storage_backend.c 51.3 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 "viralloc.h"
49
#include "internal.h"
50
#include "secret_conf.h"
51
#include "viruuid.h"
52
#include "virstoragefile.h"
D
Daniel P. Berrange 已提交
53
#include "storage_backend.h"
54
#include "virlog.h"
E
Eric Blake 已提交
55
#include "virfile.h"
56
#include "stat-time.h"
57
#include "virstring.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
        virReportSystemError(-fd,
408
                             _("Failed to create file '%s'"),
E
Eric Blake 已提交
409 410 411 412
                             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
    def->usage_type = VIR_SECRET_USAGE_TYPE_VOLUME;
489
    if (VIR_STRDUP(def->usage.volume, vol->target.path) < 0)
490
        goto cleanup;
491
    xml = virSecretDefFormat(def);
492 493 494 495 496
    virSecretDefFree(def);
    def = NULL;
    if (xml == NULL)
        goto cleanup;

497
    secret = conn->secretDriver->secretDefineXML(conn, xml, 0);
498 499 500 501 502 503
    if (secret == NULL) {
        VIR_FREE(xml);
        goto cleanup;
    }
    VIR_FREE(xml);

504
    if (virStorageGenerateQcowPassphrase(value) < 0)
505 506
        goto cleanup;

507
    if (conn->secretDriver->secretSetValue(secret, value, sizeof(value), 0) < 0)
508 509 510
        goto cleanup;

    enc_secret->type = VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE;
511
    memcpy(enc_secret->uuid, secret->uuid, VIR_UUID_BUFLEN);
512 513 514 515 516 517 518 519 520
    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) {
521
        if (ret != 0 &&
522 523
            conn->secretDriver->secretUndefine != NULL)
            conn->secretDriver->secretUndefine(secret);
524 525
        virSecretFree(secret);
    }
526
    virBufferFreeAndReset(&buf);
527 528 529 530 531
    virSecretDefFree(def);
    VIR_FREE(enc_secret);
    return ret;
}

532
static int virStorageBackendCreateExecCommand(virStoragePoolObjPtr pool,
533
                                              virStorageVolDefPtr vol,
534
                                              virCommandPtr cmd) {
535 536 537 538 539 540
    struct stat st;
    gid_t gid;
    uid_t uid;
    int filecreated = 0;

    if ((pool->def->type == VIR_STORAGE_POOL_NETFS)
541
        && (((getuid() == 0)
P
Philipp Hahn 已提交
542
             && (vol->target.perms.uid != (uid_t) -1)
543
             && (vol->target.perms.uid != 0))
P
Philipp Hahn 已提交
544
            || ((vol->target.perms.gid != (gid_t) -1)
545
                && (vol->target.perms.gid != getgid())))) {
546

547 548
        virCommandSetUID(cmd, vol->target.perms.uid);
        virCommandSetGID(cmd, vol->target.perms.gid);
549 550

        if (virCommandRun(cmd, NULL) == 0) {
551 552 553 554 555
            /* command was successfully run, check if the file was created */
            if (stat(vol->target.path, &st) >=0)
                filecreated = 1;
        }
    }
556

557 558 559
    /* don't change uid/gid if we retry */
    virCommandSetUID(cmd, -1);
    virCommandSetGID(cmd, -1);
560

561
    if (!filecreated) {
562
        if (virCommandRun(cmd, NULL) < 0) {
563 564 565
            return -1;
        }
        if (stat(vol->target.path, &st) < 0) {
566
            virReportSystemError(errno,
567
                                 _("failed to create %s"), vol->target.path);
568 569 570 571
            return -1;
        }
    }

P
Philipp Hahn 已提交
572 573 574
    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))
575
        && (chown(vol->target.path, uid, gid) < 0)) {
576
        virReportSystemError(errno,
577
                             _("cannot chown %s to (%u, %u)"),
578 579
                             vol->target.path, (unsigned int) uid,
                             (unsigned int) gid);
580 581 582
        return -1;
    }
    if (chmod(vol->target.path, vol->target.perms.mode) < 0) {
583
        virReportSystemError(errno,
584 585 586 587 588 589 590
                             _("cannot set mode of '%s' to %04o"),
                             vol->target.path, vol->target.perms.mode);
        return -1;
    }
    return 0;
}

591 592 593 594 595 596 597 598 599 600 601 602 603
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;
604
    int exitstatus;
605
    virCommandPtr cmd = virCommandNewArgList(qemuimg, "-h", NULL);
606

607 608 609
    virCommandAddEnvString(cmd, "LC_ALL=C");
    virCommandSetOutputBuffer(cmd, &help);
    virCommandClearCaps(cmd);
610

611 612 613
    /* 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 */
614
    if (virCommandRun(cmd, &exitstatus) < 0)
615 616
        goto cleanup;

617 618
    if ((start = strstr(help, " create ")) == NULL ||
        (end = strstr(start, "\n")) == NULL) {
619 620 621
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unable to parse qemu-img output '%s'"),
                       help);
622 623
        goto cleanup;
    }
624 625
    if (((tmp = strstr(start, "-F fmt")) && tmp < end) ||
        ((tmp = strstr(start, "-F backing_fmt")) && tmp < end))
626 627 628 629 630 631 632
        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:
633
    virCommandFree(cmd);
634 635 636 637
    VIR_FREE(help);
    return ret;
}

638 639 640 641 642 643 644 645
virCommandPtr
virStorageBackendCreateQemuImgCmd(virConnectPtr conn,
                                  virStoragePoolObjPtr pool,
                                  virStorageVolDefPtr vol,
                                  virStorageVolDefPtr inputvol,
                                  unsigned int flags,
                                  const char *create_tool,
                                  int imgformat)
646
{
647 648
    virCommandPtr cmd = NULL;
    bool do_encryption = (vol->target.encryption != NULL);
649
    unsigned long long int size_arg;
650
    bool preallocate = false;
651

652 653 654 655 656 657
    /* Treat output block devices as 'raw' format */
    const char *type =
        virStorageFileFormatTypeToString(vol->type == VIR_STORAGE_VOL_BLOCK ?
                                         VIR_STORAGE_FILE_RAW :
                                         vol->target.format);

658
    const char *backingType = vol->backingStore.path ?
659
        virStorageFileFormatTypeToString(vol->backingStore.format) : NULL;
660 661 662 663 664 665

    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 ?
666 667 668 669
        virStorageFileFormatTypeToString(inputvol->type == VIR_STORAGE_VOL_BLOCK ?
                                         VIR_STORAGE_FILE_RAW :
                                         inputvol->target.format) :
        NULL;
670

671 672 673 674
    virCheckFlags(VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA, NULL);

    preallocate = !!(flags & VIR_STORAGE_VOL_CREATE_PREALLOC_METADATA);

675
    if (type == NULL) {
676 677 678
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unknown storage vol type %d"),
                       vol->target.format);
679
        return NULL;
680 681
    }
    if (inputvol && inputType == NULL) {
682 683 684
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unknown storage vol type %d"),
                       inputvol->target.format);
685
        return NULL;
686
    }
687 688 689
    if (preallocate && vol->target.format != VIR_STORAGE_FILE_QCOW2) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("metadata preallocation only available with qcow2"));
690
        return NULL;
691
    }
692 693

    if (vol->backingStore.path) {
694 695
        int accessRetCode = -1;
        char *absolutePath = NULL;
696

697 698 699 700
        if (preallocate) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("metadata preallocation conflicts with backing"
                             " store"));
701
            return NULL;
702 703
        }

704 705 706 707 708 709 710
        /* 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))) {
711 712 713
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("a different backing store cannot "
                                   "be specified."));
714
            return NULL;
715 716 717
        }

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

        /* 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();
731
            return NULL;
732 733 734 735 736
        }
        accessRetCode = access(absolutePath ? absolutePath
                               : vol->backingStore.path, R_OK);
        VIR_FREE(absolutePath);
        if (accessRetCode != 0) {
737
            virReportSystemError(errno,
738 739
                                 _("inaccessible backing store volume %s"),
                                 vol->backingStore.path);
740
            return NULL;
741 742 743
        }
    }

744
    if (do_encryption) {
745 746
        virStorageEncryptionPtr enc;

747 748
        if (vol->target.format != VIR_STORAGE_FILE_QCOW &&
            vol->target.format != VIR_STORAGE_FILE_QCOW2) {
749 750 751
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("qcow volume encryption unsupported with "
                             "volume format %s"), type);
752
            return NULL;
753
        }
754 755 756
        enc = vol->target.encryption;
        if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW &&
            enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT) {
757 758 759
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("unsupported volume encryption format %d"),
                           vol->target.encryption->format);
760
            return NULL;
761
        }
762
        if (enc->nsecrets > 1) {
763 764
            virReportError(VIR_ERR_XML_ERROR, "%s",
                           _("too many secrets for qcow encryption"));
765
            return NULL;
766
        }
767 768 769
        if (enc->format == VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT ||
            enc->nsecrets == 0) {
            if (virStorageGenerateQcowEncryption(conn, vol) < 0)
770
                return NULL;
771
        }
772 773
    }

774
    /* Size in KB */
775
    size_arg = VIR_DIV_UP(vol->capacity, 1024);
776

777 778
    cmd = virCommandNew(create_tool);

779
    if (inputvol) {
780
        virCommandAddArgList(cmd, "convert", "-f", inputType, "-O", type, NULL);
781

782 783 784 785 786 787 788 789
        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");
790
        }
791
        virCommandAddArgList(cmd, inputPath, vol->target.path, NULL);
792
    } else if (vol->backingStore.path) {
793 794 795
        virCommandAddArgList(cmd, "create", "-f", type,
                             "-b", vol->backingStore.path, NULL);

796 797
        switch (imgformat) {
        case QEMU_IMG_BACKING_FORMAT_FLAG:
798
            virCommandAddArgList(cmd, "-F", backingType, NULL);
799 800
            if (do_encryption)
                virCommandAddArg(cmd, "-e");
801 802
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
803 804 805
            break;

        case QEMU_IMG_BACKING_FORMAT_OPTIONS:
806 807 808
            virCommandAddArg(cmd, "-o");
            virCommandAddArgFormat(cmd, "backing_fmt=%s%s", backingType,
                                   do_encryption ? ",encryption=on" : "");
809 810
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
811
            break;
812 813

        default:
814 815
            VIR_DEBUG("Unable to set backing store format for %s with %s",
                      vol->target.path, create_tool);
816

817 818
            if (do_encryption)
                virCommandAddArg(cmd, "-e");
819 820
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
821
        }
822
    } else {
823
        virCommandAddArgList(cmd, "create", "-f", type, NULL);
824

825 826 827 828 829 830 831 832
        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");
833
        }
834 835
        virCommandAddArg(cmd, vol->target.path);
        virCommandAddArgFormat(cmd, "%lluK", size_arg);
836
    }
837

838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874
    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;

875
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
876 877

    virCommandFree(cmd);
878
cleanup:
879
    VIR_FREE(create_tool);
880
    return ret;
881 882 883 884 885 886 887
}

/*
 * Xen removed the fully-functional qemu-img, and replaced it
 * with a partially functional qcow-create. Go figure ??!?
 */
static int
888
virStorageBackendCreateQcowCreate(virConnectPtr conn ATTRIBUTE_UNUSED,
889
                                  virStoragePoolObjPtr pool,
890
                                  virStorageVolDefPtr vol,
891
                                  virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
892
                                  unsigned int flags)
893
{
894
    int ret;
E
Eric Blake 已提交
895
    char *size;
896
    virCommandPtr cmd;
897

E
Eric Blake 已提交
898 899
    virCheckFlags(0, -1);

900
    if (inputvol) {
901 902
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("cannot copy from volume with qcow-create"));
903 904 905
        return -1;
    }

906
    if (vol->target.format != VIR_STORAGE_FILE_QCOW2) {
907 908 909
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unsupported storage vol type %d"),
                       vol->target.format);
910 911 912
        return -1;
    }
    if (vol->backingStore.path != NULL) {
913 914 915
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("copy-on-write image not supported with "
                         "qcow-create"));
916 917
        return -1;
    }
918
    if (vol->target.encryption != NULL) {
919 920 921
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       "%s", _("encrypted volumes not supported with "
                               "qcow-create"));
922 923
        return -1;
    }
924 925

    /* Size in MB - yes different units to qemu-img :-( */
926 927
    if (virAsprintf(&size, "%llu",
                    VIR_DIV_UP(vol->capacity, (1024 * 1024))) < 0) {
E
Eric Blake 已提交
928 929 930
        virReportOOMError();
        return -1;
    }
931

932
    cmd = virCommandNewArgList("qcow-create", size, vol->target.path, NULL);
933

934 935
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
    virCommandFree(cmd);
E
Eric Blake 已提交
936
    VIR_FREE(size);
937

938
    return ret;
939 940
}

941
virStorageBackendBuildVolFrom
942
virStorageBackendFSImageToolTypeToFunc(int tool_type)
943 944 945 946 947 948 949 950
{
    switch (tool_type) {
    case TOOL_KVM_IMG:
    case TOOL_QEMU_IMG:
        return virStorageBackendCreateQemuImg;
    case TOOL_QCOW_CREATE:
        return virStorageBackendCreateQcowCreate;
    default:
951 952 953
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Unknown file create tool type '%d'."),
                       tool_type);
954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980
    }

    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;
}

981
virStorageBackendBuildVolFrom
982
virStorageBackendGetBuildVolFromFunction(virStorageVolDefPtr vol,
983 984 985 986 987 988 989 990 991 992 993
                                         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 &&
994
         vol->target.format != VIR_STORAGE_FILE_RAW) ||
995
        (inputvol->type == VIR_STORAGE_VOL_FILE &&
996
         inputvol->target.format != VIR_STORAGE_FILE_RAW)) {
997

D
Daniel P. Berrange 已提交
998
        if ((tool_type = virStorageBackendFindFSImageTool(NULL)) < 0) {
999 1000 1001
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("creation of non-raw file images is "
                                   "not supported without qemu-img."));
1002 1003 1004
            return NULL;
        }

1005
        return virStorageBackendFSImageToolTypeToFunc(tool_type);
1006 1007
    }

1008 1009 1010 1011
    if (vol->type == VIR_STORAGE_VOL_BLOCK)
        return virStorageBackendCreateBlockFrom;
    else
        return virStorageBackendCreateRaw;
1012
}
1013

1014

1015 1016
virStorageBackendPtr
virStorageBackendForType(int type) {
1017
    unsigned int i;
1018
    for (i = 0; backends[i]; i++)
1019 1020 1021
        if (backends[i]->type == type)
            return backends[i];

1022 1023
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("missing backend for pool type %d"), type);
1024 1025 1026 1027
    return NULL;
}


1028 1029 1030
/*
 * Allows caller to silently ignore files with improper mode
 *
1031 1032
 * Returns -1 on error, -2 if file mode is unexpected or the
 * volume is a dangling symbolic link.
1033
 */
1034
int
1035
virStorageBackendVolOpenCheckMode(const char *path, unsigned int flags)
1036
{
1037 1038
    int fd, mode = 0;
    struct stat sb;
1039
    char *base = last_component(path);
1040

1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055
    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;
    }

1056
    if ((fd = open(path, O_RDONLY|O_NONBLOCK|O_NOCTTY)) < 0) {
1057
        if ((errno == ENOENT || errno == ELOOP) &&
1058
            S_ISLNK(sb.st_mode)) {
1059 1060 1061 1062
            VIR_WARN("ignoring dangling symlink '%s'", path);
            return -2;
        }

1063
        virReportSystemError(errno,
1064
                             _("cannot open volume '%s'"),
1065
                             path);
1066 1067 1068
        return -1;
    }

1069 1070 1071 1072 1073 1074 1075 1076
    if (fstat(fd, &sb) < 0) {
        virReportSystemError(errno,
                             _("cannot stat file '%s'"),
                             path);
        VIR_FORCE_CLOSE(fd);
        return -1;
    }

1077 1078 1079 1080 1081 1082
    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;
1083 1084 1085 1086 1087 1088 1089 1090 1091 1092
    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;
        }
    }
1093 1094

    if (!(mode & flags)) {
1095
        VIR_FORCE_CLOSE(fd);
1096
        VIR_INFO("Skipping volume '%s'", path);
1097 1098

        if (mode & VIR_STORAGE_VOL_OPEN_ERROR) {
1099 1100
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unexpected storage mode for '%s'"), path);
1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118
            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,
1119 1120
                                     unsigned long long *capacity,
                                     unsigned int openflags)
1121 1122 1123
{
    int ret, fd;

1124 1125
    if ((ret = virStorageBackendVolOpenCheckMode(target->path,
                                                 openflags)) < 0)
1126 1127 1128
        return ret;

    fd = ret;
1129
    ret = virStorageBackendUpdateVolTargetInfoFD(target,
1130 1131 1132
                                                 fd,
                                                 allocation,
                                                 capacity);
1133

1134
    VIR_FORCE_CLOSE(fd);
1135 1136 1137 1138

    return ret;
}

1139
int
1140 1141 1142
virStorageBackendUpdateVolInfoFlags(virStorageVolDefPtr vol,
                                    int withCapacity,
                                    unsigned int openflags)
1143 1144 1145
{
    int ret;

1146
    if ((ret = virStorageBackendUpdateVolTargetInfo(&vol->target,
1147 1148 1149
                                    &vol->allocation,
                                    withCapacity ? &vol->capacity : NULL,
                                    openflags)) < 0)
1150 1151 1152
        return ret;

    if (vol->backingStore.path &&
1153
        (ret = virStorageBackendUpdateVolTargetInfo(&vol->backingStore,
1154 1155
                                            NULL, NULL,
                                            VIR_STORAGE_VOL_OPEN_DEFAULT)) < 0)
1156 1157 1158 1159 1160
        return ret;

    return 0;
}

1161 1162 1163 1164 1165 1166 1167
int virStorageBackendUpdateVolInfo(virStorageVolDefPtr vol,
                                   int withCapacity)
{
    return virStorageBackendUpdateVolInfoFlags(vol, withCapacity,
                                               VIR_STORAGE_VOL_OPEN_DEFAULT);
}

1168 1169 1170 1171
/*
 * virStorageBackendUpdateVolTargetInfoFD:
 * @conn: connection to report errors on
 * @target: target definition ptr of volume to update
1172
 * @fd: fd of storage volume to update, via virStorageBackendOpenVol*
1173 1174 1175
 * @allocation: If not NULL, updated allocation information will be stored
 * @capacity: If not NULL, updated capacity info will be stored
 *
1176
 * Returns 0 for success, -1 on a legitimate error condition.
1177
 */
1178
int
1179
virStorageBackendUpdateVolTargetInfoFD(virStorageVolTargetPtr target,
1180 1181 1182
                                       int fd,
                                       unsigned long long *allocation,
                                       unsigned long long *capacity)
1183 1184
{
    struct stat sb;
1185
#if WITH_SELINUX
1186 1187 1188 1189
    security_context_t filecon = NULL;
#endif

    if (fstat(fd, &sb) < 0) {
1190
        virReportSystemError(errno,
1191
                             _("cannot stat file '%s'"),
1192
                             target->path);
1193 1194 1195
        return -1;
    }

1196 1197
    if (allocation) {
        if (S_ISREG(sb.st_mode)) {
1198
#ifndef WIN32
1199
            *allocation = (unsigned long long)sb.st_blocks *
1200
                          (unsigned long long)DEV_BSIZE;
D
Daniel P. Berrange 已提交
1201
#else
1202
            *allocation = sb.st_size;
D
Daniel P. Berrange 已提交
1203
#endif
1204 1205 1206 1207 1208
            /* Regular files may be sparse, so logical size (capacity) is not same
             * as actual allocation above
             */
            if (capacity)
                *capacity = sb.st_size;
1209 1210 1211 1212 1213
        } else if (S_ISDIR(sb.st_mode)) {
            *allocation = 0;
            if (capacity)
                *capacity = 0;

1214 1215
        } else {
            off_t end;
E
Eric Blake 已提交
1216
            /* XXX this is POSIX compliant, but doesn't work for CHAR files,
1217 1218 1219 1220 1221 1222
             * 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) {
1223
                virReportSystemError(errno,
1224 1225 1226 1227 1228 1229 1230
                                     _("cannot seek to end of file '%s'"),
                                     target->path);
                return -1;
            }
            *allocation = end;
            if (capacity)
                *capacity = end;
1231 1232 1233
        }
    }

1234 1235 1236
    target->perms.mode = sb.st_mode & S_IRWXUGO;
    target->perms.uid = sb.st_uid;
    target->perms.gid = sb.st_gid;
1237

1238 1239 1240 1241 1242 1243 1244 1245 1246
    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);

1247
    VIR_FREE(target->perms.label);
1248

1249
#if WITH_SELINUX
1250
    /* XXX: make this a security driver call */
M
Martin Kletzander 已提交
1251
    if (fgetfilecon_raw(fd, &filecon) == -1) {
1252
        if (errno != ENODATA && errno != ENOTSUP) {
1253
            virReportSystemError(errno,
1254
                                 _("cannot get file context of '%s'"),
1255
                                 target->path);
1256 1257
            return -1;
        } else {
1258
            target->perms.label = NULL;
1259 1260
        }
    } else {
1261 1262
        if (VIR_STRDUP(target->perms.label, filecon) < 0) {
            freecon(filecon);
1263 1264
            return -1;
        }
1265
        freecon(filecon);
1266 1267
    }
#else
1268
    target->perms.label = NULL;
1269 1270 1271 1272 1273
#endif

    return 0;
}

D
Dave Allan 已提交
1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308

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
1309 1310
virStorageBackendDetectBlockVolFormatFD(virStorageVolTargetPtr target,
                                        int fd)
D
Dave Allan 已提交
1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321
{
    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) {
1322
        virReportSystemError(errno,
D
Dave Allan 已提交
1323 1324 1325 1326 1327 1328
                             _("cannot seek to beginning of file '%s'"),
                             target->path);
        return -1;
    }
    bytes = saferead(fd, buffer, sizeof(buffer));
    if (bytes < 0) {
1329
        virReportSystemError(errno,
D
Dave Allan 已提交
1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348
                             _("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;
}


1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360
/*
 * 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.
1361
 *
1362
 * If 'loop' is true, we use a timeout loop to give dynamic paths
1363
 * a change to appear.
1364 1365
 */
char *
1366
virStorageBackendStablePath(virStoragePoolObjPtr pool,
1367
                            const char *devpath,
1368
                            bool loop)
1369 1370 1371
{
    DIR *dh;
    struct dirent *dent;
1372
    char *stablepath;
1373
    int opentries = 0;
1374
    int retry = 0;
1375 1376 1377 1378 1379

    /* 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/"))
1380
        goto ret_strdup;
1381

1382
    /* Skip whole thing for a pool which isn't in /dev
1383
     * so we don't mess filesystem/dir based pools
1384 1385 1386 1387
     */
    if (!STRPREFIX(pool->def->target.path, "/dev"))
        goto ret_strdup;

1388 1389 1390 1391
    /* Logical pools are under /dev but already have stable paths */
    if (pool->def->type == VIR_STORAGE_POOL_LOGICAL)
        goto ret_strdup;

1392 1393 1394
    /* 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.
1395
     */
1396
 reopen:
1397
    if ((dh = opendir(pool->def->target.path)) == NULL) {
1398
        opentries++;
1399
        if (loop && errno == ENOENT && opentries < 50) {
1400 1401 1402
            usleep(100 * 1000);
            goto reopen;
        }
1403
        virReportSystemError(errno,
1404 1405
                             _("cannot read dir '%s'"),
                             pool->def->target.path);
1406 1407 1408
        return NULL;
    }

1409 1410 1411
    /* 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
1412 1413
     * to this device node.
     *
1414
     * And it might need some time till the stable path shows
1415
     * up, so add timeout to retry here.
1416
     */
1417
 retry:
1418 1419 1420 1421
    while ((dent = readdir(dh)) != NULL) {
        if (dent->d_name[0] == '.')
            continue;

1422 1423 1424
        if (virAsprintf(&stablepath, "%s/%s",
                        pool->def->target.path,
                        dent->d_name) == -1) {
1425
            virReportOOMError();
1426 1427 1428 1429 1430 1431 1432 1433 1434
            closedir(dh);
            return NULL;
        }

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

1435
        VIR_FREE(stablepath);
1436 1437
    }

1438
    if (loop && ++retry < 100) {
1439 1440 1441 1442
        usleep(100 * 1000);
        goto retry;
    }

1443 1444
    closedir(dh);

1445
 ret_strdup:
1446 1447 1448
    /* Couldn't find any matching stable link so give back
     * the original non-stable dev path
     */
1449

1450
    ignore_value(VIR_STRDUP(stablepath, devpath));
1451 1452

    return stablepath;
1453 1454
}

D
Daniel P. Berrange 已提交
1455

1456
#ifndef WIN32
1457 1458 1459 1460
/*
 * Run an external program.
 *
 * Read its output and apply a series of regexes to each line
R
Richard W.M. Jones 已提交
1461
 * When the entire set of regexes has matched consecutively
1462 1463 1464
 * then run a callback passing in all the matches
 */
int
1465
virStorageBackendRunProgRegex(virStoragePoolObjPtr pool,
1466
                              virCommandPtr cmd,
1467 1468 1469 1470
                              int nregex,
                              const char **regex,
                              int *nvars,
                              virStorageBackendListVolRegexFunc func,
1471
                              void *data, const char *prefix)
1472
{
1473
    int fd = -1, err, ret = -1;
1474 1475 1476 1477 1478 1479 1480 1481 1482
    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 */
1483
    if (VIR_ALLOC_N(reg, nregex) < 0) {
1484
        virReportOOMError();
1485 1486 1487
        return -1;
    }

1488
    for (i = 0; i < nregex; i++) {
1489 1490 1491 1492
        err = regcomp(&reg[i], regex[i], REG_EXTENDED);
        if (err != 0) {
            char error[100];
            regerror(err, &reg[i], error, sizeof(error));
1493 1494
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Failed to compile regex %s"), error);
1495
            for (j = 0; j <= i; j++)
1496
                regfree(&reg[j]);
1497
            VIR_FREE(reg);
1498 1499 1500 1501 1502 1503 1504 1505 1506 1507
            return -1;
        }

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

    }

    /* Storage for matched variables */
1508
    if (VIR_ALLOC_N(groups, totgroups) < 0) {
1509
        virReportOOMError();
1510 1511
        goto cleanup;
    }
1512
    if (VIR_ALLOC_N(vars, maxvars+1) < 0) {
1513
        virReportOOMError();
1514 1515 1516
        goto cleanup;
    }

1517 1518
    virCommandSetOutputFD(cmd, &fd);
    if (virCommandRunAsync(cmd, NULL) < 0) {
1519 1520 1521
        goto cleanup;
    }

1522
    if ((list = VIR_FDOPEN(fd, "r")) == NULL) {
1523 1524
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("cannot read fd"));
1525 1526 1527 1528
        goto cleanup;
    }

    while (fgets(line, sizeof(line), list) != NULL) {
1529
        char *p = NULL;
1530 1531 1532 1533 1534
        /* Strip trailing newline */
        int len = strlen(line);
        if (len && line[len-1] == '\n')
            line[len-1] = '\0';

1535 1536 1537 1538 1539 1540
        /* ignore any command prefix */
        if (prefix)
            p = STRSKIP(line, prefix);
        if (!p)
            p = line;

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

                if (i == 0)
                    ngroup = 0;

                /* NULL terminate each captured group in the line */
1549
                for (j = 0; j < nvars[i]; j++) {
1550
                    /* NB vars[0] is the full pattern, so we offset j by 1 */
1551
                    p[vars[j+1].rm_eo] = '\0';
1552
                    if (VIR_STRDUP(groups[ngroup++], p + vars[j+1].rm_so) < 0)
1553 1554 1555 1556 1557
                        goto cleanup;
                }

                /* We're matching on the last regex, so callback time */
                if (i == (nregex-1)) {
1558
                    if (((*func)(pool, groups, data)) < 0)
1559 1560 1561
                        goto cleanup;

                    /* Release matches & restart to matching the first regex */
1562
                    for (j = 0; j < totgroups; j++)
1563
                        VIR_FREE(groups[j]);
1564 1565 1566 1567 1568 1569 1570
                    maxReg = 0;
                    ngroup = 0;
                }
            }
        }
    }

1571
    ret = virCommandWait(cmd, NULL);
1572
cleanup:
1573
    if (groups) {
1574
        for (j = 0; j < totgroups; j++)
1575 1576
            VIR_FREE(groups[j]);
        VIR_FREE(groups);
1577
    }
1578
    VIR_FREE(vars);
1579

1580
    for (i = 0; i < nregex; i++)
1581 1582
        regfree(&reg[i]);

1583
    VIR_FREE(reg);
1584

1585 1586
    VIR_FORCE_FCLOSE(list);
    VIR_FORCE_CLOSE(fd);
1587

1588
    return ret;
1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602
}

/*
 * 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
1603
virStorageBackendRunProgNul(virStoragePoolObjPtr pool,
1604
                            virCommandPtr cmd,
1605 1606 1607 1608 1609
                            size_t n_columns,
                            virStorageBackendListVolNulFunc func,
                            void *data)
{
    size_t n_tok = 0;
1610
    int fd = -1;
1611 1612
    FILE *fp = NULL;
    char **v;
1613
    int ret = -1;
1614 1615 1616 1617 1618
    int i;

    if (n_columns == 0)
        return -1;

1619
    if (VIR_ALLOC_N(v, n_columns) < 0) {
1620
        virReportOOMError();
1621 1622 1623 1624 1625
        return -1;
    }
    for (i = 0; i < n_columns; i++)
        v[i] = NULL;

1626 1627
    virCommandSetOutputFD(cmd, &fd);
    if (virCommandRunAsync(cmd, NULL) < 0) {
1628 1629 1630
        goto cleanup;
    }

1631
    if ((fp = VIR_FDOPEN(fd, "r")) == NULL) {
1632 1633
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("cannot open file using fd"));
1634 1635 1636 1637 1638 1639 1640 1641
        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.  */
1642
        ssize_t tok_len = getdelim(&buf, &buf_len, 0, fp);
1643 1644 1645 1646
        v[n_tok] = buf;
        if (tok_len < 0) {
            /* Maybe EOF, maybe an error.
               If n_tok > 0, then we know it's an error.  */
1647
            if (n_tok && func(pool, n_tok, v, data) < 0)
1648 1649 1650 1651 1652
                goto cleanup;
            break;
        }
        ++n_tok;
        if (n_tok == n_columns) {
1653
            if (func(pool, n_tok, v, data) < 0)
1654 1655 1656
                goto cleanup;
            n_tok = 0;
            for (i = 0; i < n_columns; i++) {
1657
                VIR_FREE(v[i]);
1658 1659 1660 1661
            }
        }
    }

1662 1663 1664
    if (feof(fp) < 0) {
        virReportSystemError(errno, "%s",
                             _("read error on pipe"));
1665 1666
        goto cleanup;
    }
1667

1668
    ret = virCommandWait(cmd, NULL);
1669 1670
 cleanup:
    for (i = 0; i < n_columns; i++)
1671 1672
        VIR_FREE(v[i]);
    VIR_FREE(v);
1673

1674 1675
    VIR_FORCE_FCLOSE(fp);
    VIR_FORCE_CLOSE(fd);
1676

1677
    return ret;
1678
}
D
Daniel P. Berrange 已提交
1679

1680
#else /* WIN32 */
D
Daniel P. Berrange 已提交
1681 1682 1683 1684 1685 1686 1687 1688 1689

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,
1690
                              void *data ATTRIBUTE_UNUSED)
D
Daniel P. Berrange 已提交
1691
{
1692 1693
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704
    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)
{
1705 1706
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1707 1708
    return -1;
}
1709
#endif /* WIN32 */