storage_backend.c 50.6 KB
Newer Older
1
/*
2
 * storage_backend.c: internal storage driver backend contract
3
 *
4
 * Copyright (C) 2007-2012 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>
D
Daniel P. Berrange 已提交
28
#if HAVE_REGEX_H
29
# include <regex.h>
D
Daniel P. Berrange 已提交
30
#endif
31
#include <sys/types.h>
E
Eric Blake 已提交
32
#include <sys/wait.h>
33
#include <unistd.h>
34 35
#include <fcntl.h>
#include <sys/stat.h>
36
#include <sys/param.h>
37
#include <dirent.h>
38
#include "dirname.h"
39 40 41 42
#ifdef __linux__
# include <sys/ioctl.h>
# include <linux/fs.h>
#endif
43 44

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

48
#include "datatypes.h"
49
#include "virterror_internal.h"
D
Daniel P. Berrange 已提交
50
#include "util.h"
51
#include "viralloc.h"
52
#include "internal.h"
53
#include "secret_conf.h"
54
#include "uuid.h"
55
#include "storage_file.h"
D
Daniel P. Berrange 已提交
56
#include "storage_backend.h"
57
#include "virlog.h"
E
Eric Blake 已提交
58
#include "virfile.h"
59
#include "stat-time.h"
60

61
#if WITH_STORAGE_LVM
62
# include "storage_backend_logical.h"
63 64
#endif
#if WITH_STORAGE_ISCSI
65
# include "storage_backend_iscsi.h"
66
#endif
67
#if WITH_STORAGE_SCSI
68
# include "storage_backend_scsi.h"
69
#endif
D
Dave Allan 已提交
70
#if WITH_STORAGE_MPATH
71
# include "storage_backend_mpath.h"
D
Dave Allan 已提交
72
#endif
73
#if WITH_STORAGE_DISK
74
# include "storage_backend_disk.h"
75 76
#endif
#if WITH_STORAGE_DIR
77
# include "storage_backend_fs.h"
78
#endif
79 80 81
#if WITH_STORAGE_RBD
# include "storage_backend_rbd.h"
#endif
82 83 84
#if WITH_STORAGE_SHEEPDOG
# include "storage_backend_sheepdog.h"
#endif
85

86 87
#define VIR_FROM_THIS VIR_FROM_STORAGE

88 89 90 91 92 93 94 95 96 97 98 99 100 101
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
102 103 104
#if WITH_STORAGE_SCSI
    &virStorageBackendSCSI,
#endif
D
Dave Allan 已提交
105 106 107
#if WITH_STORAGE_MPATH
    &virStorageBackendMpath,
#endif
108 109
#if WITH_STORAGE_DISK
    &virStorageBackendDisk,
110 111 112
#endif
#if WITH_STORAGE_RBD
    &virStorageBackendRBD,
113 114 115
#endif
#if WITH_STORAGE_SHEEPDOG
    &virStorageBackendSheepdog,
116 117 118
#endif
    NULL
};
119

120 121 122 123 124 125 126 127
static int track_allocation_progress = 0;

enum {
    TOOL_QEMU_IMG,
    TOOL_KVM_IMG,
    TOOL_QCOW_CREATE,
};

128 129 130
#define READ_BLOCK_SIZE_DEFAULT  (1024 * 1024)
#define WRITE_BLOCK_SIZE_DEFAULT (4 * 1024)

131
static int ATTRIBUTE_NONNULL(2)
132
virStorageBackendCopyToFD(virStorageVolDefPtr vol,
133 134
                          virStorageVolDefPtr inputvol,
                          int fd,
135 136
                          unsigned long long *total,
                          int is_dest_file)
137 138 139
{
    int inputfd = -1;
    int amtread = -1;
140
    int ret = 0;
141 142 143 144
    size_t rbytes = READ_BLOCK_SIZE_DEFAULT;
    size_t wbytes = 0;
    int interval;
    char *zerobuf;
145
    char *buf = NULL;
146
    struct stat st;
147

148
    if ((inputfd = open(inputvol->target.path, O_RDONLY)) < 0) {
149
        ret = -errno;
150
        virReportSystemError(errno,
151 152 153
                             _("could not open input path '%s'"),
                             inputvol->target.path);
        goto cleanup;
154 155
    }

156 157 158 159 160 161 162 163 164 165 166 167 168 169 170
#ifdef __linux__
    if (ioctl(fd, BLKBSZGET, &wbytes) < 0) {
        wbytes = 0;
    }
#endif
    if ((wbytes == 0) && fstat(fd, &st) == 0)
        wbytes = st.st_blksize;
    if (wbytes < WRITE_BLOCK_SIZE_DEFAULT)
        wbytes = WRITE_BLOCK_SIZE_DEFAULT;

    if (VIR_ALLOC_N(zerobuf, wbytes) < 0) {
        ret = -errno;
        virReportOOMError();
        goto cleanup;
    }
171

172
    if (VIR_ALLOC_N(buf, rbytes) < 0) {
173
        ret = -errno;
174
        virReportOOMError();
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 (is_dest_file && 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

E
Eric Blake 已提交
259 260
    virCheckFlags(0, -1);

261
    if ((fd = open(vol->target.path, O_RDWR)) < 0) {
262
        virReportSystemError(errno,
263 264 265 266 267 268 269 270
                             _("cannot create path '%s'"),
                             vol->target.path);
        goto cleanup;
    }

    remain = vol->allocation;

    if (inputvol) {
271
        int res = virStorageBackendCopyToFD(vol, inputvol,
272
                                            fd, &remain, 0);
273 274 275 276
        if (res < 0)
            goto cleanup;
    }

277
    if (fstat(fd, &st) == -1) {
278
        virReportSystemError(errno, _("stat of '%s' failed"),
279 280 281 282 283 284
                             vol->target.path);
        goto cleanup;
    }
    uid = (vol->target.perms.uid != st.st_uid) ? vol->target.perms.uid : -1;
    gid = (vol->target.perms.gid != st.st_gid) ? vol->target.perms.gid : -1;
    if (((uid != -1) || (gid != -1))
285
        && (fchown(fd, uid, gid) < 0)) {
286
        virReportSystemError(errno,
287
                             _("cannot chown '%s' to (%u, %u)"),
288
                             vol->target.path, uid, gid);
289 290 291
        goto cleanup;
    }
    if (fchmod(fd, vol->target.perms.mode) < 0) {
292
        virReportSystemError(errno,
293 294 295 296
                             _("cannot set mode of '%s' to %04o"),
                             vol->target.path, vol->target.perms.mode);
        goto cleanup;
    }
297
    if (VIR_CLOSE(fd) < 0) {
298
        virReportSystemError(errno,
299 300 301 302 303 304 305 306
                             _("cannot close file '%s'"),
                             vol->target.path);
        goto cleanup;
    }
    fd = -1;

    ret = 0;
cleanup:
307
    VIR_FORCE_CLOSE(fd);
308 309 310 311

    return ret;
}

E
Eric Blake 已提交
312 313 314 315
static int
createRawFile(int fd, virStorageVolDefPtr vol,
              virStorageVolDefPtr inputvol)
{
316 317
    int ret = 0;
    unsigned long long remain;
318

319 320
    /* Seek to the final size, so the capacity is available upfront
     * for progress reporting */
E
Eric Blake 已提交
321
    if (ftruncate(fd, vol->capacity) < 0) {
322
        ret = -errno;
323
        virReportSystemError(errno,
324
                             _("cannot extend file '%s'"),
E
Eric Blake 已提交
325
                             vol->target.path);
326 327 328
        goto cleanup;
    }

E
Eric Blake 已提交
329
    remain = vol->allocation;
330

E
Eric Blake 已提交
331 332
    if (inputvol) {
        ret = virStorageBackendCopyToFD(vol, inputvol, fd, &remain, 1);
333
        if (ret < 0) {
334
            goto cleanup;
335
        }
336 337 338 339 340 341 342 343 344 345 346 347 348
    }

    if (remain) {
        if (track_allocation_progress) {
            while (remain) {
                /* Allocate in chunks of 512MiB: big-enough chunk
                 * size and takes approx. 9s on ext3. A progress
                 * update every 9s is a fair-enough trade-off
                 */
                unsigned long long bytes = 512 * 1024 * 1024;

                if (bytes > remain)
                    bytes = remain;
E
Eric Blake 已提交
349
                if (safezero(fd, vol->allocation - remain, bytes) < 0) {
350
                    ret = -errno;
J
Jiri Denemark 已提交
351
                    virReportSystemError(errno, _("cannot fill file '%s'"),
E
Eric Blake 已提交
352
                                         vol->target.path);
353 354 355 356 357
                    goto cleanup;
                }
                remain -= bytes;
            }
        } else { /* No progress bars to be shown */
E
Eric Blake 已提交
358
            if (safezero(fd, 0, remain) < 0) {
359
                ret = -errno;
J
Jiri Denemark 已提交
360
                virReportSystemError(errno, _("cannot fill file '%s'"),
E
Eric Blake 已提交
361
                                     vol->target.path);
362 363 364
                goto cleanup;
            }
        }
365 366 367
    }

    if (fsync(fd) < 0) {
368
        ret = -errno;
369
        virReportSystemError(errno, _("cannot sync data to file '%s'"),
E
Eric Blake 已提交
370
                             vol->target.path);
371
        goto cleanup;
372 373
    }

374 375 376 377 378 379 380 381 382
cleanup:
    return ret;
}

int
virStorageBackendCreateRaw(virConnectPtr conn ATTRIBUTE_UNUSED,
                           virStoragePoolObjPtr pool,
                           virStorageVolDefPtr vol,
                           virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
383
                           unsigned int flags)
384 385
{
    int ret = -1;
E
Eric Blake 已提交
386 387 388 389
    int fd = -1;
    int operation_flags;

    virCheckFlags(0, -1);
390 391

    if (vol->target.encryption != NULL) {
392 393 394
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       "%s", _("storage pool does not support encrypted "
                               "volumes"));
395 396
        goto cleanup;
    }
397

L
Laine Stump 已提交
398
    operation_flags = VIR_FILE_OPEN_FORCE_MODE | VIR_FILE_OPEN_FORCE_OWNER;
E
Eric Blake 已提交
399
    if (pool->def->type == VIR_STORAGE_POOL_NETFS)
L
Laine Stump 已提交
400
        operation_flags |= VIR_FILE_OPEN_FORK;
E
Eric Blake 已提交
401

402 403
    if ((fd = virFileOpenAs(vol->target.path,
                            O_RDWR | O_CREAT | O_EXCL,
L
Laine Stump 已提交
404 405 406
                            vol->target.perms.mode,
                            vol->target.perms.uid,
                            vol->target.perms.gid,
407
                            operation_flags)) < 0) {
E
Eric Blake 已提交
408 409 410 411 412 413
        virReportSystemError(-fd,
                             _("cannot create path '%s'"),
                             vol->target.path);
        goto cleanup;
    }

414 415
    if ((ret = createRawFile(fd, vol, inputvol)) < 0)
        /* createRawFile already reported the exact error. */
E
Eric Blake 已提交
416
        ret = -1;
417 418

cleanup:
E
Eric Blake 已提交
419
    VIR_FORCE_CLOSE(fd);
420 421 422
    return ret;
}

423 424 425 426 427 428 429 430 431
static int
virStorageGenerateSecretUUID(virConnectPtr conn,
                             unsigned char *uuid)
{
    unsigned attempt;

    for (attempt = 0; attempt < 65536; attempt++) {
        virSecretPtr tmp;
        if (virUUIDGenerate(uuid) < 0) {
432 433
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("unable to generate uuid"));
434 435 436 437 438 439 440 441 442
            return -1;
        }
        tmp = conn->secretDriver->lookupByUUID(conn, uuid);
        if (tmp == NULL)
            return 0;

        virSecretFree(tmp);
    }

443 444
    virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                   _("too many conflicts when generating an uuid"));
445 446 447 448

    return -1;
}

449 450 451 452 453 454 455 456 457
static int
virStorageGenerateQcowEncryption(virConnectPtr conn,
                                 virStorageVolDefPtr vol)
{
    virSecretDefPtr def = NULL;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    virStorageEncryptionPtr enc;
    virStorageEncryptionSecretPtr enc_secret = NULL;
    virSecretPtr secret = NULL;
458
    char *xml;
459 460 461
    unsigned char value[VIR_STORAGE_QCOW_PASSPHRASE_SIZE];
    int ret = -1;

462 463 464
    if (conn->secretDriver == NULL ||
        conn->secretDriver->lookupByUUID == NULL ||
        conn->secretDriver->defineXML == NULL ||
465
        conn->secretDriver->setValue == NULL) {
466 467
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("secret storage not supported"));
468 469 470 471 472
        goto cleanup;
    }

    enc = vol->target.encryption;
    if (enc->nsecrets != 0) {
473 474
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("secrets already defined"));
475 476 477 478 479
        goto cleanup;
    }

    if (VIR_ALLOC(enc_secret) < 0 || VIR_REALLOC_N(enc->secrets, 1) < 0 ||
        VIR_ALLOC(def) < 0) {
480
        virReportOOMError();
481 482 483 484 485
        goto cleanup;
    }

    def->ephemeral = 0;
    def->private = 0;
486
    if (virStorageGenerateSecretUUID(conn, def->uuid) < 0)
487
        goto cleanup;
488

489 490 491
    def->usage_type = VIR_SECRET_USAGE_TYPE_VOLUME;
    def->usage.volume = strdup(vol->target.path);
    if (def->usage.volume == NULL) {
492
        virReportOOMError();
493 494
        goto cleanup;
    }
495
    xml = virSecretDefFormat(def);
496 497 498 499 500 501 502 503 504 505 506 507
    virSecretDefFree(def);
    def = NULL;
    if (xml == NULL)
        goto cleanup;

    secret = conn->secretDriver->defineXML(conn, xml, 0);
    if (secret == NULL) {
        VIR_FREE(xml);
        goto cleanup;
    }
    VIR_FREE(xml);

508
    if (virStorageGenerateQcowPassphrase(value) < 0)
509 510 511 512 513 514
        goto cleanup;

    if (conn->secretDriver->setValue(secret, value, sizeof(value), 0) < 0)
        goto cleanup;

    enc_secret->type = VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE;
515
    memcpy(enc_secret->uuid, secret->uuid, VIR_UUID_BUFLEN);
516 517 518 519 520 521 522 523 524
    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) {
525 526
        if (ret != 0 &&
            conn->secretDriver->undefine != NULL)
527 528 529
            conn->secretDriver->undefine(secret);
        virSecretFree(secret);
    }
530
    virBufferFreeAndReset(&buf);
531 532 533 534 535
    virSecretDefFree(def);
    VIR_FREE(enc_secret);
    return ret;
}

536 537 538 539 540
struct hookdata {
    virStorageVolDefPtr vol;
    bool skip;
};

541
static int virStorageBuildSetUIDHook(void *data) {
542 543 544 545 546
    struct hookdata *tmp = data;
    virStorageVolDefPtr vol = tmp->vol;

    if (tmp->skip)
        return 0;
547

548
    if (virSetUIDGID(vol->target.perms.uid, vol->target.perms.gid) < 0)
549
        return -1;
550

551 552 553
    return 0;
}

554
static int virStorageBackendCreateExecCommand(virStoragePoolObjPtr pool,
555
                                              virStorageVolDefPtr vol,
556
                                              virCommandPtr cmd) {
557 558 559 560
    struct stat st;
    gid_t gid;
    uid_t uid;
    int filecreated = 0;
561
    struct hookdata data = {vol, false};
562 563

    if ((pool->def->type == VIR_STORAGE_POOL_NETFS)
564 565 566 567 568
        && (((getuid() == 0)
             && (vol->target.perms.uid != -1)
             && (vol->target.perms.uid != 0))
            || ((vol->target.perms.gid != -1)
                && (vol->target.perms.gid != getgid())))) {
569 570 571 572

        virCommandSetPreExecHook(cmd, virStorageBuildSetUIDHook, &data);

        if (virCommandRun(cmd, NULL) == 0) {
573 574 575 576 577
            /* command was successfully run, check if the file was created */
            if (stat(vol->target.path, &st) >=0)
                filecreated = 1;
        }
    }
578 579 580

    data.skip = true;

581
    if (!filecreated) {
582
        if (virCommandRun(cmd, NULL) < 0) {
583 584 585
            return -1;
        }
        if (stat(vol->target.path, &st) < 0) {
586
            virReportSystemError(errno,
587
                                 _("failed to create %s"), vol->target.path);
588 589 590 591 592 593 594 595
            return -1;
        }
    }

    uid = (vol->target.perms.uid != st.st_uid) ? vol->target.perms.uid : -1;
    gid = (vol->target.perms.gid != st.st_gid) ? vol->target.perms.gid : -1;
    if (((uid != -1) || (gid != -1))
        && (chown(vol->target.path, uid, gid) < 0)) {
596
        virReportSystemError(errno,
597
                             _("cannot chown %s to (%u, %u)"),
598
                             vol->target.path, uid, gid);
599 600 601
        return -1;
    }
    if (chmod(vol->target.path, vol->target.perms.mode) < 0) {
602
        virReportSystemError(errno,
603 604 605 606 607 608 609
                             _("cannot set mode of '%s' to %04o"),
                             vol->target.path, vol->target.perms.mode);
        return -1;
    }
    return 0;
}

610 611 612 613 614 615 616 617 618 619 620 621 622
enum {
    QEMU_IMG_BACKING_FORMAT_NONE = 0,
    QEMU_IMG_BACKING_FORMAT_FLAG,
    QEMU_IMG_BACKING_FORMAT_OPTIONS,
};

static int virStorageBackendQEMUImgBackingFormat(const char *qemuimg)
{
    char *help = NULL;
    char *start;
    char *end;
    char *tmp;
    int ret = -1;
623
    int exitstatus;
624
    virCommandPtr cmd = virCommandNewArgList(qemuimg, "-h", NULL);
625

626 627 628
    virCommandAddEnvString(cmd, "LC_ALL=C");
    virCommandSetOutputBuffer(cmd, &help);
    virCommandClearCaps(cmd);
629

630 631 632
    /* 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 */
633
    if (virCommandRun(cmd, &exitstatus) < 0)
634 635
        goto cleanup;

636 637
    if ((start = strstr(help, " create ")) == NULL ||
        (end = strstr(start, "\n")) == NULL) {
638 639 640
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unable to parse qemu-img output '%s'"),
                       help);
641 642
        goto cleanup;
    }
643 644
    if (((tmp = strstr(start, "-F fmt")) && tmp < end) ||
        ((tmp = strstr(start, "-F backing_fmt")) && tmp < end))
645 646 647 648 649 650 651
        ret = QEMU_IMG_BACKING_FORMAT_FLAG;
    else if ((tmp = strstr(start, "[-o options]")) && tmp < end)
        ret = QEMU_IMG_BACKING_FORMAT_OPTIONS;
    else
        ret = QEMU_IMG_BACKING_FORMAT_NONE;

cleanup:
652
    virCommandFree(cmd);
653 654 655 656 657
    VIR_FREE(help);
    return ret;
}


658 659
static int
virStorageBackendCreateQemuImg(virConnectPtr conn,
660
                               virStoragePoolObjPtr pool,
661
                               virStorageVolDefPtr vol,
662
                               virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
663
                               unsigned int flags)
664
{
665
    int ret = -1;
666
    char *create_tool;
667
    int imgformat = -1;
668 669
    virCommandPtr cmd = NULL;
    bool do_encryption = (vol->target.encryption != NULL);
670
    unsigned long long int size_arg;
671
    bool preallocate = false;
672

673 674 675
    virCheckFlags(VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA, -1);

    preallocate = !!(flags & VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA);
E
Eric Blake 已提交
676

677
    const char *type = virStorageFileFormatTypeToString(vol->target.format);
678
    const char *backingType = vol->backingStore.path ?
679
        virStorageFileFormatTypeToString(vol->backingStore.format) : NULL;
680 681 682 683 684 685

    const char *inputBackingPath = (inputvol ? inputvol->backingStore.path
                                             : NULL);
    const char *inputPath = inputvol ? inputvol->target.path : NULL;
    /* Treat input block devices as 'raw' format */
    const char *inputType = inputPath ?
686 687 688 689
        virStorageFileFormatTypeToString(inputvol->type == VIR_STORAGE_VOL_BLOCK ?
                                         VIR_STORAGE_FILE_RAW :
                                         inputvol->target.format) :
        NULL;
690 691

    if (type == NULL) {
692 693 694
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unknown storage vol type %d"),
                       vol->target.format);
695 696 697
        return -1;
    }
    if (inputvol && inputType == NULL) {
698 699 700
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unknown storage vol type %d"),
                       inputvol->target.format);
701 702
        return -1;
    }
703 704 705 706 707
    if (preallocate && vol->target.format != VIR_STORAGE_FILE_QCOW2) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("metadata preallocation only available with qcow2"));
        return -1;
    }
708 709

    if (vol->backingStore.path) {
710 711
        int accessRetCode = -1;
        char *absolutePath = NULL;
712

713 714 715 716 717 718 719
        if (preallocate) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("metadata preallocation conflicts with backing"
                             " store"));
            return -1;
        }

720 721 722 723 724 725 726
        /* 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 &&
            (!inputBackingPath ||
             STRNEQ(inputBackingPath, vol->backingStore.path))) {
727 728 729
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("a different backing store cannot "
                                   "be specified."));
730 731 732 733
            return -1;
        }

        if (backingType == NULL) {
734 735 736
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unknown storage vol backing store type %d"),
                           vol->backingStore.format);
737 738
            return -1;
        }
739 740 741 742 743 744 745 746 747 748 749 750 751 752

        /* Convert relative backing store paths to absolute paths for access
         * validation.
         */
        if ('/' != *(vol->backingStore.path) &&
            virAsprintf(&absolutePath, "%s/%s", pool->def->target.path,
                        vol->backingStore.path) < 0) {
            virReportOOMError();
            return -1;
        }
        accessRetCode = access(absolutePath ? absolutePath
                               : vol->backingStore.path, R_OK);
        VIR_FREE(absolutePath);
        if (accessRetCode != 0) {
753
            virReportSystemError(errno,
754 755 756 757 758 759
                                 _("inaccessible backing store volume %s"),
                                 vol->backingStore.path);
            return -1;
        }
    }

760
    if (do_encryption) {
761 762
        virStorageEncryptionPtr enc;

763 764
        if (vol->target.format != VIR_STORAGE_FILE_QCOW &&
            vol->target.format != VIR_STORAGE_FILE_QCOW2) {
765 766 767
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("qcow volume encryption unsupported with "
                             "volume format %s"), type);
768 769
            return -1;
        }
770 771 772
        enc = vol->target.encryption;
        if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW &&
            enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT) {
773 774 775
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("unsupported volume encryption format %d"),
                           vol->target.encryption->format);
776 777
            return -1;
        }
778
        if (enc->nsecrets > 1) {
779 780
            virReportError(VIR_ERR_XML_ERROR, "%s",
                           _("too many secrets for qcow encryption"));
781 782
            return -1;
        }
783 784 785 786 787
        if (enc->format == VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT ||
            enc->nsecrets == 0) {
            if (virStorageGenerateQcowEncryption(conn, vol) < 0)
                return -1;
        }
788 789
    }

790
    /* Size in KB */
791
    size_arg = VIR_DIV_UP(vol->capacity, 1024);
792 793 794 795 796 797 798

    /* 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) {
799 800
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("unable to find kvm-img or qemu-img"));
801 802 803
        return -1;
    }

804 805 806 807
    imgformat = virStorageBackendQEMUImgBackingFormat(create_tool);
    if (imgformat < 0)
        goto cleanup;

808 809
    cmd = virCommandNew(create_tool);

810
    if (inputvol) {
811 812 813
        virCommandAddArgList(cmd, "convert", "-f", inputType, "-O", type,
                             inputPath, vol->target.path, NULL);

814 815 816 817 818 819 820 821
        if (imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS &&
            (do_encryption || preallocate)) {
            virCommandAddArg(cmd, "-o");
            virCommandAddArgFormat(cmd, "%s%s%s", do_encryption ? "encryption=on" : "",
                                   (do_encryption && preallocate) ? "," : "",
                                   preallocate ? "preallocation=metadata" : "");
        } else if (do_encryption) {
            virCommandAddArg(cmd, "-e");
822
        }
823
    } else if (vol->backingStore.path) {
824 825 826
        virCommandAddArgList(cmd, "create", "-f", type,
                             "-b", vol->backingStore.path, NULL);

827 828
        switch (imgformat) {
        case QEMU_IMG_BACKING_FORMAT_FLAG:
829
            virCommandAddArgList(cmd, "-F", backingType, vol->target.path,
830 831
                                 NULL);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
832 833 834

            if (do_encryption)
                virCommandAddArg(cmd, "-e");
835 836 837
            break;

        case QEMU_IMG_BACKING_FORMAT_OPTIONS:
838 839 840
            virCommandAddArg(cmd, "-o");
            virCommandAddArgFormat(cmd, "backing_fmt=%s%s", backingType,
                                   do_encryption ? ",encryption=on" : "");
841 842
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
843
            break;
844 845 846 847 848

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

849 850
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
851 852 853
            if (do_encryption)
                virCommandAddArg(cmd, "-e");
        }
854
    } else {
855
        virCommandAddArgList(cmd, "create", "-f", type,
856 857
                             vol->target.path, NULL);
        virCommandAddArgFormat(cmd, "%lluK", size_arg);
858

859 860 861 862 863 864 865 866
        if (imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS &&
            (do_encryption || preallocate)) {
            virCommandAddArg(cmd, "-o");
            virCommandAddArgFormat(cmd, "%s%s%s", do_encryption ? "encryption=on" : "",
                                   (do_encryption && preallocate) ? "," : "",
                                   preallocate ? "preallocation=metadata" : "");
        } else if (do_encryption) {
            virCommandAddArg(cmd, "-e");
867
        }
868
    }
869

870 871
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
cleanup:
872
    VIR_FREE(create_tool);
873
    virCommandFree(cmd);
874

875
    return ret;
876 877 878 879 880 881 882
}

/*
 * Xen removed the fully-functional qemu-img, and replaced it
 * with a partially functional qcow-create. Go figure ??!?
 */
static int
883
virStorageBackendCreateQcowCreate(virConnectPtr conn ATTRIBUTE_UNUSED,
884
                                  virStoragePoolObjPtr pool,
885
                                  virStorageVolDefPtr vol,
886
                                  virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
887
                                  unsigned int flags)
888
{
889
    int ret;
E
Eric Blake 已提交
890
    char *size;
891
    virCommandPtr cmd;
892

E
Eric Blake 已提交
893 894
    virCheckFlags(0, -1);

895
    if (inputvol) {
896 897
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("cannot copy from volume with qcow-create"));
898 899 900
        return -1;
    }

901
    if (vol->target.format != VIR_STORAGE_FILE_QCOW2) {
902 903 904
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unsupported storage vol type %d"),
                       vol->target.format);
905 906 907
        return -1;
    }
    if (vol->backingStore.path != NULL) {
908 909 910
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("copy-on-write image not supported with "
                         "qcow-create"));
911 912
        return -1;
    }
913
    if (vol->target.encryption != NULL) {
914 915 916
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       "%s", _("encrypted volumes not supported with "
                               "qcow-create"));
917 918
        return -1;
    }
919 920

    /* Size in MB - yes different units to qemu-img :-( */
921 922
    if (virAsprintf(&size, "%llu",
                    VIR_DIV_UP(vol->capacity, (1024 * 1024))) < 0) {
E
Eric Blake 已提交
923 924 925
        virReportOOMError();
        return -1;
    }
926

927
    cmd = virCommandNewArgList("qcow-create", size, vol->target.path, NULL);
928

929 930
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
    virCommandFree(cmd);
E
Eric Blake 已提交
931
    VIR_FREE(size);
932

933
    return ret;
934 935
}

936
virStorageBackendBuildVolFrom
937
virStorageBackendFSImageToolTypeToFunc(int tool_type)
938 939 940 941 942 943 944 945
{
    switch (tool_type) {
    case TOOL_KVM_IMG:
    case TOOL_QEMU_IMG:
        return virStorageBackendCreateQemuImg;
    case TOOL_QCOW_CREATE:
        return virStorageBackendCreateQcowCreate;
    default:
946 947 948
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Unknown file create tool type '%d'."),
                       tool_type);
949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975
    }

    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;
}

976
virStorageBackendBuildVolFrom
977
virStorageBackendGetBuildVolFromFunction(virStorageVolDefPtr vol,
978 979 980 981 982 983 984 985 986 987 988
                                         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 &&
989
         vol->target.format != VIR_STORAGE_FILE_RAW) ||
990
        (inputvol->type == VIR_STORAGE_VOL_FILE &&
991
         inputvol->target.format != VIR_STORAGE_FILE_RAW)) {
992

D
Daniel P. Berrange 已提交
993
        if ((tool_type = virStorageBackendFindFSImageTool(NULL)) < 0) {
994 995 996
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("creation of non-raw file images is "
                                   "not supported without qemu-img."));
997 998 999
            return NULL;
        }

1000
        return virStorageBackendFSImageToolTypeToFunc(tool_type);
1001 1002
    }

1003 1004 1005 1006
    if (vol->type == VIR_STORAGE_VOL_BLOCK)
        return virStorageBackendCreateBlockFrom;
    else
        return virStorageBackendCreateRaw;
1007
}
1008

1009

1010 1011
virStorageBackendPtr
virStorageBackendForType(int type) {
1012
    unsigned int i;
1013
    for (i = 0; backends[i]; i++)
1014 1015 1016
        if (backends[i]->type == type)
            return backends[i];

1017 1018
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("missing backend for pool type %d"), type);
1019 1020 1021 1022
    return NULL;
}


1023 1024 1025
/*
 * Allows caller to silently ignore files with improper mode
 *
1026 1027
 * Returns -1 on error, -2 if file mode is unexpected or the
 * volume is a dangling symbolic link.
1028
 */
1029
int
1030
virStorageBackendVolOpenCheckMode(const char *path, unsigned int flags)
1031
{
1032 1033
    int fd, mode = 0;
    struct stat sb;
1034
    char *base = last_component(path);
1035

1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050
    if (lstat(path, &sb) < 0) {
        virReportSystemError(errno,
                             _("cannot stat file '%s'"),
                             path);
        return -1;
    }

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

1051
    if ((fd = open(path, O_RDONLY|O_NONBLOCK|O_NOCTTY)) < 0) {
1052
        if ((errno == ENOENT || errno == ELOOP) &&
1053
            S_ISLNK(sb.st_mode)) {
1054 1055 1056 1057
            VIR_WARN("ignoring dangling symlink '%s'", path);
            return -2;
        }

1058
        virReportSystemError(errno,
1059
                             _("cannot open volume '%s'"),
1060
                             path);
1061 1062 1063
        return -1;
    }

1064 1065 1066 1067 1068 1069 1070 1071
    if (fstat(fd, &sb) < 0) {
        virReportSystemError(errno,
                             _("cannot stat file '%s'"),
                             path);
        VIR_FORCE_CLOSE(fd);
        return -1;
    }

1072 1073 1074 1075 1076 1077
    if (S_ISREG(sb.st_mode))
        mode = VIR_STORAGE_VOL_OPEN_REG;
    else if (S_ISCHR(sb.st_mode))
        mode = VIR_STORAGE_VOL_OPEN_CHAR;
    else if (S_ISBLK(sb.st_mode))
        mode = VIR_STORAGE_VOL_OPEN_BLOCK;
1078 1079 1080 1081 1082 1083 1084 1085 1086 1087
    else if (S_ISDIR(sb.st_mode)) {
        mode = VIR_STORAGE_VOL_OPEN_DIR;

        if (STREQ(base, ".") ||
            STREQ(base, "..")) {
            VIR_FORCE_CLOSE(fd);
            VIR_INFO("Skipping special dir '%s'", base);
            return -2;
        }
    }
1088 1089

    if (!(mode & flags)) {
1090
        VIR_FORCE_CLOSE(fd);
1091
        VIR_INFO("Skipping volume '%s'", path);
1092 1093

        if (mode & VIR_STORAGE_VOL_OPEN_ERROR) {
1094 1095
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unexpected storage mode for '%s'"), path);
1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113
            return -1;
        }

        return -2;
    }

    return fd;
}

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

int
virStorageBackendUpdateVolTargetInfo(virStorageVolTargetPtr target,
                                     unsigned long long *allocation,
1114 1115
                                     unsigned long long *capacity,
                                     unsigned int openflags)
1116 1117 1118
{
    int ret, fd;

1119 1120
    if ((ret = virStorageBackendVolOpenCheckMode(target->path,
                                                 openflags)) < 0)
1121 1122 1123
        return ret;

    fd = ret;
1124
    ret = virStorageBackendUpdateVolTargetInfoFD(target,
1125 1126 1127
                                                 fd,
                                                 allocation,
                                                 capacity);
1128

1129
    VIR_FORCE_CLOSE(fd);
1130 1131 1132 1133

    return ret;
}

1134
int
1135 1136 1137
virStorageBackendUpdateVolInfoFlags(virStorageVolDefPtr vol,
                                    int withCapacity,
                                    unsigned int openflags)
1138 1139 1140
{
    int ret;

1141
    if ((ret = virStorageBackendUpdateVolTargetInfo(&vol->target,
1142 1143 1144
                                    &vol->allocation,
                                    withCapacity ? &vol->capacity : NULL,
                                    openflags)) < 0)
1145 1146 1147
        return ret;

    if (vol->backingStore.path &&
1148
        (ret = virStorageBackendUpdateVolTargetInfo(&vol->backingStore,
1149 1150
                                            NULL, NULL,
                                            VIR_STORAGE_VOL_OPEN_DEFAULT)) < 0)
1151 1152 1153 1154 1155
        return ret;

    return 0;
}

1156 1157 1158 1159 1160 1161 1162
int virStorageBackendUpdateVolInfo(virStorageVolDefPtr vol,
                                   int withCapacity)
{
    return virStorageBackendUpdateVolInfoFlags(vol, withCapacity,
                                               VIR_STORAGE_VOL_OPEN_DEFAULT);
}

1163 1164 1165 1166
/*
 * virStorageBackendUpdateVolTargetInfoFD:
 * @conn: connection to report errors on
 * @target: target definition ptr of volume to update
1167
 * @fd: fd of storage volume to update, via virStorageBackendOpenVol*
1168 1169 1170
 * @allocation: If not NULL, updated allocation information will be stored
 * @capacity: If not NULL, updated capacity info will be stored
 *
1171
 * Returns 0 for success, -1 on a legitimate error condition.
1172
 */
1173
int
1174
virStorageBackendUpdateVolTargetInfoFD(virStorageVolTargetPtr target,
1175 1176 1177
                                       int fd,
                                       unsigned long long *allocation,
                                       unsigned long long *capacity)
1178 1179 1180 1181 1182 1183 1184
{
    struct stat sb;
#if HAVE_SELINUX
    security_context_t filecon = NULL;
#endif

    if (fstat(fd, &sb) < 0) {
1185
        virReportSystemError(errno,
1186
                             _("cannot stat file '%s'"),
1187
                             target->path);
1188 1189 1190
        return -1;
    }

1191 1192
    if (allocation) {
        if (S_ISREG(sb.st_mode)) {
1193
#ifndef WIN32
1194
            *allocation = (unsigned long long)sb.st_blocks *
1195
                          (unsigned long long)DEV_BSIZE;
D
Daniel P. Berrange 已提交
1196
#else
1197
            *allocation = sb.st_size;
D
Daniel P. Berrange 已提交
1198
#endif
1199 1200 1201 1202 1203
            /* Regular files may be sparse, so logical size (capacity) is not same
             * as actual allocation above
             */
            if (capacity)
                *capacity = sb.st_size;
1204 1205 1206 1207 1208
        } else if (S_ISDIR(sb.st_mode)) {
            *allocation = 0;
            if (capacity)
                *capacity = 0;

1209 1210
        } else {
            off_t end;
E
Eric Blake 已提交
1211
            /* XXX this is POSIX compliant, but doesn't work for CHAR files,
1212 1213 1214 1215 1216 1217
             * 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) {
1218
                virReportSystemError(errno,
1219 1220 1221 1222 1223 1224 1225
                                     _("cannot seek to end of file '%s'"),
                                     target->path);
                return -1;
            }
            *allocation = end;
            if (capacity)
                *capacity = end;
1226 1227 1228
        }
    }

1229 1230 1231
    target->perms.mode = sb.st_mode & S_IRWXUGO;
    target->perms.uid = sb.st_uid;
    target->perms.gid = sb.st_gid;
1232

1233 1234 1235 1236 1237 1238 1239 1240 1241
    if (!target->timestamps && VIR_ALLOC(target->timestamps) < 0) {
        virReportOOMError();
        return -1;
    }
    target->timestamps->atime = get_stat_atime(&sb);
    target->timestamps->btime = get_stat_birthtime(&sb);
    target->timestamps->ctime = get_stat_ctime(&sb);
    target->timestamps->mtime = get_stat_mtime(&sb);

1242
    VIR_FREE(target->perms.label);
1243 1244

#if HAVE_SELINUX
1245
    /* XXX: make this a security driver call */
M
Martin Kletzander 已提交
1246
    if (fgetfilecon_raw(fd, &filecon) == -1) {
1247
        if (errno != ENODATA && errno != ENOTSUP) {
1248
            virReportSystemError(errno,
1249
                                 _("cannot get file context of '%s'"),
1250
                                 target->path);
1251 1252
            return -1;
        } else {
1253
            target->perms.label = NULL;
1254 1255
        }
    } else {
1256
        target->perms.label = strdup(filecon);
1257
        freecon(filecon);
1258
        if (target->perms.label == NULL) {
1259
            virReportOOMError();
1260 1261
            return -1;
        }
1262 1263
    }
#else
1264
    target->perms.label = NULL;
1265 1266 1267 1268 1269
#endif

    return 0;
}

D
Dave Allan 已提交
1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304

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
1305 1306
virStorageBackendDetectBlockVolFormatFD(virStorageVolTargetPtr target,
                                        int fd)
D
Dave Allan 已提交
1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317
{
    int i;
    off_t start;
    unsigned char buffer[1024];
    ssize_t bytes;

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

    start = lseek(fd, 0, SEEK_SET);
    if (start < 0) {
1318
        virReportSystemError(errno,
D
Dave Allan 已提交
1319 1320 1321 1322 1323 1324
                             _("cannot seek to beginning of file '%s'"),
                             target->path);
        return -1;
    }
    bytes = saferead(fd, buffer, sizeof(buffer));
    if (bytes < 0) {
1325
        virReportSystemError(errno,
D
Dave Allan 已提交
1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344
                             _("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;
}


1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356
/*
 * 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.
1357
 *
1358
 * If 'loop' is true, we use a timeout loop to give dynamic paths
1359
 * a change to appear.
1360 1361
 */
char *
1362
virStorageBackendStablePath(virStoragePoolObjPtr pool,
1363
                            const char *devpath,
1364
                            bool loop)
1365 1366 1367
{
    DIR *dh;
    struct dirent *dent;
1368
    char *stablepath;
1369
    int opentries = 0;
1370
    int retry = 0;
1371 1372 1373 1374 1375

    /* 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/"))
1376
        goto ret_strdup;
1377

1378
    /* Skip whole thing for a pool which isn't in /dev
1379
     * so we don't mess filesystem/dir based pools
1380 1381 1382 1383
     */
    if (!STRPREFIX(pool->def->target.path, "/dev"))
        goto ret_strdup;

1384 1385 1386 1387
    /* Logical pools are under /dev but already have stable paths */
    if (pool->def->type == VIR_STORAGE_POOL_LOGICAL)
        goto ret_strdup;

1388 1389 1390
    /* 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.
1391
     */
1392
 reopen:
1393
    if ((dh = opendir(pool->def->target.path)) == NULL) {
1394
        opentries++;
1395
        if (loop && errno == ENOENT && opentries < 50) {
1396 1397 1398
            usleep(100 * 1000);
            goto reopen;
        }
1399
        virReportSystemError(errno,
1400 1401
                             _("cannot read dir '%s'"),
                             pool->def->target.path);
1402 1403 1404
        return NULL;
    }

1405 1406 1407
    /* 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
1408 1409
     * to this device node.
     *
1410
     * And it might need some time till the stable path shows
1411
     * up, so add timeout to retry here.
1412
     */
1413
 retry:
1414 1415 1416 1417
    while ((dent = readdir(dh)) != NULL) {
        if (dent->d_name[0] == '.')
            continue;

1418 1419 1420
        if (virAsprintf(&stablepath, "%s/%s",
                        pool->def->target.path,
                        dent->d_name) == -1) {
1421
            virReportOOMError();
1422 1423 1424 1425 1426 1427 1428 1429 1430
            closedir(dh);
            return NULL;
        }

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

1431
        VIR_FREE(stablepath);
1432 1433
    }

1434
    if (loop && ++retry < 100) {
1435 1436 1437 1438
        usleep(100 * 1000);
        goto retry;
    }

1439 1440
    closedir(dh);

1441
 ret_strdup:
1442 1443 1444
    /* Couldn't find any matching stable link so give back
     * the original non-stable dev path
     */
1445 1446 1447 1448

    stablepath = strdup(devpath);

    if (stablepath == NULL)
1449
        virReportOOMError();
1450 1451

    return stablepath;
1452 1453
}

D
Daniel P. Berrange 已提交
1454

1455
#ifndef WIN32
1456 1457 1458 1459
/*
 * Run an external program.
 *
 * Read its output and apply a series of regexes to each line
R
Richard W.M. Jones 已提交
1460
 * When the entire set of regexes has matched consecutively
1461 1462 1463
 * then run a callback passing in all the matches
 */
int
1464
virStorageBackendRunProgRegex(virStoragePoolObjPtr pool,
1465
                              virCommandPtr cmd,
1466 1467 1468 1469
                              int nregex,
                              const char **regex,
                              int *nvars,
                              virStorageBackendListVolRegexFunc func,
1470
                              void *data, const char *prefix)
1471
{
1472
    int fd = -1, err, ret = -1;
1473 1474 1475 1476 1477 1478 1479 1480 1481
    FILE *list = NULL;
    regex_t *reg;
    regmatch_t *vars = NULL;
    char line[1024];
    int maxReg = 0, i, j;
    int totgroups = 0, ngroup = 0, maxvars = 0;
    char **groups;

    /* Compile all regular expressions */
1482
    if (VIR_ALLOC_N(reg, nregex) < 0) {
1483
        virReportOOMError();
1484 1485 1486 1487 1488 1489 1490 1491
        return -1;
    }

    for (i = 0 ; i < nregex ; i++) {
        err = regcomp(&reg[i], regex[i], REG_EXTENDED);
        if (err != 0) {
            char error[100];
            regerror(err, &reg[i], error, sizeof(error));
1492 1493
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Failed to compile regex %s"), error);
1494 1495
            for (j = 0 ; j <= i ; j++)
                regfree(&reg[j]);
1496
            VIR_FREE(reg);
1497 1498 1499 1500 1501 1502 1503 1504 1505 1506
            return -1;
        }

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

    }

    /* Storage for matched variables */
1507
    if (VIR_ALLOC_N(groups, totgroups) < 0) {
1508
        virReportOOMError();
1509 1510
        goto cleanup;
    }
1511
    if (VIR_ALLOC_N(vars, maxvars+1) < 0) {
1512
        virReportOOMError();
1513 1514 1515
        goto cleanup;
    }

1516 1517
    virCommandSetOutputFD(cmd, &fd);
    if (virCommandRunAsync(cmd, NULL) < 0) {
1518 1519 1520
        goto cleanup;
    }

1521
    if ((list = VIR_FDOPEN(fd, "r")) == NULL) {
1522 1523
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("cannot read fd"));
1524 1525 1526 1527
        goto cleanup;
    }

    while (fgets(line, sizeof(line), list) != NULL) {
1528
        char *p = NULL;
1529 1530 1531 1532 1533
        /* Strip trailing newline */
        int len = strlen(line);
        if (len && line[len-1] == '\n')
            line[len-1] = '\0';

1534 1535 1536 1537 1538 1539
        /* ignore any command prefix */
        if (prefix)
            p = STRSKIP(line, prefix);
        if (!p)
            p = line;

1540
        for (i = 0 ; i <= maxReg && i < nregex ; i++) {
1541
            if (regexec(&reg[i], p, nvars[i]+1, vars, 0) == 0) {
1542 1543 1544 1545 1546 1547 1548 1549
                maxReg++;

                if (i == 0)
                    ngroup = 0;

                /* NULL terminate each captured group in the line */
                for (j = 0 ; j < nvars[i] ; j++) {
                    /* NB vars[0] is the full pattern, so we offset j by 1 */
1550
                    p[vars[j+1].rm_eo] = '\0';
1551
                    if ((groups[ngroup++] =
1552
                         strdup(p + vars[j+1].rm_so)) == NULL) {
1553
                        virReportOOMError();
1554 1555 1556 1557 1558 1559
                        goto cleanup;
                    }
                }

                /* We're matching on the last regex, so callback time */
                if (i == (nregex-1)) {
1560
                    if (((*func)(pool, groups, data)) < 0)
1561 1562 1563
                        goto cleanup;

                    /* Release matches & restart to matching the first regex */
1564
                    for (j = 0 ; j < totgroups ; j++)
1565
                        VIR_FREE(groups[j]);
1566 1567 1568 1569 1570 1571 1572
                    maxReg = 0;
                    ngroup = 0;
                }
            }
        }
    }

1573
    ret = virCommandWait(cmd, NULL);
1574
cleanup:
1575 1576
    if (groups) {
        for (j = 0 ; j < totgroups ; j++)
1577 1578
            VIR_FREE(groups[j]);
        VIR_FREE(groups);
1579
    }
1580
    VIR_FREE(vars);
1581 1582 1583 1584

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

1585
    VIR_FREE(reg);
1586

1587 1588
    VIR_FORCE_FCLOSE(list);
    VIR_FORCE_CLOSE(fd);
1589

1590
    return ret;
1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604
}

/*
 * 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
1605
virStorageBackendRunProgNul(virStoragePoolObjPtr pool,
1606
                            virCommandPtr cmd,
1607 1608 1609 1610 1611
                            size_t n_columns,
                            virStorageBackendListVolNulFunc func,
                            void *data)
{
    size_t n_tok = 0;
1612
    int fd = -1;
1613 1614
    FILE *fp = NULL;
    char **v;
1615
    int ret = -1;
1616 1617 1618 1619 1620
    int i;

    if (n_columns == 0)
        return -1;

1621
    if (VIR_ALLOC_N(v, n_columns) < 0) {
1622
        virReportOOMError();
1623 1624 1625 1626 1627
        return -1;
    }
    for (i = 0; i < n_columns; i++)
        v[i] = NULL;

1628 1629
    virCommandSetOutputFD(cmd, &fd);
    if (virCommandRunAsync(cmd, NULL) < 0) {
1630 1631 1632
        goto cleanup;
    }

1633
    if ((fp = VIR_FDOPEN(fd, "r")) == NULL) {
1634 1635
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("cannot open file using fd"));
1636 1637 1638 1639 1640 1641 1642 1643
        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.  */
1644
        ssize_t tok_len = getdelim(&buf, &buf_len, 0, fp);
1645 1646 1647 1648
        v[n_tok] = buf;
        if (tok_len < 0) {
            /* Maybe EOF, maybe an error.
               If n_tok > 0, then we know it's an error.  */
1649
            if (n_tok && func(pool, n_tok, v, data) < 0)
1650 1651 1652 1653 1654
                goto cleanup;
            break;
        }
        ++n_tok;
        if (n_tok == n_columns) {
1655
            if (func(pool, n_tok, v, data) < 0)
1656 1657 1658
                goto cleanup;
            n_tok = 0;
            for (i = 0; i < n_columns; i++) {
1659
                VIR_FREE(v[i]);
1660 1661 1662 1663
            }
        }
    }

1664 1665 1666
    if (feof(fp) < 0) {
        virReportSystemError(errno, "%s",
                             _("read error on pipe"));
1667 1668
        goto cleanup;
    }
1669

1670
    ret = virCommandWait(cmd, NULL);
1671 1672
 cleanup:
    for (i = 0; i < n_columns; i++)
1673 1674
        VIR_FREE(v[i]);
    VIR_FREE(v);
1675

1676 1677
    VIR_FORCE_FCLOSE(fp);
    VIR_FORCE_CLOSE(fd);
1678

1679
    return ret;
1680
}
D
Daniel P. Berrange 已提交
1681

1682
#else /* WIN32 */
D
Daniel P. Berrange 已提交
1683 1684 1685 1686 1687 1688 1689 1690 1691

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,
1692
                              void *data ATTRIBUTE_UNUSED)
D
Daniel P. Berrange 已提交
1693
{
1694 1695
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706
    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)
{
1707 1708
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1709 1710
    return -1;
}
1711
#endif /* WIN32 */