storage_backend.c 51.5 KB
Newer Older
1
/*
2
 * storage_backend.c: internal storage driver backend contract
3
 *
4
 * Copyright (C) 2007-2013 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2007-2008 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24 25 26
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

#include <config.h>

#include <string.h>
27
#include <stdio.h>
28
#include <regex.h>
29
#include <sys/types.h>
E
Eric Blake 已提交
30
#include <sys/wait.h>
31
#include <unistd.h>
32 33
#include <fcntl.h>
#include <sys/stat.h>
34
#include <sys/param.h>
35
#include <dirent.h>
36
#include "dirname.h"
37 38 39 40
#ifdef __linux__
# include <sys/ioctl.h>
# include <linux/fs.h>
#endif
41

42
#if WITH_SELINUX
43
# include <selinux/selinux.h>
44
#endif
45

46
#include "datatypes.h"
47
#include "virerror.h"
48
#include "virutil.h"
49
#include "viralloc.h"
50
#include "internal.h"
51
#include "secret_conf.h"
52
#include "viruuid.h"
53
#include "virstoragefile.h"
D
Daniel P. Berrange 已提交
54
#include "storage_backend.h"
55
#include "virlog.h"
E
Eric Blake 已提交
56
#include "virfile.h"
57
#include "stat-time.h"
58

59
#if WITH_STORAGE_LVM
60
# include "storage_backend_logical.h"
61 62
#endif
#if WITH_STORAGE_ISCSI
63
# include "storage_backend_iscsi.h"
64
#endif
65
#if WITH_STORAGE_SCSI
66
# include "storage_backend_scsi.h"
67
#endif
D
Dave Allan 已提交
68
#if WITH_STORAGE_MPATH
69
# include "storage_backend_mpath.h"
D
Dave Allan 已提交
70
#endif
71
#if WITH_STORAGE_DISK
72
# include "storage_backend_disk.h"
73 74
#endif
#if WITH_STORAGE_DIR
75
# include "storage_backend_fs.h"
76
#endif
77 78 79
#if WITH_STORAGE_RBD
# include "storage_backend_rbd.h"
#endif
80 81 82
#if WITH_STORAGE_SHEEPDOG
# include "storage_backend_sheepdog.h"
#endif
83

84 85
#define VIR_FROM_THIS VIR_FROM_STORAGE

86 87 88 89 90 91 92 93 94 95 96 97 98 99
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
100 101 102
#if WITH_STORAGE_SCSI
    &virStorageBackendSCSI,
#endif
D
Dave Allan 已提交
103 104 105
#if WITH_STORAGE_MPATH
    &virStorageBackendMpath,
#endif
106 107
#if WITH_STORAGE_DISK
    &virStorageBackendDisk,
108 109 110
#endif
#if WITH_STORAGE_RBD
    &virStorageBackendRBD,
111 112 113
#endif
#if WITH_STORAGE_SHEEPDOG
    &virStorageBackendSheepdog,
114 115 116
#endif
    NULL
};
117

118 119 120 121 122 123 124 125
static int track_allocation_progress = 0;

enum {
    TOOL_QEMU_IMG,
    TOOL_KVM_IMG,
    TOOL_QCOW_CREATE,
};

126 127 128
#define READ_BLOCK_SIZE_DEFAULT  (1024 * 1024)
#define WRITE_BLOCK_SIZE_DEFAULT (4 * 1024)

129
static int ATTRIBUTE_NONNULL(2)
130
virStorageBackendCopyToFD(virStorageVolDefPtr vol,
131 132
                          virStorageVolDefPtr inputvol,
                          int fd,
133 134
                          unsigned long long *total,
                          int is_dest_file)
135 136 137
{
    int inputfd = -1;
    int amtread = -1;
138
    int ret = 0;
139 140 141
    size_t rbytes = READ_BLOCK_SIZE_DEFAULT;
    size_t wbytes = 0;
    int interval;
142
    char *zerobuf = NULL;
143
    char *buf = NULL;
144
    struct stat st;
145

146
    if ((inputfd = open(inputvol->target.path, O_RDONLY)) < 0) {
147
        ret = -errno;
148
        virReportSystemError(errno,
149 150 151
                             _("could not open input path '%s'"),
                             inputvol->target.path);
        goto cleanup;
152 153
    }

154 155 156 157 158 159 160 161 162 163 164 165 166 167 168
#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;
    }
169

170
    if (VIR_ALLOC_N(buf, rbytes) < 0) {
171
        ret = -errno;
172
        virReportOOMError();
173 174 175 176 177 178
        goto cleanup;
    }

    while (amtread != 0) {
        int amtleft;

179 180
        if (*total < rbytes)
            rbytes = *total;
181

182
        if ((amtread = saferead(inputfd, buf, rbytes)) < 0) {
183
            ret = -errno;
184
            virReportSystemError(errno,
185 186 187 188
                                 _("failed reading from file '%s'"),
                                 inputvol->target.path);
            goto cleanup;
        }
189
        *total -= amtread;
190 191 192 193 194

        /* Loop over amt read in 512 byte increments, looking for sparse
         * blocks */
        amtleft = amtread;
        do {
195
            interval = ((wbytes > amtleft) ? amtleft : wbytes);
196 197
            int offset = amtread - amtleft;

198
            if (is_dest_file && memcmp(buf+offset, zerobuf, interval) == 0) {
199
                if (lseek(fd, interval, SEEK_CUR) < 0) {
200
                    ret = -errno;
201
                    virReportSystemError(errno,
202 203 204 205 206
                                         _("cannot extend file '%s'"),
                                         vol->target.path);
                    goto cleanup;
                }
            } else if (safewrite(fd, buf+offset, interval) < 0) {
207
                ret = -errno;
208
                virReportSystemError(errno,
209 210 211 212 213
                                     _("failed writing to file '%s'"),
                                     vol->target.path);
                goto cleanup;

            }
214
        } while ((amtleft -= interval) > 0);
215 216
    }

217 218 219 220 221 222 223 224
    if (fdatasync(fd) < 0) {
        ret = -errno;
        virReportSystemError(errno, _("cannot sync data to file '%s'"),
                             vol->target.path);
        goto cleanup;
    }


225
    if (VIR_CLOSE(inputfd) < 0) {
226
        ret = -errno;
227
        virReportSystemError(errno,
228 229 230 231 232 233 234
                             _("cannot close file '%s'"),
                             inputvol->target.path);
        goto cleanup;
    }
    inputfd = -1;

cleanup:
235
    VIR_FORCE_CLOSE(inputfd);
236

237
    VIR_FREE(zerobuf);
238 239
    VIR_FREE(buf);

240 241 242
    return ret;
}

243
static int
244
virStorageBackendCreateBlockFrom(virConnectPtr conn ATTRIBUTE_UNUSED,
245
                                 virStoragePoolObjPtr pool ATTRIBUTE_UNUSED,
246 247
                                 virStorageVolDefPtr vol,
                                 virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
248
                                 unsigned int flags)
249 250 251 252
{
    int fd = -1;
    int ret = -1;
    unsigned long long remain;
253 254 255
    struct stat st;
    gid_t gid;
    uid_t uid;
256

E
Eric Blake 已提交
257 258
    virCheckFlags(0, -1);

259
    if ((fd = open(vol->target.path, O_RDWR)) < 0) {
260
        virReportSystemError(errno,
261 262 263 264 265 266 267 268
                             _("cannot create path '%s'"),
                             vol->target.path);
        goto cleanup;
    }

    remain = vol->allocation;

    if (inputvol) {
269
        int res = virStorageBackendCopyToFD(vol, inputvol,
270
                                            fd, &remain, 0);
271 272 273 274
        if (res < 0)
            goto cleanup;
    }

275
    if (fstat(fd, &st) == -1) {
276
        virReportSystemError(errno, _("stat of '%s' failed"),
277 278 279
                             vol->target.path);
        goto cleanup;
    }
P
Philipp Hahn 已提交
280 281 282
    uid = (vol->target.perms.uid != st.st_uid) ? vol->target.perms.uid : (uid_t) -1;
    gid = (vol->target.perms.gid != st.st_gid) ? vol->target.perms.gid : (gid_t) -1;
    if (((uid != (uid_t) -1) || (gid != (gid_t) -1))
283
        && (fchown(fd, uid, gid) < 0)) {
284
        virReportSystemError(errno,
285
                             _("cannot chown '%s' to (%u, %u)"),
286 287
                             vol->target.path, (unsigned int) uid,
                             (unsigned int) gid);
288 289 290
        goto cleanup;
    }
    if (fchmod(fd, vol->target.perms.mode) < 0) {
291
        virReportSystemError(errno,
292 293 294 295
                             _("cannot set mode of '%s' to %04o"),
                             vol->target.path, vol->target.perms.mode);
        goto cleanup;
    }
296
    if (VIR_CLOSE(fd) < 0) {
297
        virReportSystemError(errno,
298 299 300 301 302 303 304 305
                             _("cannot close file '%s'"),
                             vol->target.path);
        goto cleanup;
    }
    fd = -1;

    ret = 0;
cleanup:
306
    VIR_FORCE_CLOSE(fd);
307 308 309 310

    return ret;
}

E
Eric Blake 已提交
311 312 313 314
static int
createRawFile(int fd, virStorageVolDefPtr vol,
              virStorageVolDefPtr inputvol)
{
315 316
    int ret = 0;
    unsigned long long remain;
317

318 319
    /* Seek to the final size, so the capacity is available upfront
     * for progress reporting */
E
Eric Blake 已提交
320
    if (ftruncate(fd, vol->capacity) < 0) {
321
        ret = -errno;
322
        virReportSystemError(errno,
323
                             _("cannot extend file '%s'"),
E
Eric Blake 已提交
324
                             vol->target.path);
325 326 327
        goto cleanup;
    }

E
Eric Blake 已提交
328
    remain = vol->allocation;
329

E
Eric Blake 已提交
330 331
    if (inputvol) {
        ret = virStorageBackendCopyToFD(vol, inputvol, fd, &remain, 1);
332
        if (ret < 0) {
333
            goto cleanup;
334
        }
335 336 337 338 339 340 341 342 343 344 345 346 347
    }

    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
348
                if (safezero(fd, vol->allocation - remain, bytes) < 0) {
349
                    ret = -errno;
J
Jiri Denemark 已提交
350
                    virReportSystemError(errno, _("cannot fill file '%s'"),
E
Eric Blake 已提交
351
                                         vol->target.path);
352 353 354 355 356
                    goto cleanup;
                }
                remain -= bytes;
            }
        } else { /* No progress bars to be shown */
E
Eric Blake 已提交
357
            if (safezero(fd, 0, remain) < 0) {
358
                ret = -errno;
J
Jiri Denemark 已提交
359
                virReportSystemError(errno, _("cannot fill file '%s'"),
E
Eric Blake 已提交
360
                                     vol->target.path);
361 362 363
                goto cleanup;
            }
        }
364 365 366
    }

    if (fsync(fd) < 0) {
367
        ret = -errno;
368
        virReportSystemError(errno, _("cannot sync data to file '%s'"),
E
Eric Blake 已提交
369
                             vol->target.path);
370
        goto cleanup;
371 372
    }

373 374 375 376 377 378 379 380 381
cleanup:
    return ret;
}

int
virStorageBackendCreateRaw(virConnectPtr conn ATTRIBUTE_UNUSED,
                           virStoragePoolObjPtr pool,
                           virStorageVolDefPtr vol,
                           virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
382
                           unsigned int flags)
383 384
{
    int ret = -1;
E
Eric Blake 已提交
385 386 387 388
    int fd = -1;
    int operation_flags;

    virCheckFlags(0, -1);
389 390

    if (vol->target.encryption != NULL) {
391 392 393
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       "%s", _("storage pool does not support encrypted "
                               "volumes"));
394 395
        goto cleanup;
    }
396

L
Laine Stump 已提交
397
    operation_flags = VIR_FILE_OPEN_FORCE_MODE | VIR_FILE_OPEN_FORCE_OWNER;
E
Eric Blake 已提交
398
    if (pool->def->type == VIR_STORAGE_POOL_NETFS)
L
Laine Stump 已提交
399
        operation_flags |= VIR_FILE_OPEN_FORK;
E
Eric Blake 已提交
400

401 402
    if ((fd = virFileOpenAs(vol->target.path,
                            O_RDWR | O_CREAT | O_EXCL,
L
Laine Stump 已提交
403 404 405
                            vol->target.perms.mode,
                            vol->target.perms.uid,
                            vol->target.perms.gid,
406
                            operation_flags)) < 0) {
E
Eric Blake 已提交
407 408 409 410 411 412
        virReportSystemError(-fd,
                             _("cannot create path '%s'"),
                             vol->target.path);
        goto cleanup;
    }

413 414
    if ((ret = createRawFile(fd, vol, inputvol)) < 0)
        /* createRawFile already reported the exact error. */
E
Eric Blake 已提交
415
        ret = -1;
416 417

cleanup:
E
Eric Blake 已提交
418
    VIR_FORCE_CLOSE(fd);
419 420 421
    return ret;
}

422 423 424 425 426 427 428 429 430
static int
virStorageGenerateSecretUUID(virConnectPtr conn,
                             unsigned char *uuid)
{
    unsigned attempt;

    for (attempt = 0; attempt < 65536; attempt++) {
        virSecretPtr tmp;
        if (virUUIDGenerate(uuid) < 0) {
431 432
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("unable to generate uuid"));
433 434
            return -1;
        }
435
        tmp = conn->secretDriver->secretLookupByUUID(conn, uuid);
436 437 438 439 440 441
        if (tmp == NULL)
            return 0;

        virSecretFree(tmp);
    }

442 443
    virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                   _("too many conflicts when generating an uuid"));
444 445 446 447

    return -1;
}

448 449 450 451 452 453 454 455 456
static int
virStorageGenerateQcowEncryption(virConnectPtr conn,
                                 virStorageVolDefPtr vol)
{
    virSecretDefPtr def = NULL;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    virStorageEncryptionPtr enc;
    virStorageEncryptionSecretPtr enc_secret = NULL;
    virSecretPtr secret = NULL;
457
    char *xml;
458 459 460
    unsigned char value[VIR_STORAGE_QCOW_PASSPHRASE_SIZE];
    int ret = -1;

461
    if (conn->secretDriver == NULL ||
462 463 464
        conn->secretDriver->secretLookupByUUID == NULL ||
        conn->secretDriver->secretDefineXML == NULL ||
        conn->secretDriver->secretSetValue == NULL) {
465 466
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("secret storage not supported"));
467 468 469 470 471
        goto cleanup;
    }

    enc = vol->target.encryption;
    if (enc->nsecrets != 0) {
472 473
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("secrets already defined"));
474 475 476 477 478
        goto cleanup;
    }

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

483
    def->ephemeral = false;
484
    def->private = false;
485
    if (virStorageGenerateSecretUUID(conn, def->uuid) < 0)
486
        goto cleanup;
487

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

500
    secret = conn->secretDriver->secretDefineXML(conn, xml, 0);
501 502 503 504 505 506
    if (secret == NULL) {
        VIR_FREE(xml);
        goto cleanup;
    }
    VIR_FREE(xml);

507
    if (virStorageGenerateQcowPassphrase(value) < 0)
508 509
        goto cleanup;

510
    if (conn->secretDriver->secretSetValue(secret, value, sizeof(value), 0) < 0)
511 512 513
        goto cleanup;

    enc_secret->type = VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE;
514
    memcpy(enc_secret->uuid, secret->uuid, VIR_UUID_BUFLEN);
515 516 517 518 519 520 521 522 523
    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) {
524
        if (ret != 0 &&
525 526
            conn->secretDriver->secretUndefine != NULL)
            conn->secretDriver->secretUndefine(secret);
527 528
        virSecretFree(secret);
    }
529
    virBufferFreeAndReset(&buf);
530 531 532 533 534
    virSecretDefFree(def);
    VIR_FREE(enc_secret);
    return ret;
}

535
static int virStorageBackendCreateExecCommand(virStoragePoolObjPtr pool,
536
                                              virStorageVolDefPtr vol,
537
                                              virCommandPtr cmd) {
538 539 540 541 542 543
    struct stat st;
    gid_t gid;
    uid_t uid;
    int filecreated = 0;

    if ((pool->def->type == VIR_STORAGE_POOL_NETFS)
544
        && (((getuid() == 0)
P
Philipp Hahn 已提交
545
             && (vol->target.perms.uid != (uid_t) -1)
546
             && (vol->target.perms.uid != 0))
P
Philipp Hahn 已提交
547
            || ((vol->target.perms.gid != (gid_t) -1)
548
                && (vol->target.perms.gid != getgid())))) {
549

550 551
        virCommandSetUID(cmd, vol->target.perms.uid);
        virCommandSetGID(cmd, vol->target.perms.gid);
552 553

        if (virCommandRun(cmd, NULL) == 0) {
554 555 556 557 558
            /* command was successfully run, check if the file was created */
            if (stat(vol->target.path, &st) >=0)
                filecreated = 1;
        }
    }
559

560 561 562
    /* don't change uid/gid if we retry */
    virCommandSetUID(cmd, -1);
    virCommandSetGID(cmd, -1);
563

564
    if (!filecreated) {
565
        if (virCommandRun(cmd, NULL) < 0) {
566 567 568
            return -1;
        }
        if (stat(vol->target.path, &st) < 0) {
569
            virReportSystemError(errno,
570
                                 _("failed to create %s"), vol->target.path);
571 572 573 574
            return -1;
        }
    }

P
Philipp Hahn 已提交
575 576 577
    uid = (vol->target.perms.uid != st.st_uid) ? vol->target.perms.uid : (uid_t) -1;
    gid = (vol->target.perms.gid != st.st_gid) ? vol->target.perms.gid : (gid_t) -1;
    if (((uid != (uid_t) -1) || (gid != (gid_t) -1))
578
        && (chown(vol->target.path, uid, gid) < 0)) {
579
        virReportSystemError(errno,
580
                             _("cannot chown %s to (%u, %u)"),
581 582
                             vol->target.path, (unsigned int) uid,
                             (unsigned int) gid);
583 584 585
        return -1;
    }
    if (chmod(vol->target.path, vol->target.perms.mode) < 0) {
586
        virReportSystemError(errno,
587 588 589 590 591 592 593
                             _("cannot set mode of '%s' to %04o"),
                             vol->target.path, vol->target.perms.mode);
        return -1;
    }
    return 0;
}

594 595 596 597 598 599 600 601 602 603 604 605 606
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;
607
    int exitstatus;
608
    virCommandPtr cmd = virCommandNewArgList(qemuimg, "-h", NULL);
609

610 611 612
    virCommandAddEnvString(cmd, "LC_ALL=C");
    virCommandSetOutputBuffer(cmd, &help);
    virCommandClearCaps(cmd);
613

614 615 616
    /* 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 */
617
    if (virCommandRun(cmd, &exitstatus) < 0)
618 619
        goto cleanup;

620 621
    if ((start = strstr(help, " create ")) == NULL ||
        (end = strstr(start, "\n")) == NULL) {
622 623 624
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unable to parse qemu-img output '%s'"),
                       help);
625 626
        goto cleanup;
    }
627 628
    if (((tmp = strstr(start, "-F fmt")) && tmp < end) ||
        ((tmp = strstr(start, "-F backing_fmt")) && tmp < end))
629 630 631 632 633 634 635
        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:
636
    virCommandFree(cmd);
637 638 639 640
    VIR_FREE(help);
    return ret;
}

641 642 643 644 645 646 647 648
virCommandPtr
virStorageBackendCreateQemuImgCmd(virConnectPtr conn,
                                  virStoragePoolObjPtr pool,
                                  virStorageVolDefPtr vol,
                                  virStorageVolDefPtr inputvol,
                                  unsigned int flags,
                                  const char *create_tool,
                                  int imgformat)
649
{
650 651
    virCommandPtr cmd = NULL;
    bool do_encryption = (vol->target.encryption != NULL);
652
    unsigned long long int size_arg;
653
    bool preallocate = false;
654

655 656 657 658 659 660
    /* Treat output block devices as 'raw' format */
    const char *type =
        virStorageFileFormatTypeToString(vol->type == VIR_STORAGE_VOL_BLOCK ?
                                         VIR_STORAGE_FILE_RAW :
                                         vol->target.format);

661
    const char *backingType = vol->backingStore.path ?
662
        virStorageFileFormatTypeToString(vol->backingStore.format) : NULL;
663 664 665 666 667 668

    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 ?
669 670 671 672
        virStorageFileFormatTypeToString(inputvol->type == VIR_STORAGE_VOL_BLOCK ?
                                         VIR_STORAGE_FILE_RAW :
                                         inputvol->target.format) :
        NULL;
673

674 675 676 677
    virCheckFlags(VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA, NULL);

    preallocate = !!(flags & VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA);

678
    if (type == NULL) {
679 680 681
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unknown storage vol type %d"),
                       vol->target.format);
682
        return NULL;
683 684
    }
    if (inputvol && inputType == NULL) {
685 686 687
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unknown storage vol type %d"),
                       inputvol->target.format);
688
        return NULL;
689
    }
690 691 692
    if (preallocate && vol->target.format != VIR_STORAGE_FILE_QCOW2) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("metadata preallocation only available with qcow2"));
693
        return NULL;
694
    }
695 696

    if (vol->backingStore.path) {
697 698
        int accessRetCode = -1;
        char *absolutePath = NULL;
699

700 701 702 703
        if (preallocate) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("metadata preallocation conflicts with backing"
                             " store"));
704
            return NULL;
705 706
        }

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
            return NULL;
718 719 720
        }

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

        /* 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();
734
            return NULL;
735 736 737 738 739
        }
        accessRetCode = access(absolutePath ? absolutePath
                               : vol->backingStore.path, R_OK);
        VIR_FREE(absolutePath);
        if (accessRetCode != 0) {
740
            virReportSystemError(errno,
741 742
                                 _("inaccessible backing store volume %s"),
                                 vol->backingStore.path);
743
            return NULL;
744 745 746
        }
    }

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
            return NULL;
756
        }
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
            return NULL;
764
        }
765
        if (enc->nsecrets > 1) {
766 767
            virReportError(VIR_ERR_XML_ERROR, "%s",
                           _("too many secrets for qcow encryption"));
768
            return NULL;
769
        }
770 771 772
        if (enc->format == VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT ||
            enc->nsecrets == 0) {
            if (virStorageGenerateQcowEncryption(conn, vol) < 0)
773
                return NULL;
774
        }
775 776
    }

777
    /* Size in KB */
778
    size_arg = VIR_DIV_UP(vol->capacity, 1024);
779

780 781
    cmd = virCommandNew(create_tool);

782
    if (inputvol) {
783
        virCommandAddArgList(cmd, "convert", "-f", inputType, "-O", type, NULL);
784

785 786 787 788 789 790 791 792
        if (imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS &&
            (do_encryption || preallocate)) {
            virCommandAddArg(cmd, "-o");
            virCommandAddArgFormat(cmd, "%s%s%s", do_encryption ? "encryption=on" : "",
                                   (do_encryption && preallocate) ? "," : "",
                                   preallocate ? "preallocation=metadata" : "");
        } else if (do_encryption) {
            virCommandAddArg(cmd, "-e");
793
        }
794
        virCommandAddArgList(cmd, inputPath, vol->target.path, NULL);
795
    } else if (vol->backingStore.path) {
796 797 798
        virCommandAddArgList(cmd, "create", "-f", type,
                             "-b", vol->backingStore.path, NULL);

799 800
        switch (imgformat) {
        case QEMU_IMG_BACKING_FORMAT_FLAG:
801
            virCommandAddArgList(cmd, "-F", backingType, NULL);
802 803
            if (do_encryption)
                virCommandAddArg(cmd, "-e");
804 805
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
806 807 808
            break;

        case QEMU_IMG_BACKING_FORMAT_OPTIONS:
809 810 811
            virCommandAddArg(cmd, "-o");
            virCommandAddArgFormat(cmd, "backing_fmt=%s%s", backingType,
                                   do_encryption ? ",encryption=on" : "");
812 813
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
814
            break;
815 816

        default:
817 818
            VIR_DEBUG("Unable to set backing store format for %s with %s",
                      vol->target.path, create_tool);
819

820 821
            if (do_encryption)
                virCommandAddArg(cmd, "-e");
822 823
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
824
        }
825
    } else {
826
        virCommandAddArgList(cmd, "create", "-f", type, NULL);
827

828 829 830 831 832 833 834 835
        if (imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS &&
            (do_encryption || preallocate)) {
            virCommandAddArg(cmd, "-o");
            virCommandAddArgFormat(cmd, "%s%s%s", do_encryption ? "encryption=on" : "",
                                   (do_encryption && preallocate) ? "," : "",
                                   preallocate ? "preallocation=metadata" : "");
        } else if (do_encryption) {
            virCommandAddArg(cmd, "-e");
836
        }
837 838
        virCommandAddArg(cmd, vol->target.path);
        virCommandAddArgFormat(cmd, "%lluK", size_arg);
839
    }
840

841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877
    return cmd;
}

static int
virStorageBackendCreateQemuImg(virConnectPtr conn,
                               virStoragePoolObjPtr pool,
                               virStorageVolDefPtr vol,
                               virStorageVolDefPtr inputvol,
                               unsigned int flags)
{
    int ret = -1;
    const char *create_tool;
    int imgformat;
    virCommandPtr cmd;

    virCheckFlags(VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA, -1);

    /* KVM is usually ahead of qemu on features, so try that first */
    create_tool = virFindFileInPath("kvm-img");
    if (!create_tool)
        create_tool = virFindFileInPath("qemu-img");

    if (!create_tool) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("unable to find kvm-img or qemu-img"));
        return -1;
    }

    imgformat = virStorageBackendQEMUImgBackingFormat(create_tool);
    if (imgformat < 0)
        goto cleanup;

    cmd = virStorageBackendCreateQemuImgCmd(conn, pool, vol, inputvol, flags,
                                            create_tool, imgformat);
    if (!cmd)
        goto cleanup;

878
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
879 880

    virCommandFree(cmd);
881
cleanup:
882
    VIR_FREE(create_tool);
883
    return ret;
884 885 886 887 888 889 890
}

/*
 * Xen removed the fully-functional qemu-img, and replaced it
 * with a partially functional qcow-create. Go figure ??!?
 */
static int
891
virStorageBackendCreateQcowCreate(virConnectPtr conn ATTRIBUTE_UNUSED,
892
                                  virStoragePoolObjPtr pool,
893
                                  virStorageVolDefPtr vol,
894
                                  virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
895
                                  unsigned int flags)
896
{
897
    int ret;
E
Eric Blake 已提交
898
    char *size;
899
    virCommandPtr cmd;
900

E
Eric Blake 已提交
901 902
    virCheckFlags(0, -1);

903
    if (inputvol) {
904 905
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("cannot copy from volume with qcow-create"));
906 907 908
        return -1;
    }

909
    if (vol->target.format != VIR_STORAGE_FILE_QCOW2) {
910 911 912
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unsupported storage vol type %d"),
                       vol->target.format);
913 914 915
        return -1;
    }
    if (vol->backingStore.path != NULL) {
916 917 918
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("copy-on-write image not supported with "
                         "qcow-create"));
919 920
        return -1;
    }
921
    if (vol->target.encryption != NULL) {
922 923 924
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       "%s", _("encrypted volumes not supported with "
                               "qcow-create"));
925 926
        return -1;
    }
927 928

    /* Size in MB - yes different units to qemu-img :-( */
929 930
    if (virAsprintf(&size, "%llu",
                    VIR_DIV_UP(vol->capacity, (1024 * 1024))) < 0) {
E
Eric Blake 已提交
931 932 933
        virReportOOMError();
        return -1;
    }
934

935
    cmd = virCommandNewArgList("qcow-create", size, vol->target.path, NULL);
936

937 938
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
    virCommandFree(cmd);
E
Eric Blake 已提交
939
    VIR_FREE(size);
940

941
    return ret;
942 943
}

944
virStorageBackendBuildVolFrom
945
virStorageBackendFSImageToolTypeToFunc(int tool_type)
946 947 948 949 950 951 952 953
{
    switch (tool_type) {
    case TOOL_KVM_IMG:
    case TOOL_QEMU_IMG:
        return virStorageBackendCreateQemuImg;
    case TOOL_QCOW_CREATE:
        return virStorageBackendCreateQcowCreate;
    default:
954 955 956
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Unknown file create tool type '%d'."),
                       tool_type);
957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983
    }

    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;
}

984
virStorageBackendBuildVolFrom
985
virStorageBackendGetBuildVolFromFunction(virStorageVolDefPtr vol,
986 987 988 989 990 991 992 993 994 995 996
                                         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 &&
997
         vol->target.format != VIR_STORAGE_FILE_RAW) ||
998
        (inputvol->type == VIR_STORAGE_VOL_FILE &&
999
         inputvol->target.format != VIR_STORAGE_FILE_RAW)) {
1000

D
Daniel P. Berrange 已提交
1001
        if ((tool_type = virStorageBackendFindFSImageTool(NULL)) < 0) {
1002 1003 1004
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("creation of non-raw file images is "
                                   "not supported without qemu-img."));
1005 1006 1007
            return NULL;
        }

1008
        return virStorageBackendFSImageToolTypeToFunc(tool_type);
1009 1010
    }

1011 1012 1013 1014
    if (vol->type == VIR_STORAGE_VOL_BLOCK)
        return virStorageBackendCreateBlockFrom;
    else
        return virStorageBackendCreateRaw;
1015
}
1016

1017

1018 1019
virStorageBackendPtr
virStorageBackendForType(int type) {
1020
    unsigned int i;
1021
    for (i = 0; backends[i]; i++)
1022 1023 1024
        if (backends[i]->type == type)
            return backends[i];

1025 1026
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("missing backend for pool type %d"), type);
1027 1028 1029 1030
    return NULL;
}


1031 1032 1033
/*
 * Allows caller to silently ignore files with improper mode
 *
1034 1035
 * Returns -1 on error, -2 if file mode is unexpected or the
 * volume is a dangling symbolic link.
1036
 */
1037
int
1038
virStorageBackendVolOpenCheckMode(const char *path, unsigned int flags)
1039
{
1040 1041
    int fd, mode = 0;
    struct stat sb;
1042
    char *base = last_component(path);
1043

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

1059
    if ((fd = open(path, O_RDONLY|O_NONBLOCK|O_NOCTTY)) < 0) {
1060
        if ((errno == ENOENT || errno == ELOOP) &&
1061
            S_ISLNK(sb.st_mode)) {
1062 1063 1064 1065
            VIR_WARN("ignoring dangling symlink '%s'", path);
            return -2;
        }

1066
        virReportSystemError(errno,
1067
                             _("cannot open volume '%s'"),
1068
                             path);
1069 1070 1071
        return -1;
    }

1072 1073 1074 1075 1076 1077 1078 1079
    if (fstat(fd, &sb) < 0) {
        virReportSystemError(errno,
                             _("cannot stat file '%s'"),
                             path);
        VIR_FORCE_CLOSE(fd);
        return -1;
    }

1080 1081 1082 1083 1084 1085
    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;
1086 1087 1088 1089 1090 1091 1092 1093 1094 1095
    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;
        }
    }
1096 1097

    if (!(mode & flags)) {
1098
        VIR_FORCE_CLOSE(fd);
1099
        VIR_INFO("Skipping volume '%s'", path);
1100 1101

        if (mode & VIR_STORAGE_VOL_OPEN_ERROR) {
1102 1103
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unexpected storage mode for '%s'"), path);
1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121
            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,
1122 1123
                                     unsigned long long *capacity,
                                     unsigned int openflags)
1124 1125 1126
{
    int ret, fd;

1127 1128
    if ((ret = virStorageBackendVolOpenCheckMode(target->path,
                                                 openflags)) < 0)
1129 1130 1131
        return ret;

    fd = ret;
1132
    ret = virStorageBackendUpdateVolTargetInfoFD(target,
1133 1134 1135
                                                 fd,
                                                 allocation,
                                                 capacity);
1136

1137
    VIR_FORCE_CLOSE(fd);
1138 1139 1140 1141

    return ret;
}

1142
int
1143 1144 1145
virStorageBackendUpdateVolInfoFlags(virStorageVolDefPtr vol,
                                    int withCapacity,
                                    unsigned int openflags)
1146 1147 1148
{
    int ret;

1149
    if ((ret = virStorageBackendUpdateVolTargetInfo(&vol->target,
1150 1151 1152
                                    &vol->allocation,
                                    withCapacity ? &vol->capacity : NULL,
                                    openflags)) < 0)
1153 1154 1155
        return ret;

    if (vol->backingStore.path &&
1156
        (ret = virStorageBackendUpdateVolTargetInfo(&vol->backingStore,
1157 1158
                                            NULL, NULL,
                                            VIR_STORAGE_VOL_OPEN_DEFAULT)) < 0)
1159 1160 1161 1162 1163
        return ret;

    return 0;
}

1164 1165 1166 1167 1168 1169 1170
int virStorageBackendUpdateVolInfo(virStorageVolDefPtr vol,
                                   int withCapacity)
{
    return virStorageBackendUpdateVolInfoFlags(vol, withCapacity,
                                               VIR_STORAGE_VOL_OPEN_DEFAULT);
}

1171 1172 1173 1174
/*
 * virStorageBackendUpdateVolTargetInfoFD:
 * @conn: connection to report errors on
 * @target: target definition ptr of volume to update
1175
 * @fd: fd of storage volume to update, via virStorageBackendOpenVol*
1176 1177 1178
 * @allocation: If not NULL, updated allocation information will be stored
 * @capacity: If not NULL, updated capacity info will be stored
 *
1179
 * Returns 0 for success, -1 on a legitimate error condition.
1180
 */
1181
int
1182
virStorageBackendUpdateVolTargetInfoFD(virStorageVolTargetPtr target,
1183 1184 1185
                                       int fd,
                                       unsigned long long *allocation,
                                       unsigned long long *capacity)
1186 1187
{
    struct stat sb;
1188
#if WITH_SELINUX
1189 1190 1191 1192
    security_context_t filecon = NULL;
#endif

    if (fstat(fd, &sb) < 0) {
1193
        virReportSystemError(errno,
1194
                             _("cannot stat file '%s'"),
1195
                             target->path);
1196 1197 1198
        return -1;
    }

1199 1200
    if (allocation) {
        if (S_ISREG(sb.st_mode)) {
1201
#ifndef WIN32
1202
            *allocation = (unsigned long long)sb.st_blocks *
1203
                          (unsigned long long)DEV_BSIZE;
D
Daniel P. Berrange 已提交
1204
#else
1205
            *allocation = sb.st_size;
D
Daniel P. Berrange 已提交
1206
#endif
1207 1208 1209 1210 1211
            /* Regular files may be sparse, so logical size (capacity) is not same
             * as actual allocation above
             */
            if (capacity)
                *capacity = sb.st_size;
1212 1213 1214 1215 1216
        } else if (S_ISDIR(sb.st_mode)) {
            *allocation = 0;
            if (capacity)
                *capacity = 0;

1217 1218
        } else {
            off_t end;
E
Eric Blake 已提交
1219
            /* XXX this is POSIX compliant, but doesn't work for CHAR files,
1220 1221 1222 1223 1224 1225
             * 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) {
1226
                virReportSystemError(errno,
1227 1228 1229 1230 1231 1232 1233
                                     _("cannot seek to end of file '%s'"),
                                     target->path);
                return -1;
            }
            *allocation = end;
            if (capacity)
                *capacity = end;
1234 1235 1236
        }
    }

1237 1238 1239
    target->perms.mode = sb.st_mode & S_IRWXUGO;
    target->perms.uid = sb.st_uid;
    target->perms.gid = sb.st_gid;
1240

1241 1242 1243 1244 1245 1246 1247 1248 1249
    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);

1250
    VIR_FREE(target->perms.label);
1251

1252
#if WITH_SELINUX
1253
    /* XXX: make this a security driver call */
M
Martin Kletzander 已提交
1254
    if (fgetfilecon_raw(fd, &filecon) == -1) {
1255
        if (errno != ENODATA && errno != ENOTSUP) {
1256
            virReportSystemError(errno,
1257
                                 _("cannot get file context of '%s'"),
1258
                                 target->path);
1259 1260
            return -1;
        } else {
1261
            target->perms.label = NULL;
1262 1263
        }
    } else {
1264
        target->perms.label = strdup(filecon);
1265
        freecon(filecon);
1266
        if (target->perms.label == NULL) {
1267
            virReportOOMError();
1268 1269
            return -1;
        }
1270 1271
    }
#else
1272
    target->perms.label = NULL;
1273 1274 1275 1276 1277
#endif

    return 0;
}

D
Dave Allan 已提交
1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312

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
1313 1314
virStorageBackendDetectBlockVolFormatFD(virStorageVolTargetPtr target,
                                        int fd)
D
Dave Allan 已提交
1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325
{
    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) {
1326
        virReportSystemError(errno,
D
Dave Allan 已提交
1327 1328 1329 1330 1331 1332
                             _("cannot seek to beginning of file '%s'"),
                             target->path);
        return -1;
    }
    bytes = saferead(fd, buffer, sizeof(buffer));
    if (bytes < 0) {
1333
        virReportSystemError(errno,
D
Dave Allan 已提交
1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352
                             _("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;
}


1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364
/*
 * 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.
1365
 *
1366
 * If 'loop' is true, we use a timeout loop to give dynamic paths
1367
 * a change to appear.
1368 1369
 */
char *
1370
virStorageBackendStablePath(virStoragePoolObjPtr pool,
1371
                            const char *devpath,
1372
                            bool loop)
1373 1374 1375
{
    DIR *dh;
    struct dirent *dent;
1376
    char *stablepath;
1377
    int opentries = 0;
1378
    int retry = 0;
1379 1380 1381 1382 1383

    /* 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/"))
1384
        goto ret_strdup;
1385

1386
    /* Skip whole thing for a pool which isn't in /dev
1387
     * so we don't mess filesystem/dir based pools
1388 1389 1390 1391
     */
    if (!STRPREFIX(pool->def->target.path, "/dev"))
        goto ret_strdup;

1392 1393 1394 1395
    /* Logical pools are under /dev but already have stable paths */
    if (pool->def->type == VIR_STORAGE_POOL_LOGICAL)
        goto ret_strdup;

1396 1397 1398
    /* 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.
1399
     */
1400
 reopen:
1401
    if ((dh = opendir(pool->def->target.path)) == NULL) {
1402
        opentries++;
1403
        if (loop && errno == ENOENT && opentries < 50) {
1404 1405 1406
            usleep(100 * 1000);
            goto reopen;
        }
1407
        virReportSystemError(errno,
1408 1409
                             _("cannot read dir '%s'"),
                             pool->def->target.path);
1410 1411 1412
        return NULL;
    }

1413 1414 1415
    /* 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
1416 1417
     * to this device node.
     *
1418
     * And it might need some time till the stable path shows
1419
     * up, so add timeout to retry here.
1420
     */
1421
 retry:
1422 1423 1424 1425
    while ((dent = readdir(dh)) != NULL) {
        if (dent->d_name[0] == '.')
            continue;

1426 1427 1428
        if (virAsprintf(&stablepath, "%s/%s",
                        pool->def->target.path,
                        dent->d_name) == -1) {
1429
            virReportOOMError();
1430 1431 1432 1433 1434 1435 1436 1437 1438
            closedir(dh);
            return NULL;
        }

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

1439
        VIR_FREE(stablepath);
1440 1441
    }

1442
    if (loop && ++retry < 100) {
1443 1444 1445 1446
        usleep(100 * 1000);
        goto retry;
    }

1447 1448
    closedir(dh);

1449
 ret_strdup:
1450 1451 1452
    /* Couldn't find any matching stable link so give back
     * the original non-stable dev path
     */
1453 1454 1455 1456

    stablepath = strdup(devpath);

    if (stablepath == NULL)
1457
        virReportOOMError();
1458 1459

    return stablepath;
1460 1461
}

D
Daniel P. Berrange 已提交
1462

1463
#ifndef WIN32
1464 1465 1466 1467
/*
 * Run an external program.
 *
 * Read its output and apply a series of regexes to each line
R
Richard W.M. Jones 已提交
1468
 * When the entire set of regexes has matched consecutively
1469 1470 1471
 * then run a callback passing in all the matches
 */
int
1472
virStorageBackendRunProgRegex(virStoragePoolObjPtr pool,
1473
                              virCommandPtr cmd,
1474 1475 1476 1477
                              int nregex,
                              const char **regex,
                              int *nvars,
                              virStorageBackendListVolRegexFunc func,
1478
                              void *data, const char *prefix)
1479
{
1480
    int fd = -1, err, ret = -1;
1481 1482 1483 1484 1485 1486 1487 1488 1489
    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 */
1490
    if (VIR_ALLOC_N(reg, nregex) < 0) {
1491
        virReportOOMError();
1492 1493 1494 1495 1496 1497 1498 1499
        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));
1500 1501
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Failed to compile regex %s"), error);
1502 1503
            for (j = 0 ; j <= i ; j++)
                regfree(&reg[j]);
1504
            VIR_FREE(reg);
1505 1506 1507 1508 1509 1510 1511 1512 1513 1514
            return -1;
        }

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

    }

    /* Storage for matched variables */
1515
    if (VIR_ALLOC_N(groups, totgroups) < 0) {
1516
        virReportOOMError();
1517 1518
        goto cleanup;
    }
1519
    if (VIR_ALLOC_N(vars, maxvars+1) < 0) {
1520
        virReportOOMError();
1521 1522 1523
        goto cleanup;
    }

1524 1525
    virCommandSetOutputFD(cmd, &fd);
    if (virCommandRunAsync(cmd, NULL) < 0) {
1526 1527 1528
        goto cleanup;
    }

1529
    if ((list = VIR_FDOPEN(fd, "r")) == NULL) {
1530 1531
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("cannot read fd"));
1532 1533 1534 1535
        goto cleanup;
    }

    while (fgets(line, sizeof(line), list) != NULL) {
1536
        char *p = NULL;
1537 1538 1539 1540 1541
        /* Strip trailing newline */
        int len = strlen(line);
        if (len && line[len-1] == '\n')
            line[len-1] = '\0';

1542 1543 1544 1545 1546 1547
        /* ignore any command prefix */
        if (prefix)
            p = STRSKIP(line, prefix);
        if (!p)
            p = line;

1548
        for (i = 0 ; i <= maxReg && i < nregex ; i++) {
1549
            if (regexec(&reg[i], p, nvars[i]+1, vars, 0) == 0) {
1550 1551 1552 1553 1554 1555 1556 1557
                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 */
1558
                    p[vars[j+1].rm_eo] = '\0';
1559
                    if ((groups[ngroup++] =
1560
                         strdup(p + vars[j+1].rm_so)) == NULL) {
1561
                        virReportOOMError();
1562 1563 1564 1565 1566 1567
                        goto cleanup;
                    }
                }

                /* We're matching on the last regex, so callback time */
                if (i == (nregex-1)) {
1568
                    if (((*func)(pool, groups, data)) < 0)
1569 1570 1571
                        goto cleanup;

                    /* Release matches & restart to matching the first regex */
1572
                    for (j = 0 ; j < totgroups ; j++)
1573
                        VIR_FREE(groups[j]);
1574 1575 1576 1577 1578 1579 1580
                    maxReg = 0;
                    ngroup = 0;
                }
            }
        }
    }

1581
    ret = virCommandWait(cmd, NULL);
1582
cleanup:
1583 1584
    if (groups) {
        for (j = 0 ; j < totgroups ; j++)
1585 1586
            VIR_FREE(groups[j]);
        VIR_FREE(groups);
1587
    }
1588
    VIR_FREE(vars);
1589 1590 1591 1592

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

1593
    VIR_FREE(reg);
1594

1595 1596
    VIR_FORCE_FCLOSE(list);
    VIR_FORCE_CLOSE(fd);
1597

1598
    return ret;
1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612
}

/*
 * 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
1613
virStorageBackendRunProgNul(virStoragePoolObjPtr pool,
1614
                            virCommandPtr cmd,
1615 1616 1617 1618 1619
                            size_t n_columns,
                            virStorageBackendListVolNulFunc func,
                            void *data)
{
    size_t n_tok = 0;
1620
    int fd = -1;
1621 1622
    FILE *fp = NULL;
    char **v;
1623
    int ret = -1;
1624 1625 1626 1627 1628
    int i;

    if (n_columns == 0)
        return -1;

1629
    if (VIR_ALLOC_N(v, n_columns) < 0) {
1630
        virReportOOMError();
1631 1632 1633 1634 1635
        return -1;
    }
    for (i = 0; i < n_columns; i++)
        v[i] = NULL;

1636 1637
    virCommandSetOutputFD(cmd, &fd);
    if (virCommandRunAsync(cmd, NULL) < 0) {
1638 1639 1640
        goto cleanup;
    }

1641
    if ((fp = VIR_FDOPEN(fd, "r")) == NULL) {
1642 1643
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("cannot open file using fd"));
1644 1645 1646 1647 1648 1649 1650 1651
        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.  */
1652
        ssize_t tok_len = getdelim(&buf, &buf_len, 0, fp);
1653 1654 1655 1656
        v[n_tok] = buf;
        if (tok_len < 0) {
            /* Maybe EOF, maybe an error.
               If n_tok > 0, then we know it's an error.  */
1657
            if (n_tok && func(pool, n_tok, v, data) < 0)
1658 1659 1660 1661 1662
                goto cleanup;
            break;
        }
        ++n_tok;
        if (n_tok == n_columns) {
1663
            if (func(pool, n_tok, v, data) < 0)
1664 1665 1666
                goto cleanup;
            n_tok = 0;
            for (i = 0; i < n_columns; i++) {
1667
                VIR_FREE(v[i]);
1668 1669 1670 1671
            }
        }
    }

1672 1673 1674
    if (feof(fp) < 0) {
        virReportSystemError(errno, "%s",
                             _("read error on pipe"));
1675 1676
        goto cleanup;
    }
1677

1678
    ret = virCommandWait(cmd, NULL);
1679 1680
 cleanup:
    for (i = 0; i < n_columns; i++)
1681 1682
        VIR_FREE(v[i]);
    VIR_FREE(v);
1683

1684 1685
    VIR_FORCE_FCLOSE(fp);
    VIR_FORCE_CLOSE(fd);
1686

1687
    return ret;
1688
}
D
Daniel P. Berrange 已提交
1689

1690
#else /* WIN32 */
D
Daniel P. Berrange 已提交
1691 1692 1693 1694 1695 1696 1697 1698 1699

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,
1700
                              void *data ATTRIBUTE_UNUSED)
D
Daniel P. Berrange 已提交
1701
{
1702 1703
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714
    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)
{
1715 1716
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1717 1718
    return -1;
}
1719
#endif /* WIN32 */