storage_backend.c 49.6 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 51
#include "util.h"
#include "memory.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 "logging.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
        }
336 337 338 339 340 341 342 343 344 345 346 347 348 349
    }

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

367 368 369
    }

    if (fsync(fd) < 0) {
370
        ret = -errno;
371
        virReportSystemError(errno, _("cannot sync data to file '%s'"),
E
Eric Blake 已提交
372
                             vol->target.path);
373
        goto cleanup;
374 375
    }

376 377 378 379 380 381 382 383 384
cleanup:
    return ret;
}

int
virStorageBackendCreateRaw(virConnectPtr conn ATTRIBUTE_UNUSED,
                           virStoragePoolObjPtr pool,
                           virStorageVolDefPtr vol,
                           virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
385
                           unsigned int flags)
386 387
{
    int ret = -1;
E
Eric Blake 已提交
388 389 390 391
    int fd = -1;
    int operation_flags;

    virCheckFlags(0, -1);
392 393

    if (vol->target.encryption != NULL) {
394 395 396
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       "%s", _("storage pool does not support encrypted "
                               "volumes"));
397 398
        goto cleanup;
    }
399

L
Laine Stump 已提交
400
    operation_flags = VIR_FILE_OPEN_FORCE_MODE | VIR_FILE_OPEN_FORCE_OWNER;
E
Eric Blake 已提交
401
    if (pool->def->type == VIR_STORAGE_POOL_NETFS)
L
Laine Stump 已提交
402
        operation_flags |= VIR_FILE_OPEN_FORK;
E
Eric Blake 已提交
403

404 405
    if ((fd = virFileOpenAs(vol->target.path,
                            O_RDWR | O_CREAT | O_EXCL,
L
Laine Stump 已提交
406 407 408
                            vol->target.perms.mode,
                            vol->target.perms.uid,
                            vol->target.perms.gid,
409
                            operation_flags)) < 0) {
E
Eric Blake 已提交
410 411 412 413 414 415
        virReportSystemError(-fd,
                             _("cannot create path '%s'"),
                             vol->target.path);
        goto cleanup;
    }

416 417
    if ((ret = createRawFile(fd, vol, inputvol)) < 0)
        /* createRawFile already reported the exact error. */
E
Eric Blake 已提交
418
        ret = -1;
419 420

cleanup:
E
Eric Blake 已提交
421
    VIR_FORCE_CLOSE(fd);
422 423 424
    return ret;
}

425 426 427 428 429 430 431 432 433
static int
virStorageGenerateSecretUUID(virConnectPtr conn,
                             unsigned char *uuid)
{
    unsigned attempt;

    for (attempt = 0; attempt < 65536; attempt++) {
        virSecretPtr tmp;
        if (virUUIDGenerate(uuid) < 0) {
434 435
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("unable to generate uuid"));
436 437 438 439 440 441 442 443 444
            return -1;
        }
        tmp = conn->secretDriver->lookupByUUID(conn, uuid);
        if (tmp == NULL)
            return 0;

        virSecretFree(tmp);
    }

445 446
    virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                   _("too many conflicts when generating an uuid"));
447 448 449 450

    return -1;
}

451 452 453 454 455 456 457 458 459
static int
virStorageGenerateQcowEncryption(virConnectPtr conn,
                                 virStorageVolDefPtr vol)
{
    virSecretDefPtr def = NULL;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    virStorageEncryptionPtr enc;
    virStorageEncryptionSecretPtr enc_secret = NULL;
    virSecretPtr secret = NULL;
460
    char *xml;
461 462 463
    unsigned char value[VIR_STORAGE_QCOW_PASSPHRASE_SIZE];
    int ret = -1;

464 465 466
    if (conn->secretDriver == NULL ||
        conn->secretDriver->lookupByUUID == NULL ||
        conn->secretDriver->defineXML == NULL ||
467
        conn->secretDriver->setValue == NULL) {
468 469
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("secret storage not supported"));
470 471 472 473 474
        goto cleanup;
    }

    enc = vol->target.encryption;
    if (enc->nsecrets != 0) {
475 476
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("secrets already defined"));
477 478 479 480 481
        goto cleanup;
    }

    if (VIR_ALLOC(enc_secret) < 0 || VIR_REALLOC_N(enc->secrets, 1) < 0 ||
        VIR_ALLOC(def) < 0) {
482
        virReportOOMError();
483 484 485 486 487
        goto cleanup;
    }

    def->ephemeral = 0;
    def->private = 0;
488
    if (virStorageGenerateSecretUUID(conn, def->uuid) < 0)
489
        goto cleanup;
490

491 492 493
    def->usage_type = VIR_SECRET_USAGE_TYPE_VOLUME;
    def->usage.volume = strdup(vol->target.path);
    if (def->usage.volume == NULL) {
494
        virReportOOMError();
495 496
        goto cleanup;
    }
497
    xml = virSecretDefFormat(def);
498 499 500 501 502 503 504 505 506 507 508 509
    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);

510
    if (virStorageGenerateQcowPassphrase(value) < 0)
511 512 513 514 515 516
        goto cleanup;

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

    enc_secret->type = VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE;
517
    memcpy(enc_secret->uuid, secret->uuid, VIR_UUID_BUFLEN);
518 519 520 521 522 523 524 525 526
    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) {
527 528
        if (ret != 0 &&
            conn->secretDriver->undefine != NULL)
529 530 531
            conn->secretDriver->undefine(secret);
        virSecretFree(secret);
    }
532
    virBufferFreeAndReset(&buf);
533 534 535 536 537
    virSecretDefFree(def);
    VIR_FREE(enc_secret);
    return ret;
}

538 539 540 541 542
struct hookdata {
    virStorageVolDefPtr vol;
    bool skip;
};

543
static int virStorageBuildSetUIDHook(void *data) {
544 545 546 547 548
    struct hookdata *tmp = data;
    virStorageVolDefPtr vol = tmp->vol;

    if (tmp->skip)
        return 0;
549

550
    if (virSetUIDGID(vol->target.perms.uid, vol->target.perms.gid) < 0)
551
        return -1;
552

553 554 555
    return 0;
}

556
static int virStorageBackendCreateExecCommand(virStoragePoolObjPtr pool,
557
                                              virStorageVolDefPtr vol,
558
                                              virCommandPtr cmd) {
559 560 561 562
    struct stat st;
    gid_t gid;
    uid_t uid;
    int filecreated = 0;
563
    struct hookdata data = {vol, false};
564 565

    if ((pool->def->type == VIR_STORAGE_POOL_NETFS)
566 567 568 569 570
        && (((getuid() == 0)
             && (vol->target.perms.uid != -1)
             && (vol->target.perms.uid != 0))
            || ((vol->target.perms.gid != -1)
                && (vol->target.perms.gid != getgid())))) {
571 572 573 574

        virCommandSetPreExecHook(cmd, virStorageBuildSetUIDHook, &data);

        if (virCommandRun(cmd, NULL) == 0) {
575 576 577 578 579
            /* command was successfully run, check if the file was created */
            if (stat(vol->target.path, &st) >=0)
                filecreated = 1;
        }
    }
580 581 582

    data.skip = true;

583
    if (!filecreated) {
584
        if (virCommandRun(cmd, NULL) < 0) {
585 586 587
            return -1;
        }
        if (stat(vol->target.path, &st) < 0) {
588
            virReportSystemError(errno,
589
                                 _("failed to create %s"), vol->target.path);
590 591 592 593 594 595 596 597
            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)) {
598
        virReportSystemError(errno,
599
                             _("cannot chown %s to (%u, %u)"),
600
                             vol->target.path, uid, gid);
601 602 603
        return -1;
    }
    if (chmod(vol->target.path, vol->target.perms.mode) < 0) {
604
        virReportSystemError(errno,
605 606 607 608 609 610 611
                             _("cannot set mode of '%s' to %04o"),
                             vol->target.path, vol->target.perms.mode);
        return -1;
    }
    return 0;
}

612 613 614 615 616 617 618 619 620 621 622 623 624
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;
625
    int exitstatus;
626
    virCommandPtr cmd = virCommandNewArgList(qemuimg, "-h", NULL);
627

628 629 630
    virCommandAddEnvString(cmd, "LC_ALL=C");
    virCommandSetOutputBuffer(cmd, &help);
    virCommandClearCaps(cmd);
631

632 633 634
    /* 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 */
635
    if (virCommandRun(cmd, &exitstatus) < 0)
636 637
        goto cleanup;

638 639
    if ((start = strstr(help, " create ")) == NULL ||
        (end = strstr(start, "\n")) == NULL) {
640 641 642
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unable to parse qemu-img output '%s'"),
                       help);
643 644
        goto cleanup;
    }
645 646
    if (((tmp = strstr(start, "-F fmt")) && tmp < end) ||
        ((tmp = strstr(start, "-F backing_fmt")) && tmp < end))
647 648 649 650 651 652 653
        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:
654
    virCommandFree(cmd);
655 656 657 658 659
    VIR_FREE(help);
    return ret;
}


660 661
static int
virStorageBackendCreateQemuImg(virConnectPtr conn,
662
                               virStoragePoolObjPtr pool,
663
                               virStorageVolDefPtr vol,
664
                               virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
665
                               unsigned int flags)
666
{
667
    int ret = -1;
668
    char *create_tool;
669
    int imgformat = -1;
670 671
    virCommandPtr cmd = NULL;
    bool do_encryption = (vol->target.encryption != NULL);
672
    unsigned long long int size_arg;
673

E
Eric Blake 已提交
674 675
    virCheckFlags(0, -1);

676
    const char *type = virStorageFileFormatTypeToString(vol->target.format);
677
    const char *backingType = vol->backingStore.path ?
678
        virStorageFileFormatTypeToString(vol->backingStore.format) : NULL;
679 680 681 682 683 684

    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 ?
685 686 687 688
        virStorageFileFormatTypeToString(inputvol->type == VIR_STORAGE_VOL_BLOCK ?
                                         VIR_STORAGE_FILE_RAW :
                                         inputvol->target.format) :
        NULL;
689 690

    if (type == NULL) {
691 692 693
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unknown storage vol type %d"),
                       vol->target.format);
694 695 696
        return -1;
    }
    if (inputvol && inputType == NULL) {
697 698 699
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unknown storage vol type %d"),
                       inputvol->target.format);
700 701 702 703
        return -1;
    }

    if (vol->backingStore.path) {
704 705
        int accessRetCode = -1;
        char *absolutePath = NULL;
706 707 708 709 710 711 712 713

        /* 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))) {
714 715 716
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("a different backing store cannot "
                                   "be specified."));
717 718 719 720
            return -1;
        }

        if (backingType == NULL) {
721 722 723
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unknown storage vol backing store type %d"),
                           vol->backingStore.format);
724 725
            return -1;
        }
726 727 728 729 730 731 732 733 734 735 736 737 738 739

        /* 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) {
740
            virReportSystemError(errno,
741 742 743 744 745 746
                                 _("inaccessible backing store volume %s"),
                                 vol->backingStore.path);
            return -1;
        }
    }

747
    if (do_encryption) {
748 749
        virStorageEncryptionPtr enc;

750 751
        if (vol->target.format != VIR_STORAGE_FILE_QCOW &&
            vol->target.format != VIR_STORAGE_FILE_QCOW2) {
752 753 754
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("qcow volume encryption unsupported with "
                             "volume format %s"), type);
755 756
            return -1;
        }
757 758 759
        enc = vol->target.encryption;
        if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW &&
            enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT) {
760 761 762
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("unsupported volume encryption format %d"),
                           vol->target.encryption->format);
763 764
            return -1;
        }
765
        if (enc->nsecrets > 1) {
766 767
            virReportError(VIR_ERR_XML_ERROR, "%s",
                           _("too many secrets for qcow encryption"));
768 769
            return -1;
        }
770 771 772 773 774
        if (enc->format == VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT ||
            enc->nsecrets == 0) {
            if (virStorageGenerateQcowEncryption(conn, vol) < 0)
                return -1;
        }
775 776
    }

777
    /* Size in KB */
778
    size_arg = VIR_DIV_UP(vol->capacity, 1024);
779 780 781 782 783 784 785

    /* 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) {
786 787
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("unable to find kvm-img or qemu-img"));
788 789 790
        return -1;
    }

791 792 793 794
    imgformat = virStorageBackendQEMUImgBackingFormat(create_tool);
    if (imgformat < 0)
        goto cleanup;

795 796
    cmd = virCommandNew(create_tool);

797
    if (inputvol) {
798 799 800 801
        virCommandAddArgList(cmd, "convert", "-f", inputType, "-O", type,
                             inputPath, vol->target.path, NULL);

        if (do_encryption) {
802
            if (imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS) {
803
                virCommandAddArgList(cmd, "-o", "encryption=on", NULL);
804
            } else {
805
                virCommandAddArg(cmd, "-e");
806 807 808
            }
        }

809
    } else if (vol->backingStore.path) {
810 811 812
        virCommandAddArgList(cmd, "create", "-f", type,
                             "-b", vol->backingStore.path, NULL);

813 814
        switch (imgformat) {
        case QEMU_IMG_BACKING_FORMAT_FLAG:
815
            virCommandAddArgList(cmd, "-F", backingType, vol->target.path,
816 817
                                 NULL);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
818 819 820

            if (do_encryption)
                virCommandAddArg(cmd, "-e");
821 822 823
            break;

        case QEMU_IMG_BACKING_FORMAT_OPTIONS:
824 825 826
            virCommandAddArg(cmd, "-o");
            virCommandAddArgFormat(cmd, "backing_fmt=%s%s", backingType,
                                   do_encryption ? ",encryption=on" : "");
827 828
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
829
            break;
830 831 832 833 834

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

835 836
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
837 838 839
            if (do_encryption)
                virCommandAddArg(cmd, "-e");
        }
840
    } else {
841
        virCommandAddArgList(cmd, "create", "-f", type,
842 843
                             vol->target.path, NULL);
        virCommandAddArgFormat(cmd, "%lluK", size_arg);
844 845

        if (do_encryption) {
846
            if (imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS) {
847
                virCommandAddArgList(cmd, "-o", "encryption=on", NULL);
848
            } else {
849
                virCommandAddArg(cmd, "-e");
850 851
            }
        }
852
    }
853

854 855
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
cleanup:
856
    VIR_FREE(create_tool);
857
    virCommandFree(cmd);
858

859
    return ret;
860 861 862 863 864 865 866
}

/*
 * Xen removed the fully-functional qemu-img, and replaced it
 * with a partially functional qcow-create. Go figure ??!?
 */
static int
867
virStorageBackendCreateQcowCreate(virConnectPtr conn ATTRIBUTE_UNUSED,
868
                                  virStoragePoolObjPtr pool,
869
                                  virStorageVolDefPtr vol,
870
                                  virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
871
                                  unsigned int flags)
872
{
873
    int ret;
E
Eric Blake 已提交
874
    char *size;
875
    virCommandPtr cmd;
876

E
Eric Blake 已提交
877 878
    virCheckFlags(0, -1);

879
    if (inputvol) {
880 881
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("cannot copy from volume with qcow-create"));
882 883 884
        return -1;
    }

885
    if (vol->target.format != VIR_STORAGE_FILE_QCOW2) {
886 887 888
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unsupported storage vol type %d"),
                       vol->target.format);
889 890 891
        return -1;
    }
    if (vol->backingStore.path != NULL) {
892 893 894
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("copy-on-write image not supported with "
                         "qcow-create"));
895 896
        return -1;
    }
897
    if (vol->target.encryption != NULL) {
898 899 900
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       "%s", _("encrypted volumes not supported with "
                               "qcow-create"));
901 902
        return -1;
    }
903 904

    /* Size in MB - yes different units to qemu-img :-( */
905 906
    if (virAsprintf(&size, "%llu",
                    VIR_DIV_UP(vol->capacity, (1024 * 1024))) < 0) {
E
Eric Blake 已提交
907 908 909
        virReportOOMError();
        return -1;
    }
910

911
    cmd = virCommandNewArgList("qcow-create", size, vol->target.path, NULL);
912

913 914
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
    virCommandFree(cmd);
E
Eric Blake 已提交
915
    VIR_FREE(size);
916

917
    return ret;
918 919
}

920
virStorageBackendBuildVolFrom
921
virStorageBackendFSImageToolTypeToFunc(int tool_type)
922 923 924 925 926 927 928 929
{
    switch (tool_type) {
    case TOOL_KVM_IMG:
    case TOOL_QEMU_IMG:
        return virStorageBackendCreateQemuImg;
    case TOOL_QCOW_CREATE:
        return virStorageBackendCreateQcowCreate;
    default:
930 931 932
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Unknown file create tool type '%d'."),
                       tool_type);
933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959
    }

    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;
}

960
virStorageBackendBuildVolFrom
961
virStorageBackendGetBuildVolFromFunction(virStorageVolDefPtr vol,
962 963 964 965 966 967 968 969 970 971 972
                                         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 &&
973
         vol->target.format != VIR_STORAGE_FILE_RAW) ||
974
        (inputvol->type == VIR_STORAGE_VOL_FILE &&
975
         inputvol->target.format != VIR_STORAGE_FILE_RAW)) {
976

D
Daniel P. Berrange 已提交
977
        if ((tool_type = virStorageBackendFindFSImageTool(NULL)) < 0) {
978 979 980
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("creation of non-raw file images is "
                                   "not supported without qemu-img."));
981 982 983
            return NULL;
        }

984
        return virStorageBackendFSImageToolTypeToFunc(tool_type);
985 986
    }

987 988 989 990
    if (vol->type == VIR_STORAGE_VOL_BLOCK)
        return virStorageBackendCreateBlockFrom;
    else
        return virStorageBackendCreateRaw;
991
}
992

993

994 995
virStorageBackendPtr
virStorageBackendForType(int type) {
996
    unsigned int i;
997
    for (i = 0; backends[i]; i++)
998 999 1000
        if (backends[i]->type == type)
            return backends[i];

1001 1002
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("missing backend for pool type %d"), type);
1003 1004 1005 1006
    return NULL;
}


1007 1008 1009
/*
 * Allows caller to silently ignore files with improper mode
 *
1010 1011
 * Returns -1 on error, -2 if file mode is unexpected or the
 * volume is a dangling symbolic link.
1012
 */
1013
int
1014
virStorageBackendVolOpenCheckMode(const char *path, unsigned int flags)
1015
{
1016 1017
    int fd, mode = 0;
    struct stat sb;
1018
    char *base = last_component(path);
1019

1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034
    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;
    }

1035
    if ((fd = open(path, O_RDONLY|O_NONBLOCK|O_NOCTTY)) < 0) {
1036
        if ((errno == ENOENT || errno == ELOOP) &&
1037
            S_ISLNK(sb.st_mode)) {
1038 1039 1040 1041
            VIR_WARN("ignoring dangling symlink '%s'", path);
            return -2;
        }

1042
        virReportSystemError(errno,
1043
                             _("cannot open volume '%s'"),
1044
                             path);
1045 1046 1047
        return -1;
    }

1048 1049 1050 1051 1052 1053 1054 1055
    if (fstat(fd, &sb) < 0) {
        virReportSystemError(errno,
                             _("cannot stat file '%s'"),
                             path);
        VIR_FORCE_CLOSE(fd);
        return -1;
    }

1056 1057 1058 1059 1060 1061
    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;
1062 1063 1064 1065 1066 1067 1068 1069 1070 1071
    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;
        }
    }
1072 1073

    if (!(mode & flags)) {
1074
        VIR_FORCE_CLOSE(fd);
1075
        VIR_INFO("Skipping volume '%s'", path);
1076 1077

        if (mode & VIR_STORAGE_VOL_OPEN_ERROR) {
1078 1079
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unexpected storage mode for '%s'"), path);
1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097
            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,
1098 1099
                                     unsigned long long *capacity,
                                     unsigned int openflags)
1100 1101 1102
{
    int ret, fd;

1103 1104
    if ((ret = virStorageBackendVolOpenCheckMode(target->path,
                                                 openflags)) < 0)
1105 1106 1107
        return ret;

    fd = ret;
1108
    ret = virStorageBackendUpdateVolTargetInfoFD(target,
1109 1110 1111
                                                 fd,
                                                 allocation,
                                                 capacity);
1112

1113
    VIR_FORCE_CLOSE(fd);
1114 1115 1116 1117

    return ret;
}

1118
int
1119 1120 1121
virStorageBackendUpdateVolInfoFlags(virStorageVolDefPtr vol,
                                    int withCapacity,
                                    unsigned int openflags)
1122 1123 1124
{
    int ret;

1125
    if ((ret = virStorageBackendUpdateVolTargetInfo(&vol->target,
1126 1127 1128
                                    &vol->allocation,
                                    withCapacity ? &vol->capacity : NULL,
                                    openflags)) < 0)
1129 1130 1131
        return ret;

    if (vol->backingStore.path &&
1132
        (ret = virStorageBackendUpdateVolTargetInfo(&vol->backingStore,
1133 1134
                                            NULL, NULL,
                                            VIR_STORAGE_VOL_OPEN_DEFAULT)) < 0)
1135 1136 1137 1138 1139
        return ret;

    return 0;
}

1140 1141 1142 1143 1144 1145 1146
int virStorageBackendUpdateVolInfo(virStorageVolDefPtr vol,
                                   int withCapacity)
{
    return virStorageBackendUpdateVolInfoFlags(vol, withCapacity,
                                               VIR_STORAGE_VOL_OPEN_DEFAULT);
}

1147 1148 1149 1150
/*
 * virStorageBackendUpdateVolTargetInfoFD:
 * @conn: connection to report errors on
 * @target: target definition ptr of volume to update
1151
 * @fd: fd of storage volume to update, via virStorageBackendOpenVol*
1152 1153 1154
 * @allocation: If not NULL, updated allocation information will be stored
 * @capacity: If not NULL, updated capacity info will be stored
 *
1155
 * Returns 0 for success, -1 on a legitimate error condition.
1156
 */
1157
int
1158
virStorageBackendUpdateVolTargetInfoFD(virStorageVolTargetPtr target,
1159 1160 1161
                                       int fd,
                                       unsigned long long *allocation,
                                       unsigned long long *capacity)
1162 1163 1164 1165 1166 1167 1168
{
    struct stat sb;
#if HAVE_SELINUX
    security_context_t filecon = NULL;
#endif

    if (fstat(fd, &sb) < 0) {
1169
        virReportSystemError(errno,
1170
                             _("cannot stat file '%s'"),
1171
                             target->path);
1172 1173 1174
        return -1;
    }

1175 1176
    if (allocation) {
        if (S_ISREG(sb.st_mode)) {
1177
#ifndef WIN32
1178
            *allocation = (unsigned long long)sb.st_blocks *
1179
                          (unsigned long long)DEV_BSIZE;
D
Daniel P. Berrange 已提交
1180
#else
1181
            *allocation = sb.st_size;
D
Daniel P. Berrange 已提交
1182
#endif
1183 1184 1185 1186 1187
            /* Regular files may be sparse, so logical size (capacity) is not same
             * as actual allocation above
             */
            if (capacity)
                *capacity = sb.st_size;
1188 1189 1190 1191 1192
        } else if (S_ISDIR(sb.st_mode)) {
            *allocation = 0;
            if (capacity)
                *capacity = 0;

1193 1194
        } else {
            off_t end;
E
Eric Blake 已提交
1195
            /* XXX this is POSIX compliant, but doesn't work for CHAR files,
1196 1197 1198 1199 1200 1201
             * 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) {
1202
                virReportSystemError(errno,
1203 1204 1205 1206 1207 1208 1209
                                     _("cannot seek to end of file '%s'"),
                                     target->path);
                return -1;
            }
            *allocation = end;
            if (capacity)
                *capacity = end;
1210 1211 1212
        }
    }

1213 1214 1215
    target->perms.mode = sb.st_mode & S_IRWXUGO;
    target->perms.uid = sb.st_uid;
    target->perms.gid = sb.st_gid;
1216

1217 1218 1219 1220 1221 1222 1223 1224 1225
    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);

1226
    VIR_FREE(target->perms.label);
1227 1228

#if HAVE_SELINUX
1229
    /* XXX: make this a security driver call */
M
Martin Kletzander 已提交
1230
    if (fgetfilecon_raw(fd, &filecon) == -1) {
1231
        if (errno != ENODATA && errno != ENOTSUP) {
1232
            virReportSystemError(errno,
1233
                                 _("cannot get file context of '%s'"),
1234
                                 target->path);
1235 1236
            return -1;
        } else {
1237
            target->perms.label = NULL;
1238 1239
        }
    } else {
1240
        target->perms.label = strdup(filecon);
1241
        freecon(filecon);
1242
        if (target->perms.label == NULL) {
1243
            virReportOOMError();
1244 1245
            return -1;
        }
1246 1247
    }
#else
1248
    target->perms.label = NULL;
1249 1250 1251 1252 1253
#endif

    return 0;
}

D
Dave Allan 已提交
1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288

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
1289 1290
virStorageBackendDetectBlockVolFormatFD(virStorageVolTargetPtr target,
                                        int fd)
D
Dave Allan 已提交
1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301
{
    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) {
1302
        virReportSystemError(errno,
D
Dave Allan 已提交
1303 1304 1305 1306 1307 1308
                             _("cannot seek to beginning of file '%s'"),
                             target->path);
        return -1;
    }
    bytes = saferead(fd, buffer, sizeof(buffer));
    if (bytes < 0) {
1309
        virReportSystemError(errno,
D
Dave Allan 已提交
1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328
                             _("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;
}


1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340
/*
 * 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.
1341
 *
1342
 * If 'loop' is true, we use a timeout loop to give dynamic paths
1343
 * a change to appear.
1344 1345
 */
char *
1346
virStorageBackendStablePath(virStoragePoolObjPtr pool,
1347
                            const char *devpath,
1348
                            bool loop)
1349 1350 1351
{
    DIR *dh;
    struct dirent *dent;
1352
    char *stablepath;
1353
    int opentries = 0;
1354
    int retry = 0;
1355 1356 1357 1358 1359

    /* 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/"))
1360
        goto ret_strdup;
1361

1362
    /* Skip whole thing for a pool which isn't in /dev
1363
     * so we don't mess filesystem/dir based pools
1364 1365 1366 1367
     */
    if (!STRPREFIX(pool->def->target.path, "/dev"))
        goto ret_strdup;

1368 1369 1370 1371
    /* Logical pools are under /dev but already have stable paths */
    if (pool->def->type == VIR_STORAGE_POOL_LOGICAL)
        goto ret_strdup;

1372 1373 1374
    /* 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.
1375
     */
1376
 reopen:
1377
    if ((dh = opendir(pool->def->target.path)) == NULL) {
1378
        opentries++;
1379
        if (loop && errno == ENOENT && opentries < 50) {
1380 1381 1382
            usleep(100 * 1000);
            goto reopen;
        }
1383
        virReportSystemError(errno,
1384 1385
                             _("cannot read dir '%s'"),
                             pool->def->target.path);
1386 1387 1388
        return NULL;
    }

1389 1390 1391
    /* 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
1392 1393
     * to this device node.
     *
1394
     * And it might need some time till the stable path shows
1395
     * up, so add timeout to retry here.
1396
     */
1397
 retry:
1398 1399 1400 1401
    while ((dent = readdir(dh)) != NULL) {
        if (dent->d_name[0] == '.')
            continue;

1402 1403 1404
        if (virAsprintf(&stablepath, "%s/%s",
                        pool->def->target.path,
                        dent->d_name) == -1) {
1405
            virReportOOMError();
1406 1407 1408 1409 1410 1411 1412 1413 1414
            closedir(dh);
            return NULL;
        }

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

1415
        VIR_FREE(stablepath);
1416 1417
    }

1418
    if (loop && ++retry < 100) {
1419 1420 1421 1422
        usleep(100 * 1000);
        goto retry;
    }

1423 1424
    closedir(dh);

1425
 ret_strdup:
1426 1427 1428
    /* Couldn't find any matching stable link so give back
     * the original non-stable dev path
     */
1429 1430 1431 1432

    stablepath = strdup(devpath);

    if (stablepath == NULL)
1433
        virReportOOMError();
1434 1435

    return stablepath;
1436 1437
}

D
Daniel P. Berrange 已提交
1438

1439
#ifndef WIN32
1440 1441 1442 1443
/*
 * Run an external program.
 *
 * Read its output and apply a series of regexes to each line
R
Richard W.M. Jones 已提交
1444
 * When the entire set of regexes has matched consecutively
1445 1446 1447
 * then run a callback passing in all the matches
 */
int
1448
virStorageBackendRunProgRegex(virStoragePoolObjPtr pool,
1449
                              virCommandPtr cmd,
1450 1451 1452 1453
                              int nregex,
                              const char **regex,
                              int *nvars,
                              virStorageBackendListVolRegexFunc func,
1454
                              void *data, const char *prefix)
1455
{
1456
    int fd = -1, err, ret = -1;
1457 1458 1459 1460 1461 1462 1463 1464 1465
    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 */
1466
    if (VIR_ALLOC_N(reg, nregex) < 0) {
1467
        virReportOOMError();
1468 1469 1470 1471 1472 1473 1474 1475
        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));
1476 1477
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Failed to compile regex %s"), error);
1478 1479
            for (j = 0 ; j <= i ; j++)
                regfree(&reg[j]);
1480
            VIR_FREE(reg);
1481 1482 1483 1484 1485 1486 1487 1488 1489 1490
            return -1;
        }

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

    }

    /* Storage for matched variables */
1491
    if (VIR_ALLOC_N(groups, totgroups) < 0) {
1492
        virReportOOMError();
1493 1494
        goto cleanup;
    }
1495
    if (VIR_ALLOC_N(vars, maxvars+1) < 0) {
1496
        virReportOOMError();
1497 1498 1499
        goto cleanup;
    }

1500 1501
    virCommandSetOutputFD(cmd, &fd);
    if (virCommandRunAsync(cmd, NULL) < 0) {
1502 1503 1504
        goto cleanup;
    }

1505
    if ((list = VIR_FDOPEN(fd, "r")) == NULL) {
1506 1507
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("cannot read fd"));
1508 1509 1510 1511
        goto cleanup;
    }

    while (fgets(line, sizeof(line), list) != NULL) {
1512
        char *p = NULL;
1513 1514 1515 1516 1517
        /* Strip trailing newline */
        int len = strlen(line);
        if (len && line[len-1] == '\n')
            line[len-1] = '\0';

1518 1519 1520 1521 1522 1523
        /* ignore any command prefix */
        if (prefix)
            p = STRSKIP(line, prefix);
        if (!p)
            p = line;

1524
        for (i = 0 ; i <= maxReg && i < nregex ; i++) {
1525
            if (regexec(&reg[i], p, nvars[i]+1, vars, 0) == 0) {
1526 1527 1528 1529 1530 1531 1532 1533
                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 */
1534
                    p[vars[j+1].rm_eo] = '\0';
1535
                    if ((groups[ngroup++] =
1536
                         strdup(p + vars[j+1].rm_so)) == NULL) {
1537
                        virReportOOMError();
1538 1539 1540 1541 1542 1543
                        goto cleanup;
                    }
                }

                /* We're matching on the last regex, so callback time */
                if (i == (nregex-1)) {
1544
                    if (((*func)(pool, groups, data)) < 0)
1545 1546 1547
                        goto cleanup;

                    /* Release matches & restart to matching the first regex */
1548
                    for (j = 0 ; j < totgroups ; j++)
1549
                        VIR_FREE(groups[j]);
1550 1551 1552 1553 1554 1555 1556
                    maxReg = 0;
                    ngroup = 0;
                }
            }
        }
    }

1557
    ret = virCommandWait(cmd, NULL);
1558
cleanup:
1559 1560
    if (groups) {
        for (j = 0 ; j < totgroups ; j++)
1561 1562
            VIR_FREE(groups[j]);
        VIR_FREE(groups);
1563
    }
1564
    VIR_FREE(vars);
1565 1566 1567 1568

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

1569
    VIR_FREE(reg);
1570

1571 1572
    VIR_FORCE_FCLOSE(list);
    VIR_FORCE_CLOSE(fd);
1573

1574
    return ret;
1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588
}

/*
 * 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
1589
virStorageBackendRunProgNul(virStoragePoolObjPtr pool,
1590
                            virCommandPtr cmd,
1591 1592 1593 1594 1595
                            size_t n_columns,
                            virStorageBackendListVolNulFunc func,
                            void *data)
{
    size_t n_tok = 0;
1596
    int fd = -1;
1597 1598
    FILE *fp = NULL;
    char **v;
1599
    int ret = -1;
1600 1601 1602 1603 1604
    int i;

    if (n_columns == 0)
        return -1;

1605
    if (VIR_ALLOC_N(v, n_columns) < 0) {
1606
        virReportOOMError();
1607 1608 1609 1610 1611
        return -1;
    }
    for (i = 0; i < n_columns; i++)
        v[i] = NULL;

1612 1613
    virCommandSetOutputFD(cmd, &fd);
    if (virCommandRunAsync(cmd, NULL) < 0) {
1614 1615 1616
        goto cleanup;
    }

1617
    if ((fp = VIR_FDOPEN(fd, "r")) == NULL) {
1618 1619
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("cannot open file using fd"));
1620 1621 1622 1623 1624 1625 1626 1627
        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.  */
1628
        ssize_t tok_len = getdelim(&buf, &buf_len, 0, fp);
1629 1630 1631 1632
        v[n_tok] = buf;
        if (tok_len < 0) {
            /* Maybe EOF, maybe an error.
               If n_tok > 0, then we know it's an error.  */
1633
            if (n_tok && func(pool, n_tok, v, data) < 0)
1634 1635 1636 1637 1638
                goto cleanup;
            break;
        }
        ++n_tok;
        if (n_tok == n_columns) {
1639
            if (func(pool, n_tok, v, data) < 0)
1640 1641 1642
                goto cleanup;
            n_tok = 0;
            for (i = 0; i < n_columns; i++) {
1643
                VIR_FREE(v[i]);
1644 1645 1646 1647
            }
        }
    }

1648 1649 1650
    if (feof(fp) < 0) {
        virReportSystemError(errno, "%s",
                             _("read error on pipe"));
1651 1652
        goto cleanup;
    }
1653

1654
    ret = virCommandWait(cmd, NULL);
1655 1656
 cleanup:
    for (i = 0; i < n_columns; i++)
1657 1658
        VIR_FREE(v[i]);
    VIR_FREE(v);
1659

1660 1661
    VIR_FORCE_FCLOSE(fp);
    VIR_FORCE_CLOSE(fd);
1662

1663
    return ret;
1664
}
D
Daniel P. Berrange 已提交
1665

1666
#else /* WIN32 */
D
Daniel P. Berrange 已提交
1667 1668 1669 1670 1671 1672 1673 1674 1675

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,
1676
                              void *data ATTRIBUTE_UNUSED)
D
Daniel P. Berrange 已提交
1677
{
1678 1679
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690
    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)
{
1691 1692
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1693 1694
    return -1;
}
1695
#endif /* WIN32 */