storage_backend.c 50.6 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
                              _("unknown storage vol type %d"),
                              inputvol->target.format);
        return -1;
    }

    if (vol->backingStore.path) {
695 696
        int accessRetCode = -1;
        char *absolutePath = NULL;
697 698 699 700 701 702 703 704

        /* 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))) {
705
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
E
Eric Blake 已提交
706
                                  "%s", _("a different backing store cannot "
707 708 709 710 711
                                          "be specified."));
            return -1;
        }

        if (backingType == NULL) {
712
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
713 714 715 716
                                  _("unknown storage vol backing store type %d"),
                                  vol->backingStore.format);
            return -1;
        }
717 718 719 720 721 722 723 724 725 726 727 728 729 730

        /* Convert relative backing store paths to absolute paths for access
         * validation.
         */
        if ('/' != *(vol->backingStore.path) &&
            virAsprintf(&absolutePath, "%s/%s", pool->def->target.path,
                        vol->backingStore.path) < 0) {
            virReportOOMError();
            return -1;
        }
        accessRetCode = access(absolutePath ? absolutePath
                               : vol->backingStore.path, R_OK);
        VIR_FREE(absolutePath);
        if (accessRetCode != 0) {
731
            virReportSystemError(errno,
732 733 734 735 736 737
                                 _("inaccessible backing store volume %s"),
                                 vol->backingStore.path);
            return -1;
        }
    }

738
    if (vol->target.encryption != NULL) {
739 740
        virStorageEncryptionPtr enc;

741 742
        if (vol->target.format != VIR_STORAGE_FILE_QCOW &&
            vol->target.format != VIR_STORAGE_FILE_QCOW2) {
743
            virStorageReportError(VIR_ERR_NO_SUPPORT,
744 745 746 747
                                  _("qcow volume encryption unsupported with "
                                    "volume format %s"), type);
            return -1;
        }
748 749 750
        enc = vol->target.encryption;
        if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW &&
            enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT) {
751
            virStorageReportError(VIR_ERR_NO_SUPPORT,
752 753 754 755
                                  _("unsupported volume encryption format %d"),
                                  vol->target.encryption->format);
            return -1;
        }
756
        if (enc->nsecrets > 1) {
757
            virStorageReportError(VIR_ERR_XML_ERROR, "%s",
758 759 760
                                  _("too many secrets for qcow encryption"));
            return -1;
        }
761 762 763 764 765
        if (enc->format == VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT ||
            enc->nsecrets == 0) {
            if (virStorageGenerateQcowEncryption(conn, vol) < 0)
                return -1;
        }
766 767
    }

768
    /* Size in KB */
769
    if (virAsprintf(&size, "%lluK", VIR_DIV_UP(vol->capacity, 1024)) < 0) {
E
Eric Blake 已提交
770 771 772
        virReportOOMError();
        goto cleanup;
    }
773 774 775 776 777 778 779

    /* 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) {
780
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
781
                              "%s", _("unable to find kvm-img or qemu-img"));
782 783 784
        return -1;
    }

785 786 787 788
    imgformat = virStorageBackendQEMUImgBackingFormat(create_tool);
    if (imgformat < 0)
        goto cleanup;

789
    if (inputvol) {
790 791 792 793 794 795 796 797
        const char *imgargv[] = {
            create_tool,
            "convert",
            "-f", inputType,
            "-O", type,
            inputPath,
            vol->target.path,
            NULL,
798 799
            NULL,
            NULL
800 801
        };

802
        if (vol->target.encryption != NULL) {
803
            if (imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS) {
804 805 806 807 808 809 810
                imgargv[8] = "-o";
                imgargv[9] = "encryption=on";
            } else {
                imgargv[8] = "-e";
            }
        }

811
        ret = virStorageBackendCreateExecCommand(pool, vol, imgargv);
812
    } else if (vol->backingStore.path) {
813 814 815 816 817 818 819 820 821 822 823 824 825
        const char *imgargv[] = {
            create_tool,
            "create",
            "-f", type,
            "-b", vol->backingStore.path,
            NULL,
            NULL,
            NULL,
            NULL,
            NULL,
            NULL
        };

826
        char *optflag = NULL;
827 828 829 830 831 832
        switch (imgformat) {
        case QEMU_IMG_BACKING_FORMAT_FLAG:
            imgargv[6] = "-F";
            imgargv[7] = backingType;
            imgargv[8] = vol->target.path;
            imgargv[9] = size;
833
            if (vol->target.encryption != NULL)
834 835 836 837 838 839 840 841
                imgargv[10] = "-e";
            break;

        case QEMU_IMG_BACKING_FORMAT_OPTIONS:
            if (virAsprintf(&optflag, "backing_fmt=%s", backingType) < 0) {
                virReportOOMError();
                goto cleanup;
            }
842 843 844 845 846 847 848 849 850 851 852

            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;
            }

853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868
            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);
869
    } else {
870 871 872 873 874 875 876 877
        /* The extra NULL field is for indicating encryption (-e). */
        const char *imgargv[] = {
            create_tool,
            "create",
            "-f", type,
            vol->target.path,
            size,
            NULL,
878
            NULL,
879 880
            NULL
        };
881 882

        if (vol->target.encryption != NULL) {
883
            if (imgformat == QEMU_IMG_BACKING_FORMAT_OPTIONS) {
884 885 886 887 888 889
                imgargv[6] = "-o";
                imgargv[7] = "encryption=on";
            } else {
                imgargv[6] = "-e";
            }
        }
890

891 892
        ret = virStorageBackendCreateExecCommand(pool, vol, imgargv);
    }
893

894
    cleanup:
E
Eric Blake 已提交
895
    VIR_FREE(size);
896
    VIR_FREE(create_tool);
897

898
    return ret;
899 900 901 902 903 904 905
}

/*
 * Xen removed the fully-functional qemu-img, and replaced it
 * with a partially functional qcow-create. Go figure ??!?
 */
static int
906
virStorageBackendCreateQcowCreate(virConnectPtr conn ATTRIBUTE_UNUSED,
907
                                  virStoragePoolObjPtr pool,
908
                                  virStorageVolDefPtr vol,
909 910
                                  virStorageVolDefPtr inputvol,
                                  unsigned int flags ATTRIBUTE_UNUSED)
911
{
912
    int ret;
E
Eric Blake 已提交
913
    char *size;
914 915 916
    const char *imgargv[4];

    if (inputvol) {
917
        virStorageReportError(VIR_ERR_INTERNAL_ERROR, "%s",
918 919 920 921
                              _("cannot copy from volume with qcow-create"));
        return -1;
    }

922
    if (vol->target.format != VIR_STORAGE_FILE_QCOW2) {
923
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
924 925 926 927 928
                              _("unsupported storage vol type %d"),
                              vol->target.format);
        return -1;
    }
    if (vol->backingStore.path != NULL) {
929
        virStorageReportError(VIR_ERR_NO_SUPPORT, "%s",
930
                              _("copy-on-write image not supported with "
931
                                      "qcow-create"));
932 933
        return -1;
    }
934
    if (vol->target.encryption != NULL) {
935
        virStorageReportError(VIR_ERR_NO_SUPPORT,
936 937 938 939
                              "%s", _("encrypted volumes not supported with "
                                      "qcow-create"));
        return -1;
    }
940 941

    /* Size in MB - yes different units to qemu-img :-( */
942 943
    if (virAsprintf(&size, "%llu",
                    VIR_DIV_UP(vol->capacity, (1024 * 1024))) < 0) {
E
Eric Blake 已提交
944 945 946
        virReportOOMError();
        return -1;
    }
947 948 949 950 951 952

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

953
    ret = virStorageBackendCreateExecCommand(pool, vol, imgargv);
954
    VIR_FREE(imgargv[0]);
E
Eric Blake 已提交
955
    VIR_FREE(size);
956

957
    return ret;
958 959
}

960
virStorageBackendBuildVolFrom
961
virStorageBackendFSImageToolTypeToFunc(int tool_type)
962 963 964 965 966 967 968 969
{
    switch (tool_type) {
    case TOOL_KVM_IMG:
    case TOOL_QEMU_IMG:
        return virStorageBackendCreateQemuImg;
    case TOOL_QCOW_CREATE:
        return virStorageBackendCreateQcowCreate;
    default:
970
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999
                              _("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;
}

1000
virStorageBackendBuildVolFrom
1001
virStorageBackendGetBuildVolFromFunction(virStorageVolDefPtr vol,
1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012
                                         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 &&
1013
         vol->target.format != VIR_STORAGE_FILE_RAW) ||
1014
        (inputvol->type == VIR_STORAGE_VOL_FILE &&
1015
         inputvol->target.format != VIR_STORAGE_FILE_RAW)) {
1016

D
Daniel P. Berrange 已提交
1017
        if ((tool_type = virStorageBackendFindFSImageTool(NULL)) < 0) {
1018
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
1019 1020 1021 1022 1023
                                  "%s", _("creation of non-raw file images is "
                                          "not supported without qemu-img."));
            return NULL;
        }

1024
        return virStorageBackendFSImageToolTypeToFunc(tool_type);
1025 1026
    }

1027 1028 1029 1030
    if (vol->type == VIR_STORAGE_VOL_BLOCK)
        return virStorageBackendCreateBlockFrom;
    else
        return virStorageBackendCreateRaw;
1031
}
1032

1033

1034 1035
virStorageBackendPtr
virStorageBackendForType(int type) {
1036
    unsigned int i;
1037
    for (i = 0; backends[i]; i++)
1038 1039 1040
        if (backends[i]->type == type)
            return backends[i];

1041
    virStorageReportError(VIR_ERR_INTERNAL_ERROR,
1042 1043 1044 1045 1046
                          _("missing backend for pool type %d"), type);
    return NULL;
}


1047 1048 1049
/*
 * Allows caller to silently ignore files with improper mode
 *
1050 1051
 * Returns -1 on error, -2 if file mode is unexpected or the
 * volume is a dangling symbolic link.
1052
 */
1053
int
1054
virStorageBackendVolOpenCheckMode(const char *path, unsigned int flags)
1055
{
1056 1057
    int fd, mode = 0;
    struct stat sb;
1058

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

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

1072 1073 1074 1075
    if (fstat(fd, &sb) < 0) {
        virReportSystemError(errno,
                             _("cannot stat file '%s'"),
                             path);
1076
        VIR_FORCE_CLOSE(fd);
1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087
        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)) {
1088
        VIR_FORCE_CLOSE(fd);
1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118

        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;
1119
    ret = virStorageBackendUpdateVolTargetInfoFD(target,
1120 1121 1122
                                                 fd,
                                                 allocation,
                                                 capacity);
1123

1124
    VIR_FORCE_CLOSE(fd);
1125 1126 1127 1128

    return ret;
}

1129
int
1130
virStorageBackendUpdateVolInfo(virStorageVolDefPtr vol,
1131 1132 1133 1134
                               int withCapacity)
{
    int ret;

1135
    if ((ret = virStorageBackendUpdateVolTargetInfo(&vol->target,
1136 1137 1138 1139 1140
                                                    &vol->allocation,
                                                    withCapacity ? &vol->capacity : NULL)) < 0)
        return ret;

    if (vol->backingStore.path &&
1141
        (ret = virStorageBackendUpdateVolTargetInfo(&vol->backingStore,
1142 1143 1144 1145 1146 1147
                                                    NULL, NULL)) < 0)
        return ret;

    return 0;
}

1148 1149 1150 1151
/*
 * virStorageBackendUpdateVolTargetInfoFD:
 * @conn: connection to report errors on
 * @target: target definition ptr of volume to update
1152
 * @fd: fd of storage volume to update, via virStorageBackendOpenVol*
1153 1154 1155
 * @allocation: If not NULL, updated allocation information will be stored
 * @capacity: If not NULL, updated capacity info will be stored
 *
1156
 * Returns 0 for success, -1 on a legitimate error condition.
1157
 */
1158
int
1159
virStorageBackendUpdateVolTargetInfoFD(virStorageVolTargetPtr target,
1160 1161 1162
                                       int fd,
                                       unsigned long long *allocation,
                                       unsigned long long *capacity)
1163 1164 1165 1166 1167 1168 1169
{
    struct stat sb;
#if HAVE_SELINUX
    security_context_t filecon = NULL;
#endif

    if (fstat(fd, &sb) < 0) {
1170
        virReportSystemError(errno,
1171
                             _("cannot stat file '%s'"),
1172
                             target->path);
1173 1174 1175
        return -1;
    }

1176 1177
    if (allocation) {
        if (S_ISREG(sb.st_mode)) {
1178
#ifndef WIN32
1179
            *allocation = (unsigned long long)sb.st_blocks *
1180
                          (unsigned long long)DEV_BSIZE;
D
Daniel P. Berrange 已提交
1181
#else
1182
            *allocation = sb.st_size;
D
Daniel P. Berrange 已提交
1183
#endif
1184 1185 1186 1187 1188 1189 1190
            /* 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;
E
Eric Blake 已提交
1191
            /* XXX this is POSIX compliant, but doesn't work for CHAR files,
1192 1193 1194 1195 1196 1197
             * 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) {
1198
                virReportSystemError(errno,
1199 1200 1201 1202 1203 1204 1205
                                     _("cannot seek to end of file '%s'"),
                                     target->path);
                return -1;
            }
            *allocation = end;
            if (capacity)
                *capacity = end;
1206 1207 1208
        }
    }

1209 1210 1211
    target->perms.mode = sb.st_mode & S_IRWXUGO;
    target->perms.uid = sb.st_uid;
    target->perms.gid = sb.st_gid;
1212

1213
    VIR_FREE(target->perms.label);
1214 1215

#if HAVE_SELINUX
1216
    /* XXX: make this a security driver call */
1217
    if (fgetfilecon(fd, &filecon) == -1) {
1218
        if (errno != ENODATA && errno != ENOTSUP) {
1219
            virReportSystemError(errno,
1220
                                 _("cannot get file context of '%s'"),
1221
                                 target->path);
1222 1223
            return -1;
        } else {
1224
            target->perms.label = NULL;
1225 1226
        }
    } else {
1227
        target->perms.label = strdup(filecon);
1228
        freecon(filecon);
1229
        if (target->perms.label == NULL) {
1230
            virReportOOMError();
1231 1232
            return -1;
        }
1233 1234
    }
#else
1235
    target->perms.label = NULL;
1236 1237 1238 1239 1240
#endif

    return 0;
}

D
Dave Allan 已提交
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 1269 1270 1271 1272 1273 1274 1275

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
1276 1277
virStorageBackendDetectBlockVolFormatFD(virStorageVolTargetPtr target,
                                        int fd)
D
Dave Allan 已提交
1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288
{
    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) {
1289
        virReportSystemError(errno,
D
Dave Allan 已提交
1290 1291 1292 1293 1294 1295
                             _("cannot seek to beginning of file '%s'"),
                             target->path);
        return -1;
    }
    bytes = saferead(fd, buffer, sizeof(buffer));
    if (bytes < 0) {
1296
        virReportSystemError(errno,
D
Dave Allan 已提交
1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315
                             _("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;
}


1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329
/*
 * 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 *
1330
virStorageBackendStablePath(virStoragePoolObjPtr pool,
1331
                            const char *devpath)
1332 1333 1334
{
    DIR *dh;
    struct dirent *dent;
1335
    char *stablepath;
1336
    int opentries = 0;
1337 1338 1339 1340 1341

    /* 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/"))
1342
        goto ret_strdup;
1343

1344 1345 1346 1347 1348 1349
    /* 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;

1350 1351 1352
    /* 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.
1353
     */
1354
 reopen:
1355
    if ((dh = opendir(pool->def->target.path)) == NULL) {
1356 1357 1358 1359 1360
        opentries++;
        if (errno == ENOENT && opentries < 50) {
            usleep(100 * 1000);
            goto reopen;
        }
1361
        virReportSystemError(errno,
1362 1363
                             _("cannot read dir '%s'"),
                             pool->def->target.path);
1364 1365 1366
        return NULL;
    }

1367 1368 1369 1370 1371
    /* 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
     */
1372 1373 1374 1375
    while ((dent = readdir(dh)) != NULL) {
        if (dent->d_name[0] == '.')
            continue;

1376 1377 1378
        if (virAsprintf(&stablepath, "%s/%s",
                        pool->def->target.path,
                        dent->d_name) == -1) {
1379
            virReportOOMError();
1380 1381 1382 1383 1384 1385 1386 1387 1388
            closedir(dh);
            return NULL;
        }

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

1389
        VIR_FREE(stablepath);
1390 1391 1392 1393
    }

    closedir(dh);

1394
 ret_strdup:
1395 1396 1397
    /* Couldn't find any matching stable link so give back
     * the original non-stable dev path
     */
1398 1399 1400 1401

    stablepath = strdup(devpath);

    if (stablepath == NULL)
1402
        virReportOOMError();
1403 1404

    return stablepath;
1405 1406
}

D
Daniel P. Berrange 已提交
1407

1408
#ifndef WIN32
1409 1410 1411 1412
/*
 * Run an external program.
 *
 * Read its output and apply a series of regexes to each line
R
Richard W.M. Jones 已提交
1413
 * When the entire set of regexes has matched consecutively
1414 1415 1416
 * then run a callback passing in all the matches
 */
int
1417
virStorageBackendRunProgRegex(virStoragePoolObjPtr pool,
1418
                              const char *const*prog,
1419 1420 1421 1422
                              int nregex,
                              const char **regex,
                              int *nvars,
                              virStorageBackendListVolRegexFunc func,
1423 1424
                              void *data,
                              int *outexit)
1425
{
1426 1427
    int fd = -1, exitstatus, err, failed = 1;
    pid_t child = 0;
1428 1429 1430 1431 1432 1433 1434 1435 1436
    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 */
1437
    if (VIR_ALLOC_N(reg, nregex) < 0) {
1438
        virReportOOMError();
1439 1440 1441 1442 1443 1444 1445 1446
        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));
1447
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
1448 1449 1450
                                  _("Failed to compile regex %s"), error);
            for (j = 0 ; j <= i ; j++)
                regfree(&reg[j]);
1451
            VIR_FREE(reg);
1452 1453 1454 1455 1456 1457 1458 1459 1460 1461
            return -1;
        }

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

    }

    /* Storage for matched variables */
1462
    if (VIR_ALLOC_N(groups, totgroups) < 0) {
1463
        virReportOOMError();
1464 1465
        goto cleanup;
    }
1466
    if (VIR_ALLOC_N(vars, maxvars+1) < 0) {
1467
        virReportOOMError();
1468 1469 1470 1471 1472
        goto cleanup;
    }


    /* Run the program and capture its output */
1473
    if (virExec(prog, NULL, NULL,
1474
                &child, -1, &fd, NULL, VIR_EXEC_NONE) < 0) {
1475 1476 1477
        goto cleanup;
    }

1478
    if ((list = VIR_FDOPEN(fd, "r")) == NULL) {
1479
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
J
Jim Meyering 已提交
1480
                              "%s", _("cannot read fd"));
1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502
        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) {
1503
                        virReportOOMError();
1504 1505 1506 1507 1508 1509
                        goto cleanup;
                    }
                }

                /* We're matching on the last regex, so callback time */
                if (i == (nregex-1)) {
1510
                    if (((*func)(pool, groups, data)) < 0)
1511 1512 1513
                        goto cleanup;

                    /* Release matches & restart to matching the first regex */
1514
                    for (j = 0 ; j < totgroups ; j++)
1515
                        VIR_FREE(groups[j]);
1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527
                    maxReg = 0;
                    ngroup = 0;
                }
            }
        }
    }

    failed = 0;

 cleanup:
    if (groups) {
        for (j = 0 ; j < totgroups ; j++)
1528 1529
            VIR_FREE(groups[j]);
        VIR_FREE(groups);
1530
    }
1531
    VIR_FREE(vars);
1532 1533 1534 1535

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

1536
    VIR_FREE(reg);
1537

1538 1539
    VIR_FORCE_FCLOSE(list);
    VIR_FORCE_CLOSE(fd);
1540 1541 1542 1543 1544 1545 1546 1547

    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) {
1548
        virReportSystemError(errno,
1549 1550
                             _("failed to wait for command '%s'"),
                             prog[0]);
1551 1552 1553
        return -1;
    } else {
        if (WIFEXITED(exitstatus)) {
1554 1555
            if (outexit != NULL)
                *outexit = WEXITSTATUS(exitstatus);
1556
        } else {
1557
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
J
Jim Meyering 已提交
1558
                                  "%s", _("command did not exit cleanly"));
1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577
            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
1578
virStorageBackendRunProgNul(virStoragePoolObjPtr pool,
1579 1580 1581 1582 1583 1584
                            const char **prog,
                            size_t n_columns,
                            virStorageBackendListVolNulFunc func,
                            void *data)
{
    size_t n_tok = 0;
1585 1586
    int fd = -1, exitstatus;
    pid_t child = 0;
1587 1588 1589 1590 1591 1592 1593 1594 1595
    FILE *fp = NULL;
    char **v;
    int err = -1;
    int w_err;
    int i;

    if (n_columns == 0)
        return -1;

1596
    if (VIR_ALLOC_N(v, n_columns) < 0) {
1597
        virReportOOMError();
1598 1599 1600 1601 1602 1603
        return -1;
    }
    for (i = 0; i < n_columns; i++)
        v[i] = NULL;

    /* Run the program and capture its output */
1604
    if (virExec(prog, NULL, NULL,
1605
                &child, -1, &fd, NULL, VIR_EXEC_NONE) < 0) {
1606 1607 1608
        goto cleanup;
    }

1609
    if ((fp = VIR_FDOPEN(fd, "r")) == NULL) {
1610
        virStorageReportError(VIR_ERR_INTERNAL_ERROR,
1611
                              "%s", _("cannot open file using fd"));
1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624
        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.  */
1625
            if (n_tok && func (pool, n_tok, v, data) < 0)
1626 1627 1628 1629 1630
                goto cleanup;
            break;
        }
        ++n_tok;
        if (n_tok == n_columns) {
1631
            if (func (pool, n_tok, v, data) < 0)
1632 1633 1634
                goto cleanup;
            n_tok = 0;
            for (i = 0; i < n_columns; i++) {
1635
                VIR_FREE(v[i]);
1636 1637 1638 1639 1640 1641 1642
            }
        }
    }

    if (feof (fp))
        err = 0;
    else
1643
        virReportSystemError(errno,
1644
                             _("read error on pipe to '%s'"), prog[0]);
1645 1646 1647

 cleanup:
    for (i = 0; i < n_columns; i++)
1648 1649
        VIR_FREE(v[i]);
    VIR_FREE(v);
1650

1651 1652
    VIR_FORCE_FCLOSE(fp);
    VIR_FORCE_CLOSE(fd);
1653 1654 1655 1656 1657 1658 1659 1660 1661

    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) {
1662
        virReportSystemError(errno,
1663 1664
                             _("failed to wait for command '%s'"),
                             prog[0]);
1665 1666 1667 1668
        return -1;
    } else {
        if (WIFEXITED(exitstatus)) {
            if (WEXITSTATUS(exitstatus) != 0) {
1669
                virStorageReportError(VIR_ERR_INTERNAL_ERROR,
1670 1671 1672 1673 1674
                                      _("non-zero exit status from command %d"),
                                      WEXITSTATUS(exitstatus));
                return -1;
            }
        } else {
1675
            virStorageReportError(VIR_ERR_INTERNAL_ERROR,
J
Jim Meyering 已提交
1676
                                  "%s", _("command did not exit cleanly"));
1677 1678 1679 1680 1681 1682
            return -1;
        }
    }

    return 0;
}
D
Daniel P. Berrange 已提交
1683

1684
#else /* WIN32 */
D
Daniel P. Berrange 已提交
1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696

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)
{
1697
    virStorageReportError(VIR_ERR_INTERNAL_ERROR, _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708
    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)
{
1709
    virStorageReportError(VIR_ERR_INTERNAL_ERROR, _("%s not implemented on Win32"), __FUNCTION__);
D
Daniel P. Berrange 已提交
1710 1711
    return -1;
}
1712
#endif /* WIN32 */