lxc_controller.c 46.2 KB
Newer Older
1
/*
2
 * Copyright (C) 2010-2012 Red Hat, Inc.
3
 * Copyright IBM Corp. 2008
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
 *
 * lxc_controller.c: linux container process controller
 *
 * Authors:
 *  David L. Leskovec <dlesko at linux.vnet.ibm.com>
 *
 * 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
21
 * License along with this library.  If not, see
O
Osier Yang 已提交
22
 * <http://www.gnu.org/licenses/>.
23 24 25 26 27
 */

#include <config.h>

#include <sys/epoll.h>
28 29
#include <sys/wait.h>
#include <sys/socket.h>
30 31
#include <sys/types.h>
#include <sys/un.h>
32 33
#include <sys/utsname.h>
#include <sys/personality.h>
34
#include <unistd.h>
35
#include <paths.h>
36
#include <errno.h>
37 38
#include <fcntl.h>
#include <signal.h>
39
#include <getopt.h>
40
#include <sys/mount.h>
E
Eric Blake 已提交
41
#include <locale.h>
42 43
#include <grp.h>
#include <sys/stat.h>
44
#include <time.h>
45

D
Daniel P. Berrange 已提交
46
#if HAVE_CAPNG
47
# include <cap-ng.h>
D
Daniel P. Berrange 已提交
48 49
#endif

50 51 52 53 54
#if HAVE_NUMACTL
# define NUMA_VERSION1_COMPATIBILITY 1
# include <numa.h>
#endif

55
#include "virterror_internal.h"
56
#include "logging.h"
57 58 59
#include "util.h"

#include "lxc_conf.h"
60
#include "lxc_container.h"
61
#include "lxc_cgroup.h"
62
#include "lxc_protocol.h"
63 64
#include "virnetdev.h"
#include "virnetdevveth.h"
65 66
#include "memory.h"
#include "util.h"
E
Eric Blake 已提交
67
#include "virfile.h"
68
#include "virpidfile.h"
69
#include "command.h"
70 71
#include "processinfo.h"
#include "nodeinfo.h"
72
#include "virrandom.h"
73
#include "rpc/virnetserver.h"
74

75 76
#define VIR_FROM_THIS VIR_FROM_LXC

77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100
typedef struct _virLXCControllerConsole virLXCControllerConsole;
typedef virLXCControllerConsole *virLXCControllerConsolePtr;
struct _virLXCControllerConsole {
    int hostWatch;
    int hostFd;  /* PTY FD in the host OS */
    bool hostClosed;
    int hostEpoll;
    bool hostBlocking;

    int contWatch;
    int contFd;  /* PTY FD in the container */
    bool contClosed;
    int contEpoll;
    bool contBlocking;

    int epollWatch;
    int epollFd; /* epoll FD for dealing with EOF */

    size_t fromHostLen;
    char fromHostBuf[1024];
    size_t fromContLen;
    char fromContBuf[1024];
};

101 102 103 104 105
typedef struct _virLXCController virLXCController;
typedef virLXCController *virLXCControllerPtr;
struct _virLXCController {
    char *name;
    virDomainDefPtr def;
106

107 108
    int handshakeFd;

109 110
    pid_t initpid;

111 112
    size_t nveths;
    char **veths;
113 114 115

    size_t nconsoles;
    virLXCControllerConsolePtr consoles;
116
    char *devptmx;
117 118 119

    size_t nloopDevs;
    int *loopDevFds;
120 121

    virSecurityManagerPtr securityManager;
122

123 124
    /* Server socket */
    virNetServerPtr server;
125 126 127 128
    virNetServerClientPtr client;
    virNetServerProgramPtr prog;
    bool inShutdown;
    int timerShutdown;
129 130
};

131 132
#include "lxc_controller_dispatch.h"

133 134
static void virLXCControllerFree(virLXCControllerPtr ctrl);

135 136 137 138 139 140 141 142 143
static void virLXCControllerQuitTimer(int timer ATTRIBUTE_UNUSED, void *opaque)
{
    virLXCControllerPtr ctrl = opaque;

    VIR_DEBUG("Triggering event loop quit");
    virNetServerQuit(ctrl->server);
}


144 145 146 147 148 149 150 151 152
static virLXCControllerPtr virLXCControllerNew(const char *name)
{
    virLXCControllerPtr ctrl = NULL;
    virCapsPtr caps = NULL;
    char *configFile = NULL;

    if (VIR_ALLOC(ctrl) < 0)
        goto no_memory;

153 154
    ctrl->timerShutdown = -1;

155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170
    if (!(ctrl->name = strdup(name)))
        goto no_memory;

    if ((caps = lxcCapsInit(NULL)) == NULL)
        goto error;

    if ((configFile = virDomainConfigFile(LXC_STATE_DIR,
                                          ctrl->name)) == NULL)
        goto error;

    if ((ctrl->def = virDomainDefParseFile(caps,
                                           configFile,
                                           1 << VIR_DOMAIN_VIRT_LXC,
                                           0)) == NULL)
        goto error;

171 172 173 174 175
    if ((ctrl->timerShutdown = virEventAddTimeout(-1,
                                                  virLXCControllerQuitTimer, ctrl,
                                                  NULL)) < 0)
        goto error;

176 177 178 179 180 181 182 183 184 185 186 187 188
cleanup:
    VIR_FREE(configFile);
    virCapabilitiesFree(caps);
    return ctrl;

no_memory:
    virReportOOMError();
error:
    virLXCControllerFree(ctrl);
    ctrl = NULL;
    goto cleanup;
}

189

190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210
static int virLXCControllerCloseLoopDevices(virLXCControllerPtr ctrl,
                                            bool force)
{
    size_t i;

    for (i = 0 ; i < ctrl->nloopDevs ; i++) {
        if (force) {
            VIR_FORCE_CLOSE(ctrl->loopDevFds[i]);
        } else {
            if (VIR_CLOSE(ctrl->loopDevFds[i]) < 0) {
                virReportSystemError(errno, "%s",
                                     _("Unable to close loop device"));
                return -1;
            }
        }
    }

    return 0;
}


211 212 213 214 215
static void virLXCControllerStopInit(virLXCControllerPtr ctrl)
{
    if (ctrl->initpid == 0)
        return;

216
    virLXCControllerCloseLoopDevices(ctrl, true);
217 218 219 220 221
    virPidAbort(ctrl->initpid);
    ctrl->initpid = 0;
}


222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237
static void virLXCControllerConsoleClose(virLXCControllerConsolePtr console)
{
    if (console->hostWatch != -1)
        virEventRemoveHandle(console->hostWatch);
    VIR_FORCE_CLOSE(console->hostFd);

    if (console->contWatch != -1)
        virEventRemoveHandle(console->contWatch);
    VIR_FORCE_CLOSE(console->contFd);

    if (console->epollWatch != -1)
        virEventRemoveHandle(console->epollWatch);
    VIR_FORCE_CLOSE(console->epollFd);
}


238 239
static void virLXCControllerFree(virLXCControllerPtr ctrl)
{
240 241
    size_t i;

242 243 244
    if (!ctrl)
        return;

245 246
    virLXCControllerStopInit(ctrl);

247 248
    virSecurityManagerFree(ctrl->securityManager);

249 250 251 252
    for (i = 0 ; i < ctrl->nveths ; i++)
        VIR_FREE(ctrl->veths[i]);
    VIR_FREE(ctrl->veths);

253 254 255 256
    for (i = 0 ; i < ctrl->nconsoles ; i++)
        virLXCControllerConsoleClose(&(ctrl->consoles[i]));
    VIR_FREE(ctrl->consoles);

257 258
    VIR_FORCE_CLOSE(ctrl->handshakeFd);

259 260
    VIR_FREE(ctrl->devptmx);

261 262 263
    virDomainDefFree(ctrl->def);
    VIR_FREE(ctrl->name);

264 265 266
    if (ctrl->timerShutdown != -1)
        virEventRemoveTimeout(ctrl->timerShutdown);

267
    virObjectUnref(ctrl->server);
268

269 270 271
    VIR_FREE(ctrl);
}

272

273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304
static int virLXCControllerAddConsole(virLXCControllerPtr ctrl,
                                      int hostFd)
{
    if (VIR_EXPAND_N(ctrl->consoles, ctrl->nconsoles, 1) < 0) {
        virReportOOMError();
        return -1;
    }
    ctrl->consoles[ctrl->nconsoles-1].hostFd = hostFd;
    ctrl->consoles[ctrl->nconsoles-1].hostWatch = -1;

    ctrl->consoles[ctrl->nconsoles-1].contFd = -1;
    ctrl->consoles[ctrl->nconsoles-1].contWatch = -1;

    ctrl->consoles[ctrl->nconsoles-1].epollFd = -1;
    ctrl->consoles[ctrl->nconsoles-1].epollWatch = -1;
    return 0;
}


static int virLXCControllerConsoleSetNonblocking(virLXCControllerConsolePtr console)
{
    if (virSetBlocking(console->hostFd, false) < 0 ||
        virSetBlocking(console->contFd, false) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to set console file descriptor non-blocking"));
        return -1;
    }

    return 0;
}


305 306 307 308 309 310 311 312 313 314 315 316
static int virLXCControllerDaemonHandshake(virLXCControllerPtr ctrl)
{
    if (lxcContainerSendContinue(ctrl->handshakeFd) < 0) {
        virReportSystemError(errno, "%s",
                             _("error sending continue signal to daemon"));
        return -1;
    }
    VIR_FORCE_CLOSE(ctrl->handshakeFd);
    return 0;
}


317 318 319
static int virLXCControllerValidateNICs(virLXCControllerPtr ctrl)
{
    if (ctrl->def->nnets != ctrl->nveths) {
320 321 322
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("expecting %d veths, but got %zu"),
                       ctrl->def->nnets, ctrl->nveths);
323 324 325 326 327 328 329
        return -1;
    }

    return 0;
}


330 331 332
static int virLXCControllerValidateConsoles(virLXCControllerPtr ctrl)
{
    if (ctrl->def->nconsoles != ctrl->nconsoles) {
333 334 335
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("expecting %d consoles, but got %zu tty file handlers"),
                       ctrl->def->nconsoles, ctrl->nconsoles);
336 337 338 339 340 341 342
        return -1;
    }

    return 0;
}


343
static int virLXCControllerSetupLoopDevice(virDomainFSDefPtr fs)
344
{
345
    int lofd;
346 347
    char *loname = NULL;

348
    if ((lofd = virFileLoopDeviceAssociate(fs->src, &loname)) < 0)
349 350 351 352 353 354 355 356 357 358 359 360 361 362 363
        return -1;

    /*
     * We now change it into a block device type, so that
     * the rest of container setup 'just works'
     */
    fs->type = VIR_DOMAIN_FS_TYPE_BLOCK;
    VIR_FREE(fs->src);
    fs->src = loname;
    loname = NULL;

    return lofd;
}


364
static int virLXCControllerSetupLoopDevices(virLXCControllerPtr ctrl)
365 366 367 368
{
    size_t i;
    int ret = -1;

369
    for (i = 0 ; i < ctrl->def->nfss ; i++) {
370 371
        int fd;

372
        if (ctrl->def->fss[i]->type != VIR_DOMAIN_FS_TYPE_FILE)
373 374
            continue;

375
        fd = virLXCControllerSetupLoopDevice(ctrl->def->fss[i]);
376 377 378 379
        if (fd < 0)
            goto cleanup;

        VIR_DEBUG("Saving loop fd %d", fd);
380
        if (VIR_EXPAND_N(ctrl->loopDevFds, ctrl->nloopDevs, 1) < 0) {
381 382 383 384
            VIR_FORCE_CLOSE(fd);
            virReportOOMError();
            goto cleanup;
        }
385
        ctrl->loopDevFds[ctrl->nloopDevs - 1] = fd;
386 387 388 389 390 391 392 393 394
    }

    VIR_DEBUG("Setup all loop devices");
    ret = 0;

cleanup:
    return ret;
}

395
#if HAVE_NUMACTL
396
static int virLXCControllerSetupNUMAPolicy(virLXCControllerPtr ctrl)
397 398 399 400 401 402 403 404 405
{
    nodemask_t mask;
    int mode = -1;
    int node = -1;
    int ret = -1;
    int i = 0;
    int maxnode = 0;
    bool warned = false;

406
    if (!ctrl->def->numatune.memory.nodemask)
407 408 409 410 411
        return 0;

    VIR_DEBUG("Setting NUMA memory policy");

    if (numa_available() < 0) {
412 413
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       "%s", _("Host kernel is not aware of NUMA."));
414 415 416 417 418 419 420
        return -1;
    }

    maxnode = numa_max_node() + 1;

    /* Convert nodemask to NUMA bitmask. */
    nodemask_zero(&mask);
421 422 423 424 425 426 427 428 429 430 431
    i = -1;
    while ((i = virBitmapNextSetBit(ctrl->def->numatune.memory.nodemask, i)) >= 0) {
        if (i > NUMA_NUM_NODES) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Host cannot support NUMA node %d"), i);
            return -1;
        }
        if (i > maxnode && !warned) {
            VIR_WARN("nodeset is out of range, there is only %d NUMA "
                     "nodes on host", maxnode);
            warned = true;
432
        }
433
        nodemask_set(&mask, i);
434 435
    }

436
    mode = ctrl->def->numatune.memory.mode;
437 438 439 440 441 442 443 444 445 446 447 448 449 450 451

    if (mode == VIR_DOMAIN_NUMATUNE_MEM_STRICT) {
        numa_set_bind_policy(1);
        numa_set_membind(&mask);
        numa_set_bind_policy(0);
    } else if (mode == VIR_DOMAIN_NUMATUNE_MEM_PREFERRED) {
        int nnodes = 0;
        for (i = 0; i < NUMA_NUM_NODES; i++) {
            if (nodemask_isset(&mask, i)) {
                node = i;
                nnodes++;
            }
        }

        if (nnodes != 1) {
452 453 454
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           "%s", _("NUMA memory tuning in 'preferred' mode "
                                   "only supports single node"));
455 456 457 458 459 460 461 462
            goto cleanup;
        }

        numa_set_bind_policy(0);
        numa_set_preferred(node);
    } else if (mode == VIR_DOMAIN_NUMATUNE_MEM_INTERLEAVE) {
        numa_set_interleave_mask(&mask);
    } else {
463 464 465
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("Unable to set NUMA policy %s"),
                       virDomainNumatuneMemModeTypeToString(mode));
466 467 468 469 470 471 472 473 474
        goto cleanup;
    }

    ret = 0;

cleanup:
    return ret;
}
#else
475
static int virLXCControllerSetupNUMAPolicy(virLXCControllerPtr ctrl)
476
{
477
    if (ctrl->def->numatune.memory.nodemask) {
478 479
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("NUMA policy is not available on this platform"));
480 481 482 483 484 485 486
        return -1;
    }

    return 0;
}
#endif

487 488 489 490

/*
 * To be run while still single threaded
 */
491
static int virLXCControllerSetupCpuAffinity(virLXCControllerPtr ctrl)
492
{
H
Hu Tao 已提交
493
    int hostcpus, maxcpu = CPU_SETSIZE;
494
    virNodeInfo nodeinfo;
H
Hu Tao 已提交
495
    virBitmapPtr cpumap, cpumapToSet;
496 497 498 499 500 501 502 503 504 505 506 507

    VIR_DEBUG("Setting CPU affinity");

    if (nodeGetInfo(NULL, &nodeinfo) < 0)
        return -1;

    /* setaffinity fails if you set bits for CPUs which
     * aren't present, so we have to limit ourselves */
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;

508 509
    cpumap = virBitmapNew(maxcpu);
    if (!cpumap)
510 511
        return -1;

H
Hu Tao 已提交
512 513
    cpumapToSet = cpumap;

514
    if (ctrl->def->cpumask) {
H
Hu Tao 已提交
515
        cpumapToSet = ctrl->def->cpumask;
516 517 518 519 520
    } else {
        /* You may think this is redundant, but we can't assume libvirtd
         * itself is running on all pCPUs, so we need to explicitly set
         * the spawned LXC instance to all pCPUs if no map is given in
         * its config file */
521
        virBitmapSetAll(cpumap);
522 523
    }

524
    /* We are presuming we are running between fork/exec of LXC
525 526 527
     * so use '0' to indicate our own process ID. No threads are
     * running at this point
     */
H
Hu Tao 已提交
528
    if (virProcessInfoSetAffinity(0 /* Self */, cpumapToSet) < 0) {
529
        virBitmapFree(cpumap);
530 531
        return -1;
    }
532
    virBitmapFree(cpumap);
533 534 535 536 537

    return 0;
}


538
/**
539 540
 * virLXCControllerSetupResourceLimits
 * @ctrl: the controller state
541 542 543 544 545 546
 *
 * Creates a cgroup for the container, moves the task inside,
 * and sets resource limits
 *
 * Returns 0 on success or -1 in case of error
 */
547
static int virLXCControllerSetupResourceLimits(virLXCControllerPtr ctrl)
548 549
{

550
    if (virLXCControllerSetupCpuAffinity(ctrl) < 0)
551 552
        return -1;

553
    if (virLXCControllerSetupNUMAPolicy(ctrl) < 0)
554 555
        return -1;

556
    return virLXCCgroupSetup(ctrl->def);
D
Dan Smith 已提交
557 558
}

559

560 561 562 563 564 565 566 567 568 569 570 571 572
static void virLXCControllerClientCloseHook(virNetServerClientPtr client)
{
    virLXCControllerPtr ctrl = virNetServerClientGetPrivateData(client);

    VIR_DEBUG("Client %p has closed", client);
    if (ctrl->client == client)
        ctrl->client = NULL;
    if (ctrl->inShutdown) {
        VIR_DEBUG("Arm timer to quit event loop");
        virEventUpdateTimeout(ctrl->timerShutdown, 0);
    }
}

573 574 575 576 577 578 579
static void virLXCControllerClientPrivateFree(void *data)
{
    VIR_FREE(data);
}

static void *virLXCControllerClientPrivateNew(virNetServerClientPtr client,
                                              void *opaque)
580 581
{
    virLXCControllerPtr ctrl = opaque;
582 583 584 585 586 587 588
    int *dummy;

    if (VIR_ALLOC(dummy) < 0) {
        virReportOOMError();
        return NULL;
    }

589 590 591
    virNetServerClientSetCloseHook(client, virLXCControllerClientCloseHook);
    VIR_DEBUG("Got new client %p", client);
    ctrl->client = client;
592
    return dummy;
593 594
}

595 596

static int virLXCControllerSetupServer(virLXCControllerPtr ctrl)
597
{
598 599
    virNetServerServicePtr svc = NULL;
    char *sockpath;
600

601 602 603 604
    if (virAsprintf(&sockpath, "%s/%s.sock",
                    LXC_STATE_DIR, ctrl->name) < 0) {
        virReportOOMError();
        return -1;
605 606
    }

607 608 609
    if (!(ctrl->server = virNetServerNew(0, 0, 0, 1,
                                         -1, 0, false,
                                         NULL,
610 611
                                         virLXCControllerClientPrivateNew,
                                         virLXCControllerClientPrivateFree,
612
                                         ctrl)))
C
Chris Lalancette 已提交
613
        goto error;
614

615 616 617 618 619 620 621
    if (!(svc = virNetServerServiceNewUNIX(sockpath,
                                           0700,
                                           0,
                                           0,
                                           false,
                                           5,
                                           NULL)))
622
        goto error;
623 624

    if (virNetServerAddService(ctrl->server, svc, NULL) < 0)
625
        goto error;
626
    virObjectUnref(svc);
627
    svc = NULL;
628

629 630 631 632 633 634
    if (!(ctrl->prog = virNetServerProgramNew(VIR_LXC_PROTOCOL_PROGRAM,
                                              VIR_LXC_PROTOCOL_PROGRAM_VERSION,
                                              virLXCProtocolProcs,
                                              virLXCProtocolNProcs)))
        goto error;

635 636 637
    virNetServerUpdateServices(ctrl->server, true);
    VIR_FREE(sockpath);
    return 0;
638 639

error:
640
    VIR_FREE(sockpath);
641
    virObjectUnref(ctrl->server);
642
    ctrl->server = NULL;
643
    virObjectUnref(svc);
644 645
    return -1;
}
646

D
Daniel P. Berrange 已提交
647 648 649 650 651 652 653 654 655

static int lxcControllerClearCapabilities(void)
{
#if HAVE_CAPNG
    int ret;

    capng_clear(CAPNG_SELECT_BOTH);

    if ((ret = capng_apply(CAPNG_SELECT_BOTH)) < 0) {
656 657
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("failed to apply capabilities: %d"), ret);
D
Daniel P. Berrange 已提交
658 659 660
        return -1;
    }
#else
661
    VIR_WARN("libcap-ng support not compiled in, unable to clear capabilities");
D
Daniel P. Berrange 已提交
662 663 664 665
#endif
    return 0;
}

666
static bool quit = false;
667
static bool wantReboot = false;
668 669 670
static virMutex lock;


671 672
static void virLXCControllerSignalChildIO(virNetServerPtr server ATTRIBUTE_UNUSED,
                                          siginfo_t *info ATTRIBUTE_UNUSED,
673
                                          void *opaque)
674
{
675
    virLXCControllerPtr ctrl = opaque;
676
    int ret;
677
    int status;
678

679
    ret = waitpid(-1, &status, WNOHANG);
680 681 682
    if (ret == ctrl->initpid) {
        virMutexLock(&lock);
        quit = true;
683 684 685
        if (WIFSIGNALED(status) &&
            WTERMSIG(status) == SIGHUP)
            wantReboot = true;
686 687
        virMutexUnlock(&lock);
    }
688 689 690
}


691
static void virLXCControllerConsoleUpdateWatch(virLXCControllerConsolePtr console)
692 693 694 695
{
    int hostEvents = 0;
    int contEvents = 0;

696
    if (!console->hostClosed || (!console->hostBlocking && console->fromContLen)) {
697 698 699 700 701
        if (console->fromHostLen < sizeof(console->fromHostBuf))
            hostEvents |= VIR_EVENT_HANDLE_READABLE;
        if (console->fromContLen)
            hostEvents |= VIR_EVENT_HANDLE_WRITABLE;
    }
702
    if (!console->contClosed || (!console->contBlocking && console->fromHostLen)) {
703 704 705 706 707 708
        if (console->fromContLen < sizeof(console->fromContBuf))
            contEvents |= VIR_EVENT_HANDLE_READABLE;
        if (console->fromHostLen)
            contEvents |= VIR_EVENT_HANDLE_WRITABLE;
    }

709 710 711
    VIR_DEBUG("Container watch %d=%d host watch %d=%d",
              console->contWatch, contEvents,
              console->hostWatch, hostEvents);
712 713
    virEventUpdateHandle(console->contWatch, contEvents);
    virEventUpdateHandle(console->hostWatch, hostEvents);
714

715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750
    if (console->hostClosed) {
        int events = EPOLLIN | EPOLLET;
        if (console->hostBlocking)
            events |= EPOLLOUT;

        if (events != console->hostEpoll) {
            struct epoll_event event;
            int action = EPOLL_CTL_ADD;
            if (console->hostEpoll)
                action = EPOLL_CTL_MOD;

            VIR_DEBUG("newHostEvents=%x oldHostEvents=%x", events, console->hostEpoll);

            event.events = events;
            event.data.fd = console->hostFd;
            if (epoll_ctl(console->epollFd, action, console->hostFd, &event) < 0) {
                VIR_DEBUG(":fail");
                virReportSystemError(errno, "%s",
                                     _("Unable to add epoll fd"));
                quit = true;
                goto cleanup;
            }
            console->hostEpoll = events;
            VIR_DEBUG("newHostEvents=%x oldHostEvents=%x", events, console->hostEpoll);
        }
    } else if (console->hostEpoll) {
        VIR_DEBUG("Stop epoll oldContEvents=%x", console->hostEpoll);
        if (epoll_ctl(console->epollFd, EPOLL_CTL_DEL, console->hostFd, NULL) < 0) {
            virReportSystemError(errno, "%s",
                                 _("Unable to remove epoll fd"));
                VIR_DEBUG(":fail");
            quit = true;
            goto cleanup;
        }
        console->hostEpoll = 0;
    }
751

752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790
    if (console->contClosed) {
        int events = EPOLLIN | EPOLLET;
        if (console->contBlocking)
            events |= EPOLLOUT;

        if (events != console->contEpoll) {
            struct epoll_event event;
            int action = EPOLL_CTL_ADD;
            if (console->contEpoll)
                action = EPOLL_CTL_MOD;

            VIR_DEBUG("newContEvents=%x oldContEvents=%x", events, console->contEpoll);

            event.events = events;
            event.data.fd = console->contFd;
            if (epoll_ctl(console->epollFd, action, console->contFd, &event) < 0) {
                virReportSystemError(errno, "%s",
                                     _("Unable to add epoll fd"));
                VIR_DEBUG(":fail");
                quit = true;
                goto cleanup;
            }
            console->contEpoll = events;
            VIR_DEBUG("newHostEvents=%x oldHostEvents=%x", events, console->contEpoll);
        }
    } else if (console->contEpoll) {
        VIR_DEBUG("Stop epoll oldContEvents=%x", console->contEpoll);
        if (epoll_ctl(console->epollFd, EPOLL_CTL_DEL, console->contFd, NULL) < 0) {
            virReportSystemError(errno, "%s",
                                 _("Unable to remove epoll fd"));
                VIR_DEBUG(":fail");
            quit = true;
            goto cleanup;
        }
        console->contEpoll = 0;
    }
cleanup:
    return;
}
791 792


793
static void virLXCControllerConsoleEPoll(int watch, int fd, int events, void *opaque)
794
{
795
    virLXCControllerConsolePtr console = opaque;
796

797 798 799 800 801 802 803 804 805 806
    virMutexLock(&lock);
    VIR_DEBUG("IO event watch=%d fd=%d events=%d fromHost=%zu fromcont=%zu",
              watch, fd, events,
              console->fromHostLen,
              console->fromContLen);

    while (1) {
        struct epoll_event event;
        int ret;
        ret = epoll_wait(console->epollFd, &event, 1, 0);
807
        if (ret < 0) {
S
Stefan Berger 已提交
808
            if (errno == EINTR)
809 810 811 812 813 814 815
                continue;
            virReportSystemError(errno, "%s",
                                 _("Unable to wait on epoll"));
            quit = true;
            goto cleanup;
        }

816 817 818 819 820 821 822
        if (ret == 0)
            break;

        VIR_DEBUG("fd=%d hostFd=%d contFd=%d hostEpoll=%x contEpoll=%x",
                  event.data.fd, console->hostFd, console->contFd,
                  console->hostEpoll, console->contEpoll);

823 824 825
        /* If we get HUP+dead PID, we just re-enable the main loop
         * which will see the PID has died and exit */
        if ((event.events & EPOLLIN)) {
826 827
            if (event.data.fd == console->hostFd) {
                console->hostClosed = false;
828
            } else {
829
                console->contClosed = false;
830
            }
831
            virLXCControllerConsoleUpdateWatch(console);
832 833 834 835 836
            break;
        }
    }

cleanup:
837
    virMutexUnlock(&lock);
838 839
}

840
static void virLXCControllerConsoleIO(int watch, int fd, int events, void *opaque)
841
{
842
    virLXCControllerConsolePtr console = opaque;
843 844

    virMutexLock(&lock);
845 846 847 848
    VIR_DEBUG("IO event watch=%d fd=%d events=%d fromHost=%zu fromcont=%zu",
              watch, fd, events,
              console->fromHostLen,
              console->fromContLen);
849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904
    if (events & VIR_EVENT_HANDLE_READABLE) {
        char *buf;
        size_t *len;
        size_t avail;
        ssize_t done;
        if (watch == console->hostWatch) {
            buf = console->fromHostBuf;
            len = &console->fromHostLen;
            avail = sizeof(console->fromHostBuf) - *len;
        } else {
            buf = console->fromContBuf;
            len = &console->fromContLen;
            avail = sizeof(console->fromContBuf) - *len;
        }
    reread:
        done = read(fd, buf + *len, avail);
        if (done == -1 && errno == EINTR)
            goto reread;
        if (done == -1 && errno != EAGAIN) {
            virReportSystemError(errno, "%s",
                                 _("Unable to read container pty"));
            goto error;
        }
        if (done > 0) {
            *len += done;
        } else {
            VIR_DEBUG("Read fd %d done %d errno %d", fd, (int)done, errno);
        }
    }

    if (events & VIR_EVENT_HANDLE_WRITABLE) {
        char *buf;
        size_t *len;
        ssize_t done;
        if (watch == console->hostWatch) {
            buf = console->fromContBuf;
            len = &console->fromContLen;
        } else {
            buf = console->fromHostBuf;
            len = &console->fromHostLen;
        }

    rewrite:
        done = write(fd, buf, *len);
        if (done == -1 && errno == EINTR)
            goto rewrite;
        if (done == -1 && errno != EAGAIN) {
            virReportSystemError(errno, "%s",
                                 _("Unable to write to container pty"));
            goto error;
        }
        if (done > 0) {
            memmove(buf, buf + done, (*len - done));
            *len -= done;
        } else {
            VIR_DEBUG("Write fd %d done %d errno %d", fd, (int)done, errno);
905 906 907 908
            if (watch == console->hostWatch)
                console->hostBlocking = true;
            else
                console->contBlocking = true;
909 910 911 912 913 914 915 916 917 918 919 920
        }
    }

    if (events & VIR_EVENT_HANDLE_HANGUP) {
        if (watch == console->hostWatch) {
            console->hostClosed = true;
        } else {
            console->contClosed = true;
        }
        VIR_DEBUG("Got EOF on %d %d", watch, fd);
    }

921
    virLXCControllerConsoleUpdateWatch(console);
922 923 924 925 926 927 928 929 930 931 932 933
    virMutexUnlock(&lock);
    return;

error:
    virEventRemoveHandle(console->contWatch);
    virEventRemoveHandle(console->hostWatch);
    console->contWatch = console->hostWatch = -1;
    quit = true;
    virMutexUnlock(&lock);
}


934
/**
935
 * lxcControllerMain
936 937
 * @serverFd: server socket fd to accept client requests
 * @clientFd: initial client which is the libvirtd daemon
938
 *
939
 * Processes I/O on consoles and the monitor
940 941 942
 *
 * Returns 0 on success or -1 in case of error
 */
943
static int virLXCControllerMain(virLXCControllerPtr ctrl)
944
{
945
    virErrorPtr err;
946
    int rc = -1;
947
    size_t i;
948 949 950 951

    if (virMutexInit(&lock) < 0)
        goto cleanup2;

952 953 954 955
    if (virNetServerAddSignalHandler(ctrl->server,
                                     SIGCHLD,
                                     virLXCControllerSignalChildIO,
                                     ctrl) < 0)
956 957
        goto cleanup;

958 959
    virResetLastError();

960 961
    for (i = 0 ; i < ctrl->nconsoles ; i++) {
        if ((ctrl->consoles[i].epollFd = epoll_create1(EPOLL_CLOEXEC)) < 0) {
962 963 964 965 966
            virReportSystemError(errno, "%s",
                                 _("Unable to create epoll fd"));
            goto cleanup;
        }

967 968 969 970 971
        if ((ctrl->consoles[i].epollWatch = virEventAddHandle(ctrl->consoles[i].epollFd,
                                                              VIR_EVENT_HANDLE_READABLE,
                                                              virLXCControllerConsoleEPoll,
                                                              &(ctrl->consoles[i]),
                                                              NULL)) < 0) {
972 973
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("Unable to watch epoll FD"));
974 975 976
            goto cleanup;
        }

977 978 979 980 981
        if ((ctrl->consoles[i].hostWatch = virEventAddHandle(ctrl->consoles[i].hostFd,
                                                             VIR_EVENT_HANDLE_READABLE,
                                                             virLXCControllerConsoleIO,
                                                             &(ctrl->consoles[i]),
                                                             NULL)) < 0) {
982 983
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("Unable to watch host console PTY"));
984 985 986
            goto cleanup;
        }

987 988 989 990 991
        if ((ctrl->consoles[i].contWatch = virEventAddHandle(ctrl->consoles[i].contFd,
                                                             VIR_EVENT_HANDLE_READABLE,
                                                             virLXCControllerConsoleIO,
                                                             &(ctrl->consoles[i]),
                                                             NULL)) < 0) {
992 993
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("Unable to watch host console PTY"));
994 995
            goto cleanup;
        }
996
    }
997

998 999 1000 1001
    virMutexLock(&lock);
    while (!quit) {
        virMutexUnlock(&lock);
        if (virEventRunDefaultImpl() < 0)
1002
            goto cleanup;
1003
        virMutexLock(&lock);
1004
    }
1005
    virMutexUnlock(&lock);
1006

1007 1008
    err = virGetLastError();
    if (!err || err->code == VIR_ERR_OK)
1009
        rc = wantReboot ? 1 : 0;
1010 1011

cleanup:
1012 1013
    virMutexDestroy(&lock);
cleanup2:
1014

1015 1016
    for (i = 0 ; i < ctrl->nconsoles ; i++)
        virLXCControllerConsoleClose(&(ctrl->consoles[i]));
1017

1018 1019 1020
    return rc;
}

1021 1022 1023


/**
1024
 * virLXCControllerMoveInterfaces
1025 1026 1027 1028 1029 1030 1031 1032
 * @nveths: number of interfaces
 * @veths: interface names
 * @container: pid of container
 *
 * Moves network interfaces into a container's namespace
 *
 * Returns 0 on success or -1 in case of error
 */
1033
static int virLXCControllerMoveInterfaces(virLXCControllerPtr ctrl)
1034
{
1035 1036 1037
    size_t i;

    for (i = 0 ; i < ctrl->nveths ; i++) {
1038
        if (virNetDevSetNamespace(ctrl->veths[i], ctrl->initpid) < 0)
1039
            return -1;
1040
    }
1041 1042 1043 1044 1045 1046

    return 0;
}


/**
1047 1048
 * virLXCControllerDeleteInterfaces:
 * @ctrl: the LXC controller
1049 1050 1051 1052 1053
 *
 * Cleans up the container interfaces by deleting the veth device pairs.
 *
 * Returns 0 on success or -1 in case of error
 */
1054
static int virLXCControllerDeleteInterfaces(virLXCControllerPtr ctrl)
1055
{
1056 1057
    size_t i;
    int ret = 0;
1058

1059 1060 1061 1062 1063 1064
    for (i = 0 ; i < ctrl->nveths ; i++) {
        if (virNetDevVethDelete(ctrl->veths[i]) < 0)
            ret = -1;
    }

    return ret;
1065 1066
}

1067

1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086
static int lxcSetPersonality(virDomainDefPtr def)
{
    struct utsname utsname;
    const char *altArch;

    uname(&utsname);

    altArch = lxcContainerGetAlt32bitArch(utsname.machine);
    if (altArch &&
        STREQ(def->os.arch, altArch)) {
        if (personality(PER_LINUX32) < 0) {
            virReportSystemError(errno, _("Unable to request personality for %s on %s"),
                                 altArch, utsname.machine);
            return -1;
        }
    }
    return 0;
}

1087
#ifndef MS_REC
1088
# define MS_REC          16384
1089 1090 1091
#endif

#ifndef MS_SLAVE
1092
# define MS_SLAVE              (1<<19)
1093
#endif
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 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138
/* Create a private tty using the private devpts at PTMX, returning
 * the master in *TTYMASTER and the name of the slave, _from the
 * perspective of the guest after remounting file systems_, in
 * *TTYNAME.  Heavily borrowed from glibc, but doesn't require that
 * devpts == "/dev/pts" */
static int
lxcCreateTty(char *ptmx, int *ttymaster, char **ttyName)
{
    int ret = -1;
    int ptyno;
    int unlock = 0;

    if ((*ttymaster = open(ptmx, O_RDWR|O_NOCTTY|O_NONBLOCK)) < 0)
        goto cleanup;

    if (ioctl(*ttymaster, TIOCSPTLCK, &unlock) < 0)
        goto cleanup;

    if (ioctl(*ttymaster, TIOCGPTN, &ptyno) < 0)
        goto cleanup;

    /* If mount() succeeded at honoring newinstance, then the kernel
     * was new enough to also honor the mode=0620,gid=5 options, which
     * guarantee that the new pty already has correct permissions; so
     * while glibc has to fstat(), fchmod(), and fchown() for older
     * kernels, we can skip those steps.  ptyno shouldn't currently be
     * anything other than 0, but let's play it safe.  */
    if (virAsprintf(ttyName, "/dev/pts/%d", ptyno) < 0) {
        virReportOOMError();
        errno = ENOMEM;
        goto cleanup;
    }

    ret = 0;

cleanup:
    if (ret != 0) {
        VIR_FORCE_CLOSE(*ttymaster);
        VIR_FREE(*ttyName);
    }

    return ret;
}

1139

1140
static int
1141
virLXCControllerSetupDevPTS(virLXCControllerPtr ctrl)
1142
{
1143
    virDomainFSDefPtr root = virDomainGetRootFilesystem(ctrl->def);
1144
    char *mount_options = NULL;
1145 1146 1147
    char *opts;
    char *devpts = NULL;
    int ret = -1;
1148

1149 1150
    if (!root) {
        if (ctrl->nconsoles != 1) {
1151 1152 1153
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Expected exactly one console, but got %zu"),
                           ctrl->nconsoles);
1154 1155 1156
            return -1;
        }
        return 0;
1157 1158
    }

1159
    VIR_DEBUG("Setting up private /dev/pts");
1160

1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180
    /*
     * If doing a chroot style setup, we need to prepare
     * a private /dev/pts for the child now, which they
     * will later move into position.
     *
     * This is complex because 'virsh console' needs to
     * use /dev/pts from the host OS, and the guest OS
     * needs to use /dev/pts from the guest.
     *
     * This means that we (libvirt_lxc) need to see and
     * use both /dev/pts instances. We're running in the
     * host OS context though and don't want to expose
     * the guest OS /dev/pts there.
     *
     * Thus we call unshare(CLONE_NS) so that we can see
     * the guest's new /dev/pts, without it becoming
     * visible to the host OS. We also put the root FS
     * into slave mode, just in case it was currently
     * marked as shared
     */
1181 1182
    mount_options = virSecurityManagerGetMountOptions(ctrl->securityManager,
                                                      ctrl->def);
1183

1184 1185 1186 1187 1188 1189
    if (!virFileExists(root->src)) {
        virReportSystemError(errno,
                             _("root source %s does not exist"),
                             root->src);
        goto cleanup;
    }
1190

1191 1192 1193 1194 1195
    if (unshare(CLONE_NEWNS) < 0) {
        virReportSystemError(errno, "%s",
                             _("Cannot unshare mount namespace"));
        goto cleanup;
    }
1196

1197 1198 1199 1200 1201
    if (mount("", "/", NULL, MS_SLAVE|MS_REC, NULL) < 0) {
        virReportSystemError(errno, "%s",
                             _("Failed to switch root mount into slave mode"));
        goto cleanup;
    }
1202

1203 1204 1205 1206 1207
    if (virAsprintf(&devpts, "%s/dev/pts", root->src) < 0 ||
        virAsprintf(&ctrl->devptmx, "%s/dev/pts/ptmx", root->src) < 0) {
        virReportOOMError();
        goto cleanup;
    }
1208

1209 1210 1211 1212 1213 1214
    if (virFileMakePath(devpts) < 0) {
        virReportSystemError(errno,
                             _("Failed to make path %s"),
                             devpts);
        goto cleanup;
    }
1215

1216 1217 1218 1219 1220 1221 1222
    /* XXX should we support gid=X for X!=5 for distros which use
     * a different gid for tty?  */
    if (virAsprintf(&opts, "newinstance,ptmxmode=0666,mode=0620,gid=5%s",
                    (mount_options ? mount_options : "")) < 0) {
        virReportOOMError();
        goto cleanup;
    }
1223

1224 1225 1226 1227 1228 1229 1230 1231
    VIR_DEBUG("Mount devpts on %s type=tmpfs flags=%x, opts=%s",
              devpts, MS_NOSUID, opts);
    if (mount("devpts", devpts, "devpts", MS_NOSUID, opts) < 0) {
        virReportSystemError(errno,
                             _("Failed to mount devpts on %s"),
                             devpts);
        goto cleanup;
    }
1232

1233 1234 1235
    if (access(ctrl->devptmx, R_OK) < 0) {
        VIR_WARN("Kernel does not support private devpts, using shared devpts");
        VIR_FREE(ctrl->devptmx);
1236 1237
    }

1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252
    ret = 0;

cleanup:
    VIR_FREE(opts);
    VIR_FREE(devpts);
    return ret;
}


static int
virLXCControllerSetupConsoles(virLXCControllerPtr ctrl,
                              char **containerTTYPaths)
{
    size_t i;

1253
    for (i = 0 ; i < ctrl->nconsoles ; i++) {
1254 1255 1256
        if (ctrl->devptmx) {
            VIR_DEBUG("Opening tty on private %s", ctrl->devptmx);
            if (lxcCreateTty(ctrl->devptmx,
1257 1258
                             &ctrl->consoles[i].contFd,
                             &containerTTYPaths[i]) < 0) {
1259 1260
                virReportSystemError(errno, "%s",
                                     _("Failed to allocate tty"));
1261
                return -1;
1262 1263 1264
            }
        } else {
            VIR_DEBUG("Opening tty on shared /dev/ptmx");
1265 1266
            if (virFileOpenTty(&ctrl->consoles[i].contFd,
                               &containerTTYPaths[i],
1267 1268 1269
                               0) < 0) {
                virReportSystemError(errno, "%s",
                                     _("Failed to allocate tty"));
1270
                return -1;
1271 1272 1273
            }
        }
    }
1274 1275 1276 1277
    return 0;
}


1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329
static void
virLXCControllerEventSend(virLXCControllerPtr ctrl,
                          int procnr,
                          xdrproc_t proc,
                          void *data)
{
    virNetMessagePtr msg;

    if (!ctrl->client)
        return;

    VIR_DEBUG("Send event %d client=%p", procnr, ctrl->client);
    if (!(msg = virNetMessageNew(false)))
        goto error;

    msg->header.prog = virNetServerProgramGetID(ctrl->prog);
    msg->header.vers = virNetServerProgramGetVersion(ctrl->prog);
    msg->header.proc = procnr;
    msg->header.type = VIR_NET_MESSAGE;
    msg->header.serial = 1;
    msg->header.status = VIR_NET_OK;

    if (virNetMessageEncodeHeader(msg) < 0)
        goto error;

    if (virNetMessageEncodePayload(msg, proc, data) < 0)
        goto error;

    VIR_DEBUG("Queue event %d %zu", procnr, msg->bufferLength);
    virNetServerClientSendMessage(ctrl->client, msg);

    xdr_free(proc, data);
    return;

error:
    virNetMessageFree(msg);
    xdr_free(proc, data);
}


static int
virLXCControllerEventSendExit(virLXCControllerPtr ctrl,
                              int exitstatus)
{
    virLXCProtocolExitEventMsg msg;

    VIR_DEBUG("Exit status %d", exitstatus);
    memset(&msg, 0, sizeof(msg));
    switch (exitstatus) {
    case 0:
        msg.status = VIR_LXC_PROTOCOL_EXIT_STATUS_SHUTDOWN;
        break;
1330 1331 1332
    case 1:
        msg.status = VIR_LXC_PROTOCOL_EXIT_STATUS_REBOOT;
        break;
1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353
    default:
        msg.status = VIR_LXC_PROTOCOL_EXIT_STATUS_ERROR;
        break;
    }

    virLXCControllerEventSend(ctrl,
                              VIR_LXC_PROTOCOL_PROC_EXIT_EVENT,
                              (xdrproc_t)xdr_virLXCProtocolExitEventMsg,
                              (void*)&msg);

    if (ctrl->client) {
        VIR_DEBUG("Waiting for client to complete dispatch");
        ctrl->inShutdown = true;
        virNetServerClientDelayedClose(ctrl->client);
        virNetServerRun(ctrl->server);
    }
    VIR_DEBUG("Client has gone away");
    return 0;
}


1354
static int
1355
virLXCControllerRun(virLXCControllerPtr ctrl)
1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382
{
    int rc = -1;
    int control[2] = { -1, -1};
    int containerhandshake[2] = { -1, -1 };
    char **containerTTYPaths = NULL;
    size_t i;

    if (VIR_ALLOC_N(containerTTYPaths, ctrl->nconsoles) < 0) {
        virReportOOMError();
        goto cleanup;
    }

    if (socketpair(PF_UNIX, SOCK_STREAM, 0, control) < 0) {
        virReportSystemError(errno, "%s",
                             _("sockpair failed"));
        goto cleanup;
    }

    if (socketpair(PF_UNIX, SOCK_STREAM, 0, containerhandshake) < 0) {
        virReportSystemError(errno, "%s",
                             _("socketpair failed"));
        goto cleanup;
    }

    if (virLXCControllerSetupLoopDevices(ctrl) < 0)
        goto cleanup;

1383
    if (virLXCControllerSetupResourceLimits(ctrl) < 0)
1384 1385 1386 1387 1388 1389 1390
        goto cleanup;

    if (virLXCControllerSetupDevPTS(ctrl) < 0)
        goto cleanup;

    if (virLXCControllerSetupConsoles(ctrl, containerTTYPaths) < 0)
        goto cleanup;
1391

1392
    if (lxcSetPersonality(ctrl->def) < 0)
1393
        goto cleanup;
1394

1395
    if ((ctrl->initpid = lxcContainerStart(ctrl->def,
1396
                                           ctrl->securityManager,
1397 1398 1399 1400
                                           ctrl->nveths,
                                           ctrl->veths,
                                           control[1],
                                           containerhandshake[1],
1401 1402
                                           containerTTYPaths,
                                           ctrl->nconsoles)) < 0)
1403
        goto cleanup;
1404
    VIR_FORCE_CLOSE(control[1]);
1405
    VIR_FORCE_CLOSE(containerhandshake[1]);
1406

1407
    if (virLXCControllerMoveInterfaces(ctrl) < 0)
1408 1409
        goto cleanup;

1410 1411 1412
    if (lxcContainerSendContinue(control[0]) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to send container continue message"));
1413
        goto cleanup;
1414
    }
1415

1416 1417 1418 1419 1420 1421
    if (lxcContainerWaitForContinue(containerhandshake[0]) < 0) {
        virReportSystemError(errno, "%s",
                             _("error receiving signal from container"));
        goto cleanup;
    }

1422 1423 1424
    /* Now the container is fully setup... */

    /* ...we can close the loop devices... */
1425 1426
    if (virLXCControllerCloseLoopDevices(ctrl, false) < 0)
        goto cleanup;
1427 1428

    /* ...and reduce our privileges */
D
Daniel P. Berrange 已提交
1429 1430 1431
    if (lxcControllerClearCapabilities() < 0)
        goto cleanup;

1432
    if (virLXCControllerDaemonHandshake(ctrl) < 0)
1433 1434
        goto cleanup;

1435 1436
    for (i = 0 ; i < ctrl->nconsoles ; i++)
        if (virLXCControllerConsoleSetNonblocking(&(ctrl->consoles[i])) < 0)
1437
            goto cleanup;
1438

1439
    rc = virLXCControllerMain(ctrl);
1440

1441 1442
    virLXCControllerEventSendExit(ctrl, rc);

1443
cleanup:
1444 1445
    VIR_FORCE_CLOSE(control[0]);
    VIR_FORCE_CLOSE(control[1]);
1446 1447
    VIR_FORCE_CLOSE(containerhandshake[0]);
    VIR_FORCE_CLOSE(containerhandshake[1]);
1448

1449 1450 1451
    for (i = 0 ; i < ctrl->nconsoles ; i++)
        VIR_FREE(containerTTYPaths[i]);
    VIR_FREE(containerTTYPaths);
1452

1453
    virLXCControllerStopInit(ctrl);
1454

1455 1456 1457 1458
    return rc;
}


1459
int main(int argc, char *argv[])
1460 1461
{
    pid_t pid;
1462 1463
    int rc = 1;
    char *name = NULL;
1464
    size_t nveths = 0;
1465
    char **veths = NULL;
1466
    int handshakeFd = -1;
1467
    int bg = 0;
1468
    const struct option options[] = {
1469 1470 1471 1472
        { "background", 0, NULL, 'b' },
        { "name",   1, NULL, 'n' },
        { "veth",   1, NULL, 'v' },
        { "console", 1, NULL, 'c' },
1473
        { "handshakefd", 1, NULL, 's' },
1474
        { "security", 1, NULL, 'S' },
1475 1476 1477
        { "help", 0, NULL, 'h' },
        { 0, 0, 0, 0 },
    };
1478 1479
    int *ttyFDs = NULL;
    size_t nttyFDs = 0;
1480
    virLXCControllerPtr ctrl = NULL;
1481
    size_t i;
1482
    const char *securityDriver = "none";
1483

E
Eric Blake 已提交
1484 1485
    if (setlocale(LC_ALL, "") == NULL ||
        bindtextdomain(PACKAGE, LOCALEDIR) == NULL ||
1486
        textdomain(PACKAGE) == NULL) {
E
Eric Blake 已提交
1487 1488 1489 1490
        fprintf(stderr, _("%s: initialization failed\n"), argv[0]);
        exit(EXIT_FAILURE);
    }

1491 1492 1493
    /* Initialize logging */
    virLogSetFromEnv();

1494 1495
    while (1) {
        int c;
1496

1497
        c = getopt_long(argc, argv, "dn:v:m:c:s:h:S:",
1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509
                       options, NULL);

        if (c == -1)
            break;

        switch (c) {
        case 'b':
            bg = 1;
            break;

        case 'n':
            if ((name = strdup(optarg)) == NULL) {
1510
                virReportOOMError();
1511
                goto cleanup;
1512
            }
1513 1514 1515 1516
            break;

        case 'v':
            if (VIR_REALLOC_N(veths, nveths+1) < 0) {
1517
                virReportOOMError();
1518
                goto cleanup;
1519
            }
1520
            if ((veths[nveths++] = strdup(optarg)) == NULL) {
1521
                virReportOOMError();
1522
                goto cleanup;
1523
            }
1524 1525 1526
            break;

        case 'c':
1527 1528 1529 1530 1531
            if (VIR_REALLOC_N(ttyFDs, nttyFDs + 1) < 0) {
                virReportOOMError();
                goto cleanup;
            }
            if (virStrToLong_i(optarg, NULL, 10, &ttyFDs[nttyFDs++]) < 0) {
1532 1533 1534 1535 1536
                fprintf(stderr, "malformed --console argument '%s'", optarg);
                goto cleanup;
            }
            break;

1537
        case 's':
1538
            if (virStrToLong_i(optarg, NULL, 10, &handshakeFd) < 0) {
1539 1540 1541 1542 1543 1544
                fprintf(stderr, "malformed --handshakefd argument '%s'",
                        optarg);
                goto cleanup;
            }
            break;

1545
        case 'S':
1546
            securityDriver = optarg;
1547 1548
            break;

1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559
        case 'h':
        case '?':
            fprintf(stderr, "\n");
            fprintf(stderr, "syntax: %s [OPTIONS]\n", argv[0]);
            fprintf(stderr, "\n");
            fprintf(stderr, "Options\n");
            fprintf(stderr, "\n");
            fprintf(stderr, "  -b, --background\n");
            fprintf(stderr, "  -n NAME, --name NAME\n");
            fprintf(stderr, "  -c FD, --console FD\n");
            fprintf(stderr, "  -v VETH, --veth VETH\n");
1560
            fprintf(stderr, "  -s FD, --handshakefd FD\n");
1561
            fprintf(stderr, "  -S NAME, --security NAME\n");
1562 1563 1564
            fprintf(stderr, "  -h, --help\n");
            fprintf(stderr, "\n");
            goto cleanup;
1565 1566 1567
        }
    }

1568 1569 1570 1571 1572
    if (name == NULL) {
        fprintf(stderr, "%s: missing --name argument for configuration\n", argv[0]);
        goto cleanup;
    }

1573
    if (handshakeFd < 0) {
1574 1575 1576 1577 1578
        fprintf(stderr, "%s: missing --handshake argument for container PTY\n",
                argv[0]);
        goto cleanup;
    }

1579
    if (getuid() != 0) {
1580 1581 1582
        fprintf(stderr, "%s: must be run as the 'root' user\n", argv[0]);
        goto cleanup;
    }
1583

1584 1585
    virEventRegisterDefaultImpl();

1586
    if (!(ctrl = virLXCControllerNew(name)))
1587
        goto cleanup;
1588

1589 1590
    ctrl->handshakeFd = handshakeFd;

1591 1592 1593 1594 1595
    if (!(ctrl->securityManager = virSecurityManagerNew(securityDriver,
                                                        LXC_DRIVER_NAME,
                                                        false, false, false)))
        goto cleanup;

1596
    VIR_DEBUG("Security model %s type %s label %s imagelabel %s",
1597 1598 1599 1600
              NULLSTR(ctrl->def->seclabels[0]->model),
              virDomainSeclabelTypeToString(ctrl->def->seclabels[0]->type),
              NULLSTR(ctrl->def->seclabels[0]->label),
              NULLSTR(ctrl->def->seclabels[0]->imagelabel));
1601

1602 1603 1604
    ctrl->veths = veths;
    ctrl->nveths = nveths;

1605 1606 1607 1608 1609 1610
    for (i = 0 ; i < nttyFDs ; i++) {
        if (virLXCControllerAddConsole(ctrl, ttyFDs[i]) < 0)
            goto cleanup;
        ttyFDs[i] = -1;
    }

1611
    if (virLXCControllerValidateNICs(ctrl) < 0)
1612
        goto cleanup;
1613

1614 1615 1616
    if (virLXCControllerValidateConsoles(ctrl) < 0)
        goto cleanup;

1617
    if (virLXCControllerSetupServer(ctrl) < 0)
1618
        goto cleanup;
1619

1620 1621 1622
    if (bg) {
        if ((pid = fork()) < 0)
            goto cleanup;
1623

1624
        if (pid > 0) {
1625
            if ((rc = virPidFileWrite(LXC_STATE_DIR, name, pid)) < 0) {
1626
                virReportSystemError(-rc,
1627 1628
                                     _("Unable to write pid file '%s/%s.pid'"),
                                     LXC_STATE_DIR, name);
1629 1630
                _exit(1);
            }
1631

1632 1633 1634 1635
            /* First child now exits, allowing original caller
             * (ie libvirtd's LXC driver to complete their
             * waitpid & continue */
            _exit(0);
1636 1637
        }

1638 1639
        /* Don't hold onto any cwd we inherit from libvirtd either */
        if (chdir("/") < 0) {
1640
            virReportSystemError(errno, "%s",
1641
                                 _("Unable to change to root dir"));
1642 1643 1644 1645
            goto cleanup;
        }

        if (setsid() < 0) {
1646
            virReportSystemError(errno, "%s",
1647
                                 _("Unable to become session leader"));
1648 1649 1650
            goto cleanup;
        }
    }
1651

1652
    rc = virLXCControllerRun(ctrl);
1653

1654
cleanup:
1655
    virPidFileDelete(LXC_STATE_DIR, name);
1656
    virLXCControllerDeleteInterfaces(ctrl);
1657 1658 1659 1660
    for (i = 0 ; i < nttyFDs ; i++)
        VIR_FORCE_CLOSE(ttyFDs[i]);
    VIR_FREE(ttyFDs);

1661
    virLXCControllerFree(ctrl);
1662

1663
    return rc < 0? EXIT_FAILURE : EXIT_SUCCESS;
1664
}