storage_backend.c 49.4 KB
Newer Older
1
/*
2
 * storage_backend.c: internal storage driver backend contract
3
 *
4
 * Copyright (C) 2007-2011 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
 * Copyright (C) 2007-2008 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307  USA
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

#include <config.h>

#include <string.h>
27
#include <stdio.h>
D
Daniel P. Berrange 已提交
28
#if HAVE_REGEX_H
29
# include <regex.h>
D
Daniel P. Berrange 已提交
30
#endif
31
#include <sys/types.h>
E
Eric Blake 已提交
32
#include <sys/wait.h>
33
#include <unistd.h>
34 35 36
#include <fcntl.h>
#include <stdint.h>
#include <sys/stat.h>
37
#include <sys/param.h>
38
#include <dirent.h>
39
#include "dirname.h"
40 41 42 43
#ifdef __linux__
# include <sys/ioctl.h>
# include <linux/fs.h>
#endif
44 45

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

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

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

81 82
#define VIR_FROM_THIS VIR_FROM_STORAGE

83 84 85 86 87 88 89 90 91 92 93 94 95 96
static virStorageBackendPtr backends[] = {
#if WITH_STORAGE_DIR
    &virStorageBackendDirectory,
#endif
#if WITH_STORAGE_FS
    &virStorageBackendFileSystem,
    &virStorageBackendNetFileSystem,
#endif
#if WITH_STORAGE_LVM
    &virStorageBackendLogical,
#endif
#if WITH_STORAGE_ISCSI
    &virStorageBackendISCSI,
#endif
97 98 99
#if WITH_STORAGE_SCSI
    &virStorageBackendSCSI,
#endif
D
Dave Allan 已提交
100 101 102
#if WITH_STORAGE_MPATH
    &virStorageBackendMpath,
#endif
103 104 105 106 107
#if WITH_STORAGE_DISK
    &virStorageBackendDisk,
#endif
    NULL
};
108

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

enum {
    TOOL_QEMU_IMG,
    TOOL_KVM_IMG,
    TOOL_QCOW_CREATE,
};

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

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

137
    if ((inputfd = open(inputvol->target.path, O_RDONLY)) < 0) {
138
        ret = -errno;
139
        virReportSystemError(errno,
140 141 142
                             _("could not open input path '%s'"),
                             inputvol->target.path);
        goto cleanup;
143 144
    }

145 146 147 148 149 150 151 152 153 154 155 156 157 158 159
#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;
    }
160

161
    if (VIR_ALLOC_N(buf, rbytes) < 0) {
162
        ret = -errno;
163
        virReportOOMError();
164 165 166 167 168 169
        goto cleanup;
    }

    while (amtread != 0) {
        int amtleft;

170 171
        if (*total < rbytes)
            rbytes = *total;
172

173
        if ((amtread = saferead(inputfd, buf, rbytes)) < 0) {
174
            ret = -errno;
175
            virReportSystemError(errno,
176 177 178 179
                                 _("failed reading from file '%s'"),
                                 inputvol->target.path);
            goto cleanup;
        }
180
        *total -= amtread;
181 182 183 184 185

        /* Loop over amt read in 512 byte increments, looking for sparse
         * blocks */
        amtleft = amtread;
        do {
186
            interval = ((wbytes > amtleft) ? amtleft : wbytes);
187 188
            int offset = amtread - amtleft;

189
            if (is_dest_file && memcmp(buf+offset, zerobuf, interval) == 0) {
190
                if (lseek(fd, interval, SEEK_CUR) < 0) {
191
                    ret = -errno;
192
                    virReportSystemError(errno,
193 194 195 196 197
                                         _("cannot extend file '%s'"),
                                         vol->target.path);
                    goto cleanup;
                }
            } else if (safewrite(fd, buf+offset, interval) < 0) {
198
                ret = -errno;
199
                virReportSystemError(errno,
200 201 202 203 204
                                     _("failed writing to file '%s'"),
                                     vol->target.path);
                goto cleanup;

            }
205
        } while ((amtleft -= interval) > 0);
206 207
    }

208 209 210 211 212 213 214 215
    if (fdatasync(fd) < 0) {
        ret = -errno;
        virReportSystemError(errno, _("cannot sync data to file '%s'"),
                             vol->target.path);
        goto cleanup;
    }


216
    if (VIR_CLOSE(inputfd) < 0) {
217
        ret = -errno;
218
        virReportSystemError(errno,
219 220 221 222 223 224 225
                             _("cannot close file '%s'"),
                             inputvol->target.path);
        goto cleanup;
    }
    inputfd = -1;

cleanup:
226
    VIR_FORCE_CLOSE(inputfd);
227

228
    VIR_FREE(zerobuf);
229 230
    VIR_FREE(buf);

231 232 233
    return ret;
}

234
static int
235
virStorageBackendCreateBlockFrom(virConnectPtr conn ATTRIBUTE_UNUSED,
236
                                 virStoragePoolObjPtr pool ATTRIBUTE_UNUSED,
237 238
                                 virStorageVolDefPtr vol,
                                 virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
239
                                 unsigned int flags)
240 241 242 243
{
    int fd = -1;
    int ret = -1;
    unsigned long long remain;
244 245 246
    struct stat st;
    gid_t gid;
    uid_t uid;
247

E
Eric Blake 已提交
248 249
    virCheckFlags(0, -1);

250
    if ((fd = open(vol->target.path, O_RDWR)) < 0) {
251
        virReportSystemError(errno,
252 253 254 255 256 257 258 259
                             _("cannot create path '%s'"),
                             vol->target.path);
        goto cleanup;
    }

    remain = vol->allocation;

    if (inputvol) {
260
        int res = virStorageBackendCopyToFD(vol, inputvol,
261
                                            fd, &remain, 0);
262 263 264 265
        if (res < 0)
            goto cleanup;
    }

266
    if (fstat(fd, &st) == -1) {
267
        virReportSystemError(errno, _("stat of '%s' failed"),
268 269 270 271 272 273
                             vol->target.path);
        goto cleanup;
    }
    uid = (vol->target.perms.uid != st.st_uid) ? vol->target.perms.uid : -1;
    gid = (vol->target.perms.gid != st.st_gid) ? vol->target.perms.gid : -1;
    if (((uid != -1) || (gid != -1))
274
        && (fchown(fd, uid, gid) < 0)) {
275
        virReportSystemError(errno,
276
                             _("cannot chown '%s' to (%u, %u)"),
277
                             vol->target.path, uid, gid);
278 279 280
        goto cleanup;
    }
    if (fchmod(fd, vol->target.perms.mode) < 0) {
281
        virReportSystemError(errno,
282 283 284 285
                             _("cannot set mode of '%s' to %04o"),
                             vol->target.path, vol->target.perms.mode);
        goto cleanup;
    }
286
    if (VIR_CLOSE(fd) < 0) {
287
        virReportSystemError(errno,
288 289 290 291 292 293 294 295
                             _("cannot close file '%s'"),
                             vol->target.path);
        goto cleanup;
    }
    fd = -1;

    ret = 0;
cleanup:
296
    VIR_FORCE_CLOSE(fd);
297 298 299 300

    return ret;
}

E
Eric Blake 已提交
301 302 303 304
static int
createRawFile(int fd, virStorageVolDefPtr vol,
              virStorageVolDefPtr inputvol)
{
305 306
    int ret = 0;
    unsigned long long remain;
307

308 309
    /* Seek to the final size, so the capacity is available upfront
     * for progress reporting */
E
Eric Blake 已提交
310
    if (ftruncate(fd, vol->capacity) < 0) {
311
        ret = -errno;
312
        virReportSystemError(errno,
313
                             _("cannot extend file '%s'"),
E
Eric Blake 已提交
314
                             vol->target.path);
315 316 317
        goto cleanup;
    }

E
Eric Blake 已提交
318
    remain = vol->allocation;
319

E
Eric Blake 已提交
320 321
    if (inputvol) {
        ret = virStorageBackendCopyToFD(vol, inputvol, fd, &remain, 1);
322
        if (ret < 0) {
323
            goto cleanup;
324
        }
325 326 327 328 329 330 331 332 333 334 335 336 337 338
    }

    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
339
                if (safezero(fd, vol->allocation - remain, bytes) < 0) {
340
                    ret = -errno;
J
Jiri Denemark 已提交
341
                    virReportSystemError(errno, _("cannot fill file '%s'"),
E
Eric Blake 已提交
342
                                         vol->target.path);
343 344 345 346 347
                    goto cleanup;
                }
                remain -= bytes;
            }
        } else { /* No progress bars to be shown */
E
Eric Blake 已提交
348
            if (safezero(fd, 0, remain) < 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
                goto cleanup;
            }
        }
355

356 357 358
    }

    if (fsync(fd) < 0) {
359
        ret = -errno;
360
        virReportSystemError(errno, _("cannot sync data to file '%s'"),
E
Eric Blake 已提交
361
                             vol->target.path);
362
        goto cleanup;
363 364
    }

365 366 367 368 369 370 371 372 373
cleanup:
    return ret;
}

int
virStorageBackendCreateRaw(virConnectPtr conn ATTRIBUTE_UNUSED,
                           virStoragePoolObjPtr pool,
                           virStorageVolDefPtr vol,
                           virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
374
                           unsigned int flags)
375 376
{
    int ret = -1;
E
Eric Blake 已提交
377 378 379 380
    int fd = -1;
    int operation_flags;

    virCheckFlags(0, -1);
381 382

    if (vol->target.encryption != NULL) {
O
Osier Yang 已提交
383
        virStorageReportError(VIR_ERR_CONFIG_UNSUPPORTED,
384 385
                              "%s", _("storage pool does not support encrypted "
                                      "volumes"));
386 387
        goto cleanup;
    }
388

L
Laine Stump 已提交
389
    operation_flags = VIR_FILE_OPEN_FORCE_MODE | VIR_FILE_OPEN_FORCE_OWNER;
E
Eric Blake 已提交
390
    if (pool->def->type == VIR_STORAGE_POOL_NETFS)
L
Laine Stump 已提交
391
        operation_flags |= VIR_FILE_OPEN_FORK;
E
Eric Blake 已提交
392

393 394
    if ((fd = virFileOpenAs(vol->target.path,
                            O_RDWR | O_CREAT | O_EXCL,
L
Laine Stump 已提交
395 396 397
                            vol->target.perms.mode,
                            vol->target.perms.uid,
                            vol->target.perms.gid,
398
                            operation_flags)) < 0) {
E
Eric Blake 已提交
399 400 401 402 403 404
        virReportSystemError(-fd,
                             _("cannot create path '%s'"),
                             vol->target.path);
        goto cleanup;
    }

405 406
    if ((ret = createRawFile(fd, vol, inputvol)) < 0)
        /* createRawFile already reported the exact error. */
E
Eric Blake 已提交
407
        ret = -1;
408 409

cleanup:
E
Eric Blake 已提交
410
    VIR_FORCE_CLOSE(fd);
411 412 413
    return ret;
}

414 415 416 417 418 419 420 421 422
static int
virStorageGenerateSecretUUID(virConnectPtr conn,
                             unsigned char *uuid)
{
    unsigned attempt;

    for (attempt = 0; attempt < 65536; attempt++) {
        virSecretPtr tmp;
        if (virUUIDGenerate(uuid) < 0) {
423 424
            virStorageReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                                  _("unable to generate uuid"));
425 426 427 428 429 430 431 432 433
            return -1;
        }
        tmp = conn->secretDriver->lookupByUUID(conn, uuid);
        if (tmp == NULL)
            return 0;

        virSecretFree(tmp);
    }

434 435
    virStorageReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                          _("too many conflicts when generating an uuid"));
436 437 438 439

    return -1;
}

440 441 442 443 444 445 446 447 448
static int
virStorageGenerateQcowEncryption(virConnectPtr conn,
                                 virStorageVolDefPtr vol)
{
    virSecretDefPtr def = NULL;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    virStorageEncryptionPtr enc;
    virStorageEncryptionSecretPtr enc_secret = NULL;
    virSecretPtr secret = NULL;
449
    char *xml;
450 451 452
    unsigned char value[VIR_STORAGE_QCOW_PASSPHRASE_SIZE];
    int ret = -1;

453 454 455
    if (conn->secretDriver == NULL ||
        conn->secretDriver->lookupByUUID == NULL ||
        conn->secretDriver->defineXML == NULL ||
456
        conn->secretDriver->setValue == NULL) {
O
Osier Yang 已提交
457
        virStorageReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
458 459 460 461 462 463
                              _("secret storage not supported"));
        goto cleanup;
    }

    enc = vol->target.encryption;
    if (enc->nsecrets != 0) {
464
        virStorageReportError(VIR_ERR_INTERNAL_ERROR, "%s",
465 466 467 468 469 470
                              _("secrets already defined"));
        goto cleanup;
    }

    if (VIR_ALLOC(enc_secret) < 0 || VIR_REALLOC_N(enc->secrets, 1) < 0 ||
        VIR_ALLOC(def) < 0) {
471
        virReportOOMError();
472 473 474 475 476
        goto cleanup;
    }

    def->ephemeral = 0;
    def->private = 0;
477
    if (virStorageGenerateSecretUUID(conn, def->uuid) < 0)
478
        goto cleanup;
479

480 481 482
    def->usage_type = VIR_SECRET_USAGE_TYPE_VOLUME;
    def->usage.volume = strdup(vol->target.path);
    if (def->usage.volume == NULL) {
483
        virReportOOMError();
484 485
        goto cleanup;
    }
486
    xml = virSecretDefFormat(def);
487 488 489 490 491 492 493 494 495 496 497 498
    virSecretDefFree(def);
    def = NULL;
    if (xml == NULL)
        goto cleanup;

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

499
    if (virStorageGenerateQcowPassphrase(value) < 0)
500 501 502 503 504 505
        goto cleanup;

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

    enc_secret->type = VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE;
506
    memcpy(enc_secret->uuid, secret->uuid, VIR_UUID_BUFLEN);
507 508 509 510 511 512 513 514 515
    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) {
516 517
        if (ret != 0 &&
            conn->secretDriver->undefine != NULL)
518 519 520
            conn->secretDriver->undefine(secret);
        virSecretFree(secret);
    }
521
    virBufferFreeAndReset(&buf);
522 523 524 525 526
    virSecretDefFree(def);
    VIR_FREE(enc_secret);
    return ret;
}

527 528 529 530 531
struct hookdata {
    virStorageVolDefPtr vol;
    bool skip;
};

532
static int virStorageBuildSetUIDHook(void *data) {
533 534 535 536 537
    struct hookdata *tmp = data;
    virStorageVolDefPtr vol = tmp->vol;

    if (tmp->skip)
        return 0;
538

539
    if (virSetUIDGID(vol->target.perms.uid, vol->target.perms.gid) < 0)
540
        return -1;
541

542 543 544
    return 0;
}

545
static int virStorageBackendCreateExecCommand(virStoragePoolObjPtr pool,
546
                                              virStorageVolDefPtr vol,
547
                                              virCommandPtr cmd) {
548 549 550 551
    struct stat st;
    gid_t gid;
    uid_t uid;
    int filecreated = 0;
552
    struct hookdata data = {vol, false};
553 554

    if ((pool->def->type == VIR_STORAGE_POOL_NETFS)
555 556 557 558 559
        && (((getuid() == 0)
             && (vol->target.perms.uid != -1)
             && (vol->target.perms.uid != 0))
            || ((vol->target.perms.gid != -1)
                && (vol->target.perms.gid != getgid())))) {
560 561 562 563

        virCommandSetPreExecHook(cmd, virStorageBuildSetUIDHook, &data);

        if (virCommandRun(cmd, NULL) == 0) {
564 565 566 567 568
            /* command was successfully run, check if the file was created */
            if (stat(vol->target.path, &st) >=0)
                filecreated = 1;
        }
    }
569 570 571

    data.skip = true;

572
    if (!filecreated) {
573
        if (virCommandRun(cmd, NULL) < 0) {
574 575 576
            return -1;
        }
        if (stat(vol->target.path, &st) < 0) {
577
            virReportSystemError(errno,
578
                                 _("failed to create %s"), vol->target.path);
579 580 581 582 583 584 585 586
            return -1;
        }
    }

    uid = (vol->target.perms.uid != st.st_uid) ? vol->target.perms.uid : -1;
    gid = (vol->target.perms.gid != st.st_gid) ? vol->target.perms.gid : -1;
    if (((uid != -1) || (gid != -1))
        && (chown(vol->target.path, uid, gid) < 0)) {
587
        virReportSystemError(errno,
588
                             _("cannot chown %s to (%u, %u)"),
589
                             vol->target.path, uid, gid);
590 591 592
        return -1;
    }
    if (chmod(vol->target.path, vol->target.perms.mode) < 0) {
593
        virReportSystemError(errno,
594 595 596 597 598 599 600
                             _("cannot set mode of '%s' to %04o"),
                             vol->target.path, vol->target.perms.mode);
        return -1;
    }
    return 0;
}

601 602 603 604 605 606 607 608 609 610 611 612 613
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;
614
    int exitstatus;
615
    virCommandPtr cmd = virCommandNewArgList(qemuimg, "-h", NULL);
616

617 618 619
    virCommandAddEnvString(cmd, "LC_ALL=C");
    virCommandSetOutputBuffer(cmd, &help);
    virCommandClearCaps(cmd);
620

621 622 623
    /* 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 */
624
    if (virCommandRun(cmd, &exitstatus) < 0)
625 626
        goto cleanup;

627 628 629 630 631 632 633
    if ((start = strstr(help, " create ")) == NULL ||
        (end = strstr(start, "\n")) == NULL) {
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
                              _("unable to parse qemu-img output '%s'"),
                              help);
        goto cleanup;
    }
634 635
    if (((tmp = strstr(start, "-F fmt")) && tmp < end) ||
        ((tmp = strstr(start, "-F backing_fmt")) && tmp < end))
636 637 638 639 640 641 642
        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:
643
    virCommandFree(cmd);
644 645 646 647 648
    VIR_FREE(help);
    return ret;
}


649 650
static int
virStorageBackendCreateQemuImg(virConnectPtr conn,
651
                               virStoragePoolObjPtr pool,
652
                               virStorageVolDefPtr vol,
653
                               virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
654
                               unsigned int flags)
655
{
656
    int ret = -1;
657
    char *create_tool;
658
    int imgformat = -1;
659 660
    virCommandPtr cmd = NULL;
    bool do_encryption = (vol->target.encryption != NULL);
661
    unsigned long long int size_arg;
662

E
Eric Blake 已提交
663 664
    virCheckFlags(0, -1);

665
    const char *type = virStorageFileFormatTypeToString(vol->target.format);
666
    const char *backingType = vol->backingStore.path ?
667
        virStorageFileFormatTypeToString(vol->backingStore.format) : NULL;
668 669 670 671 672 673

    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 ?
674 675 676 677
        virStorageFileFormatTypeToString(inputvol->type == VIR_STORAGE_VOL_BLOCK ?
                                         VIR_STORAGE_FILE_RAW :
                                         inputvol->target.format) :
        NULL;
678 679

    if (type == NULL) {
680
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
681 682 683 684 685
                              _("unknown storage vol type %d"),
                              vol->target.format);
        return -1;
    }
    if (inputvol && inputType == NULL) {
686
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
687 688 689 690 691 692
                              _("unknown storage vol type %d"),
                              inputvol->target.format);
        return -1;
    }

    if (vol->backingStore.path) {
693 694
        int accessRetCode = -1;
        char *absolutePath = NULL;
695 696 697 698 699 700 701 702

        /* 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))) {
703
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
E
Eric Blake 已提交
704
                                  "%s", _("a different backing store cannot "
705 706 707 708 709
                                          "be specified."));
            return -1;
        }

        if (backingType == NULL) {
710
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
711 712 713 714
                                  _("unknown storage vol backing store type %d"),
                                  vol->backingStore.format);
            return -1;
        }
715 716 717 718 719 720 721 722 723 724 725 726 727 728

        /* Convert relative backing store paths to absolute paths for access
         * validation.
         */
        if ('/' != *(vol->backingStore.path) &&
            virAsprintf(&absolutePath, "%s/%s", pool->def->target.path,
                        vol->backingStore.path) < 0) {
            virReportOOMError();
            return -1;
        }
        accessRetCode = access(absolutePath ? absolutePath
                               : vol->backingStore.path, R_OK);
        VIR_FREE(absolutePath);
        if (accessRetCode != 0) {
729
            virReportSystemError(errno,
730 731 732 733 734 735
                                 _("inaccessible backing store volume %s"),
                                 vol->backingStore.path);
            return -1;
        }
    }

736
    if (do_encryption) {
737 738
        virStorageEncryptionPtr enc;

739 740
        if (vol->target.format != VIR_STORAGE_FILE_QCOW &&
            vol->target.format != VIR_STORAGE_FILE_QCOW2) {
O
Osier Yang 已提交
741
            virStorageReportError(VIR_ERR_CONFIG_UNSUPPORTED,
742 743 744 745
                                  _("qcow volume encryption unsupported with "
                                    "volume format %s"), type);
            return -1;
        }
746 747 748
        enc = vol->target.encryption;
        if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW &&
            enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT) {
O
Osier Yang 已提交
749
            virStorageReportError(VIR_ERR_CONFIG_UNSUPPORTED,
750 751 752 753
                                  _("unsupported volume encryption format %d"),
                                  vol->target.encryption->format);
            return -1;
        }
754
        if (enc->nsecrets > 1) {
755
            virStorageReportError(VIR_ERR_XML_ERROR, "%s",
756 757 758
                                  _("too many secrets for qcow encryption"));
            return -1;
        }
759 760 761 762 763
        if (enc->format == VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT ||
            enc->nsecrets == 0) {
            if (virStorageGenerateQcowEncryption(conn, vol) < 0)
                return -1;
        }
764 765
    }

766
    /* Size in KB */
767
    size_arg = VIR_DIV_UP(vol->capacity, 1024);
768 769 770 771 772 773 774

    /* 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) {
775
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
776
                              "%s", _("unable to find kvm-img or qemu-img"));
777 778 779
        return -1;
    }

780 781 782 783
    imgformat = virStorageBackendQEMUImgBackingFormat(create_tool);
    if (imgformat < 0)
        goto cleanup;

784 785
    cmd = virCommandNew(create_tool);

786
    if (inputvol) {
787 788 789 790
        virCommandAddArgList(cmd, "convert", "-f", inputType, "-O", type,
                             inputPath, vol->target.path, NULL);

        if (do_encryption) {
791
            if (imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS) {
792
                virCommandAddArgList(cmd, "-o", "encryption=on", NULL);
793
            } else {
794
                virCommandAddArg(cmd, "-e");
795 796 797
            }
        }

798
    } else if (vol->backingStore.path) {
799 800 801
        virCommandAddArgList(cmd, "create", "-f", type,
                             "-b", vol->backingStore.path, NULL);

802 803
        switch (imgformat) {
        case QEMU_IMG_BACKING_FORMAT_FLAG:
804
            virCommandAddArgList(cmd, "-F", backingType, vol->target.path,
805 806
                                 NULL);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
807 808 809

            if (do_encryption)
                virCommandAddArg(cmd, "-e");
810 811 812
            break;

        case QEMU_IMG_BACKING_FORMAT_OPTIONS:
813 814 815
            virCommandAddArg(cmd, "-o");
            virCommandAddArgFormat(cmd, "backing_fmt=%s%s", backingType,
                                   do_encryption ? ",encryption=on" : "");
816 817
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
818
            break;
819 820 821 822 823

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

824 825
            virCommandAddArg(cmd, vol->target.path);
            virCommandAddArgFormat(cmd, "%lluK", size_arg);
826 827 828
            if (do_encryption)
                virCommandAddArg(cmd, "-e");
        }
829
    } else {
830
        virCommandAddArgList(cmd, "create", "-f", type,
831 832
                             vol->target.path, NULL);
        virCommandAddArgFormat(cmd, "%lluK", size_arg);
833 834

        if (do_encryption) {
835
            if (imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS) {
836
                virCommandAddArgList(cmd, "-o", "encryption=on", NULL);
837
            } else {
838
                virCommandAddArg(cmd, "-e");
839 840
            }
        }
841
    }
842

843 844
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
cleanup:
845
    VIR_FREE(create_tool);
846
    virCommandFree(cmd);
847

848
    return ret;
849 850 851 852 853 854 855
}

/*
 * Xen removed the fully-functional qemu-img, and replaced it
 * with a partially functional qcow-create. Go figure ??!?
 */
static int
856
virStorageBackendCreateQcowCreate(virConnectPtr conn ATTRIBUTE_UNUSED,
857
                                  virStoragePoolObjPtr pool,
858
                                  virStorageVolDefPtr vol,
859
                                  virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
860
                                  unsigned int flags)
861
{
862
    int ret;
E
Eric Blake 已提交
863
    char *size;
864
    virCommandPtr cmd;
865

E
Eric Blake 已提交
866 867
    virCheckFlags(0, -1);

868
    if (inputvol) {
869
        virStorageReportError(VIR_ERR_INTERNAL_ERROR, "%s",
870 871 872 873
                              _("cannot copy from volume with qcow-create"));
        return -1;
    }

874
    if (vol->target.format != VIR_STORAGE_FILE_QCOW2) {
875
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
876 877 878 879 880
                              _("unsupported storage vol type %d"),
                              vol->target.format);
        return -1;
    }
    if (vol->backingStore.path != NULL) {
O
Osier Yang 已提交
881
        virStorageReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
882
                              _("copy-on-write image not supported with "
883
                                      "qcow-create"));
884 885
        return -1;
    }
886
    if (vol->target.encryption != NULL) {
O
Osier Yang 已提交
887
        virStorageReportError(VIR_ERR_CONFIG_UNSUPPORTED,
888 889 890 891
                              "%s", _("encrypted volumes not supported with "
                                      "qcow-create"));
        return -1;
    }
892 893

    /* Size in MB - yes different units to qemu-img :-( */
894 895
    if (virAsprintf(&size, "%llu",
                    VIR_DIV_UP(vol->capacity, (1024 * 1024))) < 0) {
E
Eric Blake 已提交
896 897 898
        virReportOOMError();
        return -1;
    }
899

900
    cmd = virCommandNewArgList("qcow-create", size, vol->target.path, NULL);
901

902 903
    ret = virStorageBackendCreateExecCommand(pool, vol, cmd);
    virCommandFree(cmd);
E
Eric Blake 已提交
904
    VIR_FREE(size);
905

906
    return ret;
907 908
}

909
virStorageBackendBuildVolFrom
910
virStorageBackendFSImageToolTypeToFunc(int tool_type)
911 912 913 914 915 916 917 918
{
    switch (tool_type) {
    case TOOL_KVM_IMG:
    case TOOL_QEMU_IMG:
        return virStorageBackendCreateQemuImg;
    case TOOL_QCOW_CREATE:
        return virStorageBackendCreateQcowCreate;
    default:
919
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948
                              _("Unknown file create tool type '%d'."),
                              tool_type);
    }

    return NULL;
}

int
virStorageBackendFindFSImageTool(char **tool)
{
    int tool_type = -1;
    char *tmp = NULL;

    if ((tmp = virFindFileInPath("kvm-img")) != NULL) {
        tool_type = TOOL_KVM_IMG;
    } else if ((tmp = virFindFileInPath("qemu-img")) != NULL) {
        tool_type = TOOL_QEMU_IMG;
    } else if ((tmp = virFindFileInPath("qcow-create")) != NULL) {
        tool_type = TOOL_QCOW_CREATE;
    }

    if (tool)
        *tool = tmp;
    else
        VIR_FREE(tmp);

    return tool_type;
}

949
virStorageBackendBuildVolFrom
950
virStorageBackendGetBuildVolFromFunction(virStorageVolDefPtr vol,
951 952 953 954 955 956 957 958 959 960 961
                                         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 &&
962
         vol->target.format != VIR_STORAGE_FILE_RAW) ||
963
        (inputvol->type == VIR_STORAGE_VOL_FILE &&
964
         inputvol->target.format != VIR_STORAGE_FILE_RAW)) {
965

D
Daniel P. Berrange 已提交
966
        if ((tool_type = virStorageBackendFindFSImageTool(NULL)) < 0) {
967
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
968 969 970 971 972
                                  "%s", _("creation of non-raw file images is "
                                          "not supported without qemu-img."));
            return NULL;
        }

973
        return virStorageBackendFSImageToolTypeToFunc(tool_type);
974 975
    }

976 977 978 979
    if (vol->type == VIR_STORAGE_VOL_BLOCK)
        return virStorageBackendCreateBlockFrom;
    else
        return virStorageBackendCreateRaw;
980
}
981

982

983 984
virStorageBackendPtr
virStorageBackendForType(int type) {
985
    unsigned int i;
986
    for (i = 0; backends[i]; i++)
987 988 989
        if (backends[i]->type == type)
            return backends[i];

990
    virStorageReportError(VIR_ERR_INTERNAL_ERROR,
991 992 993 994 995
                          _("missing backend for pool type %d"), type);
    return NULL;
}


996 997 998
/*
 * Allows caller to silently ignore files with improper mode
 *
999 1000
 * Returns -1 on error, -2 if file mode is unexpected or the
 * volume is a dangling symbolic link.
1001
 */
1002
int
1003
virStorageBackendVolOpenCheckMode(const char *path, unsigned int flags)
1004
{
1005 1006
    int fd, mode = 0;
    struct stat sb;
1007
    char *base = last_component(path);
1008

1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023
    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;
    }

1024
    if ((fd = open(path, O_RDONLY|O_NONBLOCK|O_NOCTTY)) < 0) {
1025
        if ((errno == ENOENT || errno == ELOOP) &&
1026
            S_ISLNK(sb.st_mode)) {
1027 1028 1029 1030
            VIR_WARN("ignoring dangling symlink '%s'", path);
            return -2;
        }

1031
        virReportSystemError(errno,
1032
                             _("cannot open volume '%s'"),
1033
                             path);
1034 1035 1036
        return -1;
    }

1037 1038 1039 1040 1041 1042 1043 1044
    if (fstat(fd, &sb) < 0) {
        virReportSystemError(errno,
                             _("cannot stat file '%s'"),
                             path);
        VIR_FORCE_CLOSE(fd);
        return -1;
    }

1045 1046 1047 1048 1049 1050
    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;
1051 1052 1053 1054 1055 1056 1057 1058 1059 1060
    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;
        }
    }
1061 1062

    if (!(mode & flags)) {
1063
        VIR_FORCE_CLOSE(fd);
1064
        VIR_INFO("Skipping volume '%s'", path);
1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086

        if (mode & VIR_STORAGE_VOL_OPEN_ERROR) {
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
                                  _("unexpected storage mode for '%s'"), path);
            return -1;
        }

        return -2;
    }

    return fd;
}

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

int
virStorageBackendUpdateVolTargetInfo(virStorageVolTargetPtr target,
                                     unsigned long long *allocation,
1087 1088
                                     unsigned long long *capacity,
                                     unsigned int openflags)
1089 1090 1091
{
    int ret, fd;

1092 1093
    if ((ret = virStorageBackendVolOpenCheckMode(target->path,
                                                 openflags)) < 0)
1094 1095 1096
        return ret;

    fd = ret;
1097
    ret = virStorageBackendUpdateVolTargetInfoFD(target,
1098 1099 1100
                                                 fd,
                                                 allocation,
                                                 capacity);
1101

1102
    VIR_FORCE_CLOSE(fd);
1103 1104 1105 1106

    return ret;
}

1107
int
1108 1109 1110
virStorageBackendUpdateVolInfoFlags(virStorageVolDefPtr vol,
                                    int withCapacity,
                                    unsigned int openflags)
1111 1112 1113
{
    int ret;

1114
    if ((ret = virStorageBackendUpdateVolTargetInfo(&vol->target,
1115 1116 1117
                                    &vol->allocation,
                                    withCapacity ? &vol->capacity : NULL,
                                    openflags)) < 0)
1118 1119 1120
        return ret;

    if (vol->backingStore.path &&
1121
        (ret = virStorageBackendUpdateVolTargetInfo(&vol->backingStore,
1122 1123
                                            NULL, NULL,
                                            VIR_STORAGE_VOL_OPEN_DEFAULT)) < 0)
1124 1125 1126 1127 1128
        return ret;

    return 0;
}

1129 1130 1131 1132 1133 1134 1135
int virStorageBackendUpdateVolInfo(virStorageVolDefPtr vol,
                                   int withCapacity)
{
    return virStorageBackendUpdateVolInfoFlags(vol, withCapacity,
                                               VIR_STORAGE_VOL_OPEN_DEFAULT);
}

1136 1137 1138 1139
/*
 * virStorageBackendUpdateVolTargetInfoFD:
 * @conn: connection to report errors on
 * @target: target definition ptr of volume to update
1140
 * @fd: fd of storage volume to update, via virStorageBackendOpenVol*
1141 1142 1143
 * @allocation: If not NULL, updated allocation information will be stored
 * @capacity: If not NULL, updated capacity info will be stored
 *
1144
 * Returns 0 for success, -1 on a legitimate error condition.
1145
 */
1146
int
1147
virStorageBackendUpdateVolTargetInfoFD(virStorageVolTargetPtr target,
1148 1149 1150
                                       int fd,
                                       unsigned long long *allocation,
                                       unsigned long long *capacity)
1151 1152 1153 1154 1155 1156 1157
{
    struct stat sb;
#if HAVE_SELINUX
    security_context_t filecon = NULL;
#endif

    if (fstat(fd, &sb) < 0) {
1158
        virReportSystemError(errno,
1159
                             _("cannot stat file '%s'"),
1160
                             target->path);
1161 1162 1163
        return -1;
    }

1164 1165
    if (allocation) {
        if (S_ISREG(sb.st_mode)) {
1166
#ifndef WIN32
1167
            *allocation = (unsigned long long)sb.st_blocks *
1168
                          (unsigned long long)DEV_BSIZE;
D
Daniel P. Berrange 已提交
1169
#else
1170
            *allocation = sb.st_size;
D
Daniel P. Berrange 已提交
1171
#endif
1172 1173 1174 1175 1176
            /* Regular files may be sparse, so logical size (capacity) is not same
             * as actual allocation above
             */
            if (capacity)
                *capacity = sb.st_size;
1177 1178 1179 1180 1181
        } else if (S_ISDIR(sb.st_mode)) {
            *allocation = 0;
            if (capacity)
                *capacity = 0;

1182 1183
        } else {
            off_t end;
E
Eric Blake 已提交
1184
            /* XXX this is POSIX compliant, but doesn't work for CHAR files,
1185 1186 1187 1188 1189 1190
             * 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) {
1191
                virReportSystemError(errno,
1192 1193 1194 1195 1196 1197 1198
                                     _("cannot seek to end of file '%s'"),
                                     target->path);
                return -1;
            }
            *allocation = end;
            if (capacity)
                *capacity = end;
1199 1200 1201
        }
    }

1202 1203 1204
    target->perms.mode = sb.st_mode & S_IRWXUGO;
    target->perms.uid = sb.st_uid;
    target->perms.gid = sb.st_gid;
1205

1206
    VIR_FREE(target->perms.label);
1207 1208

#if HAVE_SELINUX
1209
    /* XXX: make this a security driver call */
1210
    if (fgetfilecon(fd, &filecon) == -1) {
1211
        if (errno != ENODATA && errno != ENOTSUP) {
1212
            virReportSystemError(errno,
1213
                                 _("cannot get file context of '%s'"),
1214
                                 target->path);
1215 1216
            return -1;
        } else {
1217
            target->perms.label = NULL;
1218 1219
        }
    } else {
1220
        target->perms.label = strdup(filecon);
1221
        freecon(filecon);
1222
        if (target->perms.label == NULL) {
1223
            virReportOOMError();
1224 1225
            return -1;
        }
1226 1227
    }
#else
1228
    target->perms.label = NULL;
1229 1230 1231 1232 1233
#endif

    return 0;
}

D
Dave Allan 已提交
1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268

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
1269 1270
virStorageBackendDetectBlockVolFormatFD(virStorageVolTargetPtr target,
                                        int fd)
D
Dave Allan 已提交
1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281
{
    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) {
1282
        virReportSystemError(errno,
D
Dave Allan 已提交
1283 1284 1285 1286 1287 1288
                             _("cannot seek to beginning of file '%s'"),
                             target->path);
        return -1;
    }
    bytes = saferead(fd, buffer, sizeof(buffer));
    if (bytes < 0) {
1289
        virReportSystemError(errno,
D
Dave Allan 已提交
1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308
                             _("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;
}


1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322
/*
 * Given a volume path directly in /dev/XXX, iterate over the
 * entries in the directory pool->def->target.path and find the
 * first symlink pointing to the volume path.
 *
 * If, the target.path is /dev/, then return the original volume
 * path.
 *
 * If no symlink is found, then return the original volume path
 *
 * Typically target.path is one of the /dev/disk/by-XXX dirs
 * with stable paths.
 */
char *
1323
virStorageBackendStablePath(virStoragePoolObjPtr pool,
1324
                            const char *devpath)
1325 1326 1327
{
    DIR *dh;
    struct dirent *dent;
1328
    char *stablepath;
1329
    int opentries = 0;
1330 1331 1332 1333 1334

    /* 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/"))
1335
        goto ret_strdup;
1336

1337 1338 1339 1340 1341 1342
    /* Skip whole thing for a pool which isn't in /dev
     * so we don't mess will filesystem/dir based pools
     */
    if (!STRPREFIX(pool->def->target.path, "/dev"))
        goto ret_strdup;

1343 1344 1345 1346
    /* Logical pools are under /dev but already have stable paths */
    if (pool->def->type == VIR_STORAGE_POOL_LOGICAL)
        goto ret_strdup;

1347 1348 1349
    /* 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.
1350
     */
1351
 reopen:
1352
    if ((dh = opendir(pool->def->target.path)) == NULL) {
1353 1354 1355 1356 1357
        opentries++;
        if (errno == ENOENT && opentries < 50) {
            usleep(100 * 1000);
            goto reopen;
        }
1358
        virReportSystemError(errno,
1359 1360
                             _("cannot read dir '%s'"),
                             pool->def->target.path);
1361 1362 1363
        return NULL;
    }

1364 1365 1366 1367 1368
    /* The pool is pointing somewhere like /dev/disk/by-path
     * or /dev/disk/by-id, so we need to check all symlinks in
     * the target directory and figure out which one points
     * to this device node
     */
1369 1370 1371 1372
    while ((dent = readdir(dh)) != NULL) {
        if (dent->d_name[0] == '.')
            continue;

1373 1374 1375
        if (virAsprintf(&stablepath, "%s/%s",
                        pool->def->target.path,
                        dent->d_name) == -1) {
1376
            virReportOOMError();
1377 1378 1379 1380 1381 1382 1383 1384 1385
            closedir(dh);
            return NULL;
        }

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

1386
        VIR_FREE(stablepath);
1387 1388 1389 1390
    }

    closedir(dh);

1391
 ret_strdup:
1392 1393 1394
    /* Couldn't find any matching stable link so give back
     * the original non-stable dev path
     */
1395 1396 1397 1398

    stablepath = strdup(devpath);

    if (stablepath == NULL)
1399
        virReportOOMError();
1400 1401

    return stablepath;
1402 1403
}

D
Daniel P. Berrange 已提交
1404

1405
#ifndef WIN32
1406 1407 1408 1409
/*
 * Run an external program.
 *
 * Read its output and apply a series of regexes to each line
R
Richard W.M. Jones 已提交
1410
 * When the entire set of regexes has matched consecutively
1411 1412 1413
 * then run a callback passing in all the matches
 */
int
1414
virStorageBackendRunProgRegex(virStoragePoolObjPtr pool,
1415
                              const char *const*prog,
1416 1417 1418 1419
                              int nregex,
                              const char **regex,
                              int *nvars,
                              virStorageBackendListVolRegexFunc func,
1420
                              void *data, const char *prefix)
1421
{
1422
    int fd = -1, err, ret = -1;
1423 1424 1425 1426 1427 1428 1429
    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;
1430
    virCommandPtr cmd = NULL;
1431 1432

    /* Compile all regular expressions */
1433
    if (VIR_ALLOC_N(reg, nregex) < 0) {
1434
        virReportOOMError();
1435 1436 1437 1438 1439 1440 1441 1442
        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));
1443
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
1444 1445 1446
                                  _("Failed to compile regex %s"), error);
            for (j = 0 ; j <= i ; j++)
                regfree(&reg[j]);
1447
            VIR_FREE(reg);
1448 1449 1450 1451 1452 1453 1454 1455 1456 1457
            return -1;
        }

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

    }

    /* Storage for matched variables */
1458
    if (VIR_ALLOC_N(groups, totgroups) < 0) {
1459
        virReportOOMError();
1460 1461
        goto cleanup;
    }
1462
    if (VIR_ALLOC_N(vars, maxvars+1) < 0) {
1463
        virReportOOMError();
1464 1465 1466
        goto cleanup;
    }

1467 1468 1469
    cmd = virCommandNewArgs(prog);
    virCommandSetOutputFD(cmd, &fd);
    if (virCommandRunAsync(cmd, NULL) < 0) {
1470 1471 1472
        goto cleanup;
    }

1473
    if ((list = VIR_FDOPEN(fd, "r")) == NULL) {
1474
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
J
Jim Meyering 已提交
1475
                              "%s", _("cannot read fd"));
1476 1477 1478 1479
        goto cleanup;
    }

    while (fgets(line, sizeof(line), list) != NULL) {
1480
        char *p = NULL;
1481 1482 1483 1484 1485
        /* Strip trailing newline */
        int len = strlen(line);
        if (len && line[len-1] == '\n')
            line[len-1] = '\0';

1486 1487 1488 1489 1490 1491
        /* ignore any command prefix */
        if (prefix)
            p = STRSKIP(line, prefix);
        if (!p)
            p = line;

1492
        for (i = 0 ; i <= maxReg && i < nregex ; i++) {
1493
            if (regexec(&reg[i], p, nvars[i]+1, vars, 0) == 0) {
1494 1495 1496 1497 1498 1499 1500 1501
                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 */
1502
                    p[vars[j+1].rm_eo] = '\0';
1503
                    if ((groups[ngroup++] =
1504
                         strdup(p + vars[j+1].rm_so)) == NULL) {
1505
                        virReportOOMError();
1506 1507 1508 1509 1510 1511
                        goto cleanup;
                    }
                }

                /* We're matching on the last regex, so callback time */
                if (i == (nregex-1)) {
1512
                    if (((*func)(pool, groups, data)) < 0)
1513 1514 1515
                        goto cleanup;

                    /* Release matches & restart to matching the first regex */
1516
                    for (j = 0 ; j < totgroups ; j++)
1517
                        VIR_FREE(groups[j]);
1518 1519 1520 1521 1522 1523 1524
                    maxReg = 0;
                    ngroup = 0;
                }
            }
        }
    }

1525
    ret = virCommandWait(cmd, NULL);
1526
cleanup:
1527 1528
    if (groups) {
        for (j = 0 ; j < totgroups ; j++)
1529 1530
            VIR_FREE(groups[j]);
        VIR_FREE(groups);
1531
    }
1532
    VIR_FREE(vars);
1533 1534 1535 1536

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

1537
    VIR_FREE(reg);
1538
    virCommandFree(cmd);
1539

1540 1541
    VIR_FORCE_FCLOSE(list);
    VIR_FORCE_CLOSE(fd);
1542

1543
    return ret;
1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557
}

/*
 * 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
1558
virStorageBackendRunProgNul(virStoragePoolObjPtr pool,
1559 1560 1561 1562 1563 1564
                            const char **prog,
                            size_t n_columns,
                            virStorageBackendListVolNulFunc func,
                            void *data)
{
    size_t n_tok = 0;
1565
    int fd = -1;
1566 1567
    FILE *fp = NULL;
    char **v;
1568
    int ret = -1;
1569
    int i;
1570
    virCommandPtr cmd = NULL;
1571 1572 1573 1574

    if (n_columns == 0)
        return -1;

1575
    if (VIR_ALLOC_N(v, n_columns) < 0) {
1576
        virReportOOMError();
1577 1578 1579 1580 1581
        return -1;
    }
    for (i = 0; i < n_columns; i++)
        v[i] = NULL;

1582 1583 1584
    cmd = virCommandNewArgs(prog);
    virCommandSetOutputFD(cmd, &fd);
    if (virCommandRunAsync(cmd, NULL) < 0) {
1585 1586 1587
        goto cleanup;
    }

1588
    if ((fp = VIR_FDOPEN(fd, "r")) == NULL) {
1589
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
1590
                              "%s", _("cannot open file using fd"));
1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603
        goto cleanup;
    }

    while (1) {
        char *buf = NULL;
        size_t buf_len = 0;
        /* Be careful: even when it returns -1,
           this use of getdelim allocates memory.  */
        ssize_t tok_len = getdelim (&buf, &buf_len, 0, fp);
        v[n_tok] = buf;
        if (tok_len < 0) {
            /* Maybe EOF, maybe an error.
               If n_tok > 0, then we know it's an error.  */
1604
            if (n_tok && func (pool, n_tok, v, data) < 0)
1605 1606 1607 1608 1609
                goto cleanup;
            break;
        }
        ++n_tok;
        if (n_tok == n_columns) {
1610
            if (func (pool, n_tok, v, data) < 0)
1611 1612 1613
                goto cleanup;
            n_tok = 0;
            for (i = 0; i < n_columns; i++) {
1614
                VIR_FREE(v[i]);
1615 1616 1617 1618
            }
        }
    }

1619
    if (feof (fp) < 0) {
1620
        virReportSystemError(errno,
1621
                             _("read error on pipe to '%s'"), prog[0]);
1622 1623
        goto cleanup;
    }
1624

1625
    ret = virCommandWait(cmd, NULL);
1626 1627
 cleanup:
    for (i = 0; i < n_columns; i++)
1628 1629
        VIR_FREE(v[i]);
    VIR_FREE(v);
1630
    virCommandFree(cmd);
1631

1632 1633
    VIR_FORCE_FCLOSE(fp);
    VIR_FORCE_CLOSE(fd);
1634

1635
    return ret;
1636
}
D
Daniel P. Berrange 已提交
1637

1638
#else /* WIN32 */
D
Daniel P. Berrange 已提交
1639 1640 1641 1642 1643 1644 1645 1646 1647

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,
1648
                              void *data ATTRIBUTE_UNUSED)
D
Daniel P. Berrange 已提交
1649
{
1650 1651
    virStorageReportError(VIR_ERR_INTERNAL_ERROR,
                          _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662
    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)
{
1663
    virStorageReportError(VIR_ERR_INTERNAL_ERROR, _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1664 1665
    return -1;
}
1666
#endif /* WIN32 */