static virLXCMonitorPtr virLXCProcessConnectMonitor(virLXCDriverPtr driver,
virDomainObjPtr vm)
{
virLXCMonitorPtr monitor = NULL;
virLXCDriverConfigPtr cfg = virLXCDriverGetConfig(driver);
if (virSecurityManagerSetSocketLabel(driver->securityManager, vm->def) < 0)
goto cleanup;
/* Hold an extra reference because we can't allow 'vm' to be
* deleted while the monitor is active */
virObjectRef(vm);
monitor = virLXCMonitorNew(vm, cfg->stateDir, &monitorCallbacks);
if (monitor == NULL)
virObjectUnref(vm);
if (virSecurityManagerClearSocketLabel(driver->securityManager, vm->def) < 0) {
if (monitor) {
virObjectUnref(monitor);
monitor = NULL;
}
goto cleanup;
}
cleanup:
virObjectUnref(cfg);
return monitor;
}
static void virLXCProcessMonitorInitNotify(virLXCMonitorPtr mon ATTRIBUTE_UNUSED,
pid_t initpid,
virDomainObjPtr vm)
{
virLXCDriverPtr driver = lxc_driver;
virLXCDomainObjPrivatePtr priv;
virLXCDriverConfigPtr cfg = virLXCDriverGetConfig(driver);
ino_t inode;
virObjectLock(vm);
priv = vm->privateData;
priv->initpid = initpid;
if (virLXCProcessGetNsInode(initpid, "pid", &inode) < 0) {
virErrorPtr err = virGetLastError();
VIR_WARN("Cannot obtain pid NS inode for %llu: %s",
(unsigned long long)initpid,
err && err->message ? err->message : "<unknown>");
virResetLastError();
inode = 0;
}
virDomainAuditInit(vm, initpid, inode);
if (virDomainSaveStatus(lxc_driver->xmlopt, cfg->stateDir, vm) < 0)
VIR_WARN("Cannot update XML with PID for LXC %s", vm->def->name);
virObjectUnlock(vm);
virObjectUnref(cfg);
}
char *virLXCProcessSetupInterfaceDirect(virConnectPtr conn,
virDomainDefPtr def,
virDomainNetDefPtr net)
{
char *ret = NULL;
char *res_ifname = NULL;
virLXCDriverPtr driver = conn->privateData;
virNetDevBandwidthPtr bw;
virNetDevVPortProfilePtr prof;
virLXCDriverConfigPtr cfg = virLXCDriverGetConfig(driver);
const char *linkdev = virDomainNetGetActualDirectDev(net);
unsigned int macvlan_create_flags = VIR_NETDEV_MACVLAN_CREATE_IFUP;
/* XXX how todo bandwidth controls ?
* Since the 'net-ifname' is about to be moved to a different
* namespace & renamed, there will be no host side visible
* interface for the container to attach rules to
*/
bw = virDomainNetGetActualBandwidth(net);
if (bw) {
virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
_("Unable to set network bandwidth on direct interfaces"));
return NULL;
}
/* XXX how todo port profiles ?
* Although we can do the association during container
* startup, at shutdown we are unable to disassociate
* because the macvlan device was moved to the container
* and automagically dies when the container dies. So
* we have no dev to perform disassociation with.
*/
prof = virDomainNetGetActualVirtPortProfile(net);
if (prof) {
virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
_("Unable to set port profile on direct interfaces"));
return NULL;
}
if (virNetDevMacVLanCreateWithVPortProfile(
net->ifname, &net->mac,
linkdev,
virDomainNetGetActualDirectMode(net),
false, def->uuid,
prof,
&res_ifname,
VIR_NETDEV_VPORT_PROFILE_OP_CREATE,
cfg->stateDir,
macvlan_create_flags) < 0)
goto cleanup;
ret = res_ifname;
cleanup:
virObjectUnref(cfg);
return ret;
}
static virCommandPtr
virLXCProcessBuildControllerCmd(virLXCDriverPtr driver,
virDomainObjPtr vm,
int nveths,
char **veths,
int *ttyFDs,
size_t nttyFDs,
int *files,
size_t nfiles,
int handshakefd)
{
size_t i;
char *filterstr;
char *outputstr;
virCommandPtr cmd;
virLXCDriverConfigPtr cfg = virLXCDriverGetConfig(driver);
cmd = virCommandNew(vm->def->emulator);
/* The controller may call ip command, so we have to retain PATH. */
virCommandAddEnvPassBlockSUID(cmd, "PATH", "/bin:/usr/bin");
virCommandAddEnvFormat(cmd, "LIBVIRT_DEBUG=%d",
virLogGetDefaultPriority());
if (virLogGetNbFilters() > 0) {
filterstr = virLogGetFilters();
if (!filterstr) {
virReportOOMError();
goto cleanup;
}
virCommandAddEnvPair(cmd, "LIBVIRT_LOG_FILTERS", filterstr);
VIR_FREE(filterstr);
}
if (cfg->log_libvirtd) {
if (virLogGetNbOutputs() > 0) {
outputstr = virLogGetOutputs();
if (!outputstr) {
virReportOOMError();
goto cleanup;
}
virCommandAddEnvPair(cmd, "LIBVIRT_LOG_OUTPUTS", outputstr);
VIR_FREE(outputstr);
}
} else {
virCommandAddEnvFormat(cmd,
"LIBVIRT_LOG_OUTPUTS=%d:stderr",
virLogGetDefaultPriority());
}
virCommandAddArgList(cmd, "--name", vm->def->name, NULL);
for (i = 0; i < nttyFDs; i++) {
virCommandAddArg(cmd, "--console");
virCommandAddArgFormat(cmd, "%d", ttyFDs[i]);
virCommandPassFD(cmd, ttyFDs[i], 0);
}
for (i = 0; i < nfiles; i++) {
virCommandAddArg(cmd, "--passfd");
virCommandAddArgFormat(cmd, "%d", files[i]);
virCommandPassFD(cmd, files[i], 0);
}
virCommandAddArgPair(cmd, "--security",
virSecurityManagerGetModel(driver->securityManager));
virCommandAddArg(cmd, "--handshake");
virCommandAddArgFormat(cmd, "%d", handshakefd);
virCommandAddArg(cmd, "--background");
for (i = 0; i < nveths; i++) {
virCommandAddArgList(cmd, "--veth", veths[i], NULL);
}
virCommandPassFD(cmd, handshakefd, 0);
return cmd;
cleanup:
virCommandFree(cmd);
virObjectUnref(cfg);
return NULL;
}
/**
* virLXCProcessCleanup:
* @driver: pointer to driver structure
* @vm: pointer to VM to clean up
* @reason: reason for switching the VM to shutoff state
*
* Cleanout resources associated with the now dead VM
*
*/
static void virLXCProcessCleanup(virLXCDriverPtr driver,
virDomainObjPtr vm,
virDomainShutoffReason reason)
{
size_t i;
virLXCDomainObjPrivatePtr priv = vm->privateData;
virNetDevVPortProfilePtr vport = NULL;
virLXCDriverConfigPtr cfg = virLXCDriverGetConfig(driver);
VIR_DEBUG("Stopping VM name=%s pid=%d reason=%d",
vm->def->name, (int)vm->pid, (int)reason);
/* now that we know it's stopped call the hook if present */
if (virHookPresent(VIR_HOOK_DRIVER_LXC)) {
char *xml = virDomainDefFormat(vm->def, 0);
/* we can't stop the operation even if the script raised an error */
virHookCall(VIR_HOOK_DRIVER_LXC, vm->def->name,
VIR_HOOK_LXC_OP_STOPPED, VIR_HOOK_SUBOP_END,
NULL, xml, NULL);
VIR_FREE(xml);
}
/* Stop autodestroy in case guest is restarted */
virCloseCallbacksUnset(driver->closeCallbacks, vm,
lxcProcessAutoDestroy);
if (priv->monitor) {
virLXCMonitorClose(priv->monitor);
virObjectUnref(priv->monitor);
priv->monitor = NULL;
}
virPidFileDelete(cfg->stateDir, vm->def->name);
virDomainDeleteConfig(cfg->stateDir, NULL, vm);
virDomainObjSetState(vm, VIR_DOMAIN_SHUTOFF, reason);
vm->pid = -1;
vm->def->id = -1;
if (virAtomicIntDecAndTest(&driver->nactive) && driver->inhibitCallback)
driver->inhibitCallback(false, driver->inhibitOpaque);
virLXCDomainReAttachHostDevices(driver, vm->def);
for (i = 0; i < vm->def->nnets; i++) {
virDomainNetDefPtr iface = vm->def->nets[i];
vport = virDomainNetGetActualVirtPortProfile(iface);
if (iface->ifname) {
if (vport &&
vport->virtPortType == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH)
ignore_value(virNetDevOpenvswitchRemovePort(
virDomainNetGetActualBridgeName(iface),
iface->ifname));
ignore_value(virNetDevVethDelete(iface->ifname));
}
networkReleaseActualDevice(vm->def, iface);
}
virDomainConfVMNWFilterTeardown(vm);
if (priv->cgroup) {
virCgroupRemove(priv->cgroup);
virCgroupFree(&priv->cgroup);
}
/* Get machined to terminate the machine as it may not have cleaned it
* properly. See https://bugs.freedesktop.org/show_bug.cgi?id=68370 for
* the bug we are working around here.
*/
virSystemdTerminateMachine(vm->def->name, "lxc", true);
/* The "release" hook cleans up additional resources */
if (virHookPresent(VIR_HOOK_DRIVER_LXC)) {
char *xml = virDomainDefFormat(vm->def, 0);
/* we can't stop the operation even if the script raised an error */
virHookCall(VIR_HOOK_DRIVER_LXC, vm->def->name,
VIR_HOOK_LXC_OP_RELEASE, VIR_HOOK_SUBOP_END,
NULL, xml, NULL);
VIR_FREE(xml);
}
if (vm->newDef) {
virDomainDefFree(vm->def);
vm->def = vm->newDef;
vm->def->id = -1;
vm->newDef = NULL;
}
virObjectUnref(cfg);
}
/**
* virLXCProcessStart:
* @conn: pointer to connection
* @driver: pointer to driver structure
* @vm: pointer to virtual machine structure
* @autoDestroy: mark the domain for auto destruction
* @reason: reason for switching vm to running state
*
* Starts a vm
*
* Returns 0 on success or -1 in case of error
*/
int virLXCProcessStart(virConnectPtr conn,
virLXCDriverPtr driver,
virDomainObjPtr vm,
unsigned int nfiles, int *files,
bool autoDestroy,
virDomainRunningReason reason)
{
int rc = -1, r;
size_t nttyFDs = 0;
int *ttyFDs = NULL;
size_t i;
char *logfile = NULL;
int logfd = -1;
size_t nveths = 0;
char **veths = NULL;
int handshakefds[2] = { -1, -1 };
off_t pos = -1;
char ebuf[1024];
char *timestamp;
virCommandPtr cmd = NULL;
virLXCDomainObjPrivatePtr priv = vm->privateData;
virCapsPtr caps = NULL;
virErrorPtr err = NULL;
virLXCDriverConfigPtr cfg = virLXCDriverGetConfig(driver);
virCgroupPtr selfcgroup;
int status;
if (virCgroupNewSelf(&selfcgroup) < 0)
return -1;
if (!virCgroupHasController(selfcgroup,
VIR_CGROUP_CONTROLLER_CPUACCT)) {
virCgroupFree(&selfcgroup);
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("Unable to find 'cpuacct' cgroups controller mount"));
return -1;
}
if (!virCgroupHasController(selfcgroup,
VIR_CGROUP_CONTROLLER_DEVICES)) {
virCgroupFree(&selfcgroup);
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("Unable to find 'devices' cgroups controller mount"));
return -1;
}
if (!virCgroupHasController(selfcgroup,
VIR_CGROUP_CONTROLLER_MEMORY)) {
virCgroupFree(&selfcgroup);
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("Unable to find 'memory' cgroups controller mount"));
return -1;
}
virCgroupFree(&selfcgroup);
if (virFileMakePath(cfg->logDir) < 0) {
virReportSystemError(errno,
_("Cannot create log directory '%s'"),
cfg->logDir);
return -1;
}
if (!vm->def->resource) {
virDomainResourceDefPtr res;
if (VIR_ALLOC(res) < 0)
goto cleanup;
if (VIR_STRDUP(res->partition, "/machine") < 0) {
VIR_FREE(res);
goto cleanup;
}
vm->def->resource = res;
}
if (virAsprintf(&logfile, "%s/%s.log",
cfg->logDir, vm->def->name) < 0)
return -1;
if (!(caps = virLXCDriverGetCapabilities(driver, false)))
goto cleanup;
/* Do this up front, so any part of the startup process can add
* runtime state to vm->def that won't be persisted. This let's us
* report implicit runtime defaults in the XML, like vnc listen/socket
*/
VIR_DEBUG("Setting current domain def as transient");
if (virDomainObjSetDefTransient(caps, driver->xmlopt, vm, true) < 0)
goto cleanup;
/* Run an early hook to set-up missing devices */
if (virHookPresent(VIR_HOOK_DRIVER_LXC)) {
char *xml = virDomainDefFormat(vm->def, 0);
int hookret;
hookret = virHookCall(VIR_HOOK_DRIVER_LXC, vm->def->name,
VIR_HOOK_LXC_OP_PREPARE, VIR_HOOK_SUBOP_BEGIN,
NULL, xml, NULL);
VIR_FREE(xml);
/*
* If the script raised an error abort the launch
*/
if (hookret < 0)
goto cleanup;
}
if (virLXCProcessEnsureRootFS(vm) < 0)
goto cleanup;
/* Must be run before security labelling */
VIR_DEBUG("Preparing host devices");
if (virLXCPrepareHostDevices(driver, vm->def) < 0)
goto cleanup;
/* Here we open all the PTYs we need on the host OS side.
* The LXC controller will open the guest OS side PTYs
* and forward I/O between them.
*/
nttyFDs = vm->def->nconsoles;
if (VIR_ALLOC_N(ttyFDs, nttyFDs) < 0)
goto cleanup;
for (i = 0; i < vm->def->nconsoles; i++)
ttyFDs[i] = -1;
/* If you are using a SecurityDriver with dynamic labelling,
then generate a security label for isolation */
VIR_DEBUG("Generating domain security label (if required)");
if (vm->def->nseclabels &&
vm->def->seclabels[0]->type == VIR_DOMAIN_SECLABEL_DEFAULT)
vm->def->seclabels[0]->type = VIR_DOMAIN_SECLABEL_NONE;
if (virSecurityManagerGenLabel(driver->securityManager, vm->def) < 0) {
virDomainAuditSecurityLabel(vm, false);
goto cleanup;
}
virDomainAuditSecurityLabel(vm, true);
VIR_DEBUG("Setting domain security labels");
if (virSecurityManagerSetAllLabel(driver->securityManager,
vm->def, NULL) < 0)
goto cleanup;
for (i = 0; i < vm->def->nconsoles; i++) {
char *ttyPath;
if (vm->def->consoles[i]->source.type != VIR_DOMAIN_CHR_TYPE_PTY) {
virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
_("Only PTY console types are supported"));
goto cleanup;
}
if (virFileOpenTty(&ttyFDs[i], &ttyPath, 1) < 0) {
virReportSystemError(errno, "%s",
_("Failed to allocate tty"));
goto cleanup;
}
VIR_FREE(vm->def->consoles[i]->source.data.file.path);
vm->def->consoles[i]->source.data.file.path = ttyPath;
VIR_FREE(vm->def->consoles[i]->info.alias);
if (virAsprintf(&vm->def->consoles[i]->info.alias, "console%zu", i) < 0)
goto cleanup;
}
if (virLXCProcessSetupInterfaces(conn, vm->def, &nveths, &veths) < 0)
goto cleanup;
/* Save the configuration for the controller */
if (virDomainSaveConfig(cfg->stateDir, vm->def) < 0)
goto cleanup;
if ((logfd = open(logfile, O_WRONLY | O_APPEND | O_CREAT,
S_IRUSR|S_IWUSR)) < 0) {
virReportSystemError(errno,
_("Failed to open '%s'"),
logfile);
goto cleanup;
}
if (pipe(handshakefds) < 0) {
virReportSystemError(errno, "%s",
_("Unable to create pipe"));
goto cleanup;
}
if (!(cmd = virLXCProcessBuildControllerCmd(driver,
vm,
nveths, veths,
ttyFDs, nttyFDs,
files, nfiles,
handshakefds[1])))
goto cleanup;
virCommandSetOutputFD(cmd, &logfd);
virCommandSetErrorFD(cmd, &logfd);
/* now that we know it is about to start call the hook if present */
if (virHookPresent(VIR_HOOK_DRIVER_LXC)) {
char *xml = virDomainDefFormat(vm->def, 0);
int hookret;
hookret = virHookCall(VIR_HOOK_DRIVER_LXC, vm->def->name,
VIR_HOOK_LXC_OP_START, VIR_HOOK_SUBOP_BEGIN,
NULL, xml, NULL);
VIR_FREE(xml);
/*
* If the script raised an error abort the launch
*/
if (hookret < 0)
goto cleanup;
}
/* Log timestamp */
if ((timestamp = virTimeStringNow()) == NULL)
goto cleanup;
if (safewrite(logfd, timestamp, strlen(timestamp)) < 0 ||
safewrite(logfd, START_POSTFIX, strlen(START_POSTFIX)) < 0) {
VIR_WARN("Unable to write timestamp to logfile: %s",
virStrerror(errno, ebuf, sizeof(ebuf)));
}
VIR_FREE(timestamp);
/* Log generated command line */
virCommandWriteArgLog(cmd, logfd);
if ((pos = lseek(logfd, 0, SEEK_END)) < 0)
VIR_WARN("Unable to seek to end of logfile: %s",
virStrerror(errno, ebuf, sizeof(ebuf)));
virCommandRawStatus(cmd);
if (virCommandRun(cmd, &status) < 0)
goto cleanup;
if (status != 0) {
if (virLXCProcessReadLogOutput(vm, logfile, pos, ebuf,
sizeof(ebuf)) <= 0) {
if (WIFEXITED(status))
snprintf(ebuf, sizeof(ebuf), _("unexpected exit status %d"),
WEXITSTATUS(status));
else
snprintf(ebuf, sizeof(ebuf), "%s", _("terminated abnormally"));
}
virReportError(VIR_ERR_INTERNAL_ERROR,
_("guest failed to start: %s"), ebuf);
goto cleanup;
}
if (VIR_CLOSE(handshakefds[1]) < 0) {
virReportSystemError(errno, "%s", _("could not close handshake fd"));
goto cleanup;
}
/* Connect to the controller as a client *first* because
* this will block until the child has written their
* pid file out to disk & created their cgroup */
if (!(priv->monitor = virLXCProcessConnectMonitor(driver, vm))) {
/* Intentionally overwrite the real monitor error message,
* since a better one is almost always found in the logs
*/
if (virLXCProcessReadLogOutput(vm, logfile, pos, ebuf, sizeof(ebuf)) > 0) {
virResetLastError();
virReportError(VIR_ERR_INTERNAL_ERROR,
_("guest failed to start: %s"), ebuf);
}
goto cleanup;
}
/* And get its pid */
if ((r = virPidFileRead(cfg->stateDir, vm->def->name, &vm->pid)) < 0) {
if (virLXCProcessReadLogOutput(vm, logfile, pos, ebuf, sizeof(ebuf)) > 0)
virReportError(VIR_ERR_INTERNAL_ERROR,
_("guest failed to start: %s"), ebuf);
else
virReportSystemError(-r,
_("Failed to read pid file %s/%s.pid"),
cfg->stateDir, vm->def->name);
goto cleanup;
}
if (virCgroupNewDetectMachine(vm->def->name, "lxc", vm->pid,
vm->def->resource ?
vm->def->resource->partition :
NULL,
-1, &priv->cgroup) < 0)
goto error;
if (!priv->cgroup) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("No valid cgroup for machine %s"),
vm->def->name);
goto error;
}
priv->stopReason = VIR_DOMAIN_EVENT_STOPPED_FAILED;
priv->wantReboot = false;
vm->def->id = vm->pid;
virDomainObjSetState(vm, VIR_DOMAIN_RUNNING, reason);
priv->doneStopEvent = false;
if (virAtomicIntInc(&driver->nactive) == 1 && driver->inhibitCallback)
driver->inhibitCallback(true, driver->inhibitOpaque);
if (lxcContainerWaitForContinue(handshakefds[0]) < 0) {
char out[1024];
if (!(virLXCProcessReadLogOutput(vm, logfile, pos, out, 1024) < 0)) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("guest failed to start: %s"), out);
}
goto error;
}
if (autoDestroy &&
virCloseCallbacksSet(driver->closeCallbacks, vm,
conn, lxcProcessAutoDestroy) < 0)
goto error;
if (virDomainObjSetDefTransient(caps, driver->xmlopt,
vm, false) < 0)
goto error;
/* Write domain status to disk.
*
* XXX: Earlier we wrote the plain "live" domain XML to this
* location for the benefit of libvirt_lxc. We're now overwriting
* it with the live status XML instead. This is a (currently
* harmless) inconsistency we should fix one day */
if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm) < 0)
goto error;
/* finally we can call the 'started' hook script if any */
if (virHookPresent(VIR_HOOK_DRIVER_LXC)) {
char *xml = virDomainDefFormat(vm->def, 0);
int hookret;
hookret = virHookCall(VIR_HOOK_DRIVER_LXC, vm->def->name,
VIR_HOOK_LXC_OP_STARTED, VIR_HOOK_SUBOP_BEGIN,
NULL, xml, NULL);
VIR_FREE(xml);
/*
* If the script raised an error abort the launch
*/
if (hookret < 0)
goto error;
}
rc = 0;
cleanup:
if (rc != 0 && !err)
err = virSaveLastError();
virCommandFree(cmd);
if (VIR_CLOSE(logfd) < 0) {
virReportSystemError(errno, "%s", _("could not close logfile"));
rc = -1;
}
for (i = 0; i < nveths; i++) {
if (rc != 0 && veths[i])
ignore_value(virNetDevVethDelete(veths[i]));
VIR_FREE(veths[i]);
}
if (rc != 0) {
if (vm->newDef) {
virDomainDefFree(vm->newDef);
vm->newDef = NULL;
}
if (priv->monitor) {
virObjectUnref(priv->monitor);
priv->monitor = NULL;
}
virDomainConfVMNWFilterTeardown(vm);
virSecurityManagerRestoreAllLabel(driver->securityManager,
vm->def, false);
virSecurityManagerReleaseLabel(driver->securityManager, vm->def);
/* Clear out dynamically assigned labels */
if (vm->def->nseclabels &&
vm->def->seclabels[0]->type == VIR_DOMAIN_SECLABEL_DYNAMIC) {
VIR_FREE(vm->def->seclabels[0]->model);
VIR_FREE(vm->def->seclabels[0]->label);
VIR_FREE(vm->def->seclabels[0]->imagelabel);
}
}
for (i = 0; i < nttyFDs; i++)
VIR_FORCE_CLOSE(ttyFDs[i]);
VIR_FREE(ttyFDs);
VIR_FORCE_CLOSE(handshakefds[0]);
VIR_FORCE_CLOSE(handshakefds[1]);
VIR_FREE(logfile);
virObjectUnref(cfg);
virObjectUnref(caps);
if (err) {
virSetError(err);
virFreeError(err);
}
return rc;
error:
err = virSaveLastError();
virLXCProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED);
goto cleanup;
}
