storage_backend.c 49.1 KB
Newer Older
1
/*
2
 * storage_backend.c: internal storage driver backend contract
3
 *
4
 * Copyright (C) 2007-2011 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
 * 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
 * License along with this library; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307  USA
 *
 * 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 36
#include <fcntl.h>
#include <stdint.h>
#include <sys/stat.h>
37
#include <sys/param.h>
38
#include <dirent.h>
39
#include "dirname.h"
40 41 42 43
#ifdef __linux__
# include <sys/ioctl.h>
# include <linux/fs.h>
#endif
44 45

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

49
#include "datatypes.h"
50
#include "virterror_internal.h"
D
Daniel P. Berrange 已提交
51 52
#include "util.h"
#include "memory.h"
53
#include "internal.h"
54
#include "secret_conf.h"
55
#include "uuid.h"
56
#include "storage_file.h"
D
Daniel P. Berrange 已提交
57
#include "storage_backend.h"
58
#include "logging.h"
E
Eric Blake 已提交
59
#include "virfile.h"
60
#include "command.h"
61

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

81 82
#define VIR_FROM_THIS VIR_FROM_STORAGE

83 84 85 86 87 88 89 90 91 92 93 94 95 96
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
97 98 99
#if WITH_STORAGE_SCSI
    &virStorageBackendSCSI,
#endif
D
Dave Allan 已提交
100 101 102
#if WITH_STORAGE_MPATH
    &virStorageBackendMpath,
#endif
103 104 105 106 107
#if WITH_STORAGE_DISK
    &virStorageBackendDisk,
#endif
    NULL
};
108

109 110 111 112 113 114 115 116
static int track_allocation_progress = 0;

enum {
    TOOL_QEMU_IMG,
    TOOL_KVM_IMG,
    TOOL_QCOW_CREATE,
};

117 118 119
#define READ_BLOCK_SIZE_DEFAULT  (1024 * 1024)
#define WRITE_BLOCK_SIZE_DEFAULT (4 * 1024)

120 121
static int ATTRIBUTE_NONNULL (2)
virStorageBackendCopyToFD(virStorageVolDefPtr vol,
122 123
                          virStorageVolDefPtr inputvol,
                          int fd,
124 125
                          unsigned long long *total,
                          int is_dest_file)
126 127 128
{
    int inputfd = -1;
    int amtread = -1;
129
    int ret = 0;
130
    unsigned long long remain;
131 132 133 134
    size_t rbytes = READ_BLOCK_SIZE_DEFAULT;
    size_t wbytes = 0;
    int interval;
    char *zerobuf;
135
    char *buf = NULL;
136
    struct stat st;
137

138
    if ((inputfd = open(inputvol->target.path, O_RDONLY)) < 0) {
139
        ret = -errno;
140
        virReportSystemError(errno,
141 142 143
                             _("could not open input path '%s'"),
                             inputvol->target.path);
        goto cleanup;
144 145
    }

146 147 148 149 150 151 152 153 154 155 156 157 158 159 160
#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;
    }
161

162
    if (VIR_ALLOC_N(buf, rbytes) < 0) {
163
        ret = -errno;
164
        virReportOOMError();
165 166 167 168 169 170 171 172
        goto cleanup;
    }

    remain = *total;

    while (amtread != 0) {
        int amtleft;

173 174
        if (remain < rbytes)
            rbytes = remain;
175

176
        if ((amtread = saferead(inputfd, buf, rbytes)) < 0) {
177
            ret = -errno;
178
            virReportSystemError(errno,
179 180 181 182 183 184 185 186 187 188
                                 _("failed reading from file '%s'"),
                                 inputvol->target.path);
            goto cleanup;
        }
        remain -= amtread;

        /* Loop over amt read in 512 byte increments, looking for sparse
         * blocks */
        amtleft = amtread;
        do {
189
            interval = ((wbytes > amtleft) ? amtleft : wbytes);
190 191
            int offset = amtread - amtleft;

192
            if (is_dest_file && memcmp(buf+offset, zerobuf, interval) == 0) {
193
                if (lseek(fd, interval, SEEK_CUR) < 0) {
194
                    ret = -errno;
195
                    virReportSystemError(errno,
196 197 198 199 200
                                         _("cannot extend file '%s'"),
                                         vol->target.path);
                    goto cleanup;
                }
            } else if (safewrite(fd, buf+offset, interval) < 0) {
201
                ret = -errno;
202
                virReportSystemError(errno,
203 204 205 206 207
                                     _("failed writing to file '%s'"),
                                     vol->target.path);
                goto cleanup;

            }
208
        } while ((amtleft -= interval) > 0);
209 210
    }

211 212 213 214 215 216 217 218
    if (fdatasync(fd) < 0) {
        ret = -errno;
        virReportSystemError(errno, _("cannot sync data to file '%s'"),
                             vol->target.path);
        goto cleanup;
    }


219
    if (VIR_CLOSE(inputfd) < 0) {
220
        ret = -errno;
221
        virReportSystemError(errno,
222 223 224 225 226 227 228 229 230
                             _("cannot close file '%s'"),
                             inputvol->target.path);
        goto cleanup;
    }
    inputfd = -1;

    *total -= remain;

cleanup:
231
    VIR_FORCE_CLOSE(inputfd);
232

233
    VIR_FREE(zerobuf);
234 235
    VIR_FREE(buf);

236 237 238
    return ret;
}

239
static int
240
virStorageBackendCreateBlockFrom(virConnectPtr conn ATTRIBUTE_UNUSED,
241
                                 virStoragePoolObjPtr pool ATTRIBUTE_UNUSED,
242 243
                                 virStorageVolDefPtr vol,
                                 virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
244
                                 unsigned int flags)
245 246 247 248
{
    int fd = -1;
    int ret = -1;
    unsigned long long remain;
249 250 251
    struct stat st;
    gid_t gid;
    uid_t uid;
252

E
Eric Blake 已提交
253 254
    virCheckFlags(0, -1);

255
    if ((fd = open(vol->target.path, O_RDWR)) < 0) {
256
        virReportSystemError(errno,
257 258 259 260 261 262 263 264
                             _("cannot create path '%s'"),
                             vol->target.path);
        goto cleanup;
    }

    remain = vol->allocation;

    if (inputvol) {
265
        int res = virStorageBackendCopyToFD(vol, inputvol,
266
                                            fd, &remain, 0);
267 268 269 270
        if (res < 0)
            goto cleanup;
    }

271
    if (fstat(fd, &st) == -1) {
272
        virReportSystemError(errno, _("stat of '%s' failed"),
273 274 275 276 277 278
                             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))
279
        && (fchown(fd, uid, gid) < 0)) {
280
        virReportSystemError(errno,
281
                             _("cannot chown '%s' to (%u, %u)"),
282
                             vol->target.path, uid, gid);
283 284 285
        goto cleanup;
    }
    if (fchmod(fd, vol->target.perms.mode) < 0) {
286
        virReportSystemError(errno,
287 288 289 290
                             _("cannot set mode of '%s' to %04o"),
                             vol->target.path, vol->target.perms.mode);
        goto cleanup;
    }
291
    if (VIR_CLOSE(fd) < 0) {
292
        virReportSystemError(errno,
293 294 295 296 297 298 299 300
                             _("cannot close file '%s'"),
                             vol->target.path);
        goto cleanup;
    }
    fd = -1;

    ret = 0;
cleanup:
301
    VIR_FORCE_CLOSE(fd);
302 303 304 305

    return ret;
}

E
Eric Blake 已提交
306 307 308 309
static int
createRawFile(int fd, virStorageVolDefPtr vol,
              virStorageVolDefPtr inputvol)
{
310 311
    int ret = 0;
    unsigned long long remain;
312

313 314
    /* Seek to the final size, so the capacity is available upfront
     * for progress reporting */
E
Eric Blake 已提交
315
    if (ftruncate(fd, vol->capacity) < 0) {
316
        ret = -errno;
317
        virReportSystemError(errno,
318
                             _("cannot extend file '%s'"),
E
Eric Blake 已提交
319
                             vol->target.path);
320 321 322
        goto cleanup;
    }

E
Eric Blake 已提交
323
    remain = vol->allocation;
324

E
Eric Blake 已提交
325 326
    if (inputvol) {
        ret = virStorageBackendCopyToFD(vol, inputvol, fd, &remain, 1);
327
        if (ret < 0) {
328
            goto cleanup;
329
        }
330 331 332 333 334 335 336 337 338 339 340 341 342 343
    }

    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
344
                if (safezero(fd, vol->allocation - remain, bytes) < 0) {
345
                    ret = -errno;
J
Jiri Denemark 已提交
346
                    virReportSystemError(errno, _("cannot fill file '%s'"),
E
Eric Blake 已提交
347
                                         vol->target.path);
348 349 350 351 352
                    goto cleanup;
                }
                remain -= bytes;
            }
        } else { /* No progress bars to be shown */
E
Eric Blake 已提交
353
            if (safezero(fd, 0, remain) < 0) {
354
                ret = -errno;
J
Jiri Denemark 已提交
355
                virReportSystemError(errno, _("cannot fill file '%s'"),
E
Eric Blake 已提交
356
                                     vol->target.path);
357 358 359
                goto cleanup;
            }
        }
360

361 362 363
    }

    if (fsync(fd) < 0) {
364
        ret = -errno;
365
        virReportSystemError(errno, _("cannot sync data to file '%s'"),
E
Eric Blake 已提交
366
                             vol->target.path);
367
        goto cleanup;
368 369
    }

370 371 372 373 374 375 376 377 378
cleanup:
    return ret;
}

int
virStorageBackendCreateRaw(virConnectPtr conn ATTRIBUTE_UNUSED,
                           virStoragePoolObjPtr pool,
                           virStorageVolDefPtr vol,
                           virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
379
                           unsigned int flags)
380 381
{
    int ret = -1;
E
Eric Blake 已提交
382 383 384 385 386 387
    int fd = -1;
    uid_t uid;
    gid_t gid;
    int operation_flags;

    virCheckFlags(0, -1);
388 389

    if (vol->target.encryption != NULL) {
O
Osier Yang 已提交
390
        virStorageReportError(VIR_ERR_CONFIG_UNSUPPORTED,
391 392
                              "%s", _("storage pool does not support encrypted "
                                      "volumes"));
393 394
        goto cleanup;
    }
395

E
Eric Blake 已提交
396 397
    uid = (vol->target.perms.uid == -1) ? getuid() : vol->target.perms.uid;
    gid = (vol->target.perms.gid == -1) ? getgid() : vol->target.perms.gid;
398
    operation_flags = VIR_FILE_OPEN_FORCE_PERMS;
E
Eric Blake 已提交
399
    if (pool->def->type == VIR_STORAGE_POOL_NETFS)
400
        operation_flags |= VIR_FILE_OPEN_AS_UID;
E
Eric Blake 已提交
401

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

412 413
    if ((ret = createRawFile(fd, vol, inputvol)) < 0)
        /* createRawFile already reported the exact error. */
E
Eric Blake 已提交
414
        ret = -1;
415 416

cleanup:
E
Eric Blake 已提交
417
    VIR_FORCE_CLOSE(fd);
418 419 420
    return ret;
}

421 422 423 424 425 426 427 428 429
static int
virStorageGenerateSecretUUID(virConnectPtr conn,
                             unsigned char *uuid)
{
    unsigned attempt;

    for (attempt = 0; attempt < 65536; attempt++) {
        virSecretPtr tmp;
        if (virUUIDGenerate(uuid) < 0) {
430 431
            virStorageReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                                  _("unable to generate uuid"));
432 433 434 435 436 437 438 439 440
            return -1;
        }
        tmp = conn->secretDriver->lookupByUUID(conn, uuid);
        if (tmp == NULL)
            return 0;

        virSecretFree(tmp);
    }

441 442
    virStorageReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                          _("too many conflicts when generating an uuid"));
443 444 445 446

    return -1;
}

447 448 449 450 451 452 453 454 455
static int
virStorageGenerateQcowEncryption(virConnectPtr conn,
                                 virStorageVolDefPtr vol)
{
    virSecretDefPtr def = NULL;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    virStorageEncryptionPtr enc;
    virStorageEncryptionSecretPtr enc_secret = NULL;
    virSecretPtr secret = NULL;
456
    char *xml;
457 458 459
    unsigned char value[VIR_STORAGE_QCOW_PASSPHRASE_SIZE];
    int ret = -1;

460 461 462
    if (conn->secretDriver == NULL ||
        conn->secretDriver->lookupByUUID == NULL ||
        conn->secretDriver->defineXML == NULL ||
463
        conn->secretDriver->setValue == NULL) {
O
Osier Yang 已提交
464
        virStorageReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
465 466 467 468 469 470
                              _("secret storage not supported"));
        goto cleanup;
    }

    enc = vol->target.encryption;
    if (enc->nsecrets != 0) {
471
        virStorageReportError(VIR_ERR_INTERNAL_ERROR, "%s",
472 473 474 475 476 477
                              _("secrets already defined"));
        goto cleanup;
    }

    if (VIR_ALLOC(enc_secret) < 0 || VIR_REALLOC_N(enc->secrets, 1) < 0 ||
        VIR_ALLOC(def) < 0) {
478
        virReportOOMError();
479 480 481 482 483
        goto cleanup;
    }

    def->ephemeral = 0;
    def->private = 0;
484
    if (virStorageGenerateSecretUUID(conn, def->uuid) < 0)
485
        goto cleanup;
486

487 488 489
    def->usage_type = VIR_SECRET_USAGE_TYPE_VOLUME;
    def->usage.volume = strdup(vol->target.path);
    if (def->usage.volume == NULL) {
490
        virReportOOMError();
491 492
        goto cleanup;
    }
493
    xml = virSecretDefFormat(def);
494 495 496 497 498 499 500 501 502 503 504 505
    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);

506
    if (virStorageGenerateQcowPassphrase(value) < 0)
507 508 509 510 511 512
        goto cleanup;

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

    enc_secret->type = VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE;
513
    memcpy(enc_secret->uuid, secret->uuid, VIR_UUID_BUFLEN);
514 515 516 517 518 519 520 521 522
    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) {
523 524
        if (ret != 0 &&
            conn->secretDriver->undefine != NULL)
525 526 527
            conn->secretDriver->undefine(secret);
        virSecretFree(secret);
    }
528
    virBufferFreeAndReset(&buf);
529 530 531 532 533
    virSecretDefFree(def);
    VIR_FREE(enc_secret);
    return ret;
}

534 535 536 537 538
struct hookdata {
    virStorageVolDefPtr vol;
    bool skip;
};

539
static int virStorageBuildSetUIDHook(void *data) {
540 541 542 543 544
    struct hookdata *tmp = data;
    virStorageVolDefPtr vol = tmp->vol;

    if (tmp->skip)
        return 0;
545

546
    if (virSetUIDGID(vol->target.perms.uid, vol->target.perms.gid) < 0)
547
        return -1;
548

549 550 551
    return 0;
}

552
static int virStorageBackendCreateExecCommand(virStoragePoolObjPtr pool,
553
                                              virStorageVolDefPtr vol,
554
                                              virCommandPtr cmd) {
555 556 557 558
    struct stat st;
    gid_t gid;
    uid_t uid;
    int filecreated = 0;
559
    struct hookdata data = {vol, false};
560 561

    if ((pool->def->type == VIR_STORAGE_POOL_NETFS)
562 563 564 565 566
        && (((getuid() == 0)
             && (vol->target.perms.uid != -1)
             && (vol->target.perms.uid != 0))
            || ((vol->target.perms.gid != -1)
                && (vol->target.perms.gid != getgid())))) {
567 568 569 570

        virCommandSetPreExecHook(cmd, virStorageBuildSetUIDHook, &data);

        if (virCommandRun(cmd, NULL) == 0) {
571 572 573 574 575
            /* command was successfully run, check if the file was created */
            if (stat(vol->target.path, &st) >=0)
                filecreated = 1;
        }
    }
576 577 578

    data.skip = true;

579
    if (!filecreated) {
580
        if (virCommandRun(cmd, NULL) < 0) {
581 582 583
            return -1;
        }
        if (stat(vol->target.path, &st) < 0) {
584
            virReportSystemError(errno,
585
                                 _("failed to create %s"), vol->target.path);
586 587 588 589 590 591 592 593
            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)) {
594
        virReportSystemError(errno,
595
                             _("cannot chown %s to (%u, %u)"),
596
                             vol->target.path, uid, gid);
597 598 599
        return -1;
    }
    if (chmod(vol->target.path, vol->target.perms.mode) < 0) {
600
        virReportSystemError(errno,
601 602 603 604 605 606 607
                             _("cannot set mode of '%s' to %04o"),
                             vol->target.path, vol->target.perms.mode);
        return -1;
    }
    return 0;
}

608 609 610 611 612 613 614 615 616 617 618 619 620
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;
621
    int exitstatus;
622
    virCommandPtr cmd = virCommandNewArgList(qemuimg, "-h", NULL);
623

624 625 626
    virCommandAddEnvString(cmd, "LC_ALL=C");
    virCommandSetOutputBuffer(cmd, &help);
    virCommandClearCaps(cmd);
627

628 629 630
    /* 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 */
631
    if (virCommandRun(cmd, &exitstatus) < 0)
632 633
        goto cleanup;

634 635 636 637 638 639 640
    if ((start = strstr(help, " create ")) == NULL ||
        (end = strstr(start, "\n")) == NULL) {
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
                              _("unable to parse qemu-img output '%s'"),
                              help);
        goto cleanup;
    }
641 642
    if (((tmp = strstr(start, "-F fmt")) && tmp < end) ||
        ((tmp = strstr(start, "-F backing_fmt")) && tmp < end))
643 644 645 646 647 648 649
        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:
650
    virCommandFree(cmd);
651 652 653 654 655
    VIR_FREE(help);
    return ret;
}


656 657
static int
virStorageBackendCreateQemuImg(virConnectPtr conn,
658
                               virStoragePoolObjPtr pool,
659
                               virStorageVolDefPtr vol,
660
                               virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
661
                               unsigned int flags)
662
{
663
    int ret = -1;
664
    char *create_tool;
665
    int imgformat = -1;
666 667
    virCommandPtr cmd = NULL;
    bool do_encryption = (vol->target.encryption != NULL);
668
    unsigned long long int size_arg;
669

E
Eric Blake 已提交
670 671
    virCheckFlags(0, -1);

672
    const char *type = virStorageFileFormatTypeToString(vol->target.format);
673
    const char *backingType = vol->backingStore.path ?
674
        virStorageFileFormatTypeToString(vol->backingStore.format) : NULL;
675 676 677 678 679 680

    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 ?
681 682 683 684
        virStorageFileFormatTypeToString(inputvol->type == VIR_STORAGE_VOL_BLOCK ?
                                         VIR_STORAGE_FILE_RAW :
                                         inputvol->target.format) :
        NULL;
685 686

    if (type == NULL) {
687
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
688 689 690 691 692
                              _("unknown storage vol type %d"),
                              vol->target.format);
        return -1;
    }
    if (inputvol && inputType == NULL) {
693
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
694 695 696 697 698 699
                              _("unknown storage vol type %d"),
                              inputvol->target.format);
        return -1;
    }

    if (vol->backingStore.path) {
700 701
        int accessRetCode = -1;
        char *absolutePath = NULL;
702 703 704 705 706 707 708 709

        /* 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))) {
710
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
E
Eric Blake 已提交
711
                                  "%s", _("a different backing store cannot "
712 713 714 715 716
                                          "be specified."));
            return -1;
        }

        if (backingType == NULL) {
717
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
718 719 720 721
                                  _("unknown storage vol backing store type %d"),
                                  vol->backingStore.format);
            return -1;
        }
722 723 724 725 726 727 728 729 730 731 732 733 734 735

        /* 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) {
736
            virReportSystemError(errno,
737 738 739 740 741 742
                                 _("inaccessible backing store volume %s"),
                                 vol->backingStore.path);
            return -1;
        }
    }

743
    if (do_encryption) {
744 745
        virStorageEncryptionPtr enc;

746 747
        if (vol->target.format != VIR_STORAGE_FILE_QCOW &&
            vol->target.format != VIR_STORAGE_FILE_QCOW2) {
O
Osier Yang 已提交
748
            virStorageReportError(VIR_ERR_CONFIG_UNSUPPORTED,
749 750 751 752
                                  _("qcow volume encryption unsupported with "
                                    "volume format %s"), type);
            return -1;
        }
753 754 755
        enc = vol->target.encryption;
        if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW &&
            enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT) {
O
Osier Yang 已提交
756
            virStorageReportError(VIR_ERR_CONFIG_UNSUPPORTED,
757 758 759 760
                                  _("unsupported volume encryption format %d"),
                                  vol->target.encryption->format);
            return -1;
        }
761
        if (enc->nsecrets > 1) {
762
            virStorageReportError(VIR_ERR_XML_ERROR, "%s",
763 764 765
                                  _("too many secrets for qcow encryption"));
            return -1;
        }
766 767 768 769 770
        if (enc->format == VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT ||
            enc->nsecrets == 0) {
            if (virStorageGenerateQcowEncryption(conn, vol) < 0)
                return -1;
        }
771 772
    }

773
    /* Size in KB */
774
    size_arg = VIR_DIV_UP(vol->capacity, 1024);
775 776 777 778 779 780 781

    /* 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) {
782
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
783
                              "%s", _("unable to find kvm-img or qemu-img"));
784 785 786
        return -1;
    }

787 788 789 790
    imgformat = virStorageBackendQEMUImgBackingFormat(create_tool);
    if (imgformat < 0)
        goto cleanup;

791 792
    cmd = virCommandNew(create_tool);

793
    if (inputvol) {
794 795 796 797
        virCommandAddArgList(cmd, "convert", "-f", inputType, "-O", type,
                             inputPath, vol->target.path, NULL);

        if (do_encryption) {
798
            if (imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS) {
799
                virCommandAddArgList(cmd, "-o", "encryption=on", NULL);
800
            } else {
801
                virCommandAddArg(cmd, "-e");
802 803 804
            }
        }

805
    } else if (vol->backingStore.path) {
806 807 808
        virCommandAddArgList(cmd, "create", "-f", type,
                             "-b", vol->backingStore.path, NULL);

809 810
        switch (imgformat) {
        case QEMU_IMG_BACKING_FORMAT_FLAG:
811
            virCommandAddArgList(cmd, "-F", backingType, vol->target.path,
812 813
                                 NULL);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
814 815 816

            if (do_encryption)
                virCommandAddArg(cmd, "-e");
817 818 819
            break;

        case QEMU_IMG_BACKING_FORMAT_OPTIONS:
820 821 822
            virCommandAddArg(cmd, "-o");
            virCommandAddArgFormat(cmd, "backing_fmt=%s%s", backingType,
                                   do_encryption ? ",encryption=on" : "");
823 824
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
825
            break;
826 827 828 829 830

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

831 832
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
833 834 835
            if (do_encryption)
                virCommandAddArg(cmd, "-e");
        }
836
    } else {
837
        virCommandAddArgList(cmd, "create", "-f", type,
838 839
                             vol->target.path, NULL);
        virCommandAddArgFormat(cmd, "%lluK", size_arg);
840 841

        if (do_encryption) {
842
            if (imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS) {
843
                virCommandAddArgList(cmd, "-o", "encryption=on", NULL);
844
            } else {
845
                virCommandAddArg(cmd, "-e");
846 847
            }
        }
848
    }
849

850 851
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
cleanup:
852
    VIR_FREE(create_tool);
853
    virCommandFree(cmd);
854

855
    return ret;
856 857 858 859 860 861 862
}

/*
 * Xen removed the fully-functional qemu-img, and replaced it
 * with a partially functional qcow-create. Go figure ??!?
 */
static int
863
virStorageBackendCreateQcowCreate(virConnectPtr conn ATTRIBUTE_UNUSED,
864
                                  virStoragePoolObjPtr pool,
865
                                  virStorageVolDefPtr vol,
866
                                  virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
867
                                  unsigned int flags)
868
{
869
    int ret;
E
Eric Blake 已提交
870
    char *size;
871
    virCommandPtr cmd;
872

E
Eric Blake 已提交
873 874
    virCheckFlags(0, -1);

875
    if (inputvol) {
876
        virStorageReportError(VIR_ERR_INTERNAL_ERROR, "%s",
877 878 879 880
                              _("cannot copy from volume with qcow-create"));
        return -1;
    }

881
    if (vol->target.format != VIR_STORAGE_FILE_QCOW2) {
882
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
883 884 885 886 887
                              _("unsupported storage vol type %d"),
                              vol->target.format);
        return -1;
    }
    if (vol->backingStore.path != NULL) {
O
Osier Yang 已提交
888
        virStorageReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
889
                              _("copy-on-write image not supported with "
890
                                      "qcow-create"));
891 892
        return -1;
    }
893
    if (vol->target.encryption != NULL) {
O
Osier Yang 已提交
894
        virStorageReportError(VIR_ERR_CONFIG_UNSUPPORTED,
895 896 897 898
                              "%s", _("encrypted volumes not supported with "
                                      "qcow-create"));
        return -1;
    }
899 900

    /* Size in MB - yes different units to qemu-img :-( */
901 902
    if (virAsprintf(&size, "%llu",
                    VIR_DIV_UP(vol->capacity, (1024 * 1024))) < 0) {
E
Eric Blake 已提交
903 904 905
        virReportOOMError();
        return -1;
    }
906

907
    cmd = virCommandNewArgList("qcow-create", size, vol->target.path, NULL);
908

909 910
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
    virCommandFree(cmd);
E
Eric Blake 已提交
911
    VIR_FREE(size);
912

913
    return ret;
914 915
}

916
virStorageBackendBuildVolFrom
917
virStorageBackendFSImageToolTypeToFunc(int tool_type)
918 919 920 921 922 923 924 925
{
    switch (tool_type) {
    case TOOL_KVM_IMG:
    case TOOL_QEMU_IMG:
        return virStorageBackendCreateQemuImg;
    case TOOL_QCOW_CREATE:
        return virStorageBackendCreateQcowCreate;
    default:
926
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955
                              _("Unknown file create tool type '%d'."),
                              tool_type);
    }

    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;
}

956
virStorageBackendBuildVolFrom
957
virStorageBackendGetBuildVolFromFunction(virStorageVolDefPtr vol,
958 959 960 961 962 963 964 965 966 967 968
                                         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 &&
969
         vol->target.format != VIR_STORAGE_FILE_RAW) ||
970
        (inputvol->type == VIR_STORAGE_VOL_FILE &&
971
         inputvol->target.format != VIR_STORAGE_FILE_RAW)) {
972

D
Daniel P. Berrange 已提交
973
        if ((tool_type = virStorageBackendFindFSImageTool(NULL)) < 0) {
974
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
975 976 977 978 979
                                  "%s", _("creation of non-raw file images is "
                                          "not supported without qemu-img."));
            return NULL;
        }

980
        return virStorageBackendFSImageToolTypeToFunc(tool_type);
981 982
    }

983 984 985 986
    if (vol->type == VIR_STORAGE_VOL_BLOCK)
        return virStorageBackendCreateBlockFrom;
    else
        return virStorageBackendCreateRaw;
987
}
988

989

990 991
virStorageBackendPtr
virStorageBackendForType(int type) {
992
    unsigned int i;
993
    for (i = 0; backends[i]; i++)
994 995 996
        if (backends[i]->type == type)
            return backends[i];

997
    virStorageReportError(VIR_ERR_INTERNAL_ERROR,
998 999 1000 1001 1002
                          _("missing backend for pool type %d"), type);
    return NULL;
}


1003 1004 1005
/*
 * Allows caller to silently ignore files with improper mode
 *
1006 1007
 * Returns -1 on error, -2 if file mode is unexpected or the
 * volume is a dangling symbolic link.
1008
 */
1009
int
1010
virStorageBackendVolOpenCheckMode(const char *path, unsigned int flags)
1011
{
1012 1013
    int fd, mode = 0;
    struct stat sb;
1014
    char *base = last_component(path);
1015

1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030
    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;
    }

1031
    if ((fd = open(path, O_RDONLY|O_NONBLOCK|O_NOCTTY)) < 0) {
1032
        if ((errno == ENOENT || errno == ELOOP) &&
1033
            S_ISLNK(sb.st_mode)) {
1034 1035 1036 1037
            VIR_WARN("ignoring dangling symlink '%s'", path);
            return -2;
        }

1038
        virReportSystemError(errno,
1039
                             _("cannot open volume '%s'"),
1040
                             path);
1041 1042 1043
        return -1;
    }

1044 1045 1046 1047 1048 1049
    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;
1050 1051 1052 1053 1054 1055 1056 1057 1058 1059
    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;
        }
    }
1060 1061

    if (!(mode & flags)) {
1062
        VIR_FORCE_CLOSE(fd);
1063
        VIR_INFO("Skipping volume '%s'", path);
1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085

        if (mode & VIR_STORAGE_VOL_OPEN_ERROR) {
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
                                  _("unexpected storage mode for '%s'"), path);
            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,
1086 1087
                                     unsigned long long *capacity,
                                     unsigned int openflags)
1088 1089 1090
{
    int ret, fd;

1091 1092
    if ((ret = virStorageBackendVolOpenCheckMode(target->path,
                                                 openflags)) < 0)
1093 1094 1095
        return ret;

    fd = ret;
1096
    ret = virStorageBackendUpdateVolTargetInfoFD(target,
1097 1098 1099
                                                 fd,
                                                 allocation,
                                                 capacity);
1100

1101
    VIR_FORCE_CLOSE(fd);
1102 1103 1104 1105

    return ret;
}

1106
int
1107 1108 1109
virStorageBackendUpdateVolInfoFlags(virStorageVolDefPtr vol,
                                    int withCapacity,
                                    unsigned int openflags)
1110 1111 1112
{
    int ret;

1113
    if ((ret = virStorageBackendUpdateVolTargetInfo(&vol->target,
1114 1115 1116
                                    &vol->allocation,
                                    withCapacity ? &vol->capacity : NULL,
                                    openflags)) < 0)
1117 1118 1119
        return ret;

    if (vol->backingStore.path &&
1120
        (ret = virStorageBackendUpdateVolTargetInfo(&vol->backingStore,
1121 1122
                                            NULL, NULL,
                                            VIR_STORAGE_VOL_OPEN_DEFAULT)) < 0)
1123 1124 1125 1126 1127
        return ret;

    return 0;
}

1128 1129 1130 1131 1132 1133 1134
int virStorageBackendUpdateVolInfo(virStorageVolDefPtr vol,
                                   int withCapacity)
{
    return virStorageBackendUpdateVolInfoFlags(vol, withCapacity,
                                               VIR_STORAGE_VOL_OPEN_DEFAULT);
}

1135 1136 1137 1138
/*
 * virStorageBackendUpdateVolTargetInfoFD:
 * @conn: connection to report errors on
 * @target: target definition ptr of volume to update
1139
 * @fd: fd of storage volume to update, via virStorageBackendOpenVol*
1140 1141 1142
 * @allocation: If not NULL, updated allocation information will be stored
 * @capacity: If not NULL, updated capacity info will be stored
 *
1143
 * Returns 0 for success, -1 on a legitimate error condition.
1144
 */
1145
int
1146
virStorageBackendUpdateVolTargetInfoFD(virStorageVolTargetPtr target,
1147 1148 1149
                                       int fd,
                                       unsigned long long *allocation,
                                       unsigned long long *capacity)
1150 1151 1152 1153 1154 1155 1156
{
    struct stat sb;
#if HAVE_SELINUX
    security_context_t filecon = NULL;
#endif

    if (fstat(fd, &sb) < 0) {
1157
        virReportSystemError(errno,
1158
                             _("cannot stat file '%s'"),
1159
                             target->path);
1160 1161 1162
        return -1;
    }

1163 1164
    if (allocation) {
        if (S_ISREG(sb.st_mode)) {
1165
#ifndef WIN32
1166
            *allocation = (unsigned long long)sb.st_blocks *
1167
                          (unsigned long long)DEV_BSIZE;
D
Daniel P. Berrange 已提交
1168
#else
1169
            *allocation = sb.st_size;
D
Daniel P. Berrange 已提交
1170
#endif
1171 1172 1173 1174 1175
            /* Regular files may be sparse, so logical size (capacity) is not same
             * as actual allocation above
             */
            if (capacity)
                *capacity = sb.st_size;
1176 1177 1178 1179 1180
        } else if (S_ISDIR(sb.st_mode)) {
            *allocation = 0;
            if (capacity)
                *capacity = 0;

1181 1182
        } else {
            off_t end;
E
Eric Blake 已提交
1183
            /* XXX this is POSIX compliant, but doesn't work for CHAR files,
1184 1185 1186 1187 1188 1189
             * 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) {
1190
                virReportSystemError(errno,
1191 1192 1193 1194 1195 1196 1197
                                     _("cannot seek to end of file '%s'"),
                                     target->path);
                return -1;
            }
            *allocation = end;
            if (capacity)
                *capacity = end;
1198 1199 1200
        }
    }

1201 1202 1203
    target->perms.mode = sb.st_mode & S_IRWXUGO;
    target->perms.uid = sb.st_uid;
    target->perms.gid = sb.st_gid;
1204

1205
    VIR_FREE(target->perms.label);
1206 1207

#if HAVE_SELINUX
1208
    /* XXX: make this a security driver call */
1209
    if (fgetfilecon(fd, &filecon) == -1) {
1210
        if (errno != ENODATA && errno != ENOTSUP) {
1211
            virReportSystemError(errno,
1212
                                 _("cannot get file context of '%s'"),
1213
                                 target->path);
1214 1215
            return -1;
        } else {
1216
            target->perms.label = NULL;
1217 1218
        }
    } else {
1219
        target->perms.label = strdup(filecon);
1220
        freecon(filecon);
1221
        if (target->perms.label == NULL) {
1222
            virReportOOMError();
1223 1224
            return -1;
        }
1225 1226
    }
#else
1227
    target->perms.label = NULL;
1228 1229 1230 1231 1232
#endif

    return 0;
}

D
Dave Allan 已提交
1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267

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
1268 1269
virStorageBackendDetectBlockVolFormatFD(virStorageVolTargetPtr target,
                                        int fd)
D
Dave Allan 已提交
1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280
{
    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) {
1281
        virReportSystemError(errno,
D
Dave Allan 已提交
1282 1283 1284 1285 1286 1287
                             _("cannot seek to beginning of file '%s'"),
                             target->path);
        return -1;
    }
    bytes = saferead(fd, buffer, sizeof(buffer));
    if (bytes < 0) {
1288
        virReportSystemError(errno,
D
Dave Allan 已提交
1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307
                             _("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;
}


1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321
/*
 * 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.
 */
char *
1322
virStorageBackendStablePath(virStoragePoolObjPtr pool,
1323
                            const char *devpath)
1324 1325 1326
{
    DIR *dh;
    struct dirent *dent;
1327
    char *stablepath;
1328
    int opentries = 0;
1329 1330 1331 1332 1333

    /* 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/"))
1334
        goto ret_strdup;
1335

1336 1337 1338 1339 1340 1341
    /* Skip whole thing for a pool which isn't in /dev
     * so we don't mess will filesystem/dir based pools
     */
    if (!STRPREFIX(pool->def->target.path, "/dev"))
        goto ret_strdup;

1342 1343 1344
    /* 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.
1345
     */
1346
 reopen:
1347
    if ((dh = opendir(pool->def->target.path)) == NULL) {
1348 1349 1350 1351 1352
        opentries++;
        if (errno == ENOENT && opentries < 50) {
            usleep(100 * 1000);
            goto reopen;
        }
1353
        virReportSystemError(errno,
1354 1355
                             _("cannot read dir '%s'"),
                             pool->def->target.path);
1356 1357 1358
        return NULL;
    }

1359 1360 1361 1362 1363
    /* 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
     * to this device node
     */
1364 1365 1366 1367
    while ((dent = readdir(dh)) != NULL) {
        if (dent->d_name[0] == '.')
            continue;

1368 1369 1370
        if (virAsprintf(&stablepath, "%s/%s",
                        pool->def->target.path,
                        dent->d_name) == -1) {
1371
            virReportOOMError();
1372 1373 1374 1375 1376 1377 1378 1379 1380
            closedir(dh);
            return NULL;
        }

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

1381
        VIR_FREE(stablepath);
1382 1383 1384 1385
    }

    closedir(dh);

1386
 ret_strdup:
1387 1388 1389
    /* Couldn't find any matching stable link so give back
     * the original non-stable dev path
     */
1390 1391 1392 1393

    stablepath = strdup(devpath);

    if (stablepath == NULL)
1394
        virReportOOMError();
1395 1396

    return stablepath;
1397 1398
}

D
Daniel P. Berrange 已提交
1399

1400
#ifndef WIN32
1401 1402 1403 1404
/*
 * Run an external program.
 *
 * Read its output and apply a series of regexes to each line
R
Richard W.M. Jones 已提交
1405
 * When the entire set of regexes has matched consecutively
1406 1407 1408
 * then run a callback passing in all the matches
 */
int
1409
virStorageBackendRunProgRegex(virStoragePoolObjPtr pool,
1410
                              const char *const*prog,
1411 1412 1413 1414
                              int nregex,
                              const char **regex,
                              int *nvars,
                              virStorageBackendListVolRegexFunc func,
1415
                              void *data, const char *prefix)
1416
{
1417
    int fd = -1, err, ret = -1;
1418 1419 1420 1421 1422 1423 1424
    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;
1425
    virCommandPtr cmd = NULL;
1426 1427

    /* Compile all regular expressions */
1428
    if (VIR_ALLOC_N(reg, nregex) < 0) {
1429
        virReportOOMError();
1430 1431 1432 1433 1434 1435 1436 1437
        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));
1438
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
1439 1440 1441
                                  _("Failed to compile regex %s"), error);
            for (j = 0 ; j <= i ; j++)
                regfree(&reg[j]);
1442
            VIR_FREE(reg);
1443 1444 1445 1446 1447 1448 1449 1450 1451 1452
            return -1;
        }

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

    }

    /* Storage for matched variables */
1453
    if (VIR_ALLOC_N(groups, totgroups) < 0) {
1454
        virReportOOMError();
1455 1456
        goto cleanup;
    }
1457
    if (VIR_ALLOC_N(vars, maxvars+1) < 0) {
1458
        virReportOOMError();
1459 1460 1461
        goto cleanup;
    }

1462 1463 1464
    cmd = virCommandNewArgs(prog);
    virCommandSetOutputFD(cmd, &fd);
    if (virCommandRunAsync(cmd, NULL) < 0) {
1465 1466 1467
        goto cleanup;
    }

1468
    if ((list = VIR_FDOPEN(fd, "r")) == NULL) {
1469
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
J
Jim Meyering 已提交
1470
                              "%s", _("cannot read fd"));
1471 1472 1473 1474
        goto cleanup;
    }

    while (fgets(line, sizeof(line), list) != NULL) {
1475
        char *p = NULL;
1476 1477 1478 1479 1480
        /* Strip trailing newline */
        int len = strlen(line);
        if (len && line[len-1] == '\n')
            line[len-1] = '\0';

1481 1482 1483 1484 1485 1486
        /* ignore any command prefix */
        if (prefix)
            p = STRSKIP(line, prefix);
        if (!p)
            p = line;

1487
        for (i = 0 ; i <= maxReg && i < nregex ; i++) {
1488
            if (regexec(&reg[i], p, nvars[i]+1, vars, 0) == 0) {
1489 1490 1491 1492 1493 1494 1495 1496
                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 */
1497
                    p[vars[j+1].rm_eo] = '\0';
1498
                    if ((groups[ngroup++] =
1499
                         strdup(p + vars[j+1].rm_so)) == NULL) {
1500
                        virReportOOMError();
1501 1502 1503 1504 1505 1506
                        goto cleanup;
                    }
                }

                /* We're matching on the last regex, so callback time */
                if (i == (nregex-1)) {
1507
                    if (((*func)(pool, groups, data)) < 0)
1508 1509 1510
                        goto cleanup;

                    /* Release matches & restart to matching the first regex */
1511
                    for (j = 0 ; j < totgroups ; j++)
1512
                        VIR_FREE(groups[j]);
1513 1514 1515 1516 1517 1518 1519
                    maxReg = 0;
                    ngroup = 0;
                }
            }
        }
    }

1520
    ret = virCommandWait(cmd, NULL);
1521
cleanup:
1522 1523
    if (groups) {
        for (j = 0 ; j < totgroups ; j++)
1524 1525
            VIR_FREE(groups[j]);
        VIR_FREE(groups);
1526
    }
1527
    VIR_FREE(vars);
1528 1529 1530 1531

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

1532
    VIR_FREE(reg);
1533
    virCommandFree(cmd);
1534

1535 1536
    VIR_FORCE_FCLOSE(list);
    VIR_FORCE_CLOSE(fd);
1537

1538
    return ret;
1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552
}

/*
 * 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
1553
virStorageBackendRunProgNul(virStoragePoolObjPtr pool,
1554 1555 1556 1557 1558 1559
                            const char **prog,
                            size_t n_columns,
                            virStorageBackendListVolNulFunc func,
                            void *data)
{
    size_t n_tok = 0;
1560
    int fd = -1;
1561 1562
    FILE *fp = NULL;
    char **v;
1563
    int ret = -1;
1564
    int i;
1565
    virCommandPtr cmd = NULL;
1566 1567 1568 1569

    if (n_columns == 0)
        return -1;

1570
    if (VIR_ALLOC_N(v, n_columns) < 0) {
1571
        virReportOOMError();
1572 1573 1574 1575 1576
        return -1;
    }
    for (i = 0; i < n_columns; i++)
        v[i] = NULL;

1577 1578 1579
    cmd = virCommandNewArgs(prog);
    virCommandSetOutputFD(cmd, &fd);
    if (virCommandRunAsync(cmd, NULL) < 0) {
1580 1581 1582
        goto cleanup;
    }

1583
    if ((fp = VIR_FDOPEN(fd, "r")) == NULL) {
1584
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
1585
                              "%s", _("cannot open file using fd"));
1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598
        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.  */
        ssize_t tok_len = getdelim (&buf, &buf_len, 0, fp);
        v[n_tok] = buf;
        if (tok_len < 0) {
            /* Maybe EOF, maybe an error.
               If n_tok > 0, then we know it's an error.  */
1599
            if (n_tok && func (pool, n_tok, v, data) < 0)
1600 1601 1602 1603 1604
                goto cleanup;
            break;
        }
        ++n_tok;
        if (n_tok == n_columns) {
1605
            if (func (pool, n_tok, v, data) < 0)
1606 1607 1608
                goto cleanup;
            n_tok = 0;
            for (i = 0; i < n_columns; i++) {
1609
                VIR_FREE(v[i]);
1610 1611 1612 1613
            }
        }
    }

1614
    if (feof (fp) < 0) {
1615
        virReportSystemError(errno,
1616
                             _("read error on pipe to '%s'"), prog[0]);
1617 1618
        goto cleanup;
    }
1619

1620
    ret = virCommandWait(cmd, NULL);
1621 1622
 cleanup:
    for (i = 0; i < n_columns; i++)
1623 1624
        VIR_FREE(v[i]);
    VIR_FREE(v);
1625
    virCommandFree(cmd);
1626

1627 1628
    VIR_FORCE_FCLOSE(fp);
    VIR_FORCE_CLOSE(fd);
1629

1630
    return ret;
1631
}
D
Daniel P. Berrange 已提交
1632

1633
#else /* WIN32 */
D
Daniel P. Berrange 已提交
1634 1635 1636 1637 1638 1639 1640 1641 1642

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,
1643
                              void *data ATTRIBUTE_UNUSED)
D
Daniel P. Berrange 已提交
1644
{
1645 1646
    virStorageReportError(VIR_ERR_INTERNAL_ERROR,
                          _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657
    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)
{
1658
    virStorageReportError(VIR_ERR_INTERNAL_ERROR, _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1659 1660
    return -1;
}
1661
#endif /* WIN32 */