storage_backend.c 50.1 KB
Newer Older
1
/*
2
 * storage_backend.c: internal storage driver backend contract
3
 *
4
 * Copyright (C) 2007-2011 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
 * Copyright (C) 2007-2008 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307  USA
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

#include <config.h>

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

#if HAVE_SELINUX
45
# include <selinux/selinux.h>
46
#endif
47

48
#include "datatypes.h"
49
#include "virterror_internal.h"
D
Daniel P. Berrange 已提交
50 51
#include "util.h"
#include "memory.h"
52
#include "internal.h"
53
#include "secret_conf.h"
54
#include "uuid.h"
55
#include "storage_file.h"
D
Daniel P. Berrange 已提交
56
#include "storage_backend.h"
57
#include "logging.h"
58
#include "files.h"
59
#include "command.h"
60

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

80 81
#define VIR_FROM_THIS VIR_FROM_STORAGE

82 83 84 85 86 87 88 89 90 91 92 93 94 95
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
96 97 98
#if WITH_STORAGE_SCSI
    &virStorageBackendSCSI,
#endif
D
Dave Allan 已提交
99 100 101
#if WITH_STORAGE_MPATH
    &virStorageBackendMpath,
#endif
102 103 104 105 106
#if WITH_STORAGE_DISK
    &virStorageBackendDisk,
#endif
    NULL
};
107

108 109 110 111 112 113 114 115
static int track_allocation_progress = 0;

enum {
    TOOL_QEMU_IMG,
    TOOL_KVM_IMG,
    TOOL_QCOW_CREATE,
};

116 117 118
#define READ_BLOCK_SIZE_DEFAULT  (1024 * 1024)
#define WRITE_BLOCK_SIZE_DEFAULT (4 * 1024)

119 120
static int ATTRIBUTE_NONNULL (2)
virStorageBackendCopyToFD(virStorageVolDefPtr vol,
121 122
                          virStorageVolDefPtr inputvol,
                          int fd,
123 124
                          unsigned long long *total,
                          int is_dest_file)
125 126 127
{
    int inputfd = -1;
    int amtread = -1;
128
    int ret = 0;
129
    unsigned long long remain;
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();
164 165 166 167 168 169 170 171
        goto cleanup;
    }

    remain = *total;

    while (amtread != 0) {
        int amtleft;

172 173
        if (remain < rbytes)
            rbytes = remain;
174

175
        if ((amtread = saferead(inputfd, buf, rbytes)) < 0) {
176
            ret = -errno;
177
            virReportSystemError(errno,
178 179 180 181 182 183 184 185 186 187
                                 _("failed reading from file '%s'"),
                                 inputvol->target.path);
            goto cleanup;
        }
        remain -= amtread;

        /* Loop over amt read in 512 byte increments, looking for sparse
         * blocks */
        amtleft = amtread;
        do {
188
            interval = ((wbytes > amtleft) ? amtleft : wbytes);
189 190
            int offset = amtread - amtleft;

191
            if (is_dest_file && memcmp(buf+offset, zerobuf, interval) == 0) {
192
                if (lseek(fd, interval, SEEK_CUR) < 0) {
193
                    ret = -errno;
194
                    virReportSystemError(errno,
195 196 197 198 199
                                         _("cannot extend file '%s'"),
                                         vol->target.path);
                    goto cleanup;
                }
            } else if (safewrite(fd, buf+offset, interval) < 0) {
200
                ret = -errno;
201
                virReportSystemError(errno,
202 203 204 205 206
                                     _("failed writing to file '%s'"),
                                     vol->target.path);
                goto cleanup;

            }
207
        } while ((amtleft -= interval) > 0);
208 209
    }

210
    if (VIR_CLOSE(inputfd) < 0) {
211
        ret = -errno;
212
        virReportSystemError(errno,
213 214 215 216 217 218 219 220 221
                             _("cannot close file '%s'"),
                             inputvol->target.path);
        goto cleanup;
    }
    inputfd = -1;

    *total -= remain;

cleanup:
222
    VIR_FORCE_CLOSE(inputfd);
223

224
    VIR_FREE(zerobuf);
225 226
    VIR_FREE(buf);

227 228 229
    return ret;
}

230
static int
231
virStorageBackendCreateBlockFrom(virConnectPtr conn ATTRIBUTE_UNUSED,
232
                                 virStoragePoolObjPtr pool ATTRIBUTE_UNUSED,
233 234 235 236 237 238 239
                                 virStorageVolDefPtr vol,
                                 virStorageVolDefPtr inputvol,
                                 unsigned int flags ATTRIBUTE_UNUSED)
{
    int fd = -1;
    int ret = -1;
    unsigned long long remain;
240 241 242
    struct stat st;
    gid_t gid;
    uid_t uid;
243 244

    if ((fd = open(vol->target.path, O_RDWR)) < 0) {
245
        virReportSystemError(errno,
246 247 248 249 250 251 252 253
                             _("cannot create path '%s'"),
                             vol->target.path);
        goto cleanup;
    }

    remain = vol->allocation;

    if (inputvol) {
254
        int res = virStorageBackendCopyToFD(vol, inputvol,
255
                                            fd, &remain, 0);
256 257 258 259
        if (res < 0)
            goto cleanup;
    }

260
    if (fstat(fd, &st) == -1) {
261
        virReportSystemError(errno, _("stat of '%s' failed"),
262 263 264 265 266 267
                             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))
268
        && (fchown(fd, uid, gid) < 0)) {
269
        virReportSystemError(errno,
270
                             _("cannot chown '%s' to (%u, %u)"),
271
                             vol->target.path, uid, gid);
272 273 274
        goto cleanup;
    }
    if (fchmod(fd, vol->target.perms.mode) < 0) {
275
        virReportSystemError(errno,
276 277 278 279
                             _("cannot set mode of '%s' to %04o"),
                             vol->target.path, vol->target.perms.mode);
        goto cleanup;
    }
280
    if (VIR_CLOSE(fd) < 0) {
281
        virReportSystemError(errno,
282 283 284 285 286 287 288 289
                             _("cannot close file '%s'"),
                             vol->target.path);
        goto cleanup;
    }
    fd = -1;

    ret = 0;
cleanup:
290
    VIR_FORCE_CLOSE(fd);
291 292 293 294

    return ret;
}

E
Eric Blake 已提交
295 296 297 298
static int
createRawFile(int fd, virStorageVolDefPtr vol,
              virStorageVolDefPtr inputvol)
{
299 300
    int ret = 0;
    unsigned long long remain;
301

302 303
    /* Seek to the final size, so the capacity is available upfront
     * for progress reporting */
E
Eric Blake 已提交
304
    if (ftruncate(fd, vol->capacity) < 0) {
305
        ret = -errno;
306
        virReportSystemError(errno,
307
                             _("cannot extend file '%s'"),
E
Eric Blake 已提交
308
                             vol->target.path);
309 310 311
        goto cleanup;
    }

E
Eric Blake 已提交
312
    remain = vol->allocation;
313

E
Eric Blake 已提交
314 315
    if (inputvol) {
        ret = virStorageBackendCopyToFD(vol, inputvol, fd, &remain, 1);
316
        if (ret < 0) {
317
            goto cleanup;
318
        }
319 320 321 322 323 324 325 326 327 328 329 330 331 332
    }

    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
333
                if (safezero(fd, 0, vol->allocation - remain, bytes) != 0) {
334
                    ret = -errno;
J
Jiri Denemark 已提交
335
                    virReportSystemError(errno, _("cannot fill file '%s'"),
E
Eric Blake 已提交
336
                                         vol->target.path);
337 338 339 340 341
                    goto cleanup;
                }
                remain -= bytes;
            }
        } else { /* No progress bars to be shown */
342
            if (safezero(fd, 0, 0, remain) != 0) {
343
                ret = -errno;
J
Jiri Denemark 已提交
344
                virReportSystemError(errno, _("cannot fill file '%s'"),
E
Eric Blake 已提交
345
                                     vol->target.path);
346 347 348
                goto cleanup;
            }
        }
349

350 351 352
    }

    if (fsync(fd) < 0) {
353
        ret = -errno;
354
        virReportSystemError(errno, _("cannot sync data to file '%s'"),
E
Eric Blake 已提交
355
                             vol->target.path);
356
        goto cleanup;
357 358
    }

359 360 361 362 363 364 365 366 367
cleanup:
    return ret;
}

int
virStorageBackendCreateRaw(virConnectPtr conn ATTRIBUTE_UNUSED,
                           virStoragePoolObjPtr pool,
                           virStorageVolDefPtr vol,
                           virStorageVolDefPtr inputvol,
E
Eric Blake 已提交
368
                           unsigned int flags)
369 370
{
    int ret = -1;
E
Eric Blake 已提交
371 372 373 374 375 376
    int fd = -1;
    uid_t uid;
    gid_t gid;
    int operation_flags;

    virCheckFlags(0, -1);
377 378 379 380 381

    if (vol->target.encryption != NULL) {
        virStorageReportError(VIR_ERR_NO_SUPPORT,
                              "%s", _("storage pool does not support encrypted "
                                      "volumes"));
382 383
        goto cleanup;
    }
384

E
Eric Blake 已提交
385 386
    uid = (vol->target.perms.uid == -1) ? getuid() : vol->target.perms.uid;
    gid = (vol->target.perms.gid == -1) ? getgid() : vol->target.perms.gid;
387
    operation_flags = VIR_FILE_OPEN_FORCE_PERMS;
E
Eric Blake 已提交
388
    if (pool->def->type == VIR_STORAGE_POOL_NETFS)
389
        operation_flags |= VIR_FILE_OPEN_AS_UID;
E
Eric Blake 已提交
390

391 392 393 394
    if ((fd = virFileOpenAs(vol->target.path,
                            O_RDWR | O_CREAT | O_EXCL,
                            vol->target.perms.mode, uid, gid,
                            operation_flags)) < 0) {
E
Eric Blake 已提交
395 396 397 398 399 400
        virReportSystemError(-fd,
                             _("cannot create path '%s'"),
                             vol->target.path);
        goto cleanup;
    }

E
Eric Blake 已提交
401
    if ((ret = createRawFile(fd, vol, inputvol)) < 0) {
E
Eric Blake 已提交
402 403 404 405
        virReportSystemError(-fd,
                             _("cannot create path '%s'"),
                             vol->target.path);
        ret = -1;
406
    }
407 408

cleanup:
E
Eric Blake 已提交
409
    VIR_FORCE_CLOSE(fd);
410 411 412
    return ret;
}

413 414 415 416 417 418 419 420 421
static int
virStorageGenerateSecretUUID(virConnectPtr conn,
                             unsigned char *uuid)
{
    unsigned attempt;

    for (attempt = 0; attempt < 65536; attempt++) {
        virSecretPtr tmp;
        if (virUUIDGenerate(uuid) < 0) {
422 423
            virStorageReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                                  _("unable to generate uuid"));
424 425 426 427 428 429 430 431 432
            return -1;
        }
        tmp = conn->secretDriver->lookupByUUID(conn, uuid);
        if (tmp == NULL)
            return 0;

        virSecretFree(tmp);
    }

433 434
    virStorageReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                          _("too many conflicts when generating an uuid"));
435 436 437 438

    return -1;
}

439 440 441 442 443 444 445 446 447
static int
virStorageGenerateQcowEncryption(virConnectPtr conn,
                                 virStorageVolDefPtr vol)
{
    virSecretDefPtr def = NULL;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    virStorageEncryptionPtr enc;
    virStorageEncryptionSecretPtr enc_secret = NULL;
    virSecretPtr secret = NULL;
448
    char *xml;
449 450 451
    unsigned char value[VIR_STORAGE_QCOW_PASSPHRASE_SIZE];
    int ret = -1;

452 453 454
    if (conn->secretDriver == NULL ||
        conn->secretDriver->lookupByUUID == NULL ||
        conn->secretDriver->defineXML == NULL ||
455
        conn->secretDriver->setValue == NULL) {
456
        virStorageReportError(VIR_ERR_NO_SUPPORT, "%s",
457 458 459 460 461 462
                              _("secret storage not supported"));
        goto cleanup;
    }

    enc = vol->target.encryption;
    if (enc->nsecrets != 0) {
463
        virStorageReportError(VIR_ERR_INTERNAL_ERROR, "%s",
464 465 466 467 468 469
                              _("secrets already defined"));
        goto cleanup;
    }

    if (VIR_ALLOC(enc_secret) < 0 || VIR_REALLOC_N(enc->secrets, 1) < 0 ||
        VIR_ALLOC(def) < 0) {
470
        virReportOOMError();
471 472 473 474 475
        goto cleanup;
    }

    def->ephemeral = 0;
    def->private = 0;
476
    if (virStorageGenerateSecretUUID(conn, def->uuid) < 0)
477
        goto cleanup;
478

479 480 481
    def->usage_type = VIR_SECRET_USAGE_TYPE_VOLUME;
    def->usage.volume = strdup(vol->target.path);
    if (def->usage.volume == NULL) {
482
        virReportOOMError();
483 484
        goto cleanup;
    }
485
    xml = virSecretDefFormat(def);
486 487 488 489 490 491 492 493 494 495 496 497
    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);

498
    if (virStorageGenerateQcowPassphrase(value) < 0)
499 500 501 502 503 504
        goto cleanup;

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

    enc_secret->type = VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE;
505
    memcpy(enc_secret->uuid, secret->uuid, VIR_UUID_BUFLEN);
506 507 508 509 510 511 512 513 514
    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) {
515 516
        if (ret != 0 &&
            conn->secretDriver->undefine != NULL)
517 518 519
            conn->secretDriver->undefine(secret);
        virSecretFree(secret);
    }
520
    virBufferFreeAndReset(&buf);
521 522 523 524 525
    virSecretDefFree(def);
    VIR_FREE(enc_secret);
    return ret;
}

526 527 528
static int virStorageBuildSetUIDHook(void *data) {
    virStorageVolDefPtr vol = data;

529
    if ((vol->target.perms.gid != -1)
530
        && (setgid(vol->target.perms.gid) != 0)) {
531
        virReportSystemError(errno,
532 533 534 535
                             _("Cannot set gid to %u before creating %s"),
                             vol->target.perms.gid, vol->target.path);
        return -1;
    }
536
    if ((vol->target.perms.uid != -1)
537
        && (setuid(vol->target.perms.uid) != 0)) {
538
        virReportSystemError(errno,
539 540 541 542 543 544 545
                             _("Cannot set uid to %u before creating %s"),
                             vol->target.perms.uid, vol->target.path);
        return -1;
    }
    return 0;
}

546
static int virStorageBackendCreateExecCommand(virStoragePoolObjPtr pool,
547 548 549 550 551 552 553 554
                                              virStorageVolDefPtr vol,
                                              const char **cmdargv) {
    struct stat st;
    gid_t gid;
    uid_t uid;
    int filecreated = 0;

    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
        if (virRunWithHook(cmdargv,
561 562 563 564 565 566 567
                           virStorageBuildSetUIDHook, vol, NULL) == 0) {
            /* command was successfully run, check if the file was created */
            if (stat(vol->target.path, &st) >=0)
                filecreated = 1;
        }
    }
    if (!filecreated) {
568
        if (virRun(cmdargv, NULL) < 0) {
569 570 571
            return -1;
        }
        if (stat(vol->target.path, &st) < 0) {
572
            virReportSystemError(errno,
573 574 575 576 577 578 579 580 581 582
                                 _("%s failed to create %s"),
                                 cmdargv[0], vol->target.path);
            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)) {
583
        virReportSystemError(errno,
584
                             _("cannot chown %s to (%u, %u)"),
585
                             vol->target.path, uid, gid);
586 587 588
        return -1;
    }
    if (chmod(vol->target.path, vol->target.perms.mode) < 0) {
589
        virReportSystemError(errno,
590 591 592 593 594 595 596
                             _("cannot set mode of '%s' to %04o"),
                             vol->target.path, vol->target.perms.mode);
        return -1;
    }
    return 0;
}

597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620
enum {
    QEMU_IMG_BACKING_FORMAT_NONE = 0,
    QEMU_IMG_BACKING_FORMAT_FLAG,
    QEMU_IMG_BACKING_FORMAT_OPTIONS,
};

static int virStorageBackendQEMUImgBackingFormat(const char *qemuimg)
{
    const char *const qemuarg[] = { qemuimg, "-h", NULL };
    const char *const qemuenv[] = { "LC_ALL=C", NULL };
    pid_t child = 0;
    int status;
    int newstdout = -1;
    char *help = NULL;
    enum { MAX_HELP_OUTPUT_SIZE = 1024*8 };
    char *start;
    char *end;
    char *tmp;
    int ret = -1;

    if (virExec(qemuarg, qemuenv, NULL,
                &child, -1, &newstdout, NULL, VIR_EXEC_CLEAR_CAPS) < 0)
        goto cleanup;

E
Eric Blake 已提交
621
    if (virFileReadLimFD(newstdout, MAX_HELP_OUTPUT_SIZE, &help) < 0) {
622 623 624 625 626 627 628 629
        virReportSystemError(errno,
                             _("Unable to read '%s -h' output"),
                             qemuimg);
        goto cleanup;
    }

    start = strstr(help, " create ");
    end = strstr(start, "\n");
630 631
    if (((tmp = strstr(start, "-F fmt")) && tmp < end) ||
        ((tmp = strstr(start, "-F backing_fmt")) && tmp < end))
632 633 634 635 636 637 638 639
        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:
    VIR_FREE(help);
640
    VIR_FORCE_CLOSE(newstdout);
641
    if (child) {
642 643 644 645 646 647
        while (waitpid(child, &status, 0) == -1 && errno == EINTR);
        if (status) {
            tmp = virCommandTranslateStatus(status);
            VIR_WARN("Unexpected status, qemu probably failed: %s",
                     NULLSTR(tmp));
            VIR_FREE(tmp);
648 649 650 651 652 653 654
        }
    }

    return ret;
}


655 656
static int
virStorageBackendCreateQemuImg(virConnectPtr conn,
657
                               virStoragePoolObjPtr pool,
658
                               virStorageVolDefPtr vol,
659 660
                               virStorageVolDefPtr inputvol,
                               unsigned int flags ATTRIBUTE_UNUSED)
661
{
662
    int ret = -1;
E
Eric Blake 已提交
663
    char *size = NULL;
664
    char *create_tool;
665
    int imgformat = -1;
666

667
    const char *type = virStorageFileFormatTypeToString(vol->target.format);
668
    const char *backingType = vol->backingStore.path ?
669
        virStorageFileFormatTypeToString(vol->backingStore.format) : NULL;
670 671 672 673 674 675

    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 ?
676 677 678 679
        virStorageFileFormatTypeToString(inputvol->type == VIR_STORAGE_VOL_BLOCK ?
                                         VIR_STORAGE_FILE_RAW :
                                         inputvol->target.format) :
        NULL;
680 681

    if (type == NULL) {
682
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
683 684 685 686 687
                              _("unknown storage vol type %d"),
                              vol->target.format);
        return -1;
    }
    if (inputvol && inputType == NULL) {
688
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
689 690 691 692 693 694 695 696 697 698 699 700 701 702
                              _("unknown storage vol type %d"),
                              inputvol->target.format);
        return -1;
    }

    if (vol->backingStore.path) {

        /* 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,
704 705 706 707 708 709
                                  "%s", _("a different backing store can not "
                                          "be specified."));
            return -1;
        }

        if (backingType == NULL) {
710
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
711 712 713 714 715
                                  _("unknown storage vol backing store type %d"),
                                  vol->backingStore.format);
            return -1;
        }
        if (access(vol->backingStore.path, R_OK) != 0) {
716
            virReportSystemError(errno,
717 718 719 720 721 722
                                 _("inaccessible backing store volume %s"),
                                 vol->backingStore.path);
            return -1;
        }
    }

723
    if (vol->target.encryption != NULL) {
724 725
        virStorageEncryptionPtr enc;

726 727
        if (vol->target.format != VIR_STORAGE_FILE_QCOW &&
            vol->target.format != VIR_STORAGE_FILE_QCOW2) {
728
            virStorageReportError(VIR_ERR_NO_SUPPORT,
729 730 731 732
                                  _("qcow volume encryption unsupported with "
                                    "volume format %s"), type);
            return -1;
        }
733 734 735
        enc = vol->target.encryption;
        if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW &&
            enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT) {
736
            virStorageReportError(VIR_ERR_NO_SUPPORT,
737 738 739 740
                                  _("unsupported volume encryption format %d"),
                                  vol->target.encryption->format);
            return -1;
        }
741
        if (enc->nsecrets > 1) {
742
            virStorageReportError(VIR_ERR_XML_ERROR, "%s",
743 744 745
                                  _("too many secrets for qcow encryption"));
            return -1;
        }
746 747 748 749 750
        if (enc->format == VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT ||
            enc->nsecrets == 0) {
            if (virStorageGenerateQcowEncryption(conn, vol) < 0)
                return -1;
        }
751 752
    }

753
    /* Size in KB */
754
    if (virAsprintf(&size, "%lluK", VIR_DIV_UP(vol->capacity, 1024)) < 0) {
E
Eric Blake 已提交
755 756 757
        virReportOOMError();
        goto cleanup;
    }
758 759 760 761 762 763 764

    /* 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) {
765
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
766
                              "%s", _("unable to find kvm-img or qemu-img"));
767 768 769
        return -1;
    }

770 771 772 773
    imgformat = virStorageBackendQEMUImgBackingFormat(create_tool);
    if (imgformat < 0)
        goto cleanup;

774
    if (inputvol) {
775 776 777 778 779 780 781 782
        const char *imgargv[] = {
            create_tool,
            "convert",
            "-f", inputType,
            "-O", type,
            inputPath,
            vol->target.path,
            NULL,
783 784
            NULL,
            NULL
785 786
        };

787
        if (vol->target.encryption != NULL) {
788
            if (imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS) {
789 790 791 792 793 794 795
                imgargv[8] = "-o";
                imgargv[9] = "encryption=on";
            } else {
                imgargv[8] = "-e";
            }
        }

796
        ret = virStorageBackendCreateExecCommand(pool, vol, imgargv);
797
    } else if (vol->backingStore.path) {
798 799 800 801 802 803 804 805 806 807 808 809 810
        const char *imgargv[] = {
            create_tool,
            "create",
            "-f", type,
            "-b", vol->backingStore.path,
            NULL,
            NULL,
            NULL,
            NULL,
            NULL,
            NULL
        };

811
        char *optflag = NULL;
812 813 814 815 816 817
        switch (imgformat) {
        case QEMU_IMG_BACKING_FORMAT_FLAG:
            imgargv[6] = "-F";
            imgargv[7] = backingType;
            imgargv[8] = vol->target.path;
            imgargv[9] = size;
818
            if (vol->target.encryption != NULL)
819 820 821 822 823 824 825 826
                imgargv[10] = "-e";
            break;

        case QEMU_IMG_BACKING_FORMAT_OPTIONS:
            if (virAsprintf(&optflag, "backing_fmt=%s", backingType) < 0) {
                virReportOOMError();
                goto cleanup;
            }
827 828 829 830 831 832 833 834 835 836 837

            if (vol->target.encryption != NULL) {
                char *tmp = NULL;
                if (virAsprintf(&tmp, "%s,%s", optflag, "encryption=on") < 0) {
                    virReportOOMError();
                    goto cleanup;
                }
                VIR_FREE(optflag);
                optflag = tmp;
            }

838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853
            imgargv[6] = "-o";
            imgargv[7] = optflag;
            imgargv[8] = vol->target.path;
            imgargv[9] = size;

        default:
            VIR_INFO("Unable to set backing store format for %s with %s",
                     vol->target.path, create_tool);
            imgargv[6] = vol->target.path;
            imgargv[7] = size;
            if (vol->target.encryption != NULL)
                imgargv[8] = "-e";
        }

        ret = virStorageBackendCreateExecCommand(pool, vol, imgargv);
        VIR_FREE(optflag);
854
    } else {
855 856 857 858 859 860 861 862
        /* The extra NULL field is for indicating encryption (-e). */
        const char *imgargv[] = {
            create_tool,
            "create",
            "-f", type,
            vol->target.path,
            size,
            NULL,
863
            NULL,
864 865
            NULL
        };
866 867

        if (vol->target.encryption != NULL) {
868
            if (imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS) {
869 870 871 872 873 874
                imgargv[6] = "-o";
                imgargv[7] = "encryption=on";
            } else {
                imgargv[6] = "-e";
            }
        }
875

876 877
        ret = virStorageBackendCreateExecCommand(pool, vol, imgargv);
    }
878

879
    cleanup:
E
Eric Blake 已提交
880
    VIR_FREE(size);
881
    VIR_FREE(create_tool);
882

883
    return ret;
884 885 886 887 888 889 890
}

/*
 * Xen removed the fully-functional qemu-img, and replaced it
 * with a partially functional qcow-create. Go figure ??!?
 */
static int
891
virStorageBackendCreateQcowCreate(virConnectPtr conn ATTRIBUTE_UNUSED,
892
                                  virStoragePoolObjPtr pool,
893
                                  virStorageVolDefPtr vol,
894 895
                                  virStorageVolDefPtr inputvol,
                                  unsigned int flags ATTRIBUTE_UNUSED)
896
{
897
    int ret;
E
Eric Blake 已提交
898
    char *size;
899 900 901
    const char *imgargv[4];

    if (inputvol) {
902
        virStorageReportError(VIR_ERR_INTERNAL_ERROR, "%s",
903 904 905 906
                              _("cannot copy from volume with qcow-create"));
        return -1;
    }

907
    if (vol->target.format != VIR_STORAGE_FILE_QCOW2) {
908
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
909 910 911 912 913
                              _("unsupported storage vol type %d"),
                              vol->target.format);
        return -1;
    }
    if (vol->backingStore.path != NULL) {
914
        virStorageReportError(VIR_ERR_NO_SUPPORT, "%s",
915
                              _("copy-on-write image not supported with "
916
                                      "qcow-create"));
917 918
        return -1;
    }
919
    if (vol->target.encryption != NULL) {
920
        virStorageReportError(VIR_ERR_NO_SUPPORT,
921 922 923 924
                              "%s", _("encrypted volumes not supported with "
                                      "qcow-create"));
        return -1;
    }
925 926

    /* Size in MB - yes different units to qemu-img :-( */
927 928
    if (virAsprintf(&size, "%llu",
                    VIR_DIV_UP(vol->capacity, (1024 * 1024))) < 0) {
E
Eric Blake 已提交
929 930 931
        virReportOOMError();
        return -1;
    }
932 933 934 935 936 937

    imgargv[0] = virFindFileInPath("qcow-create");
    imgargv[1] = size;
    imgargv[2] = vol->target.path;
    imgargv[3] = NULL;

938
    ret = virStorageBackendCreateExecCommand(pool, vol, imgargv);
939
    VIR_FREE(imgargv[0]);
E
Eric Blake 已提交
940
    VIR_FREE(size);
941

942
    return ret;
943 944
}

945
virStorageBackendBuildVolFrom
946
virStorageBackendFSImageToolTypeToFunc(int tool_type)
947 948 949 950 951 952 953 954
{
    switch (tool_type) {
    case TOOL_KVM_IMG:
    case TOOL_QEMU_IMG:
        return virStorageBackendCreateQemuImg;
    case TOOL_QCOW_CREATE:
        return virStorageBackendCreateQcowCreate;
    default:
955
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984
                              _("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;
}

985
virStorageBackendBuildVolFrom
986
virStorageBackendGetBuildVolFromFunction(virStorageVolDefPtr vol,
987 988 989 990 991 992 993 994 995 996 997
                                         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 &&
998
         vol->target.format != VIR_STORAGE_FILE_RAW) ||
999
        (inputvol->type == VIR_STORAGE_VOL_FILE &&
1000
         inputvol->target.format != VIR_STORAGE_FILE_RAW)) {
1001

D
Daniel P. Berrange 已提交
1002
        if ((tool_type = virStorageBackendFindFSImageTool(NULL)) < 0) {
1003
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
1004 1005 1006 1007 1008
                                  "%s", _("creation of non-raw file images is "
                                          "not supported without qemu-img."));
            return NULL;
        }

1009
        return virStorageBackendFSImageToolTypeToFunc(tool_type);
1010 1011
    }

1012 1013 1014 1015
    if (vol->type == VIR_STORAGE_VOL_BLOCK)
        return virStorageBackendCreateBlockFrom;
    else
        return virStorageBackendCreateRaw;
1016
}
1017

1018

1019 1020
virStorageBackendPtr
virStorageBackendForType(int type) {
1021
    unsigned int i;
1022
    for (i = 0; backends[i]; i++)
1023 1024 1025
        if (backends[i]->type == type)
            return backends[i];

1026
    virStorageReportError(VIR_ERR_INTERNAL_ERROR,
1027 1028 1029 1030 1031
                          _("missing backend for pool type %d"), type);
    return NULL;
}


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

1044
    if ((fd = open(path, O_RDONLY|O_NONBLOCK|O_NOCTTY)) < 0) {
1045 1046 1047 1048 1049 1050
        if ((errno == ENOENT || errno == ELOOP) &&
            lstat(path, &sb) == 0) {
            VIR_WARN("ignoring dangling symlink '%s'", path);
            return -2;
        }

1051
        virReportSystemError(errno,
1052
                             _("cannot open volume '%s'"),
1053
                             path);
1054 1055 1056
        return -1;
    }

1057 1058 1059 1060
    if (fstat(fd, &sb) < 0) {
        virReportSystemError(errno,
                             _("cannot stat file '%s'"),
                             path);
1061
        VIR_FORCE_CLOSE(fd);
1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072
        return -1;
    }

    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;

    if (!(mode & flags)) {
1073
        VIR_FORCE_CLOSE(fd);
1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103

        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,
                                     unsigned long long *capacity)
{
    int ret, fd;

    if ((ret = virStorageBackendVolOpen(target->path)) < 0)
        return ret;

    fd = ret;
1104
    ret = virStorageBackendUpdateVolTargetInfoFD(target,
1105 1106 1107
                                                 fd,
                                                 allocation,
                                                 capacity);
1108

1109
    VIR_FORCE_CLOSE(fd);
1110 1111 1112 1113

    return ret;
}

1114
int
1115
virStorageBackendUpdateVolInfo(virStorageVolDefPtr vol,
1116 1117 1118 1119
                               int withCapacity)
{
    int ret;

1120
    if ((ret = virStorageBackendUpdateVolTargetInfo(&vol->target,
1121 1122 1123 1124 1125
                                                    &vol->allocation,
                                                    withCapacity ? &vol->capacity : NULL)) < 0)
        return ret;

    if (vol->backingStore.path &&
1126
        (ret = virStorageBackendUpdateVolTargetInfo(&vol->backingStore,
1127 1128 1129 1130 1131 1132
                                                    NULL, NULL)) < 0)
        return ret;

    return 0;
}

1133 1134 1135 1136
/*
 * virStorageBackendUpdateVolTargetInfoFD:
 * @conn: connection to report errors on
 * @target: target definition ptr of volume to update
1137
 * @fd: fd of storage volume to update, via virStorageBackendOpenVol*
1138 1139 1140
 * @allocation: If not NULL, updated allocation information will be stored
 * @capacity: If not NULL, updated capacity info will be stored
 *
1141
 * Returns 0 for success, -1 on a legitimate error condition.
1142
 */
1143
int
1144
virStorageBackendUpdateVolTargetInfoFD(virStorageVolTargetPtr target,
1145 1146 1147
                                       int fd,
                                       unsigned long long *allocation,
                                       unsigned long long *capacity)
1148 1149 1150 1151 1152 1153 1154
{
    struct stat sb;
#if HAVE_SELINUX
    security_context_t filecon = NULL;
#endif

    if (fstat(fd, &sb) < 0) {
1155
        virReportSystemError(errno,
1156
                             _("cannot stat file '%s'"),
1157
                             target->path);
1158 1159 1160
        return -1;
    }

1161 1162
    if (allocation) {
        if (S_ISREG(sb.st_mode)) {
1163
#ifndef WIN32
1164
            *allocation = (unsigned long long)sb.st_blocks *
1165
                          (unsigned long long)DEV_BSIZE;
D
Daniel P. Berrange 已提交
1166
#else
1167
            *allocation = sb.st_size;
D
Daniel P. Berrange 已提交
1168
#endif
1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182
            /* Regular files may be sparse, so logical size (capacity) is not same
             * as actual allocation above
             */
            if (capacity)
                *capacity = sb.st_size;
        } else {
            off_t end;
            /* XXX this is POSIX compliant, but doesn't work for for CHAR files,
             * 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) {
1183
                virReportSystemError(errno,
1184 1185 1186 1187 1188 1189 1190
                                     _("cannot seek to end of file '%s'"),
                                     target->path);
                return -1;
            }
            *allocation = end;
            if (capacity)
                *capacity = end;
1191 1192 1193
        }
    }

1194 1195 1196
    target->perms.mode = sb.st_mode & S_IRWXUGO;
    target->perms.uid = sb.st_uid;
    target->perms.gid = sb.st_gid;
1197

1198
    VIR_FREE(target->perms.label);
1199 1200

#if HAVE_SELINUX
1201
    /* XXX: make this a security driver call */
1202
    if (fgetfilecon(fd, &filecon) == -1) {
1203
        if (errno != ENODATA && errno != ENOTSUP) {
1204
            virReportSystemError(errno,
1205
                                 _("cannot get file context of '%s'"),
1206
                                 target->path);
1207 1208
            return -1;
        } else {
1209
            target->perms.label = NULL;
1210 1211
        }
    } else {
1212
        target->perms.label = strdup(filecon);
1213
        freecon(filecon);
1214
        if (target->perms.label == NULL) {
1215
            virReportOOMError();
1216 1217
            return -1;
        }
1218 1219
    }
#else
1220
    target->perms.label = NULL;
1221 1222 1223 1224 1225
#endif

    return 0;
}

D
Dave Allan 已提交
1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260

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
1261 1262
virStorageBackendDetectBlockVolFormatFD(virStorageVolTargetPtr target,
                                        int fd)
D
Dave Allan 已提交
1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273
{
    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) {
1274
        virReportSystemError(errno,
D
Dave Allan 已提交
1275 1276 1277 1278 1279 1280
                             _("cannot seek to beginning of file '%s'"),
                             target->path);
        return -1;
    }
    bytes = saferead(fd, buffer, sizeof(buffer));
    if (bytes < 0) {
1281
        virReportSystemError(errno,
D
Dave Allan 已提交
1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300
                             _("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;
}


1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314
/*
 * 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 *
1315
virStorageBackendStablePath(virStoragePoolObjPtr pool,
1316
                            const char *devpath)
1317 1318 1319
{
    DIR *dh;
    struct dirent *dent;
1320
    char *stablepath;
1321
    int opentries = 0;
1322 1323 1324 1325 1326

    /* 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/"))
1327
        goto ret_strdup;
1328

1329 1330 1331 1332 1333 1334
    /* 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;

1335 1336 1337
    /* 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.
1338
     */
1339
 reopen:
1340
    if ((dh = opendir(pool->def->target.path)) == NULL) {
1341 1342 1343 1344 1345
        opentries++;
        if (errno == ENOENT && opentries < 50) {
            usleep(100 * 1000);
            goto reopen;
        }
1346
        virReportSystemError(errno,
1347 1348
                             _("cannot read dir '%s'"),
                             pool->def->target.path);
1349 1350 1351
        return NULL;
    }

1352 1353 1354 1355 1356
    /* 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
     */
1357 1358 1359 1360
    while ((dent = readdir(dh)) != NULL) {
        if (dent->d_name[0] == '.')
            continue;

1361 1362 1363
        if (virAsprintf(&stablepath, "%s/%s",
                        pool->def->target.path,
                        dent->d_name) == -1) {
1364
            virReportOOMError();
1365 1366 1367 1368 1369 1370 1371 1372 1373
            closedir(dh);
            return NULL;
        }

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

1374
        VIR_FREE(stablepath);
1375 1376 1377 1378
    }

    closedir(dh);

1379
 ret_strdup:
1380 1381 1382
    /* Couldn't find any matching stable link so give back
     * the original non-stable dev path
     */
1383 1384 1385 1386

    stablepath = strdup(devpath);

    if (stablepath == NULL)
1387
        virReportOOMError();
1388 1389

    return stablepath;
1390 1391
}

D
Daniel P. Berrange 已提交
1392

1393
#ifndef WIN32
1394 1395 1396 1397
/*
 * Run an external program.
 *
 * Read its output and apply a series of regexes to each line
R
Richard W.M. Jones 已提交
1398
 * When the entire set of regexes has matched consecutively
1399 1400 1401
 * then run a callback passing in all the matches
 */
int
1402
virStorageBackendRunProgRegex(virStoragePoolObjPtr pool,
1403
                              const char *const*prog,
1404 1405 1406 1407
                              int nregex,
                              const char **regex,
                              int *nvars,
                              virStorageBackendListVolRegexFunc func,
1408 1409
                              void *data,
                              int *outexit)
1410
{
1411 1412
    int fd = -1, exitstatus, err, failed = 1;
    pid_t child = 0;
1413 1414 1415 1416 1417 1418 1419 1420 1421
    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 */
1422
    if (VIR_ALLOC_N(reg, nregex) < 0) {
1423
        virReportOOMError();
1424 1425 1426 1427 1428 1429 1430 1431
        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));
1432
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
1433 1434 1435
                                  _("Failed to compile regex %s"), error);
            for (j = 0 ; j <= i ; j++)
                regfree(&reg[j]);
1436
            VIR_FREE(reg);
1437 1438 1439 1440 1441 1442 1443 1444 1445 1446
            return -1;
        }

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

    }

    /* Storage for matched variables */
1447
    if (VIR_ALLOC_N(groups, totgroups) < 0) {
1448
        virReportOOMError();
1449 1450
        goto cleanup;
    }
1451
    if (VIR_ALLOC_N(vars, maxvars+1) < 0) {
1452
        virReportOOMError();
1453 1454 1455 1456 1457
        goto cleanup;
    }


    /* Run the program and capture its output */
1458
    if (virExec(prog, NULL, NULL,
1459
                &child, -1, &fd, NULL, VIR_EXEC_NONE) < 0) {
1460 1461 1462
        goto cleanup;
    }

1463
    if ((list = VIR_FDOPEN(fd, "r")) == NULL) {
1464
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
J
Jim Meyering 已提交
1465
                              "%s", _("cannot read fd"));
1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487
        goto cleanup;
    }

    while (fgets(line, sizeof(line), list) != NULL) {
        /* Strip trailing newline */
        int len = strlen(line);
        if (len && line[len-1] == '\n')
            line[len-1] = '\0';

        for (i = 0 ; i <= maxReg && i < nregex ; i++) {
            if (regexec(&reg[i], line, nvars[i]+1, vars, 0) == 0) {
                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 */
                    line[vars[j+1].rm_eo] = '\0';
                    if ((groups[ngroup++] =
                         strdup(line + vars[j+1].rm_so)) == NULL) {
1488
                        virReportOOMError();
1489 1490 1491 1492 1493 1494
                        goto cleanup;
                    }
                }

                /* We're matching on the last regex, so callback time */
                if (i == (nregex-1)) {
1495
                    if (((*func)(pool, groups, data)) < 0)
1496 1497 1498
                        goto cleanup;

                    /* Release matches & restart to matching the first regex */
1499
                    for (j = 0 ; j < totgroups ; j++)
1500
                        VIR_FREE(groups[j]);
1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512
                    maxReg = 0;
                    ngroup = 0;
                }
            }
        }
    }

    failed = 0;

 cleanup:
    if (groups) {
        for (j = 0 ; j < totgroups ; j++)
1513 1514
            VIR_FREE(groups[j]);
        VIR_FREE(groups);
1515
    }
1516
    VIR_FREE(vars);
1517 1518 1519 1520

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

1521
    VIR_FREE(reg);
1522

1523 1524
    VIR_FORCE_FCLOSE(list);
    VIR_FORCE_CLOSE(fd);
1525 1526 1527 1528 1529 1530 1531 1532

    while ((err = waitpid(child, &exitstatus, 0) == -1) && errno == EINTR);

    /* Don't bother checking exit status if we already failed */
    if (failed)
        return -1;

    if (err == -1) {
1533
        virReportSystemError(errno,
1534 1535
                             _("failed to wait for command '%s'"),
                             prog[0]);
1536 1537 1538
        return -1;
    } else {
        if (WIFEXITED(exitstatus)) {
1539 1540
            if (outexit != NULL)
                *outexit = WEXITSTATUS(exitstatus);
1541
        } else {
1542
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
J
Jim Meyering 已提交
1543
                                  "%s", _("command did not exit cleanly"));
1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562
            return -1;
        }
    }

    return 0;
}

/*
 * 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
1563
virStorageBackendRunProgNul(virStoragePoolObjPtr pool,
1564 1565 1566 1567 1568 1569
                            const char **prog,
                            size_t n_columns,
                            virStorageBackendListVolNulFunc func,
                            void *data)
{
    size_t n_tok = 0;
1570 1571
    int fd = -1, exitstatus;
    pid_t child = 0;
1572 1573 1574 1575 1576 1577 1578 1579 1580
    FILE *fp = NULL;
    char **v;
    int err = -1;
    int w_err;
    int i;

    if (n_columns == 0)
        return -1;

1581
    if (VIR_ALLOC_N(v, n_columns) < 0) {
1582
        virReportOOMError();
1583 1584 1585 1586 1587 1588
        return -1;
    }
    for (i = 0; i < n_columns; i++)
        v[i] = NULL;

    /* Run the program and capture its output */
1589
    if (virExec(prog, NULL, NULL,
1590
                &child, -1, &fd, NULL, VIR_EXEC_NONE) < 0) {
1591 1592 1593
        goto cleanup;
    }

1594
    if ((fp = VIR_FDOPEN(fd, "r")) == NULL) {
1595
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
1596
                              "%s", _("cannot open file using fd"));
1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609
        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.  */
1610
            if (n_tok && func (pool, n_tok, v, data) < 0)
1611 1612 1613 1614 1615
                goto cleanup;
            break;
        }
        ++n_tok;
        if (n_tok == n_columns) {
1616
            if (func (pool, n_tok, v, data) < 0)
1617 1618 1619
                goto cleanup;
            n_tok = 0;
            for (i = 0; i < n_columns; i++) {
1620
                VIR_FREE(v[i]);
1621 1622 1623 1624 1625 1626 1627
            }
        }
    }

    if (feof (fp))
        err = 0;
    else
1628
        virReportSystemError(errno,
1629
                             _("read error on pipe to '%s'"), prog[0]);
1630 1631 1632

 cleanup:
    for (i = 0; i < n_columns; i++)
1633 1634
        VIR_FREE(v[i]);
    VIR_FREE(v);
1635

1636 1637
    VIR_FORCE_FCLOSE(fp);
    VIR_FORCE_CLOSE(fd);
1638 1639 1640 1641 1642 1643 1644 1645 1646

    while ((w_err = waitpid (child, &exitstatus, 0) == -1) && errno == EINTR)
        /* empty */ ;

    /* Don't bother checking exit status if we already failed */
    if (err < 0)
        return -1;

    if (w_err == -1) {
1647
        virReportSystemError(errno,
1648 1649
                             _("failed to wait for command '%s'"),
                             prog[0]);
1650 1651 1652 1653
        return -1;
    } else {
        if (WIFEXITED(exitstatus)) {
            if (WEXITSTATUS(exitstatus) != 0) {
1654
                virStorageReportError(VIR_ERR_INTERNAL_ERROR,
1655 1656 1657 1658 1659
                                      _("non-zero exit status from command %d"),
                                      WEXITSTATUS(exitstatus));
                return -1;
            }
        } else {
1660
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
J
Jim Meyering 已提交
1661
                                  "%s", _("command did not exit cleanly"));
1662 1663 1664 1665 1666 1667
            return -1;
        }
    }

    return 0;
}
D
Daniel P. Berrange 已提交
1668

1669
#else /* WIN32 */
D
Daniel P. Berrange 已提交
1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681

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,
                              void *data ATTRIBUTE_UNUSED,
                              int *outexit ATTRIBUTE_UNUSED)
{
1682
    virStorageReportError(VIR_ERR_INTERNAL_ERROR, _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693
    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)
{
1694
    virStorageReportError(VIR_ERR_INTERNAL_ERROR, _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1695 1696
    return -1;
}
1697
#endif /* WIN32 */