     * race.  */
1451
    if ((fd = open(path, O_RDONLY|O_NONBLOCK|O_NOCTTY)) < 0) {
1452
        if ((errno == ENOENT || errno == ELOOP) &&
1453
            S_ISLNK(sb->st_mode) && noerror) {
1454 1455
            VIR_WARN("ignoring dangling symlink '%s'", path);
            return -2;
1456
        }
1457
        if (errno == ENOENT && noerror) {
1458 1459
            VIR_WARN("ignoring missing file '%s'", path);
            return -2;
1460
        }
1461 1462 1463 1464
        if (errno == ENXIO && noerror) {
            VIR_WARN("ignoring missing fifo '%s'", path);
            return -2;
        }
1465 1466 1467 1468
        if ((errno == EACCES || errno == EPERM) && noerror) {
            VIR_WARN("ignoring permission error for '%s'", path);
            return -2;
        }
1469

1470
        virReportSystemError(errno, _("cannot open volume '%s'"), path);
1471 1472 1473
        return -1;
    }

E
Eric Blake 已提交
1474
    if (fstat(fd, sb) < 0) {
1475
        virReportSystemError(errno, _("cannot stat file '%s'"), path);
1476 1477 1478 1479
        VIR_FORCE_CLOSE(fd);
        return -1;
    }

1480
    if (S_ISREG(sb->st_mode)) {
1481
        mode = VIR_STORAGE_VOL_OPEN_REG;
1482
    } else if (S_ISCHR(sb->st_mode)) {
1483
        mode = VIR_STORAGE_VOL_OPEN_CHAR;
1484
    } else if (S_ISBLK(sb->st_mode)) {
1485
        mode = VIR_STORAGE_VOL_OPEN_BLOCK;
1486
    } else if (S_ISDIR(sb->st_mode)) {
1487 1488 1489 1490 1491
        mode = VIR_STORAGE_VOL_OPEN_DIR;

        if (STREQ(base, ".") ||
            STREQ(base, "..")) {
            VIR_FORCE_CLOSE(fd);
1492 1493 1494 1495 1496 1497 1498
            if (noerror) {
                VIR_INFO("Skipping special dir '%s'", base);
                return -2;
            }
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Cannot use volume path '%s'"), path);
            return -1;
1499
        }
1500 1501
    } else {
        VIR_FORCE_CLOSE(fd);
1502 1503 1504 1505 1506 1507 1508
        if (noerror) {
            VIR_WARN("ignoring unexpected type for file '%s'", path);
            return -2;
        }
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unexpected type for file '%s'"), path);
        return -1;
1509 1510 1511
    }

    if (virSetBlocking(fd, true) < 0) {
1512
        VIR_FORCE_CLOSE(fd);
1513 1514
        virReportSystemError(errno, _("unable to set blocking mode for '%s'"),
                             path);
1515
        return -1;
1516
    }
1517 1518

    if (!(mode & flags)) {
1519
        VIR_FORCE_CLOSE(fd);
1520 1521 1522
        if (noerror) {
            VIR_INFO("Skipping volume '%s'", path);
            return -2;
1523 1524
        }

1525 1526 1527
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unexpected storage mode for '%s'"), path);
        return -1;
1528 1529 1530 1531 1532 1533
    }

    return fd;
}

int
1534
virStorageBackendUpdateVolTargetInfo(virStorageSourcePtr target,
1535
                                     bool withBlockVolFormat,
1536
                                     unsigned int openflags)
1537
{
1538
    int ret, fd = -1;
E
Eric Blake 已提交
1539
    struct stat sb;
1540 1541 1542
    virStorageSourcePtr meta = NULL;
    char *buf = NULL;
    ssize_t len = VIR_STORAGE_MAX_HEADER;
1543

1544
    if ((ret = virStorageBackendVolOpen(target->path, &sb, openflags)) < 0)
1545
        goto cleanup;
1546
    fd = ret;
1547

1548
    if ((ret = virStorageBackendUpdateVolTargetInfoFD(target, fd, &sb)) < 0)
1549 1550
        goto cleanup;

1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576
    if (target->type == VIR_STORAGE_VOL_FILE &&
        target->format != VIR_STORAGE_FILE_NONE) {
        if (S_ISDIR(sb.st_mode)) {
            ret = 0;
            goto cleanup;
        }

        if (lseek(fd, 0, SEEK_SET) == (off_t)-1) {
            virReportSystemError(errno, _("cannot seek to start of '%s'"), target->path);
            goto cleanup;
        }

        if ((len = virFileReadHeaderFD(fd, len, &buf)) < 0) {
            virReportSystemError(errno, _("cannot read header '%s'"), target->path);
            goto cleanup;
        }

        if (!(meta = virStorageFileGetMetadataFromBuf(target->path, buf, len, target->format,
                                                      NULL))) {
            goto cleanup;
        }

        if (meta->capacity)
            target->capacity = meta->capacity;
    }

1577 1578 1579 1580 1581
    if (withBlockVolFormat) {
        if ((ret = virStorageBackendDetectBlockVolFormatFD(target, fd)) < 0)
            goto cleanup;
    }

1582
 cleanup:
1583
    virStorageSourceFree(meta);
1584
    VIR_FORCE_CLOSE(fd);
1585
    VIR_FREE(buf);
1586 1587 1588
    return ret;
}

1589
int
1590
virStorageBackendUpdateVolInfo(virStorageVolDefPtr vol,
1591
                               bool withBlockVolFormat,
1592
                               unsigned int openflags)
1593 1594 1595
{
    int ret;

1596
    if ((ret = virStorageBackendUpdateVolTargetInfo(&vol->target,
1597 1598
                                                    withBlockVolFormat,
                                                    openflags)) < 0)
1599 1600
        return ret;

1601 1602
    if (vol->target.backingStore &&
        (ret = virStorageBackendUpdateVolTargetInfo(vol->target.backingStore,
1603
                                                    withBlockVolFormat,
1604 1605
                                                    VIR_STORAGE_VOL_OPEN_DEFAULT |
                                                    VIR_STORAGE_VOL_OPEN_NOERROR) < 0))
1606 1607 1608 1609 1610
        return ret;

    return 0;
}

1611 1612 1613
/*
 * virStorageBackendUpdateVolTargetInfoFD:
 * @target: target definition ptr of volume to update
1614 1615
 * @fd: fd of storage volume to update, via virStorageBackendOpenVol*, or -1
 * @sb: details about file (must match @fd, if that is provided)
1616
 *
1617
 * Returns 0 for success, -1 on a legitimate error condition.
1618
 */
1619
int
1620
virStorageBackendUpdateVolTargetInfoFD(virStorageSourcePtr target,
1621
                                       int fd,
1622
                                       struct stat *sb)
1623
{
1624
#if WITH_SELINUX
1625 1626 1627
    security_context_t filecon = NULL;
#endif

1628
    if (S_ISREG(sb->st_mode)) {
1629
#ifndef WIN32
1630 1631
        target->allocation = (unsigned long long)sb->st_blocks *
            (unsigned long long)DEV_BSIZE;
D
Daniel P. Berrange 已提交
1632
#else
1633
        target->allocation = sb->st_size;
D
Daniel P. Berrange 已提交
1634
#endif
1635 1636 1637
        /* Regular files may be sparse, so logical size (capacity) is not same
         * as actual allocation above
         */
1638
        target->capacity = sb->st_size;
1639 1640
    } else if (S_ISDIR(sb->st_mode)) {
        target->allocation = 0;
1641
        target->capacity = 0;
1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654
    } else if (fd >= 0) {
        off_t end;
        /* XXX this is POSIX compliant, but doesn't work for CHAR files,
         * 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) {
            virReportSystemError(errno,
                                 _("cannot seek to end of file '%s'"),
                                 target->path);
            return -1;
1655
        }
1656
        target->allocation = end;
1657
        target->capacity = end;
1658 1659
    }

1660 1661 1662 1663 1664
    if (!target->perms && VIR_ALLOC(target->perms) < 0)
        return -1;
    target->perms->mode = sb->st_mode & S_IRWXUGO;
    target->perms->uid = sb->st_uid;
    target->perms->gid = sb->st_gid;
1665

1666
    if (!target->timestamps && VIR_ALLOC(target->timestamps) < 0)
1667
        return -1;
E
Eric Blake 已提交
1668 1669 1670 1671
    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);
1672

1673
    VIR_FREE(target->perms->label);
1674

1675
#if WITH_SELINUX
1676
    /* XXX: make this a security driver call */
1677 1678 1679 1680 1681 1682 1683 1684
    if (fd >= 0) {
        if (fgetfilecon_raw(fd, &filecon) == -1) {
            if (errno != ENODATA && errno != ENOTSUP) {
                virReportSystemError(errno,
                                     _("cannot get file context of '%s'"),
                                     target->path);
                return -1;
            }
1685
        } else {
1686
            if (VIR_STRDUP(target->perms->label, filecon) < 0) {
1687 1688 1689
                freecon(filecon);
                return -1;
            }
1690
            freecon(filecon);
1691
        }
1692 1693 1694 1695 1696 1697
    }
#endif

    return 0;
}

1698 1699 1700 1701 1702 1703
bool
virStorageBackendPoolPathIsStable(const char *path)
{
    if (path == NULL || STREQ(path, "/dev") || STREQ(path, "/dev/"))
        return false;

1704
    if (!STRPREFIX(path, "/dev/"))
1705 1706 1707 1708
        return false;

    return true;
}
D
Dave Allan 已提交
1709

1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721
/*
 * 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.
1722
 *
1723
 * If 'loop' is true, we use a timeout loop to give dynamic paths
1724
 * a change to appear.
1725 1726
 */
char *
1727
virStorageBackendStablePath(virStoragePoolObjPtr pool,
1728
                            const char *devpath,
1729
                            bool loop)
1730 1731 1732
{
    DIR *dh;
    struct dirent *dent;
1733
    char *stablepath;
1734
    int opentries = 0;
1735
    int retry = 0;
E
Eric Blake 已提交
1736
    int direrr;
1737

1738
    /* Logical pools are under /dev but already have stable paths */
1739 1740
    if (pool->def->type == VIR_STORAGE_POOL_LOGICAL ||
        !virStorageBackendPoolPathIsStable(pool->def->target.path))
1741 1742
        goto ret_strdup;

1743 1744 1745
    /* 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.
1746
     */
1747
 reopen:
1748
    if ((dh = opendir(pool->def->target.path)) == NULL) {
1749
        opentries++;
1750
        if (loop && errno == ENOENT && opentries < 50) {
1751 1752 1753
            usleep(100 * 1000);
            goto reopen;
        }
1754
        virReportSystemError(errno,
1755 1756
                             _("cannot read dir '%s'"),
                             pool->def->target.path);
1757 1758 1759
        return NULL;
    }

1760 1761 1762
    /* 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
1763 1764
     * to this device node.
     *
1765
     * And it might need some time till the stable path shows
E
Eric Blake 已提交
1766 1767
     * up, so add timeout to retry here.  Ignore readdir failures,
     * since we have a fallback.
1768
     */
1769
 retry:
E
Eric Blake 已提交
1770
    while ((direrr = virDirRead(dh, &dent, NULL)) > 0) {
1771 1772 1773
        if (dent->d_name[0] == '.')
            continue;

1774 1775 1776
        if (virAsprintf(&stablepath, "%s/%s",
                        pool->def->target.path,
                        dent->d_name) == -1) {
1777 1778 1779 1780 1781 1782 1783 1784 1785
            closedir(dh);
            return NULL;
        }

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

1786
        VIR_FREE(stablepath);
1787 1788
    }

E
Eric Blake 已提交
1789
    if (!direrr && loop && ++retry < 100) {
1790 1791 1792 1793
        usleep(100 * 1000);
        goto retry;
    }

1794 1795
    closedir(dh);

1796
 ret_strdup:
1797 1798 1799
    /* Couldn't find any matching stable link so give back
     * the original non-stable dev path
     */
1800

1801
    ignore_value(VIR_STRDUP(stablepath, devpath));
1802 1803

    return stablepath;
1804
}
1805

1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818
int
virStorageBackendVolUploadLocal(virConnectPtr conn ATTRIBUTE_UNUSED,
                                virStoragePoolObjPtr pool ATTRIBUTE_UNUSED,
                                virStorageVolDefPtr vol,
                                virStreamPtr stream,
                                unsigned long long offset,
                                unsigned long long len,
                                unsigned int flags)
{
    virCheckFlags(0, -1);

    /* Not using O_CREAT because the file is required to already exist at
     * this point */
1819 1820
    return virFDStreamOpenBlockDevice(stream, vol->target.path,
                                      offset, len, O_WRONLY);
1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833
}

int
virStorageBackendVolDownloadLocal(virConnectPtr conn ATTRIBUTE_UNUSED,
                                  virStoragePoolObjPtr pool ATTRIBUTE_UNUSED,
                                  virStorageVolDefPtr vol,
                                  virStreamPtr stream,
                                  unsigned long long offset,
                                  unsigned long long len,
                                  unsigned int flags)
{
    virCheckFlags(0, -1);

1834 1835
    return virFDStreamOpenBlockDevice(stream, vol->target.path,
                                      offset, len, O_RDONLY);
1836 1837
}

1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040

/* If the volume we're wiping is already a sparse file, we simply
 * truncate and extend it to its original size, filling it with
 * zeroes.  This behavior is guaranteed by POSIX:
 *
 * http://www.opengroup.org/onlinepubs/9699919799/functions/ftruncate.html
 *
 * If fildes refers to a regular file, the ftruncate() function shall
 * cause the size of the file to be truncated to length. If the size
 * of the file previously exceeded length, the extra data shall no
 * longer be available to reads on the file. If the file previously
 * was smaller than this size, ftruncate() shall increase the size of
 * the file. If the file size is increased, the extended area shall
 * appear as if it were zero-filled.
 */
static int
virStorageBackendVolZeroSparseFileLocal(virStorageVolDefPtr vol,
                                        off_t size,
                                        int fd)
{
    int ret = -1;

    ret = ftruncate(fd, 0);
    if (ret == -1) {
        virReportSystemError(errno,
                             _("Failed to truncate volume with "
                               "path '%s' to 0 bytes"),
                             vol->target.path);
        return ret;
    }

    ret = ftruncate(fd, size);
    if (ret == -1) {
        virReportSystemError(errno,
                             _("Failed to truncate volume with "
                               "path '%s' to %ju bytes"),
                             vol->target.path, (uintmax_t)size);
    }

    return ret;
}


static int
virStorageBackendWipeExtentLocal(virStorageVolDefPtr vol,
                                 int fd,
                                 off_t extent_start,
                                 off_t extent_length,
                                 char *writebuf,
                                 size_t writebuf_length,
                                 size_t *bytes_wiped)
{
    int ret = -1, written = 0;
    off_t remaining = 0;
    size_t write_size = 0;

    VIR_DEBUG("extent logical start: %ju len: %ju",
              (uintmax_t)extent_start, (uintmax_t)extent_length);

    if ((ret = lseek(fd, extent_start, SEEK_SET)) < 0) {
        virReportSystemError(errno,
                             _("Failed to seek to position %ju in volume "
                               "with path '%s'"),
                             (uintmax_t)extent_start, vol->target.path);
        goto cleanup;
    }

    remaining = extent_length;
    while (remaining > 0) {

        write_size = (writebuf_length < remaining) ? writebuf_length : remaining;
        written = safewrite(fd, writebuf, write_size);
        if (written < 0) {
            virReportSystemError(errno,
                                 _("Failed to write %zu bytes to "
                                   "storage volume with path '%s'"),
                                 write_size, vol->target.path);

            goto cleanup;
        }

        *bytes_wiped += written;
        remaining -= written;
    }

    if (fdatasync(fd) < 0) {
        ret = -errno;
        virReportSystemError(errno,
                             _("cannot sync data to volume with path '%s'"),
                             vol->target.path);
        goto cleanup;
    }

    VIR_DEBUG("Wrote %zu bytes to volume with path '%s'",
              *bytes_wiped, vol->target.path);

    ret = 0;

 cleanup:
    return ret;
}


int
virStorageBackendVolWipeLocal(virConnectPtr conn ATTRIBUTE_UNUSED,
                              virStoragePoolObjPtr pool ATTRIBUTE_UNUSED,
                              virStorageVolDefPtr vol,
                              unsigned int algorithm,
                              unsigned int flags)
{
    int ret = -1, fd = -1;
    struct stat st;
    char *writebuf = NULL;
    size_t bytes_wiped = 0;
    virCommandPtr cmd = NULL;

    virCheckFlags(0, -1);

    VIR_DEBUG("Wiping volume with path '%s' and algorithm %u",
              vol->target.path, algorithm);

    fd = open(vol->target.path, O_RDWR);
    if (fd == -1) {
        virReportSystemError(errno,
                             _("Failed to open storage volume with path '%s'"),
                             vol->target.path);
        goto cleanup;
    }

    if (fstat(fd, &st) == -1) {
        virReportSystemError(errno,
                             _("Failed to stat storage volume with path '%s'"),
                             vol->target.path);
        goto cleanup;
    }

    if (algorithm != VIR_STORAGE_VOL_WIPE_ALG_ZERO) {
        const char *alg_char ATTRIBUTE_UNUSED = NULL;
        switch (algorithm) {
        case VIR_STORAGE_VOL_WIPE_ALG_NNSA:
            alg_char = "nnsa";
            break;
        case VIR_STORAGE_VOL_WIPE_ALG_DOD:
            alg_char = "dod";
            break;
        case VIR_STORAGE_VOL_WIPE_ALG_BSI:
            alg_char = "bsi";
            break;
        case VIR_STORAGE_VOL_WIPE_ALG_GUTMANN:
            alg_char = "gutmann";
            break;
        case VIR_STORAGE_VOL_WIPE_ALG_SCHNEIER:
            alg_char = "schneier";
            break;
        case VIR_STORAGE_VOL_WIPE_ALG_PFITZNER7:
            alg_char = "pfitzner7";
            break;
        case VIR_STORAGE_VOL_WIPE_ALG_PFITZNER33:
            alg_char = "pfitzner33";
            break;
        case VIR_STORAGE_VOL_WIPE_ALG_RANDOM:
            alg_char = "random";
            break;
        default:
            virReportError(VIR_ERR_INVALID_ARG,
                           _("unsupported algorithm %d"),
                           algorithm);
        }
        cmd = virCommandNew(SCRUB);
        virCommandAddArgList(cmd, "-f", "-p", alg_char,
                             vol->target.path, NULL);

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

        ret = 0;
        goto cleanup;
    } else {
        if (S_ISREG(st.st_mode) && st.st_blocks < (st.st_size / DEV_BSIZE)) {
            ret = virStorageBackendVolZeroSparseFileLocal(vol, st.st_size, fd);
        } else {

            if (VIR_ALLOC_N(writebuf, st.st_blksize) < 0)
                goto cleanup;

            ret = virStorageBackendWipeExtentLocal(vol,
                                                   fd,
                                                   0,
                                                   vol->target.allocation,
                                                   writebuf,
                                                   st.st_blksize,
                                                   &bytes_wiped);
        }
    }

 cleanup:
    virCommandFree(cmd);
    VIR_FREE(writebuf);
    VIR_FORCE_CLOSE(fd);
    return ret;
}


2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128
#ifdef GLUSTER_CLI
int
virStorageBackendFindGlusterPoolSources(const char *host,
                                        int pooltype,
                                        virStoragePoolSourceListPtr list)
{
    char *outbuf = NULL;
    virCommandPtr cmd = NULL;
    xmlDocPtr doc = NULL;
    xmlXPathContextPtr ctxt = NULL;
    xmlNodePtr *nodes = NULL;
    virStoragePoolSource *src = NULL;
    size_t i;
    int nnodes;
    int rc;

    int ret = -1;

    cmd = virCommandNewArgList(GLUSTER_CLI,
                               "--xml",
                               "--log-file=/dev/null",
                               "volume", "info", "all", NULL);

    virCommandAddArgFormat(cmd, "--remote-host=%s", host);
    virCommandSetOutputBuffer(cmd, &outbuf);

    if (virCommandRun(cmd, &rc) < 0)
        goto cleanup;

    if (rc != 0) {
        VIR_INFO("failed to query host '%s' for gluster volumes: %s",
                 host, outbuf);
        ret = 0;
        goto cleanup;
    }

    if (!(doc = virXMLParseStringCtxt(outbuf, _("(gluster_cli_output)"),
                                      &ctxt)))
        goto cleanup;

    if ((nnodes = virXPathNodeSet("//volumes/volume", ctxt, &nodes)) <= 0) {
        VIR_INFO("no gluster volumes available on '%s'", host);
        ret = 0;
        goto cleanup;
    }

    for (i = 0; i < nnodes; i++) {
        ctxt->node = nodes[i];

        if (!(src = virStoragePoolSourceListNewSource(list)))
            goto cleanup;

        if (!(src->dir = virXPathString("string(//name)", ctxt))) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("failed to extract gluster volume name"));
            goto cleanup;
        }

        if (VIR_ALLOC_N(src->hosts, 1) < 0)
            goto cleanup;
        src->nhost = 1;

        if (VIR_STRDUP(src->hosts[0].name, host) < 0)
            goto cleanup;

        src->format = pooltype;
    }

    ret = 0;

 cleanup:
    VIR_FREE(nodes);
    xmlXPathFreeContext(ctxt);
    xmlFreeDoc(doc);
    VIR_FREE(outbuf);
    virCommandFree(cmd);
    return ret;
}
#else /* #ifdef GLUSTER_CLI */
int
virStorageBackendFindGlusterPoolSources(const char *host ATTRIBUTE_UNUSED,
                                        int pooltype ATTRIBUTE_UNUSED,
                                        virStoragePoolSourceListPtr list ATTRIBUTE_UNUSED)
{
    VIR_INFO("gluster cli tool not installed");
    return 0;
}
#endif /* #ifdef GLUSTER_CLI */