libvirt/src/vz/vz_driver.c

4175 lines
114 KiB
C
Raw Normal View History

/*
* vz_driver.c: core driver functions for managing
* Parallels Cloud Server hosts
*
* Copyright (C) 2014-2015 Red Hat, Inc.
* Copyright (C) 2012 Parallels, Inc.
*
* 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, see
* <http://www.gnu.org/licenses/>.
*
*/
#include <config.h>
#include <sys/types.h>
#include <sys/poll.h>
#include <stdarg.h>
#include <unistd.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <pwd.h>
#include <sys/wait.h>
#include <sys/time.h>
#include <sys/statvfs.h>
#include "datatypes.h"
#include "virerror.h"
2012-12-12 18:06:53 +00:00
#include "viralloc.h"
2012-12-12 17:59:27 +00:00
#include "virlog.h"
#include "vircommand.h"
#include "configmake.h"
#include "virfile.h"
#include "virpidfile.h"
#include "virstoragefile.h"
#include "virstring.h"
#include "cpu/cpu.h"
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
#include "virtypedparam.h"
#include "virhostmem.h"
#include "virhostcpu.h"
#include "viraccessapicheck.h"
#include "vz_driver.h"
#include "vz_utils.h"
#include "vz_sdk.h"
#define VIR_FROM_THIS VIR_FROM_PARALLELS
VIR_LOG_INIT("parallels.parallels_driver");
#define PRLCTL "prlctl"
#define VZ_STATEDIR RUNSTATEDIR "/libvirt/vz"
static virClassPtr vzDriverClass;
static bool vz_driver_privileged;
/* pid file FD, ensures two copies of the driver can't use the same root */
static int vz_driver_lock_fd = -1;
static virMutex vz_driver_lock;
static vzDriverPtr vz_driver;
static vzConnPtr vz_conn_list;
static vzDriverPtr
vzDriverObjNew(void);
static int
vzCapsAddGuestDomain(virCapsPtr caps,
virDomainOSType ostype,
virArch arch,
const char * emulator,
virDomainVirtType virt_type)
{
virCapsGuestPtr guest;
if ((guest = virCapabilitiesAddGuest(caps, ostype, arch, emulator,
NULL, 0, NULL)) == NULL)
return -1;
if (virCapabilitiesAddGuestDomain(guest, virt_type,
NULL, NULL, 0, NULL) == NULL)
return -1;
return 0;
}
static virCapsPtr
vzBuildCapabilities(void)
{
virCapsPtr caps = NULL;
virNodeInfo nodeinfo;
virDomainOSType ostypes[] = {
VIR_DOMAIN_OSTYPE_HVM,
VIR_DOMAIN_OSTYPE_EXE
};
virArch archs[] = { VIR_ARCH_I686, VIR_ARCH_X86_64 };
const char *const emulators[] = { "vz", "parallels"};
virDomainVirtType virt_types[] = {
VIR_DOMAIN_VIRT_VZ,
VIR_DOMAIN_VIRT_PARALLELS
};
size_t i, j, k;
if ((caps = virCapabilitiesNew(virArchFromHost(),
capabilities: use bool instead of int While preparing to add a capability for active commit, I noticed that the existing code was abusing int for boolean values. * src/conf/capabilities.h (_virCapsGuestFeature, _virCapsHost) (virCapabilitiesNew, virCapabilitiesAddGuestFeature): Improve types. * src/conf/capabilities.c (virCapabilitiesNew) (virCapabilitiesAddGuestFeature): Adjust signature. * src/bhyve/bhyve_capabilities.c (virBhyveCapsBuild): Update clients. * src/esx/esx_driver.c (esxCapsInit): Likewise. * src/libxl/libxl_conf.c (libxlMakeCapabilities): Likewise. * src/lxc/lxc_conf.c (virLXCDriverCapsInit): Likewise. * src/openvz/openvz_conf.c (openvzCapsInit): Likewise. * src/parallels/parallels_driver.c (parallelsBuildCapabilities): Likewise. * src/phyp/phyp_driver.c (phypCapsInit): Likewise. * src/qemu/qemu_capabilities.c (virQEMUCapsInit) (virQEMUCapsInitGuestFromBinary): Likewise. * src/security/virt-aa-helper.c (get_definition): Likewise. * src/test/test_driver.c (testBuildCapabilities): Likewise. * src/uml/uml_conf.c (umlCapsInit): Likewise. * src/vbox/vbox_tmpl.c (vboxCapsInit): Likewise. * src/vmware/vmware_conf.c (vmwareCapsInit): Likewise. * src/xen/xen_hypervisor.c (xenHypervisorBuildCapabilities): Likewise. * src/xenapi/xenapi_driver.c (getCapsObject): Likewise. * tests/qemucaps2xmltest.c (testGetCaps): Likewise. * tests/testutils.c (virTestGenericCapsInit): Likewise. * tests/testutilslxc.c (testLXCCapsInit): Likewise. * tests/testutilsqemu.c (testQemuCapsInit): Likewise. * tests/testutilsxen.c (testXenCapsInit): Likewise. * tests/vircaps2xmltest.c (buildVirCapabilities): Likewise. * tests/vircapstest.c (buildNUMATopology): Likewise. * tests/vmx2xmltest.c (testCapsInit): Likewise. * tests/xml2vmxtest.c (testCapsInit): Likewise. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-07-14 12:56:13 +00:00
false, false)) == NULL)
return NULL;
if (virCapabilitiesInitNUMA(caps) < 0)
goto error;
if (virCapabilitiesInitCaches(caps) < 0)
goto error;
verify(G_N_ELEMENTS(archs) == G_N_ELEMENTS(emulators));
for (i = 0; i < G_N_ELEMENTS(ostypes); i++)
for (j = 0; j < G_N_ELEMENTS(archs); j++)
for (k = 0; k < G_N_ELEMENTS(emulators); k++)
if (vzCapsAddGuestDomain(caps, ostypes[i], archs[j],
emulators[k], virt_types[k]) < 0)
goto error;
if (virCapabilitiesGetNodeInfo(&nodeinfo))
goto error;
if (!(caps->host.cpu = virCPUGetHost(caps->host.arch, VIR_CPU_TYPE_HOST,
&nodeinfo, NULL)))
goto error;
if (virCapabilitiesAddHostMigrateTransport(caps, "vzmigr") < 0)
goto error;
return caps;
error:
virObjectUnref(caps);
return NULL;
}
static void vzDriverDispose(void * obj)
{
vzDriverPtr driver = obj;
prlsdkDisconnect(driver);
virObjectUnref(driver->domains);
virObjectUnref(driver->caps);
virObjectUnref(driver->xmlopt);
virObjectUnref(driver->domainEventState);
virSysinfoDefFree(driver->hostsysinfo);
}
static int vzDriverOnceInit(void)
{
if (!VIR_CLASS_NEW(vzDriver, virClassForObjectLockable()))
return -1;
return 0;
}
VIR_ONCE_GLOBAL_INIT(vzDriver);
vzDriverPtr
vzGetDriverConnection(void)
{
if (!vz_driver_privileged) {
virReportError(VIR_ERR_INTERNAL_ERROR,
"%s", _("vz state driver is not active"));
return NULL;
}
virMutexLock(&vz_driver_lock);
if (!vz_driver)
vz_driver = vzDriverObjNew();
virObjectRef(vz_driver);
virMutexUnlock(&vz_driver_lock);
return vz_driver;
}
void
vzDestroyDriverConnection(void)
{
vzDriverPtr driver;
vzConnPtr privconn_list;
virMutexLock(&vz_driver_lock);
driver = g_steal_pointer(&vz_driver);
privconn_list = g_steal_pointer(&vz_conn_list);
virMutexUnlock(&vz_driver_lock);
while (privconn_list) {
vzConnPtr privconn = privconn_list;
privconn_list = privconn->next;
virConnectCloseCallbackDataCall(privconn->closeCallback,
VIR_CONNECT_CLOSE_REASON_EOF);
}
virObjectUnref(driver);
}
static char *
vzConnectGetCapabilities(virConnectPtr conn)
{
vzConnPtr privconn = conn->privateData;
if (virConnectGetCapabilitiesEnsureACL(conn) < 0)
return NULL;
return virCapabilitiesFormatXML(privconn->driver->caps);
}
static int
vzDomainDefAddDefaultInputDevices(virDomainDefPtr def)
{
if (def->ngraphics == 0)
return 0;
int bus = IS_CT(def) ? VIR_DOMAIN_INPUT_BUS_PARALLELS :
VIR_DOMAIN_INPUT_BUS_PS2;
if (virDomainDefMaybeAddInput(def,
VIR_DOMAIN_INPUT_TYPE_MOUSE,
bus) < 0)
return -1;
if (virDomainDefMaybeAddInput(def,
VIR_DOMAIN_INPUT_TYPE_KBD,
bus) < 0)
return -1;
return 0;
}
static int
vzDomainDefPostParse(virDomainDefPtr def,
virCapsPtr caps G_GNUC_UNUSED,
unsigned int parseFlags G_GNUC_UNUSED,
void *opaque G_GNUC_UNUSED,
void *parseOpaque G_GNUC_UNUSED)
{
if (vzDomainDefAddDefaultInputDevices(def) < 0)
return -1;
return 0;
}
static int
vzDomainDefValidate(const virDomainDef *def,
virCapsPtr caps G_GNUC_UNUSED,
void *opaque)
{
if (vzCheckUnsupportedControllers(def, opaque) < 0)
return -1;
return 0;
}
static int
vzDomainDeviceDefPostParse(virDomainDeviceDefPtr dev,
const virDomainDef *def,
virCapsPtr caps G_GNUC_UNUSED,
unsigned int parseFlags G_GNUC_UNUSED,
void *opaque G_GNUC_UNUSED,
void *parseOpaque G_GNUC_UNUSED)
{
if (dev->type == VIR_DOMAIN_DEVICE_NET &&
(dev->data.net->type == VIR_DOMAIN_NET_TYPE_NETWORK ||
dev->data.net->type == VIR_DOMAIN_NET_TYPE_BRIDGE) &&
dev->data.net->model == VIR_DOMAIN_NET_MODEL_UNKNOWN &&
def->os.type == VIR_DOMAIN_OSTYPE_HVM)
dev->data.net->model = VIR_DOMAIN_NET_MODEL_E1000;
return 0;
}
static int
vzDomainDeviceDefValidate(const virDomainDeviceDef *dev,
const virDomainDef *def,
void *opaque G_GNUC_UNUSED)
{
if (dev->type == VIR_DOMAIN_DEVICE_DISK)
return vzCheckUnsupportedDisk(def, dev->data.disk, opaque);
else if (dev->type == VIR_DOMAIN_DEVICE_GRAPHICS)
return vzCheckUnsupportedGraphics(dev->data.graphics);
return 0;
}
static virDomainXMLPrivateDataCallbacks vzDomainXMLPrivateDataCallbacksPtr = {
.alloc = vzDomObjAlloc,
.free = vzDomObjFree,
};
static virDomainDefParserConfig vzDomainDefParserConfig = {
.macPrefix = {0x42, 0x1C, 0x00},
.domainPostParseCallback = vzDomainDefPostParse,
.devicesPostParseCallback = vzDomainDeviceDefPostParse,
.domainValidateCallback = vzDomainDefValidate,
.deviceValidateCallback = vzDomainDeviceDefValidate,
};
static vzDriverPtr
vzDriverObjNew(void)
{
vzDriverPtr driver;
if (vzDriverInitialize() < 0)
return NULL;
if (!(driver = virObjectLockableNew(vzDriverClass)))
return NULL;
vzDomainDefParserConfig.priv = &driver->vzCaps;
if (!(driver->caps = vzBuildCapabilities()) ||
!(driver->xmlopt = virDomainXMLOptionNew(&vzDomainDefParserConfig,
&vzDomainXMLPrivateDataCallbacksPtr,
NULL, NULL, NULL)) ||
!(driver->domains = virDomainObjListNew()) ||
!(driver->domainEventState = virObjectEventStateNew()) ||
(vzInitVersion(driver) < 0) ||
(prlsdkConnect(driver) < 0)) {
virObjectUnref(driver);
return NULL;
}
driver->hostsysinfo = virSysinfoRead();
ignore_value(prlsdkLoadDomains(driver));
/* As far as waitDomainJob finally calls virReportErrorHelper
* and we are not going to report it, reset it expicitly*/
virResetLastError();
return driver;
}
static virDrvOpenStatus
vzConnectOpen(virConnectPtr conn,
virConnectAuthPtr auth G_GNUC_UNUSED,
virConfPtr conf G_GNUC_UNUSED,
unsigned int flags)
{
vzDriverPtr driver = NULL;
vzConnPtr privconn = NULL;
virCheckFlags(VIR_CONNECT_RO, VIR_DRV_OPEN_ERROR);
/* From this point on, the connection is for us. */
if (STRNEQ(conn->uri->path, "/system")) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Unexpected Virtuozzo URI path '%s', try vz:///system"),
conn->uri->path);
return VIR_DRV_OPEN_ERROR;
}
if (virConnectOpenEnsureACL(conn) < 0)
return VIR_DRV_OPEN_ERROR;
if (!(driver = vzGetDriverConnection()))
return VIR_DRV_OPEN_ERROR;
if (VIR_ALLOC(privconn) < 0)
goto error;
conn->privateData = privconn;
privconn->driver = driver;
if (!(privconn->closeCallback = virNewConnectCloseCallbackData()))
goto error;
virMutexLock(&vz_driver_lock);
privconn->next = vz_conn_list;
vz_conn_list = privconn;
virMutexUnlock(&vz_driver_lock);
return VIR_DRV_OPEN_SUCCESS;
error:
conn->privateData = NULL;
virObjectUnref(driver);
VIR_FREE(privconn);
return VIR_DRV_OPEN_ERROR;
}
static int
vzConnectClose(virConnectPtr conn)
{
vzConnPtr curr, *prev = &vz_conn_list;
vzConnPtr privconn = conn->privateData;
if (!privconn)
return 0;
virMutexLock(&vz_driver_lock);
for (curr = vz_conn_list; curr; prev = &curr->next, curr = curr->next) {
if (curr == privconn) {
*prev = curr->next;
break;
}
}
virMutexUnlock(&vz_driver_lock);
virObjectUnref(privconn->closeCallback);
virObjectUnref(privconn->driver);
VIR_FREE(privconn);
conn->privateData = NULL;
return 0;
}
static int
vzConnectGetVersion(virConnectPtr conn, unsigned long *hvVer)
{
vzConnPtr privconn = conn->privateData;
if (virConnectGetVersionEnsureACL(conn) < 0)
return -1;
*hvVer = privconn->driver->vzVersion;
return 0;
}
static char *vzConnectGetHostname(virConnectPtr conn)
{
if (virConnectGetHostnameEnsureACL(conn) < 0)
return NULL;
return virGetHostname();
}
static char *
vzConnectGetSysinfo(virConnectPtr conn, unsigned int flags)
{
vzConnPtr privconn = conn->privateData;
vzDriverPtr driver = privconn->driver;
virBuffer buf = VIR_BUFFER_INITIALIZER;
virCheckFlags(0, NULL);
if (virConnectGetSysinfoEnsureACL(conn) < 0)
return NULL;
if (!driver->hostsysinfo) {
virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
_("Host SMBIOS information is not available"));
return NULL;
}
if (virSysinfoFormat(&buf, driver->hostsysinfo) < 0)
return NULL;
return virBufferContentAndReset(&buf);
}
static int
vzConnectListDomains(virConnectPtr conn, int *ids, int maxids)
{
vzConnPtr privconn = conn->privateData;
if (virConnectListDomainsEnsureACL(conn) < 0)
return -1;
return virDomainObjListGetActiveIDs(privconn->driver->domains, ids,
maxids, virConnectListDomainsCheckACL,
conn);
}
static int
vzConnectNumOfDomains(virConnectPtr conn)
{
vzConnPtr privconn = conn->privateData;
if (virConnectNumOfDomainsEnsureACL(conn) < 0)
return -1;
return virDomainObjListNumOfDomains(privconn->driver->domains, true,
virConnectNumOfDomainsCheckACL, conn);
}
static int
vzConnectListDefinedDomains(virConnectPtr conn, char **const names, int maxnames)
{
vzConnPtr privconn = conn->privateData;
if (virConnectListDefinedDomainsEnsureACL(conn) < 0)
return -1;
memset(names, 0, sizeof(*names) * maxnames);
return virDomainObjListGetInactiveNames(privconn->driver->domains, names,
maxnames,
virConnectListDefinedDomainsCheckACL,
conn);
}
static int
vzConnectNumOfDefinedDomains(virConnectPtr conn)
{
vzConnPtr privconn = conn->privateData;
if (virConnectNumOfDefinedDomainsEnsureACL(conn) < 0)
return -1;
return virDomainObjListNumOfDomains(privconn->driver->domains, false,
virConnectNumOfDefinedDomainsCheckACL,
conn);
}
static int
vzConnectListAllDomains(virConnectPtr conn,
virDomainPtr **domains,
unsigned int flags)
{
vzConnPtr privconn = conn->privateData;
virCheckFlags(VIR_CONNECT_LIST_DOMAINS_FILTERS_ALL, -1);
if (virConnectListAllDomainsEnsureACL(conn) < 0)
return -1;
return virDomainObjListExport(privconn->driver->domains, conn, domains,
virConnectListAllDomainsCheckACL, flags);
}
static virDomainPtr
vzDomainLookupByID(virConnectPtr conn, int id)
{
vzConnPtr privconn = conn->privateData;
virDomainPtr ret = NULL;
virDomainObjPtr dom;
dom = virDomainObjListFindByID(privconn->driver->domains, id);
if (dom == NULL) {
virReportError(VIR_ERR_NO_DOMAIN, NULL);
return NULL;
}
if (virDomainLookupByIDEnsureACL(conn, dom->def) < 0)
goto cleanup;
ret = virGetDomain(conn, dom->def->name, dom->def->uuid, dom->def->id);
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static virDomainPtr
vzDomainLookupByUUID(virConnectPtr conn, const unsigned char *uuid)
{
vzConnPtr privconn = conn->privateData;
virDomainPtr ret = NULL;
virDomainObjPtr dom;
dom = virDomainObjListFindByUUID(privconn->driver->domains, uuid);
if (dom == NULL) {
char uuidstr[VIR_UUID_STRING_BUFLEN];
virUUIDFormat(uuid, uuidstr);
virReportError(VIR_ERR_NO_DOMAIN,
_("no domain with matching uuid '%s'"), uuidstr);
return NULL;
}
if (virDomainLookupByUUIDEnsureACL(conn, dom->def) < 0)
goto cleanup;
ret = virGetDomain(conn, dom->def->name, dom->def->uuid, dom->def->id);
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static virDomainPtr
vzDomainLookupByName(virConnectPtr conn, const char *name)
{
vzConnPtr privconn = conn->privateData;
virDomainPtr ret = NULL;
virDomainObjPtr dom;
dom = virDomainObjListFindByName(privconn->driver->domains, name);
if (dom == NULL) {
virReportError(VIR_ERR_NO_DOMAIN,
_("no domain with matching name '%s'"), name);
return NULL;
}
if (virDomainLookupByNameEnsureACL(conn, dom->def) < 0)
goto cleanup;
ret = virGetDomain(conn, dom->def->name, dom->def->uuid, dom->def->id);
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzDomainGetInfo(virDomainPtr domain, virDomainInfoPtr info)
{
virDomainObjPtr dom;
vzDomObjPtr privdom;
int ret = -1;
if (!(dom = vzDomObjFromDomain(domain)))
goto cleanup;
if (virDomainGetInfoEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
info->state = virDomainObjGetState(dom, NULL);
info->memory = dom->def->mem.cur_balloon;
info->maxMem = virDomainDefGetMemoryTotal(dom->def);
info->nrVirtCpu = virDomainDefGetVcpus(dom->def);
info->cpuTime = 0;
privdom = dom->privateData;
if (PRL_INVALID_HANDLE != privdom->stats && virDomainObjIsActive(dom)) {
unsigned long long vtime;
size_t i;
for (i = 0; i < virDomainDefGetVcpus(dom->def); ++i) {
if (prlsdkGetVcpuStats(privdom->stats, i, &vtime) < 0) {
virReportError(VIR_ERR_OPERATION_FAILED, "%s",
_("cannot read cputime for domain"));
goto cleanup;
}
info->cpuTime += vtime;
}
}
ret = 0;
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static char *
vzDomainGetOSType(virDomainPtr domain)
{
virDomainObjPtr dom;
char *ret = NULL;
if (!(dom = vzDomObjFromDomain(domain)))
return NULL;
if (virDomainGetOSTypeEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
ret = g_strdup(virDomainOSTypeToString(dom->def->os.type));
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzDomainIsPersistent(virDomainPtr domain)
{
virDomainObjPtr dom;
int ret = -1;
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainIsPersistentEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
ret = 1;
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzDomainGetState(virDomainPtr domain,
int *state, int *reason, unsigned int flags)
{
virDomainObjPtr dom;
int ret = -1;
virCheckFlags(0, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainGetStateEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
*state = virDomainObjGetState(dom, reason);
ret = 0;
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static char *
vzDomainGetXMLDesc(virDomainPtr domain, unsigned int flags)
{
vzConnPtr privconn = domain->conn->privateData;
virDomainDefPtr def;
virDomainObjPtr dom;
char *ret = NULL;
domain: Fix unknown flags diagnosis in virDomainGetXMLDesc Many drivers had a comment that they did not validate the incoming 'flags' to virDomainGetXMLDesc() because they were relying on virDomainDefFormat() to do it instead. This used to be the case (at least since 461e0f1a and friends in 0.9.4 added unknown flag checking in general), but regressed in commit 0ecd6851 (1.2.12), when all of the drivers were changed to pass 'flags' through the new helper virDomainDefFormatConvertXMLFlags(). Since this helper silently ignores unknown flags, we need to implement flag checking in each driver instead. Annoyingly, this means that any new flag values added will silently be ignored when targeting an older libvirt, rather than our usual practice of loudly diagnosing an unsupported flag. Add comments in domain_conf.[ch] to remind us to be extra vigilant about the impact when adding flags (a new flag to add data is safe if the older server omitting the requested data doesn't break things in the newer client; a new flag to suppress data rather than enhancing the existing VIR_DOMAIN_XML_SECURE may form a data leak or even a security hole). In the qemu driver, there are multiple callers all funnelling to qemuDomainDefFormatBufInternal(); many of them already validated flags (and often only a subset of the full set of possible flags), but for ease of maintenance, we can also check flags at the common helper function. Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: John Ferlan <jferlan@redhat.com>
2019-02-14 20:25:01 +00:00
virCheckFlags(VIR_DOMAIN_XML_COMMON_FLAGS, NULL);
if (!(dom = vzDomObjFromDomain(domain)))
return NULL;
if (virDomainGetXMLDescEnsureACL(domain->conn, dom->def, flags) < 0)
goto cleanup;
def = (flags & VIR_DOMAIN_XML_INACTIVE) &&
dom->newDef ? dom->newDef : dom->def;
ret = virDomainDefFormat(def, privconn->driver->caps, flags);
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzDomainGetAutostart(virDomainPtr domain, int *autostart)
{
virDomainObjPtr dom;
int ret = -1;
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainGetAutostartEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
*autostart = dom->autostart;
ret = 0;
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzEnsureDomainExists(virDomainObjPtr dom)
{
char uuidstr[VIR_UUID_STRING_BUFLEN];
if (!dom->removing)
return 0;
virUUIDFormat(dom->def->uuid, uuidstr);
virReportError(VIR_ERR_NO_DOMAIN,
_("no domain with matching uuid '%s' (%s)"),
uuidstr, dom->def->name);
return -1;
}
static virDomainPtr
vzDomainDefineXMLFlags(virConnectPtr conn, const char *xml, unsigned int flags)
{
vzConnPtr privconn = conn->privateData;
virDomainPtr retdom = NULL;
virDomainDefPtr def;
virDomainObjPtr dom = NULL;
unsigned int parse_flags = VIR_DOMAIN_DEF_PARSE_INACTIVE;
vzDriverPtr driver = privconn->driver;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
bool job = false;
virCheckFlags(VIR_DOMAIN_DEFINE_VALIDATE, NULL);
if (flags & VIR_DOMAIN_DEFINE_VALIDATE)
parse_flags |= VIR_DOMAIN_DEF_PARSE_VALIDATE_SCHEMA;
if ((def = virDomainDefParseString(xml, driver->caps, driver->xmlopt,
NULL, parse_flags)) == NULL)
goto cleanup;
if (virXMLCheckIllegalChars("name", def->name, "\n") < 0)
goto cleanup;
if (virDomainDefineXMLFlagsEnsureACL(conn, def) < 0)
goto cleanup;
dom = virDomainObjListFindByUUID(driver->domains, def->uuid);
if (dom == NULL) {
virResetLastError();
if (def->os.type == VIR_DOMAIN_OSTYPE_HVM) {
if (prlsdkCreateVm(driver, def))
goto cleanup;
} else if (def->os.type == VIR_DOMAIN_OSTYPE_EXE) {
if (prlsdkCreateCt(conn, def))
goto cleanup;
} else {
virReportError(VIR_ERR_INVALID_ARG,
_("Unsupported OS type: %s"),
virDomainOSTypeToString(def->os.type));
goto cleanup;
}
if (!(dom = prlsdkAddDomainByUUID(driver, def->uuid)))
goto cleanup;
} else {
int state, reason;
state = virDomainObjGetState(dom, &reason);
if (state == VIR_DOMAIN_SHUTOFF &&
reason == VIR_DOMAIN_SHUTOFF_SAVED) {
/* PCS doesn't store domain config in managed save state file.
* It's forbidden to change config for VMs in this state.
* It's possible to change config for containers, but after
* restoring domain will have that new config, not a config,
* which domain had at the moment of virDomainManagedSave.
*
* So forbid this operation, if config is changed. If it's
* not changed - just do nothing. */
if (!virDomainDefCheckABIStability(dom->def, def, driver->xmlopt)) {
virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED, "%s",
_("Can't change domain configuration "
"in managed save state"));
goto cleanup;
}
} else {
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
if (prlsdkApplyConfig(driver, dom, def))
goto cleanup;
if (prlsdkUpdateDomain(driver, dom))
goto cleanup;
}
}
retdom = virGetDomain(conn, def->name, def->uuid, def->id);
cleanup:
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
virDomainDefFree(def);
return retdom;
}
static virDomainPtr
vzDomainDefineXML(virConnectPtr conn, const char *xml)
{
return vzDomainDefineXMLFlags(conn, xml, 0);
}
static int
vzNodeGetInfo(virConnectPtr conn,
virNodeInfoPtr nodeinfo)
{
if (virNodeGetInfoEnsureACL(conn) < 0)
return -1;
return virCapabilitiesGetNodeInfo(nodeinfo);
}
static int vzConnectIsEncrypted(virConnectPtr conn G_GNUC_UNUSED)
{
/* Encryption is not relevant / applicable to way we talk to PCS */
return 0;
}
static int vzConnectIsSecure(virConnectPtr conn G_GNUC_UNUSED)
{
/* We run CLI tools directly so this is secure */
return 1;
}
static int vzConnectIsAlive(virConnectPtr conn G_GNUC_UNUSED)
{
return 1;
}
static char *
vzConnectBaselineCPU(virConnectPtr conn,
const char **xmlCPUs,
unsigned int ncpus,
unsigned int flags)
{
virCPUDefPtr *cpus = NULL;
virCPUDefPtr cpu = NULL;
char *cpustr = NULL;
virCheckFlags(VIR_CONNECT_BASELINE_CPU_EXPAND_FEATURES, NULL);
if (virConnectBaselineCPUEnsureACL(conn) < 0)
return NULL;
if (!(cpus = virCPUDefListParse(xmlCPUs, ncpus, VIR_CPU_TYPE_HOST)))
goto cleanup;
if (!(cpu = virCPUBaseline(VIR_ARCH_NONE, cpus, ncpus, NULL, NULL, false)))
goto cleanup;
if ((flags & VIR_CONNECT_BASELINE_CPU_EXPAND_FEATURES) &&
virCPUExpandFeatures(cpus[0]->arch, cpu) < 0)
goto cleanup;
cpustr = virCPUDefFormat(cpu, NULL);
cleanup:
virCPUDefListFree(cpus);
virCPUDefFree(cpu);
return cpustr;
}
static int
vzDomainGetVcpus(virDomainPtr domain,
virVcpuInfoPtr info,
int maxinfo,
unsigned char *cpumaps,
int maplen)
{
virDomainObjPtr dom = NULL;
size_t i;
int ret = -1;
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainGetVcpusEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (!virDomainObjIsActive(dom)) {
virReportError(VIR_ERR_OPERATION_INVALID,
"%s",
_("cannot list vcpu pinning for an inactive domain"));
goto cleanup;
}
if (maxinfo >= 1) {
if (info != NULL) {
vzDomObjPtr privdom;
memset(info, 0, sizeof(*info) * maxinfo);
privdom = dom->privateData;
for (i = 0; i < maxinfo; i++) {
info[i].number = i;
info[i].state = VIR_VCPU_RUNNING;
if (prlsdkGetVcpuStats(privdom->stats, i, &info[i].cpuTime) < 0)
goto cleanup;
}
}
if (cpumaps != NULL) {
memset(cpumaps, 0, maplen * maxinfo);
for (i = 0; i < maxinfo; i++)
virBitmapToDataBuf(dom->def->cpumask,
VIR_GET_CPUMAP(cpumaps, maplen, i),
maplen);
}
}
ret = maxinfo;
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzNodeGetCPUMap(virConnectPtr conn,
unsigned char **cpumap,
unsigned int *online,
unsigned int flags)
{
if (virNodeGetCPUMapEnsureACL(conn) < 0)
return -1;
return virHostCPUGetMap(cpumap, online, flags);
}
static int
vzConnectDomainEventRegisterAny(virConnectPtr conn,
virDomainPtr domain,
int eventID,
virConnectDomainEventGenericCallback callback,
void *opaque,
virFreeCallback freecb)
{
int ret = -1;
vzConnPtr privconn = conn->privateData;
if (virConnectDomainEventRegisterAnyEnsureACL(conn) < 0)
return -1;
if (virDomainEventStateRegisterID(conn,
privconn->driver->domainEventState,
domain, eventID,
callback, opaque, freecb, &ret) < 0)
ret = -1;
return ret;
}
static int
vzConnectDomainEventDeregisterAny(virConnectPtr conn,
int callbackID)
{
vzConnPtr privconn = conn->privateData;
if (virConnectDomainEventDeregisterAnyEnsureACL(conn) < 0)
return -1;
if (virObjectEventStateDeregisterID(conn,
privconn->driver->domainEventState,
events: Avoid double free possibility on remote call failure If a remote call fails during event registration (more than likely from a network failure or remote libvirtd restart timed just right), then when calling the virObjectEventStateDeregisterID we don't want to call the registered @freecb function because that breaks our contract that we would only call it after succesfully returning. If the @freecb routine were called, it could result in a double free from properly coded applications that free their opaque data on failure to register, as seen in the following details: Program terminated with signal 6, Aborted. #0 0x00007fc45cba15d7 in raise #1 0x00007fc45cba2cc8 in abort #2 0x00007fc45cbe12f7 in __libc_message #3 0x00007fc45cbe86d3 in _int_free #4 0x00007fc45d8d292c in PyDict_Fini #5 0x00007fc45d94f46a in Py_Finalize #6 0x00007fc45d960735 in Py_Main #7 0x00007fc45cb8daf5 in __libc_start_main #8 0x0000000000400721 in _start The double dereference of 'pyobj_cbData' is triggered in the following way: (1) libvirt_virConnectDomainEventRegisterAny is invoked. (2) the event is successfully added to the event callback list (virDomainEventStateRegisterClient in remoteConnectDomainEventRegisterAny returns 1 which means ok). (3) when function remoteConnectDomainEventRegisterAny is hit, network connection disconnected coincidently (or libvirtd is restarted) in the context of function 'call' then the connection is lost and the function 'call' failed, the branch virObjectEventStateDeregisterID is therefore taken. (4) 'pyobj_conn' is dereferenced the 1st time in libvirt_virConnectDomainEventFreeFunc. (5) 'pyobj_cbData' (refered to pyobj_conn) is dereferenced the 2nd time in libvirt_virConnectDomainEventRegisterAny. (6) the double free error is triggered. Resolve this by adding a @doFreeCb boolean in order to avoid calling the freeCb in virObjectEventStateDeregisterID for any remote call failure in a remoteConnect*EventRegister* API. For remoteConnect*EventDeregister* calls, the passed value would be true indicating they should run the freecb if it exists; whereas, it's false for the remote call failure path. Patch based on the investigation and initial patch posted by fangying <fangying1@huawei.com>.
2017-06-14 11:32:15 +00:00
callbackID, true) < 0)
return -1;
return 0;
}
static int
vzDomainSuspend(virDomainPtr domain)
{
vzConnPtr privconn = domain->conn->privateData;
virDomainObjPtr dom;
int ret = -1;
bool job = false;
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainSuspendEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
if (prlsdkPause(dom) < 0)
goto cleanup;
if (prlsdkUpdateDomain(privconn->driver, dom) < 0)
goto cleanup;
ret = 0;
cleanup:
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzDomainResume(virDomainPtr domain)
{
vzConnPtr privconn = domain->conn->privateData;
virDomainObjPtr dom;
int ret = -1;
bool job = false;
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainResumeEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
if (prlsdkResume(dom) < 0)
goto cleanup;
if (prlsdkUpdateDomain(privconn->driver, dom) < 0)
goto cleanup;
ret = 0;
cleanup:
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzDomainCreateWithFlags(virDomainPtr domain, unsigned int flags)
{
vzConnPtr privconn = domain->conn->privateData;
virDomainObjPtr dom;
int ret = -1;
bool job = false;
virCheckFlags(0, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainCreateWithFlagsEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
if (prlsdkStart(dom) < 0)
goto cleanup;
if (prlsdkUpdateDomain(privconn->driver, dom) < 0)
goto cleanup;
ret = 0;
cleanup:
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzDomainDestroyFlags(virDomainPtr domain, unsigned int flags)
{
vzConnPtr privconn = domain->conn->privateData;
virDomainObjPtr dom;
int ret = -1;
bool job = false;
virCheckFlags(0, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainDestroyFlagsEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
if (prlsdkKill(dom) < 0)
goto cleanup;
if (prlsdkUpdateDomain(privconn->driver, dom) < 0)
goto cleanup;
ret = 0;
cleanup:
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzDomainDestroy(virDomainPtr dom)
{
return vzDomainDestroyFlags(dom, 0);
}
static int
vzDomainShutdownFlags(virDomainPtr domain, unsigned int flags)
{
vzConnPtr privconn = domain->conn->privateData;
virDomainObjPtr dom;
int ret = -1;
bool job = false;
virCheckFlags(0, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainShutdownFlagsEnsureACL(domain->conn, dom->def, flags) < 0)
goto cleanup;
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
if (prlsdkStop(dom) < 0)
goto cleanup;
if (prlsdkUpdateDomain(privconn->driver, dom) < 0)
goto cleanup;
ret = 0;
cleanup:
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static int vzDomainShutdown(virDomainPtr dom)
{
return vzDomainShutdownFlags(dom, 0);
}
static int
vzDomainReboot(virDomainPtr domain, unsigned int flags)
{
vzConnPtr privconn = domain->conn->privateData;
virDomainObjPtr dom;
int ret = -1;
bool job = false;
virCheckFlags(0, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainRebootEnsureACL(domain->conn, dom->def, flags) < 0)
goto cleanup;
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
if (prlsdkRestart(dom) < 0)
goto cleanup;
if (prlsdkUpdateDomain(privconn->driver, dom) < 0)
goto cleanup;
ret = 0;
cleanup:
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static int vzDomainIsActive(virDomainPtr domain)
{
virDomainObjPtr dom = NULL;
int ret = -1;
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainIsActiveEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
ret = virDomainObjIsActive(dom);
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzDomainCreate(virDomainPtr domain)
{
return vzDomainCreateWithFlags(domain, 0);
}
static int
vzDomainUndefineFlags(virDomainPtr domain,
unsigned int flags)
{
vzConnPtr privconn = domain->conn->privateData;
virDomainObjPtr dom = NULL;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
int ret = -1;
bool job = false;
virCheckFlags(VIR_DOMAIN_UNDEFINE_MANAGED_SAVE |
VIR_DOMAIN_UNDEFINE_SNAPSHOTS_METADATA, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainUndefineFlagsEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
ret = prlsdkUnregisterDomain(privconn->driver, dom, flags);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
cleanup:
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzDomainUndefine(virDomainPtr domain)
{
return vzDomainUndefineFlags(domain, 0);
}
static int
vzDomainHasManagedSaveImage(virDomainPtr domain, unsigned int flags)
{
virDomainObjPtr dom = NULL;
int state, reason;
int ret = -1;
virCheckFlags(0, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainHasManagedSaveImageEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
state = virDomainObjGetState(dom, &reason);
if (state == VIR_DOMAIN_SHUTOFF && reason == VIR_DOMAIN_SHUTOFF_SAVED)
ret = 1;
else
ret = 0;
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzDomainManagedSave(virDomainPtr domain, unsigned int flags)
{
vzConnPtr privconn = domain->conn->privateData;
virDomainObjPtr dom = NULL;
int state, reason;
int ret = -1;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
bool job = false;
virCheckFlags(VIR_DOMAIN_SAVE_RUNNING |
VIR_DOMAIN_SAVE_PAUSED, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainManagedSaveEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
state = virDomainObjGetState(dom, &reason);
if (state == VIR_DOMAIN_RUNNING && (flags & VIR_DOMAIN_SAVE_PAUSED) &&
prlsdkPause(dom) < 0)
goto cleanup;
if (prlsdkSuspend(dom) < 0)
goto cleanup;
if (prlsdkUpdateDomain(privconn->driver, dom) < 0)
goto cleanup;
ret = 0;
cleanup:
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzDomainManagedSaveRemove(virDomainPtr domain, unsigned int flags)
{
virDomainObjPtr dom = NULL;
int state, reason;
int ret = -1;
virCheckFlags(0, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainManagedSaveRemoveEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
state = virDomainObjGetState(dom, &reason);
if (!(state == VIR_DOMAIN_SHUTOFF && reason == VIR_DOMAIN_SHUTOFF_SAVED))
goto cleanup;
ret = prlsdkDomainManagedSaveRemove(dom);
cleanup:
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
virDomainObjEndAPI(&dom);
return ret;
}
static int vzCheckConfigUpdateFlags(virDomainObjPtr dom, unsigned int *flags)
{
if (virDomainObjUpdateModificationImpact(dom, flags) < 0)
return -1;
if (!(*flags & VIR_DOMAIN_AFFECT_CONFIG)) {
virReportError(VIR_ERR_OPERATION_INVALID, "%s",
_("domain config update needs VIR_DOMAIN_AFFECT_CONFIG "
"flag to be set"));
return -1;
}
if (virDomainObjIsActive(dom) && !(*flags & VIR_DOMAIN_AFFECT_LIVE)) {
virReportError(VIR_ERR_OPERATION_INVALID, "%s",
_("Updates on a running domain need "
"VIR_DOMAIN_AFFECT_LIVE flag"));
return -1;
}
return 0;
}
static int vzDomainAttachDeviceFlags(virDomainPtr domain, const char *xml,
unsigned int flags)
{
int ret = -1;
vzConnPtr privconn = domain->conn->privateData;
virDomainDeviceDefPtr dev = NULL;
virDomainObjPtr dom = NULL;
vzDriverPtr driver = privconn->driver;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
bool job = false;
virCheckFlags(VIR_DOMAIN_AFFECT_LIVE |
VIR_DOMAIN_AFFECT_CONFIG, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (vzCheckConfigUpdateFlags(dom, &flags) < 0)
goto cleanup;
if (virDomainAttachDeviceFlagsEnsureACL(domain->conn, dom->def, flags) < 0)
goto cleanup;
dev = virDomainDeviceDefParse(xml, dom->def, driver->caps,
driver->xmlopt, NULL, VIR_DOMAIN_XML_INACTIVE);
if (dev == NULL)
goto cleanup;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
if (prlsdkAttachDevice(driver, dom, dev) < 0)
goto cleanup;
if (prlsdkUpdateDomain(driver, dom) < 0)
goto cleanup;
ret = 0;
cleanup:
virDomainDeviceDefFree(dev);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static int vzDomainAttachDevice(virDomainPtr domain, const char *xml)
{
return vzDomainAttachDeviceFlags(domain, xml,
VIR_DOMAIN_AFFECT_CONFIG | VIR_DOMAIN_AFFECT_LIVE);
}
static int vzDomainDetachDeviceFlags(virDomainPtr domain, const char *xml,
unsigned int flags)
{
int ret = -1;
vzConnPtr privconn = domain->conn->privateData;
virDomainDeviceDefPtr dev = NULL;
virDomainObjPtr dom = NULL;
vzDriverPtr driver = privconn->driver;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
bool job = false;
virCheckFlags(VIR_DOMAIN_AFFECT_LIVE |
VIR_DOMAIN_AFFECT_CONFIG, -1);
dom = vzDomObjFromDomain(domain);
if (dom == NULL)
return -1;
if (vzCheckConfigUpdateFlags(dom, &flags) < 0)
goto cleanup;
if (virDomainDetachDeviceFlagsEnsureACL(domain->conn, dom->def, flags) < 0)
goto cleanup;
dev = virDomainDeviceDefParse(xml, dom->def, driver->caps,
driver->xmlopt, NULL,
VIR_DOMAIN_XML_INACTIVE |
VIR_DOMAIN_DEF_PARSE_SKIP_VALIDATE);
if (dev == NULL)
goto cleanup;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
if (prlsdkDetachDevice(driver, dom, dev) < 0)
goto cleanup;
if (prlsdkUpdateDomain(driver, dom) < 0)
goto cleanup;
ret = 0;
cleanup:
virDomainDeviceDefFree(dev);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static int vzDomainDetachDevice(virDomainPtr domain, const char *xml)
{
return vzDomainDetachDeviceFlags(domain, xml,
VIR_DOMAIN_AFFECT_CONFIG | VIR_DOMAIN_AFFECT_LIVE);
}
static int
vzDomainSetUserPassword(virDomainPtr domain,
const char *user,
const char *password,
unsigned int flags)
{
virDomainObjPtr dom = NULL;
int ret = -1;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
bool job = false;
virCheckFlags(0, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainSetUserPasswordEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
ret = prlsdkDomainSetUserPassword(dom, user, password);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
cleanup:
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static int vzDomainUpdateDeviceFlags(virDomainPtr domain,
const char *xml,
unsigned int flags)
{
int ret = -1;
vzConnPtr privconn = domain->conn->privateData;
virDomainObjPtr dom = NULL;
virDomainDeviceDefPtr dev = NULL;
vzDriverPtr driver = privconn->driver;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
bool job = false;
virCheckFlags(VIR_DOMAIN_AFFECT_LIVE |
VIR_DOMAIN_AFFECT_CONFIG, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainUpdateDeviceFlagsEnsureACL(domain->conn, dom->def, flags) < 0)
goto cleanup;
if (vzCheckConfigUpdateFlags(dom, &flags) < 0)
goto cleanup;
if (!(dev = virDomainDeviceDefParse(xml, dom->def, driver->caps,
driver->xmlopt, NULL,
VIR_DOMAIN_XML_INACTIVE)))
goto cleanup;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
if (prlsdkUpdateDevice(driver, dom, dev) < 0)
goto cleanup;
if (prlsdkUpdateDomain(driver, dom) < 0)
goto cleanup;
ret = 0;
cleanup:
virDomainDeviceDefFree(dev);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static unsigned long long
vzDomainGetMaxMemory(virDomainPtr domain)
{
virDomainObjPtr dom = NULL;
int ret = -1;
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainGetMaxMemoryEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
ret = virDomainDefGetMemoryTotal(dom->def);
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
static int
vzDomainBlockStatsImpl(virDomainObjPtr dom,
const char *path,
virDomainBlockStatsPtr stats)
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
{
vzDomObjPtr privdom = dom->privateData;
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
size_t i;
int idx;
if (*path) {
if ((idx = virDomainDiskIndexByName(dom->def, path, false)) < 0) {
virReportError(VIR_ERR_INVALID_ARG, _("invalid path: %s"), path);
return -1;
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
}
if (prlsdkGetBlockStats(privdom->stats,
dom->def->disks[idx],
stats,
IS_CT(dom->def)) < 0)
return -1;
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
} else {
virDomainBlockStatsStruct s;
#define PARALLELS_ZERO_STATS(VAR, TYPE, NAME) \
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
stats->VAR = 0;
PARALLELS_BLOCK_STATS_FOREACH(PARALLELS_ZERO_STATS)
#undef PARALLELS_ZERO_STATS
for (i = 0; i < dom->def->ndisks; i++) {
if (prlsdkGetBlockStats(privdom->stats,
dom->def->disks[i],
&s,
IS_CT(dom->def)) < 0)
return -1;
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
#define PARALLELS_SUM_STATS(VAR, TYPE, NAME) \
if (s.VAR != -1) \
stats->VAR += s.VAR;
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
PARALLELS_BLOCK_STATS_FOREACH(PARALLELS_SUM_STATS)
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
#undef PARALLELS_SUM_STATS
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
}
}
stats->errs = -1;
return 0;
}
static int
vzDomainBlockStats(virDomainPtr domain,
const char *path,
virDomainBlockStatsPtr stats)
{
virDomainObjPtr dom;
int ret = -1;
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainBlockStatsEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (vzDomainBlockStatsImpl(dom, path, stats) < 0)
goto cleanup;
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
ret = 0;
cleanup:
virDomainObjEndAPI(&dom);
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
return ret;
}
static int
vzDomainBlockStatsToParams(virDomainBlockStatsPtr stats,
virTypedParameterPtr params,
int *nparams)
{
size_t i;
if (*nparams == 0) {
#define PARALLELS_COUNT_STATS(VAR, TYPE, NAME) \
if ((stats->VAR) != -1) \
++*nparams;
PARALLELS_BLOCK_STATS_FOREACH(PARALLELS_COUNT_STATS)
#undef PARALLELS_COUNT_STATS
return 0;
}
i = 0;
#define PARALLELS_BLOCK_STATS_ASSIGN_PARAM(VAR, TYPE, NAME) \
if (i < *nparams && (stats->VAR) != -1) { \
if (virTypedParameterAssign(params + i, TYPE, \
VIR_TYPED_PARAM_LLONG, (stats->VAR)) < 0) \
return -1; \
i++; \
}
PARALLELS_BLOCK_STATS_FOREACH(PARALLELS_BLOCK_STATS_ASSIGN_PARAM)
#undef PARALLELS_BLOCK_STATS_ASSIGN_PARAM
*nparams = i;
return 0;
}
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
static int
vzDomainBlockStatsFlags(virDomainPtr domain,
const char *path,
virTypedParameterPtr params,
int *nparams,
unsigned int flags)
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
{
virDomainBlockStatsStruct stats;
virDomainObjPtr dom;
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
int ret = -1;
virCheckFlags(VIR_TYPED_PARAM_STRING_OKAY, -1);
/* We don't return strings, and thus trivially support this flag. */
flags &= ~VIR_TYPED_PARAM_STRING_OKAY;
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainBlockStatsFlagsEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (vzDomainBlockStatsImpl(dom, path, &stats) < 0)
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
goto cleanup;
if (vzDomainBlockStatsToParams(&stats, params, nparams) < 0)
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
goto cleanup;
ret = 0;
cleanup:
virDomainObjEndAPI(&dom);
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
return ret;
}
static int
vzDomainInterfaceStats(virDomainPtr domain,
const char *device,
virDomainInterfaceStatsPtr stats)
{
virDomainObjPtr dom = NULL;
vzDomObjPtr privdom;
int ret = -1;
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainInterfaceStatsEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
privdom = dom->privateData;
ret = prlsdkGetNetStats(privdom->stats, privdom->sdkdom, device, stats);
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
parallels: add block device statistics to driver Statistics provided through PCS SDK. As we have only async interface in SDK we need to be subscribed to statistics in order to get it. Trivial solution on every stat request to subscribe, wait event and then unsubscribe will lead to significant delays in case of a number of successive requests, as the event will be delivered on next PCS server notify cycle. On the other hand we don't want to keep unnesessary subscribtion. So we take an hibrid solution to subcsribe on first request and then keep a subscription while requests are active. We populate cache of statistics on subscribtion events and use this cache to serve libvirts requests. * Cache details. Cache is just handle to last arrived event, we call this cache as if this handle is valid it is used to serve synchronous statistics requests. We use number of successive events count to detect that user lost interest to statistics. We reset this count to 0 on every request. If more than PARALLELS_STATISTICS_DROP_COUNT successive events arrive we unsubscribe. Special value of -1 of this counter is used to differentiate between subscribed/unsubscribed state to protect from delayed events. Values of PARALLELS_STATISTICS_DROP_COUNT and PARALLELS_STATISTICS_TIMEOUT are just drop-ins, choosen without special consideration. * Thread safety issues Use parallelsDomObjFromDomainRef in parallelsDomainBlockStats as we could wait on domain lock down on stack in prlsdkGetStatsParam and if we won't keep reference we could get dangling pointer on return from wait. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@parallels.com>
2015-06-09 07:35:53 +00:00
static int
vzDomainMemoryStats(virDomainPtr domain,
virDomainMemoryStatPtr stats,
unsigned int nr_stats,
unsigned int flags)
{
virDomainObjPtr dom = NULL;
vzDomObjPtr privdom;
int ret = -1;
virCheckFlags(0, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainMemoryStatsEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
privdom = dom->privateData;
ret = prlsdkGetMemoryStats(privdom->stats, stats, nr_stats);
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzDomainGetVcpusFlags(virDomainPtr domain,
unsigned int flags)
{
virDomainObjPtr dom;
int ret = -1;
virCheckFlags(VIR_DOMAIN_AFFECT_LIVE |
VIR_DOMAIN_AFFECT_CONFIG |
VIR_DOMAIN_VCPU_MAXIMUM, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainGetVcpusFlagsEnsureACL(domain->conn, dom->def, flags) < 0)
goto cleanup;
if (flags & VIR_DOMAIN_VCPU_MAXIMUM)
ret = virDomainDefGetVcpusMax(dom->def);
else
ret = virDomainDefGetVcpus(dom->def);
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static int vzDomainGetMaxVcpus(virDomainPtr domain)
{
return vzDomainGetVcpusFlags(domain, (VIR_DOMAIN_AFFECT_LIVE |
VIR_DOMAIN_VCPU_MAXIMUM));
}
static int vzDomainIsUpdated(virDomainPtr domain)
{
virDomainObjPtr dom;
int ret = -1;
/* As far as VZ domains are always updated (e.g. current==persistent),
* we just check for domain existence */
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainIsUpdatedEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
ret = 0;
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static int vzConnectGetMaxVcpus(virConnectPtr conn,
const char *type)
{
if (virConnectGetMaxVcpusEnsureACL(conn) < 0)
return -1;
/* As far as we have no limitation for containers
* we report maximum */
if (type == NULL || STRCASEEQ(type, "vz") || STRCASEEQ(type, "parallels"))
return 1028;
virReportError(VIR_ERR_INVALID_ARG,
_("unknown type '%s'"), type);
return -1;
}
static int
vzNodeGetCPUStats(virConnectPtr conn,
int cpuNum,
virNodeCPUStatsPtr params,
int *nparams,
unsigned int flags)
{
if (virNodeGetCPUStatsEnsureACL(conn) < 0)
return -1;
return virHostCPUGetStats(cpuNum, params, nparams, flags);
}
static int
vzNodeGetMemoryStats(virConnectPtr conn,
int cellNum,
virNodeMemoryStatsPtr params,
int *nparams,
unsigned int flags)
{
if (virNodeGetMemoryStatsEnsureACL(conn) < 0)
return -1;
return virHostMemGetStats(cellNum, params, nparams, flags);
}
static int
vzNodeGetCellsFreeMemory(virConnectPtr conn,
unsigned long long *freeMems,
int startCell,
int maxCells)
{
if (virNodeGetCellsFreeMemoryEnsureACL(conn) < 0)
return -1;
return virHostMemGetCellsFree(freeMems, startCell, maxCells);
}
static unsigned long long
vzNodeGetFreeMemory(virConnectPtr conn)
{
unsigned long long freeMem;
if (virNodeGetFreeMemoryEnsureACL(conn) < 0)
return -1;
if (virHostMemGetInfo(NULL, &freeMem) < 0)
return 0;
return freeMem;
}
static int
vzConnectRegisterCloseCallback(virConnectPtr conn,
virConnectCloseFunc cb,
void *opaque,
virFreeCallback freecb)
{
vzConnPtr privconn = conn->privateData;
int ret = -1;
if (virConnectRegisterCloseCallbackEnsureACL(conn) < 0)
return -1;
virObjectLock(privconn->driver);
if (virConnectCloseCallbackDataGetCallback(privconn->closeCallback) != NULL) {
virReportError(VIR_ERR_OPERATION_INVALID, "%s",
_("A close callback is already registered"));
goto cleanup;
}
virConnectCloseCallbackDataRegister(privconn->closeCallback, conn, cb,
opaque, freecb);
ret = 0;
cleanup:
virObjectUnlock(privconn->driver);
return ret;
}
static int
vzConnectUnregisterCloseCallback(virConnectPtr conn, virConnectCloseFunc cb)
{
vzConnPtr privconn = conn->privateData;
int ret = -1;
if (virConnectUnregisterCloseCallbackEnsureACL(conn) < 0)
return -1;
virObjectLock(privconn->driver);
if (virConnectCloseCallbackDataGetCallback(privconn->closeCallback) != cb) {
virReportError(VIR_ERR_OPERATION_INVALID, "%s",
_("A different callback was requested"));
goto cleanup;
}
virConnectCloseCallbackDataUnregister(privconn->closeCallback, cb);
ret = 0;
cleanup:
virObjectUnlock(privconn->driver);
return ret;
}
static int vzDomainSetMemoryFlags(virDomainPtr domain, unsigned long memory,
unsigned int flags)
{
virDomainObjPtr dom = NULL;
int ret = -1;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
bool job = false;
virCheckFlags(VIR_DOMAIN_AFFECT_LIVE |
VIR_DOMAIN_AFFECT_CONFIG, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (vzCheckConfigUpdateFlags(dom, &flags) < 0)
goto cleanup;
if (virDomainSetMemoryFlagsEnsureACL(domain->conn, dom->def, flags) < 0)
goto cleanup;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
ret = prlsdkSetMemsize(dom, memory >> 10);
cleanup:
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static int vzDomainSetMemory(virDomainPtr domain, unsigned long memory)
{
virDomainObjPtr dom = NULL;
int ret = -1;
bool job = false;
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainSetMemoryEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
ret = prlsdkSetMemsize(dom, memory >> 10);
cleanup:
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static virDomainMomentObjPtr
vzSnapObjFromName(virDomainSnapshotObjListPtr snapshots, const char *name)
{
virDomainMomentObjPtr snap = NULL;
snap = virDomainSnapshotFindByName(snapshots, name);
if (!snap)
virReportError(VIR_ERR_NO_DOMAIN_SNAPSHOT,
_("no domain snapshot with matching name '%s'"), name);
return snap;
}
static virDomainMomentObjPtr
vzSnapObjFromSnapshot(virDomainSnapshotObjListPtr snapshots,
virDomainSnapshotPtr snapshot)
{
return vzSnapObjFromName(snapshots, snapshot->name);
}
static int
vzDomainSnapshotNum(virDomainPtr domain, unsigned int flags)
{
virDomainObjPtr dom;
virDomainSnapshotObjListPtr snapshots = NULL;
int n = -1;
virCheckFlags(VIR_DOMAIN_SNAPSHOT_LIST_ROOTS |
VIR_DOMAIN_SNAPSHOT_FILTERS_ALL, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainSnapshotNumEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (!(snapshots = prlsdkLoadSnapshots(dom)))
goto cleanup;
n = virDomainSnapshotObjListNum(snapshots, NULL, flags);
cleanup:
virDomainSnapshotObjListFree(snapshots);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
virDomainObjEndAPI(&dom);
return n;
}
static int
vzDomainSnapshotListNames(virDomainPtr domain,
char **names,
int nameslen,
unsigned int flags)
{
virDomainObjPtr dom;
virDomainSnapshotObjListPtr snapshots = NULL;
int n = -1;
virCheckFlags(VIR_DOMAIN_SNAPSHOT_LIST_ROOTS |
VIR_DOMAIN_SNAPSHOT_FILTERS_ALL, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainSnapshotListNamesEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (!(snapshots = prlsdkLoadSnapshots(dom)))
goto cleanup;
n = virDomainSnapshotObjListGetNames(snapshots, NULL, names, nameslen, flags);
cleanup:
virDomainSnapshotObjListFree(snapshots);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
virDomainObjEndAPI(&dom);
return n;
}
static int
vzDomainListAllSnapshots(virDomainPtr domain,
virDomainSnapshotPtr **snaps,
unsigned int flags)
{
virDomainObjPtr dom;
virDomainSnapshotObjListPtr snapshots = NULL;
int n = -1;
virCheckFlags(VIR_DOMAIN_SNAPSHOT_LIST_ROOTS |
VIR_DOMAIN_SNAPSHOT_FILTERS_ALL, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainListAllSnapshotsEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (!(snapshots = prlsdkLoadSnapshots(dom)))
goto cleanup;
n = virDomainListSnapshots(snapshots, NULL, domain, snaps, flags);
cleanup:
virDomainSnapshotObjListFree(snapshots);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
virDomainObjEndAPI(&dom);
return n;
}
static char *
vzDomainSnapshotGetXMLDesc(virDomainSnapshotPtr snapshot, unsigned int flags)
{
virDomainObjPtr dom;
char *xml = NULL;
virDomainMomentObjPtr snap;
char uuidstr[VIR_UUID_STRING_BUFLEN];
virDomainSnapshotObjListPtr snapshots = NULL;
vzConnPtr privconn = snapshot->domain->conn->privateData;
virCheckFlags(VIR_DOMAIN_SNAPSHOT_XML_SECURE, NULL);
if (!(dom = vzDomObjFromDomain(snapshot->domain)))
return NULL;
if (virDomainSnapshotGetXMLDescEnsureACL(snapshot->domain->conn, dom->def, flags) < 0)
goto cleanup;
if (!(snapshots = prlsdkLoadSnapshots(dom)))
goto cleanup;
if (!(snap = vzSnapObjFromSnapshot(snapshots, snapshot)))
goto cleanup;
virUUIDFormat(snapshot->domain->uuid, uuidstr);
xml = virDomainSnapshotDefFormat(uuidstr, virDomainSnapshotObjGetDef(snap),
privconn->driver->caps,
privconn->driver->xmlopt,
virDomainSnapshotFormatConvertXMLFlags(flags));
cleanup:
virDomainSnapshotObjListFree(snapshots);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
virDomainObjEndAPI(&dom);
return xml;
}
static int
vzDomainSnapshotNumChildren(virDomainSnapshotPtr snapshot, unsigned int flags)
{
virDomainObjPtr dom;
virDomainMomentObjPtr snap;
virDomainSnapshotObjListPtr snapshots = NULL;
int n = -1;
virCheckFlags(VIR_DOMAIN_SNAPSHOT_LIST_DESCENDANTS |
VIR_DOMAIN_SNAPSHOT_FILTERS_ALL, -1);
if (!(dom = vzDomObjFromDomain(snapshot->domain)))
return -1;
if (virDomainSnapshotNumChildrenEnsureACL(snapshot->domain->conn, dom->def) < 0)
goto cleanup;
if (!(snapshots = prlsdkLoadSnapshots(dom)))
goto cleanup;
if (!(snap = vzSnapObjFromSnapshot(snapshots, snapshot)))
goto cleanup;
n = virDomainSnapshotObjListNum(snapshots, snap, flags);
cleanup:
virDomainSnapshotObjListFree(snapshots);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
virDomainObjEndAPI(&dom);
return n;
}
static int
vzDomainSnapshotListChildrenNames(virDomainSnapshotPtr snapshot,
char **names,
int nameslen,
unsigned int flags)
{
virDomainObjPtr dom;
virDomainMomentObjPtr snap;
virDomainSnapshotObjListPtr snapshots = NULL;
int n = -1;
virCheckFlags(VIR_DOMAIN_SNAPSHOT_LIST_DESCENDANTS |
VIR_DOMAIN_SNAPSHOT_FILTERS_ALL, -1);
if (!(dom = vzDomObjFromDomain(snapshot->domain)))
return -1;
if (virDomainSnapshotListChildrenNamesEnsureACL(snapshot->domain->conn, dom->def) < 0)
goto cleanup;
if (!(snapshots = prlsdkLoadSnapshots(dom)))
goto cleanup;
if (!(snap = vzSnapObjFromSnapshot(snapshots, snapshot)))
goto cleanup;
n = virDomainSnapshotObjListGetNames(snapshots, snap, names, nameslen, flags);
cleanup:
virDomainSnapshotObjListFree(snapshots);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
virDomainObjEndAPI(&dom);
return n;
}
static int
vzDomainSnapshotListAllChildren(virDomainSnapshotPtr snapshot,
virDomainSnapshotPtr **snaps,
unsigned int flags)
{
virDomainObjPtr dom;
virDomainMomentObjPtr snap;
virDomainSnapshotObjListPtr snapshots = NULL;
int n = -1;
virCheckFlags(VIR_DOMAIN_SNAPSHOT_LIST_DESCENDANTS |
VIR_DOMAIN_SNAPSHOT_FILTERS_ALL, -1);
if (!(dom = vzDomObjFromDomain(snapshot->domain)))
return -1;
if (virDomainSnapshotListAllChildrenEnsureACL(snapshot->domain->conn, dom->def) < 0)
goto cleanup;
if (!(snapshots = prlsdkLoadSnapshots(dom)))
goto cleanup;
if (!(snap = vzSnapObjFromSnapshot(snapshots, snapshot)))
goto cleanup;
n = virDomainListSnapshots(snapshots, snap, snapshot->domain, snaps, flags);
cleanup:
virDomainSnapshotObjListFree(snapshots);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
virDomainObjEndAPI(&dom);
return n;
}
static virDomainSnapshotPtr
vzDomainSnapshotLookupByName(virDomainPtr domain,
const char *name,
unsigned int flags)
{
virDomainObjPtr dom;
virDomainMomentObjPtr snap;
virDomainSnapshotPtr snapshot = NULL;
virDomainSnapshotObjListPtr snapshots = NULL;
virCheckFlags(0, NULL);
if (!(dom = vzDomObjFromDomain(domain)))
return NULL;
if (virDomainSnapshotLookupByNameEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (!(snapshots = prlsdkLoadSnapshots(dom)))
goto cleanup;
if (!(snap = vzSnapObjFromName(snapshots, name)))
goto cleanup;
snapshot = virGetDomainSnapshot(domain, snap->def->name);
cleanup:
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
virDomainObjEndAPI(&dom);
virDomainSnapshotObjListFree(snapshots);
return snapshot;
}
static int
vzDomainHasCurrentSnapshot(virDomainPtr domain, unsigned int flags)
{
virDomainObjPtr dom;
virDomainSnapshotObjListPtr snapshots = NULL;
int ret = -1;
virCheckFlags(0, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainHasCurrentSnapshotEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (!(snapshots = prlsdkLoadSnapshots(dom)))
goto cleanup;
ret = virDomainSnapshotGetCurrent(snapshots) != NULL;
cleanup:
virDomainSnapshotObjListFree(snapshots);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
virDomainObjEndAPI(&dom);
return ret;
}
static virDomainSnapshotPtr
vzDomainSnapshotGetParent(virDomainSnapshotPtr snapshot, unsigned int flags)
{
virDomainObjPtr dom;
virDomainMomentObjPtr snap;
virDomainSnapshotPtr parent = NULL;
virDomainSnapshotObjListPtr snapshots = NULL;
virCheckFlags(0, NULL);
if (!(dom = vzDomObjFromDomain(snapshot->domain)))
return NULL;
if (virDomainSnapshotGetParentEnsureACL(snapshot->domain->conn, dom->def) < 0)
goto cleanup;
if (!(snapshots = prlsdkLoadSnapshots(dom)))
goto cleanup;
if (!(snap = vzSnapObjFromSnapshot(snapshots, snapshot)))
goto cleanup;
if (!snap->def->parent_name) {
virReportError(VIR_ERR_NO_DOMAIN_SNAPSHOT,
_("snapshot '%s' does not have a parent"),
snap->def->name);
goto cleanup;
}
parent = virGetDomainSnapshot(snapshot->domain, snap->def->parent_name);
cleanup:
virDomainSnapshotObjListFree(snapshots);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
virDomainObjEndAPI(&dom);
return parent;
}
static virDomainSnapshotPtr
vzDomainSnapshotCurrent(virDomainPtr domain, unsigned int flags)
{
virDomainObjPtr dom;
virDomainSnapshotPtr snapshot = NULL;
virDomainSnapshotObjListPtr snapshots = NULL;
virDomainMomentObjPtr current;
virCheckFlags(0, NULL);
if (!(dom = vzDomObjFromDomain(domain)))
return NULL;
if (virDomainSnapshotCurrentEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (!(snapshots = prlsdkLoadSnapshots(dom)))
goto cleanup;
if (!(current = virDomainSnapshotGetCurrent(snapshots))) {
virReportError(VIR_ERR_NO_DOMAIN_SNAPSHOT, "%s",
_("the domain does not have a current snapshot"));
goto cleanup;
}
snapshot = virGetDomainSnapshot(domain, current->def->name);
cleanup:
virDomainSnapshotObjListFree(snapshots);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
virDomainObjEndAPI(&dom);
return snapshot;
}
static int
vzDomainSnapshotIsCurrent(virDomainSnapshotPtr snapshot, unsigned int flags)
{
virDomainObjPtr dom;
int ret = -1;
virDomainSnapshotObjListPtr snapshots = NULL;
virDomainMomentObjPtr current;
virCheckFlags(0, -1);
if (!(dom = vzDomObjFromDomain(snapshot->domain)))
return -1;
if (virDomainSnapshotIsCurrentEnsureACL(snapshot->domain->conn, dom->def) < 0)
goto cleanup;
if (!(snapshots = prlsdkLoadSnapshots(dom)))
goto cleanup;
current = virDomainSnapshotGetCurrent(snapshots);
ret = current && STREQ(snapshot->name, current->def->name);
cleanup:
virDomainSnapshotObjListFree(snapshots);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzDomainSnapshotHasMetadata(virDomainSnapshotPtr snapshot,
unsigned int flags)
{
virDomainObjPtr dom;
int ret = -1;
virDomainMomentObjPtr snap;
virDomainSnapshotObjListPtr snapshots = NULL;
virCheckFlags(0, -1);
if (!(dom = vzDomObjFromDomain(snapshot->domain)))
return -1;
if (virDomainSnapshotHasMetadataEnsureACL(snapshot->domain->conn, dom->def) < 0)
goto cleanup;
if (!(snapshots = prlsdkLoadSnapshots(dom)))
goto cleanup;
if (!(snap = vzSnapObjFromSnapshot(snapshots, snapshot)))
goto cleanup;
ret = 1;
cleanup:
virDomainSnapshotObjListFree(snapshots);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
virDomainObjEndAPI(&dom);
return ret;
}
static virDomainSnapshotPtr
vzDomainSnapshotCreateXML(virDomainPtr domain,
const char *xmlDesc,
unsigned int flags)
{
virDomainSnapshotPtr snapshot = NULL;
virDomainObjPtr dom;
vzConnPtr privconn = domain->conn->privateData;
vzDriverPtr driver = privconn->driver;
unsigned int parse_flags = VIR_DOMAIN_SNAPSHOT_PARSE_DISKS;
virDomainSnapshotObjListPtr snapshots = NULL;
virDomainMomentObjPtr current;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
bool job = false;
g_autoptr(virDomainSnapshotDef) def = NULL;
virCheckFlags(VIR_DOMAIN_SNAPSHOT_CREATE_VALIDATE, NULL);
if (!(dom = vzDomObjFromDomain(domain)))
return NULL;
if (virDomainSnapshotCreateXMLEnsureACL(domain->conn, dom->def, flags) < 0)
goto cleanup;
if (flags & VIR_DOMAIN_SNAPSHOT_CREATE_VALIDATE)
parse_flags |= VIR_DOMAIN_SNAPSHOT_PARSE_VALIDATE;
if (!(def = virDomainSnapshotDefParseString(xmlDesc, driver->caps,
driver->xmlopt, NULL, NULL,
snapshot: Drop virDomainSnapshotDef.current The only use for the 'current' member of virDomainSnapshotDef was with the PARSE/FORMAT_INTERNAL flag for controlling an internal-use <active> element marking whether a particular snapshot definition was current, and even then, only by the qemu driver on output, and by qemu and test driver on input. But this duplicates vm->snapshot_current, and gets in the way of potential simplifications to have qemu store a single file for all snapshots rather than one file per snapshot. Get rid of the member by adding a bool* parameter during parse (ignored if the PARSE_INTERNAL flag is not set), and by adding a new flag during format (if FORMAT_INTERNAL is set, the value printed in <active> depends on the new FORMAT_CURRENT). Then update the qemu driver accordingly, which involves hoisting assignments to vm->current_snapshot to occur prior to any point where a snapshot XML file is written (although qemu kept vm->current_snapshot and snapshot->def_current in sync by the end of the function, they were not always identical in the middle of functions, so the shuffling gets a bit interesting). Later patches will clean up some of that confusing churn to vm->current_snapshot. Note: even if later patches refactor qemu to no longer use FORMAT_INTERNAL for output (by storing bulk snapshot XML instead), we will always need PARSE_INTERNAL for input (because on upgrade, a new libvirt still has to parse XML left from a previous libvirt). Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: John Ferlan <jferlan@redhat.com>
2019-03-19 03:56:19 +00:00
parse_flags)))
goto cleanup;
if (def->ndisks > 0) {
virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
_("configuring disks is not supported for vz snapshots"));
goto cleanup;
}
if (def->memory) {
virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
_("configuring memory location is not supported"));
goto cleanup;
}
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
/* snaphot name is ignored, it will be set to auto generated by sdk uuid */
if (prlsdkCreateSnapshot(dom, def->parent.description) < 0)
goto cleanup;
if (!(snapshots = prlsdkLoadSnapshots(dom)))
goto cleanup;
if (!(current = virDomainSnapshotGetCurrent(snapshots))) {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("can't find created snapshot"));
goto cleanup;
}
/* hopefully new current snapshot is newly created one */
snapshot = virGetDomainSnapshot(domain, current->def->name);
cleanup:
virDomainSnapshotObjListFree(snapshots);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return snapshot;
}
static int
vzDomainSnapshotDelete(virDomainSnapshotPtr snapshot, unsigned int flags)
{
virDomainObjPtr dom;
int ret = -1;
virCheckFlags(VIR_DOMAIN_SNAPSHOT_DELETE_CHILDREN, -1);
if (!(dom = vzDomObjFromDomain(snapshot->domain)))
return -1;
if (virDomainSnapshotDeleteEnsureACL(snapshot->domain->conn, dom->def) < 0)
goto cleanup;
ret = prlsdkDeleteSnapshot(dom, snapshot->name,
flags & VIR_DOMAIN_SNAPSHOT_DELETE_CHILDREN);
cleanup:
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzDomainRevertToSnapshot(virDomainSnapshotPtr snapshot, unsigned int flags)
{
virDomainObjPtr dom;
int ret = -1;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
bool job = false;
virCheckFlags(VIR_DOMAIN_SNAPSHOT_REVERT_PAUSED, -1);
if (!(dom = vzDomObjFromDomain(snapshot->domain)))
return -1;
if (virDomainRevertToSnapshotEnsureACL(snapshot->domain->conn, dom->def) < 0)
goto cleanup;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
ret = prlsdkSwitchToSnapshot(dom, snapshot->name,
flags & VIR_DOMAIN_SNAPSHOT_REVERT_PAUSED);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
cleanup:
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
enum vzMigrationCookieFeatures {
VZ_MIGRATION_COOKIE_SESSION_UUID = (1 << 0),
VZ_MIGRATION_COOKIE_DOMAIN_UUID = (1 << 1),
VZ_MIGRATION_COOKIE_DOMAIN_NAME = (1 << 1),
};
typedef struct _vzMigrationCookie vzMigrationCookie;
typedef vzMigrationCookie *vzMigrationCookiePtr;
struct _vzMigrationCookie {
unsigned char *session_uuid;
unsigned char *uuid;
char *name;
};
static void
vzMigrationCookieFree(vzMigrationCookiePtr mig)
{
if (!mig)
return;
VIR_FREE(mig->session_uuid);
VIR_FREE(mig->uuid);
VIR_FREE(mig->name);
VIR_FREE(mig);
}
static int
vzBakeCookie(vzDriverPtr driver,
virDomainObjPtr dom,
char **cookieout, int *cookieoutlen,
unsigned int flags)
{
char uuidstr[VIR_UUID_STRING_BUFLEN];
virBuffer buf = VIR_BUFFER_INITIALIZER;
if (!cookieout || !cookieoutlen) {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("Migration cookie parameters are not provided."));
return -1;
}
*cookieout = NULL;
*cookieoutlen = 0;
virBufferAddLit(&buf, "<vz-migration>\n");
virBufferAdjustIndent(&buf, 2);
if (flags & VZ_MIGRATION_COOKIE_SESSION_UUID) {
virUUIDFormat(driver->session_uuid, uuidstr);
virBufferAsprintf(&buf, "<session-uuid>%s</session-uuid>\n", uuidstr);
}
if (flags & VZ_MIGRATION_COOKIE_DOMAIN_UUID) {
unsigned char fakeuuid[VIR_UUID_BUFLEN] = { 0 };
/* if dom is NULL just pass some parsable uuid for backward compat.
* It is not used by peer */
virUUIDFormat(dom ? dom->def->uuid : fakeuuid, uuidstr);
virBufferAsprintf(&buf, "<uuid>%s</uuid>\n", uuidstr);
}
if (flags & VZ_MIGRATION_COOKIE_DOMAIN_NAME) {
/* if dom is NULL just pass some name for backward compat.
* It is not used by peer */
virBufferAsprintf(&buf, "<name>%s</name>\n", dom ? dom->def->name :
"__fakename__");
}
virBufferAdjustIndent(&buf, -2);
virBufferAddLit(&buf, "</vz-migration>\n");
*cookieout = virBufferContentAndReset(&buf);
*cookieoutlen = strlen(*cookieout) + 1;
return 0;
}
static vzMigrationCookiePtr
vzEatCookie(const char *cookiein, int cookieinlen, unsigned int flags)
{
xmlDocPtr doc = NULL;
xmlXPathContextPtr ctx = NULL;
char *tmp = NULL;
vzMigrationCookiePtr mig = NULL;
if (VIR_ALLOC(mig) < 0)
return NULL;
if (!cookiein || cookieinlen <= 0 || cookiein[cookieinlen - 1] != '\0') {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("Invalid migration cookie"));
goto error;
}
if (!(doc = virXMLParseStringCtxt(cookiein,
_("(_migration_cookie)"), &ctx)))
goto error;
if ((flags & VZ_MIGRATION_COOKIE_SESSION_UUID)
&& (!(tmp = virXPathString("string(./session-uuid[1])", ctx))
|| (VIR_ALLOC_N(mig->session_uuid, VIR_UUID_BUFLEN) < 0)
|| (virUUIDParse(tmp, mig->session_uuid) < 0))) {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("missing or malformed session-uuid element "
"in migration data"));
VIR_FREE(tmp);
goto error;
}
VIR_FREE(tmp);
if ((flags & VZ_MIGRATION_COOKIE_DOMAIN_UUID)
&& (!(tmp = virXPathString("string(./uuid[1])", ctx))
|| (VIR_ALLOC_N(mig->uuid, VIR_UUID_BUFLEN) < 0)
|| (virUUIDParse(tmp, mig->uuid) < 0))) {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("missing or malformed uuid element in migration data"));
VIR_FREE(tmp);
goto error;
}
VIR_FREE(tmp);
if ((flags & VZ_MIGRATION_COOKIE_DOMAIN_NAME)
&& !(mig->name = virXPathString("string(./name[1])", ctx))) {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("missing name element in migration data"));
goto error;
}
cleanup:
xmlXPathFreeContext(ctx);
xmlFreeDoc(doc);
return mig;
error:
vzMigrationCookieFree(mig);
mig = NULL;
goto cleanup;
}
#define VZ_MIGRATION_FLAGS (VIR_MIGRATE_PAUSED | \
VIR_MIGRATE_PEER2PEER | \
VIR_MIGRATE_LIVE | \
VIR_MIGRATE_UNDEFINE_SOURCE | \
VIR_MIGRATE_PERSIST_DEST | \
VIR_MIGRATE_NON_SHARED_INC)
#define VZ_MIGRATION_PARAMETERS \
VIR_MIGRATE_PARAM_DEST_XML, VIR_TYPED_PARAM_STRING, \
VIR_MIGRATE_PARAM_URI, VIR_TYPED_PARAM_STRING, \
VIR_MIGRATE_PARAM_DEST_NAME, VIR_TYPED_PARAM_STRING, \
VIR_MIGRATE_PARAM_BANDWIDTH, VIR_TYPED_PARAM_ULLONG, \
NULL
static char *
vzDomainMigrateBeginStep(virDomainObjPtr dom,
vzDriverPtr driver,
virTypedParameterPtr params,
int nparams,
char **cookieout,
int *cookieoutlen)
{
/* we can't do this check via VZ_MIGRATION_PARAMETERS as on preparation
* step domain xml will be passed via this parameter and it is a common
* style to use single allowed parameter list definition in all steps */
if (virTypedParamsGet(params, nparams, VIR_MIGRATE_PARAM_DEST_XML)) {
virReportError(VIR_ERR_OPERATION_UNSUPPORTED, "%s",
_("Changing destination XML is not supported"));
return NULL;
}
/* session uuid, domain uuid and domain name are for backward compat */
if (vzBakeCookie(driver, dom, cookieout, cookieoutlen,
VZ_MIGRATION_COOKIE_SESSION_UUID
| VZ_MIGRATION_COOKIE_DOMAIN_UUID
| VZ_MIGRATION_COOKIE_DOMAIN_NAME) < 0)
return NULL;
return virDomainDefFormat(dom->def, driver->caps,
VIR_DOMAIN_XML_MIGRATABLE);
}
static char *
vzDomainMigrateBegin3Params(virDomainPtr domain,
virTypedParameterPtr params,
int nparams,
char **cookieout,
int *cookieoutlen,
unsigned int flags)
{
char *xml = NULL;
virDomainObjPtr dom = NULL;
vzConnPtr privconn = domain->conn->privateData;
unsigned long long bandwidth = 0;
virCheckFlags(VZ_MIGRATION_FLAGS, NULL);
if (virTypedParamsValidate(params, nparams, VZ_MIGRATION_PARAMETERS) < 0)
goto cleanup;
if (virTypedParamsGetULLong(params, nparams, VIR_MIGRATE_PARAM_BANDWIDTH,
&bandwidth) < 0)
goto cleanup;
if (bandwidth > 0) {
virReportError(VIR_ERR_OPERATION_UNSUPPORTED, "%s",
_("Bandwidth rate limiting is not supported"));
goto cleanup;
}
if (!(dom = vzDomObjFromDomain(domain)))
goto cleanup;
if (virDomainMigrateBegin3ParamsEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
xml = vzDomainMigrateBeginStep(dom, privconn->driver, params, nparams,
cookieout, cookieoutlen);
cleanup:
virDomainObjEndAPI(&dom);
return xml;
}
static char*
vzMigrationCreateURI(void)
{
char *hostname = NULL;
char *uri = NULL;
if (!(hostname = virGetHostname()))
goto cleanup;
if (STRPREFIX(hostname, "localhost")) {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("hostname on destination resolved to localhost,"
" but migration requires an FQDN"));
goto cleanup;
}
if (virAsprintf(&uri, "vzmigr://%s", hostname) < 0)
goto cleanup;
cleanup:
VIR_FREE(hostname);
return uri;
}
static int
vzDomainMigratePrepare3Params(virConnectPtr conn,
virTypedParameterPtr params,
int nparams,
const char *cookiein G_GNUC_UNUSED,
int cookieinlen G_GNUC_UNUSED,
char **cookieout,
int *cookieoutlen,
char **uri_out,
unsigned int flags)
{
vzConnPtr privconn = conn->privateData;
vzDriverPtr driver = privconn->driver;
const char *miguri = NULL;
const char *dname = NULL;
const char *dom_xml = NULL;
virDomainDefPtr def = NULL;
int ret = -1;
virCheckFlags(VZ_MIGRATION_FLAGS, -1);
if (virTypedParamsValidate(params, nparams, VZ_MIGRATION_PARAMETERS) < 0)
goto cleanup;
if (virTypedParamsGetString(params, nparams,
VIR_MIGRATE_PARAM_URI, &miguri) < 0 ||
virTypedParamsGetString(params, nparams,
VIR_MIGRATE_PARAM_DEST_XML, &dom_xml) < 0 ||
virTypedParamsGetString(params, nparams,
VIR_MIGRATE_PARAM_DEST_NAME, &dname) < 0)
goto cleanup;
/* We must set uri_out if miguri is not set. This is direct
* managed migration requirement */
if (!miguri && !(*uri_out = vzMigrationCreateURI()))
goto cleanup;
/* domain uuid and domain name are for backward compat */
if (vzBakeCookie(privconn->driver, NULL,
cookieout, cookieoutlen,
VZ_MIGRATION_COOKIE_SESSION_UUID
| VZ_MIGRATION_COOKIE_DOMAIN_UUID
| VZ_MIGRATION_COOKIE_DOMAIN_NAME) < 0)
goto cleanup;
if (!(def = virDomainDefParseString(dom_xml, driver->caps, driver->xmlopt,
NULL,
VIR_DOMAIN_DEF_PARSE_INACTIVE)))
goto cleanup;
if (dname) {
VIR_FREE(def->name);
def->name = g_strdup(dname);
}
if (virDomainMigratePrepare3ParamsEnsureACL(conn, def) < 0)
goto cleanup;
ret = 0;
cleanup:
virDomainDefFree(def);
return ret;
}
static int
vzConnectSupportsFeature(virConnectPtr conn G_GNUC_UNUSED, int feature)
{
if (virConnectSupportsFeatureEnsureACL(conn) < 0)
return -1;
switch ((virDrvFeature) feature) {
case VIR_DRV_FEATURE_MIGRATION_PARAMS:
case VIR_DRV_FEATURE_MIGRATION_P2P:
return 1;
case VIR_DRV_FEATURE_FD_PASSING:
case VIR_DRV_FEATURE_MIGRATE_CHANGE_PROTECTION:
case VIR_DRV_FEATURE_MIGRATION_DIRECT:
case VIR_DRV_FEATURE_MIGRATION_OFFLINE:
case VIR_DRV_FEATURE_MIGRATION_V1:
case VIR_DRV_FEATURE_MIGRATION_V2:
case VIR_DRV_FEATURE_MIGRATION_V3:
case VIR_DRV_FEATURE_PROGRAM_KEEPALIVE:
case VIR_DRV_FEATURE_REMOTE:
case VIR_DRV_FEATURE_REMOTE_CLOSE_CALLBACK:
case VIR_DRV_FEATURE_REMOTE_EVENT_CALLBACK:
case VIR_DRV_FEATURE_TYPED_PARAM_STRING:
case VIR_DRV_FEATURE_XML_MIGRATABLE:
default:
return 0;
}
}
static virURIPtr
vzParseVzURI(const char *uri_str)
{
virURIPtr uri = NULL;
if (!(uri = virURIParse(uri_str)))
goto error;
if (!uri->scheme || !uri->server) {
virReportError(VIR_ERR_INVALID_ARG,
_("scheme and host are mandatory vz migration URI: %s"),
uri_str);
goto error;
}
if (uri->user || uri->path || uri->query || uri->fragment) {
virReportError(VIR_ERR_INVALID_ARG,
_("only scheme, host and port are supported in "
"vz migration URI: %s"), uri_str);
goto error;
}
if (STRNEQ(uri->scheme, "vzmigr")) {
virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED,
_("unsupported scheme %s in migration URI %s"),
uri->scheme, uri_str);
goto error;
}
return uri;
error:
virURIFree(uri);
return NULL;
}
static int
vzDomainMigratePerformStep(virDomainObjPtr dom,
vzDriverPtr driver,
virTypedParameterPtr params,
int nparams,
const char *cookiein,
int cookieinlen,
unsigned int flags)
{
int ret = -1;
vzDomObjPtr privdom = dom->privateData;
virURIPtr vzuri = NULL;
const char *miguri = NULL;
const char *dname = NULL;
vzMigrationCookiePtr mig = NULL;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
bool job = false;
if (virTypedParamsGetString(params, nparams,
VIR_MIGRATE_PARAM_URI, &miguri) < 0 ||
virTypedParamsGetString(params, nparams,
VIR_MIGRATE_PARAM_DEST_NAME, &dname) < 0)
goto cleanup;
if (!miguri) {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("migrate uri is not set"));
goto cleanup;
}
if (!(mig = vzEatCookie(cookiein, cookieinlen,
VZ_MIGRATION_COOKIE_SESSION_UUID)))
goto cleanup;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
privdom->job.hasProgress = true;
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
if (!(vzuri = vzParseVzURI(miguri)))
goto cleanup;
if (prlsdkMigrate(dom, vzuri, mig->session_uuid, dname, flags) < 0)
goto cleanup;
virDomainObjListRemove(driver->domains, dom);
ret = 0;
cleanup:
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
if (job)
vzDomainObjEndJob(dom);
virURIFree(vzuri);
vzMigrationCookieFree(mig);
return ret;
}
static int
vzDomainMigratePerformP2P(virDomainObjPtr dom,
vzDriverPtr driver,
const char *dconnuri,
virTypedParameterPtr orig_params,
int nparams,
unsigned int flags)
{
virDomainPtr ddomain = NULL;
char *uri = NULL;
char *cookiein = NULL;
char *cookieout = NULL;
char *dom_xml = NULL;
int cookieinlen = 0;
int cookieoutlen = 0;
virErrorPtr orig_err = NULL;
int cancelled = 1;
virConnectPtr dconn = NULL;
virTypedParameterPtr params = NULL;
int ret = -1;
int maxparams = nparams;
if (virTypedParamsCopy(&params, orig_params, nparams) < 0)
return -1;
if (!(dconn = virConnectOpen(dconnuri)))
goto done;
if (!(dom_xml = vzDomainMigrateBeginStep(dom, driver, params, nparams,
&cookieout, &cookieoutlen)))
goto done;
if (virTypedParamsAddString(&params, &nparams, &maxparams,
VIR_MIGRATE_PARAM_DEST_XML, dom_xml) < 0)
goto done;
cookiein = g_steal_pointer(&cookieout);
cookieinlen = cookieoutlen;
cookieoutlen = 0;
virObjectUnlock(dom);
ret = dconn->driver->domainMigratePrepare3Params
(dconn, params, nparams, cookiein, cookieinlen,
&cookieout, &cookieoutlen, &uri, flags);
virObjectLock(dom);
if (ret < 0)
goto done;
ret = -1;
/* preparation step was successful, thus on any error we must perform
* finish step to finalize migration on target
*/
if (uri && virTypedParamsReplaceString(&params, &nparams,
VIR_MIGRATE_PARAM_URI, uri) < 0) {
virErrorPreserveLast(&orig_err);
goto finish;
}
VIR_FREE(cookiein);
cookiein = g_steal_pointer(&cookieout);
cookieinlen = cookieoutlen;
cookieoutlen = 0;
if (vzDomainMigratePerformStep(dom, driver, params, nparams, cookiein,
cookieinlen, flags) < 0) {
virErrorPreserveLast(&orig_err);
goto finish;
}
cancelled = 0;
finish:
if (virTypedParamsGetString(params, nparams,
VIR_MIGRATE_PARAM_DEST_NAME, NULL) <= 0 &&
virTypedParamsReplaceString(&params, &nparams,
VIR_MIGRATE_PARAM_DEST_NAME,
dom->def->name) < 0)
goto done;
virObjectUnlock(dom);
ddomain = dconn->driver->domainMigrateFinish3Params(dconn, params, nparams,
NULL, 0, NULL, NULL,
flags, cancelled);
virObjectLock(dom);
if (ddomain)
ret = 0;
virObjectUnref(ddomain);
/* confirm step is NOOP thus no need to call it */
done:
virErrorRestore(&orig_err);
VIR_FREE(dom_xml);
VIR_FREE(uri);
VIR_FREE(cookiein);
VIR_FREE(cookieout);
virTypedParamsFree(params, nparams);
virObjectUnref(dconn);
return ret;
}
static int
vzDomainMigratePerform3Params(virDomainPtr domain,
const char *dconnuri,
virTypedParameterPtr params,
int nparams,
const char *cookiein,
int cookieinlen,
char **cookieout G_GNUC_UNUSED,
int *cookieoutlen G_GNUC_UNUSED,
unsigned int flags)
{
int ret = -1;
virDomainObjPtr dom;
vzConnPtr privconn = domain->conn->privateData;
virCheckFlags(VZ_MIGRATION_FLAGS, -1);
if (virTypedParamsValidate(params, nparams, VZ_MIGRATION_PARAMETERS) < 0)
return -1;
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainMigratePerform3ParamsEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (flags & VIR_MIGRATE_PEER2PEER)
ret = vzDomainMigratePerformP2P(dom, privconn->driver, dconnuri,
params, nparams, flags);
else
ret = vzDomainMigratePerformStep(dom, privconn->driver, params, nparams,
cookiein, cookieinlen, flags);
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static virDomainPtr
vzDomainMigrateFinish3Params(virConnectPtr dconn,
virTypedParameterPtr params,
int nparams,
const char *cookiein G_GNUC_UNUSED,
int cookieinlen G_GNUC_UNUSED,
char **cookieout G_GNUC_UNUSED,
int *cookieoutlen G_GNUC_UNUSED,
unsigned int flags,
int cancelled)
{
virDomainObjPtr dom = NULL;
virDomainPtr domain = NULL;
vzConnPtr privconn = dconn->privateData;
vzDriverPtr driver = privconn->driver;
const char *name = NULL;
virCheckFlags(VZ_MIGRATION_FLAGS, NULL);
if (virTypedParamsValidate(params, nparams, VZ_MIGRATION_PARAMETERS) < 0)
return NULL;
if (cancelled)
return NULL;
if (virTypedParamsGetString(params, nparams,
VIR_MIGRATE_PARAM_DEST_NAME, &name) < 0)
return NULL;
if (!(dom = prlsdkAddDomainByName(driver, name)))
goto cleanup;
/* At first glace at may look strange that we add domain and
* then check ACL but we touch only cache and not real system state */
if (virDomainMigrateFinish3ParamsEnsureACL(dconn, dom->def) < 0)
goto cleanup;
domain = virGetDomain(dconn, dom->def->name, dom->def->uuid, dom->def->id);
cleanup:
/* In this situation we have to restore domain on source. But the migration
* is already finished. */
if (!domain)
VIR_WARN("Can't provide domain '%s' after successful migration.", name);
vz: make vz driver more responsive Current vz driver implementation is not usable when it comes to long runnig operations. Migration or saving a domain blocks all other operations even query ones which are expecteted to be available. This patch addresses this problem. All vz driver API calls fall into next 3 groups: 1. only query domain cache (virDomainObj, vz cache statistic) examples are vzDomainGetState, vzDomainGetXMLDesc etc. 2. use thread shared sdkdom object examples are vzDomainSetMemoryFlags, vzDomainAttachDevice etc. 3. use no thread shared sdkdom object nor domain cache examples are vzDomainSnapshotListNames, vzDomainSnapshotGetXMLDesc etc API calls from group 1 don't need to be changed as they hold domain lock only for short period of time. These calls [1] are easily distinguished. They query domain object thru libvirt common code or query vz sdk statistics handle thru vz sdk sync operations. vzDomainInterfaceStats is the only exception. It uses sdkdom object to convert interface name to its vz sdk stack index which could not be saved in domain cache. Interface statistics is available thru this stack index as a key rather than name. As a result we can have accidental 'not known interface' errors on quering intrerface stats. The reason is that in the process of updating domain configuration we drop all devices and then recreate them again in sdkdom object and domain lock can be dropped meanwhile (to remove networks for existing bridged interfaces and(or) (re)create new ones). We can fix this by changing the way we support bridged interfaces or by reordering operations and changing bridged networks beforehand. Anyway this is better than moving this API call into 2 group and making it an exclusive job. As to API calls from group 2, first thread shared sdkdom object needs to be explained. vz sdk has only one handle for a given domain, thus threads need exclusive access to operate on it. These calls are fixed to drop and reacquire domain lock on any lengthy operations - namely waiting the result of async vz sdk operation. As lock is dropped we need to take extra reference to domain object if it is not taken already as domain object can be deleted from list while lock is dropped. As this operations use thread shared sdkdom object, the simplest way to make calls from group 2 be consistent to each other is to make them mutually exclusive. This is done by taking/releasing job condition thru calling correspondent job routine. This approach makes group 1 and group 2 calls consistent to each other too. Not all calls of group 2 change the domain cache but those that do update it thru prlsdkUpdateDomain which holds the lock thoughout the update. API calls from group [2] are easily distinguished too. They use beginEdit/commit to change domain configuration (vzDomainSetMemoryFlags) or/and update domain cache from sdkdom at the end of operation (vzDomainSuspend). There is a known issue however. Frankly speaking it was introduced by ealier patch '[PATCH 6/9] vz: cleanup loading domain code' from a different series. The patch significantly reduced amount of time when the driver lock is held when creating domain from API call or as a result of domain added event from vz sdk. The problem is these two paths race on using thread shared sdkdom as we don't have libvirt domain object and can not lock on it. However this don't invalidates the patch as we can't use the former approach of preadding domain into the list as we need name at least and name is not given by event. Anyway i'm against adding half baked object into the list. Eventually this race can be fixed by extra measures. As to current situation races with different configurations are unlikely and race when adding domain thru vz driver and simultaneous event from vz sdk is not dangerous as configuration is the same. The last group [3] is API calls that need only sdkdom object to make vz sdk call and don't change thread shared sdkdom object or domain cache in any way. For now these are mostly domain snapshot API calls. The changes are similar to those of group 2 - they add extra reference and drop/reacquire the lock on waiting vz async call result. One can simply take the immutable sdkdom object from the cache and drop the lock for the rest of operations but the chosen approach makes implementation of these API calls somewhat similar to those of from group 2 and thus a bit futureproof. As calls of group 3 don't need vz driver domain/vz sdk cache in any way, they are consistent with respect to API calls from groups 1 and 3. There is another exception. Calls to make-snapshot/revert-to-snapshot/migrate are moved to group 2. That is they are made mutually exclusive. The reason is that libvirt API supports control/query only for one job per domain and these are jobs that are likely to be queried/aborted. Appendix. [1] API calls that only query domain cache. (marked [*] are included for a different reason) .domainLookupByID = vzDomainLookupByID, /* 0.10.0 */ .domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */ .domainLookupByName = vzDomainLookupByName, /* 0.10.0 */ .domainGetOSType = vzDomainGetOSType, /* 0.10.0 */ .domainGetInfo = vzDomainGetInfo, /* 0.10.0 */ .domainGetState = vzDomainGetState, /* 0.10.0 */ .domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */ .domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */ .domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */ .domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */ .domainIsActive = vzDomainIsActive, /* 1.2.10 */ .domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */ .domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */ .domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */ .domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */ .domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */ .domainBlockStats = vzDomainBlockStats, /* 1.2.17 */ .domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */ .domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */ [*] .domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */ .domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */ .domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */ [2] API calls that use thread shared sdkdom object (marked [*] are included for a different reason) .domainSuspend = vzDomainSuspend, /* 0.10.0 */ .domainResume = vzDomainResume, /* 0.10.0 */ .domainDestroy = vzDomainDestroy, /* 0.10.0 */ .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ (update part) .domainUndefine = vzDomainUndefine, /* 1.2.10 */ .domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */ .domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */ .domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */ .domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */ .domainSetUserPassword = vzDomainSetUserPassword, /* 1.3.6 */ .domainManagedSave = vzDomainManagedSave, /* 1.2.14 */ .domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */ .domainSetMemory = vzDomainSetMemory, /* 1.3.4 */ .domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */ [*] .domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */ [*] .domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */ [*] .domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */ prlsdkHandleVmConfigEvent [3] API calls that do not use thread shared sdkdom object .domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */ .domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */ .domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */ .domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */ .domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */ .domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */ .domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */ .domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */ .domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */ .domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */ .domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */ .domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */ .domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */ .domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */ .domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */ [4] Known issues. 1. accidental errors on getting network statistics 2. race with simultaneous use of thread shared domain object on paths of adding domain thru API and adding domain on vz sdk domain added event.
2016-07-20 15:00:33 +00:00
virDomainObjEndAPI(&dom);
return domain;
}
static int
vzDomainMigrateConfirm3Params(virDomainPtr domain G_GNUC_UNUSED,
virTypedParameterPtr params,
int nparams,
const char *cookiein G_GNUC_UNUSED,
int cookieinlen G_GNUC_UNUSED,
unsigned int flags,
int cancelled G_GNUC_UNUSED)
{
virCheckFlags(VZ_MIGRATION_FLAGS, -1);
if (virTypedParamsValidate(params, nparams, VZ_MIGRATION_PARAMETERS) < 0)
return -1;
return 0;
}
static int
vzDomainGetJobInfoImpl(virDomainObjPtr dom, virDomainJobInfoPtr info)
{
vzDomObjPtr privdom = dom->privateData;
vzDomainJobObjPtr job = &privdom->job;
memset(info, 0, sizeof(*info));
if (!job->active || !job->hasProgress)
return 0;
if (vzDomainJobUpdateTime(job) < 0)
return -1;
info->type = VIR_DOMAIN_JOB_UNBOUNDED;
info->dataTotal = 100;
info->dataProcessed = job->progress;
info->dataRemaining = 100 - job->progress;
info->timeElapsed = job->elapsed;
return 0;
}
static int
vzDomainGetJobInfo(virDomainPtr domain, virDomainJobInfoPtr info)
{
virDomainObjPtr dom;
int ret = -1;
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainGetJobInfoEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
ret = vzDomainGetJobInfoImpl(dom, info);
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
static int
vzDomainJobInfoToParams(virDomainJobInfoPtr info,
int *type,
virTypedParameterPtr *params,
int *nparams)
{
virTypedParameterPtr par = NULL;
int maxpar = 0;
int npar = 0;
if (virTypedParamsAddULLong(&par, &npar, &maxpar,
VIR_DOMAIN_JOB_TIME_ELAPSED,
info->timeElapsed) < 0 ||
virTypedParamsAddULLong(&par, &npar, &maxpar,
VIR_DOMAIN_JOB_DATA_TOTAL,
info->dataTotal) < 0 ||
virTypedParamsAddULLong(&par, &npar, &maxpar,
VIR_DOMAIN_JOB_DATA_PROCESSED,
info->dataProcessed) < 0 ||
virTypedParamsAddULLong(&par, &npar, &maxpar,
VIR_DOMAIN_JOB_DATA_REMAINING,
info->dataRemaining) < 0)
goto error;
*type = info->type;
*params = par;
*nparams = npar;
return 0;
error:
virTypedParamsFree(par, npar);
return -1;
}
static int
vzDomainGetJobStats(virDomainPtr domain,
int *type,
virTypedParameterPtr *params,
int *nparams,
unsigned int flags)
{
virDomainJobInfo info;
virDomainObjPtr dom;
int ret = -1;
virCheckFlags(0, -1);
if (!(dom = vzDomObjFromDomain(domain)))
return -1;
if (virDomainGetJobStatsEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (vzDomainGetJobInfoImpl(dom, &info) < 0)
goto cleanup;
if (info.type == VIR_DOMAIN_JOB_NONE) {
*type = VIR_DOMAIN_JOB_NONE;
*params = NULL;
*nparams = 0;
ret = 0;
goto cleanup;
}
ret = vzDomainJobInfoToParams(&info, type, params, nparams);
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
#define VZ_ADD_STAT_PARAM_UUL(group, field, counter) \
do { \
if (stat.field != -1) { \
snprintf(param_name, VIR_TYPED_PARAM_FIELD_LENGTH, \
group ".%zu." counter, i); \
if (virTypedParamsAddULLong(&record->params, \
&record->nparams, \
maxparams, \
param_name, \
stat.field) < 0) \
return -1; \
} \
} while (0)
static int
vzDomainGetBlockStats(virDomainObjPtr dom,
virDomainStatsRecordPtr record,
int *maxparams)
{
vzDomObjPtr privdom = dom->privateData;
size_t i;
char param_name[VIR_TYPED_PARAM_FIELD_LENGTH];
if (virTypedParamsAddUInt(&record->params,
&record->nparams,
maxparams,
"block.count",
dom->def->ndisks) < 0)
return -1;
for (i = 0; i < dom->def->ndisks; i++) {
virDomainBlockStatsStruct stat;
virDomainDiskDefPtr disk = dom->def->disks[i];
if (prlsdkGetBlockStats(privdom->stats,
disk,
&stat,
IS_CT(dom->def)) < 0)
return -1;
snprintf(param_name, VIR_TYPED_PARAM_FIELD_LENGTH,
"block.%zu.name", i);
if (virTypedParamsAddString(&record->params,
&record->nparams,
maxparams,
param_name,
disk->dst) < 0)
return -1;
if (virStorageSourceIsLocalStorage(disk->src) && disk->src->path) {
snprintf(param_name, VIR_TYPED_PARAM_FIELD_LENGTH,
"block.%zu.path", i);
if (virTypedParamsAddString(&record->params,
&record->nparams,
maxparams,
param_name,
disk->src->path) < 0)
return -1;
}
VZ_ADD_STAT_PARAM_UUL("block", rd_req, "rd.reqs");
VZ_ADD_STAT_PARAM_UUL("block", rd_bytes, "rd.bytes");
VZ_ADD_STAT_PARAM_UUL("block", wr_req, "wr.reqs");
VZ_ADD_STAT_PARAM_UUL("block", wr_bytes, "wr.bytes");
if (disk->device == VIR_DOMAIN_DISK_DEVICE_DISK) {
snprintf(param_name, VIR_TYPED_PARAM_FIELD_LENGTH,
"block.%zu.capacity", i);
if (virTypedParamsAddULLong(&record->params,
&record->nparams,
maxparams,
param_name,
disk->src->capacity) < 0)
return -1;
}
}
return 0;
}
2016-12-12 07:56:35 +00:00
static int
vzDomainGetNetStats(virDomainObjPtr dom,
virDomainStatsRecordPtr record,
int *maxparams)
{
vzDomObjPtr privdom = dom->privateData;
size_t i;
char param_name[VIR_TYPED_PARAM_FIELD_LENGTH];
if (virTypedParamsAddUInt(&record->params,
&record->nparams,
maxparams,
"net.count",
dom->def->nnets) < 0)
return -1;
for (i = 0; i < dom->def->nnets; i++) {
virDomainInterfaceStatsStruct stat;
virDomainNetDefPtr net = dom->def->nets[i];
if (prlsdkGetNetStats(privdom->stats, privdom->sdkdom, net->ifname,
&stat) < 0)
return -1;
snprintf(param_name, VIR_TYPED_PARAM_FIELD_LENGTH, "net.%zu.name", i);
if (virTypedParamsAddString(&record->params,
&record->nparams,
maxparams,
param_name,
net->ifname) < 0)
return -1;
VZ_ADD_STAT_PARAM_UUL("net", rx_bytes, "rx.bytes");
VZ_ADD_STAT_PARAM_UUL("net", rx_packets, "rx.pkts");
VZ_ADD_STAT_PARAM_UUL("net", tx_bytes, "tx.bytes");
VZ_ADD_STAT_PARAM_UUL("net", tx_packets, "tx.pkts");
}
return 0;
}
2016-12-12 07:56:36 +00:00
static int
vzDomainGetVCPUStats(virDomainObjPtr dom,
virDomainStatsRecordPtr record,
int *maxparams)
{
vzDomObjPtr privdom = dom->privateData;
size_t i;
char param_name[VIR_TYPED_PARAM_FIELD_LENGTH];
if (virTypedParamsAddUInt(&record->params,
&record->nparams,
maxparams,
"vcpu.current",
virDomainDefGetVcpus(dom->def)) < 0)
return -1;
if (virTypedParamsAddUInt(&record->params,
&record->nparams,
maxparams,
"vcpu.maximum",
virDomainDefGetVcpusMax(dom->def)) < 0)
return -1;
for (i = 0; i < virDomainDefGetVcpusMax(dom->def); i++) {
int state = dom->def->vcpus[i]->online ? VIR_VCPU_RUNNING :
VIR_VCPU_OFFLINE;
unsigned long long time;
snprintf(param_name, VIR_TYPED_PARAM_FIELD_LENGTH, "vcpu.%zu.state", i);
if (virTypedParamsAddInt(&record->params,
&record->nparams,
maxparams,
param_name,
state) < 0)
return -1;
if (prlsdkGetVcpuStats(privdom->stats, i, &time) < 0)
return -1;
snprintf(param_name, VIR_TYPED_PARAM_FIELD_LENGTH, "vcpu.%zu.time", i);
if (virTypedParamsAddULLong(&record->params,
&record->nparams,
maxparams,
param_name,
time) < 0)
return -1;
}
return 0;
}
static int
vzDomainGetBalloonStats(virDomainObjPtr dom,
virDomainStatsRecordPtr record,
int *maxparams)
{
vzDomObjPtr privdom = dom->privateData;
virDomainMemoryStatStruct stats[VIR_DOMAIN_MEMORY_STAT_NR];
size_t i;
int n;
if (virTypedParamsAddULLong(&record->params,
&record->nparams,
maxparams,
"balloon.maximum",
virDomainDefGetMemoryTotal(dom->def)) < 0)
return -1;
if (virTypedParamsAddULLong(&record->params,
&record->nparams,
maxparams,
"balloon.current",
virDomainDefGetMemoryTotal(dom->def)) < 0)
return -1;
n = prlsdkGetMemoryStats(privdom->stats, stats, VIR_DOMAIN_MEMORY_STAT_NR);
if (n < 0)
return -1;
#define STORE_MEM_RECORD(TAG, NAME) \
if (stats[i].tag == VIR_DOMAIN_MEMORY_STAT_ ##TAG) \
if (virTypedParamsAddULLong(&record->params, \
&record->nparams, \
maxparams, \
"balloon." NAME, \
stats[i].val) < 0) \
return -1;
for (i = 0; i < n; i++) {
STORE_MEM_RECORD(SWAP_IN, "swap_in")
STORE_MEM_RECORD(SWAP_OUT, "swap_out")
STORE_MEM_RECORD(MAJOR_FAULT, "major_fault")
STORE_MEM_RECORD(MINOR_FAULT, "minor_fault")
STORE_MEM_RECORD(AVAILABLE, "available")
STORE_MEM_RECORD(UNUSED, "unused")
}
#undef STORE_MEM_RECORD
return 0;
}
static int
vzDomainGetStateStats(virDomainObjPtr dom,
virDomainStatsRecordPtr record,
int *maxparams)
{
if (virTypedParamsAddInt(&record->params,
&record->nparams,
maxparams,
"state.state",
dom->state.state) < 0)
return -1;
if (virTypedParamsAddInt(&record->params,
&record->nparams,
maxparams,
"state.reason",
dom->state.reason) < 0)
return -1;
return 0;
}
static virDomainStatsRecordPtr
vzDomainGetAllStats(virConnectPtr conn,
virDomainObjPtr dom)
{
virDomainStatsRecordPtr stat;
int maxparams = 0;
if (VIR_ALLOC(stat) < 0)
return NULL;
if (vzDomainGetStateStats(dom, stat, &maxparams) < 0)
goto error;
if (vzDomainGetBlockStats(dom, stat, &maxparams) < 0)
goto error;
2016-12-12 07:56:35 +00:00
if (vzDomainGetNetStats(dom, stat, &maxparams) < 0)
goto error;
2016-12-12 07:56:36 +00:00
if (vzDomainGetVCPUStats(dom, stat, &maxparams) < 0)
goto error;
if (vzDomainGetBalloonStats(dom, stat, &maxparams) < 0)
goto error;
if (!(stat->dom = virGetDomain(conn, dom->def->name, dom->def->uuid, dom->def->id)))
goto error;
return stat;
error:
virTypedParamsFree(stat->params, stat->nparams);
virObjectUnref(stat->dom);
VIR_FREE(stat);
return NULL;
}
static int
vzConnectGetAllDomainStats(virConnectPtr conn,
virDomainPtr *domains,
unsigned int ndomains,
unsigned int stats,
virDomainStatsRecordPtr **retStats,
unsigned int flags)
{
vzConnPtr privconn = conn->privateData;
vzDriverPtr driver = privconn->driver;
unsigned int lflags = flags & (VIR_CONNECT_LIST_DOMAINS_FILTERS_ACTIVE |
VIR_CONNECT_LIST_DOMAINS_FILTERS_PERSISTENT |
VIR_CONNECT_LIST_DOMAINS_FILTERS_STATE);
unsigned int supported = VIR_DOMAIN_STATS_STATE |
VIR_DOMAIN_STATS_VCPU |
VIR_DOMAIN_STATS_INTERFACE |
VIR_DOMAIN_STATS_BALLOON |
VIR_DOMAIN_STATS_BLOCK;
virDomainObjPtr *doms = NULL;
size_t ndoms;
virDomainStatsRecordPtr *tmpstats = NULL;
int nstats = 0;
int ret = -1;
size_t i;
virCheckFlags(VIR_CONNECT_LIST_DOMAINS_FILTERS_ACTIVE |
VIR_CONNECT_LIST_DOMAINS_FILTERS_PERSISTENT |
VIR_CONNECT_LIST_DOMAINS_FILTERS_STATE |
VIR_CONNECT_GET_ALL_DOMAINS_STATS_ENFORCE_STATS, -1);
if (virConnectGetAllDomainStatsEnsureACL(conn) < 0)
return -1;
if (!stats) {
stats = supported;
} else if ((flags & VIR_CONNECT_GET_ALL_DOMAINS_STATS_ENFORCE_STATS) &&
(stats & ~supported)) {
virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED,
_("Stats types bits 0x%x are not supported by this daemon"),
stats & ~supported);
return -1;
}
if (ndomains) {
if (virDomainObjListConvert(driver->domains, conn, domains, ndomains, &doms,
&ndoms, virConnectGetAllDomainStatsCheckACL,
lflags, true) < 0)
return -1;
} else {
if (virDomainObjListCollect(driver->domains, conn, &doms, &ndoms,
virConnectGetAllDomainStatsCheckACL,
lflags) < 0)
return -1;
}
if (VIR_ALLOC_N(tmpstats, ndoms + 1) < 0)
goto cleanup;
for (i = 0; i < ndoms; i++) {
virDomainStatsRecordPtr tmp;
virDomainObjPtr dom = doms[i];
virObjectLock(dom);
tmp = vzDomainGetAllStats(conn, dom);
virObjectUnlock(dom);
if (!tmp)
goto cleanup;
tmpstats[nstats++] = tmp;
}
*retStats = tmpstats;
tmpstats = NULL;
ret = nstats;
cleanup:
virDomainStatsRecordListFree(tmpstats);
virObjectListFreeCount(doms, ndoms);
return ret;
}
#undef VZ_ADD_STAT_PARAM_UUL
2017-01-31 07:44:12 +00:00
static int
vzDomainAbortJob(virDomainPtr domain)
{
virDomainObjPtr dom;
int ret = -1;
if (!(dom = vzDomObjFromDomain(domain)))
2017-01-31 07:44:12 +00:00
return -1;
if (virDomainAbortJobEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
ret = prlsdkCancelJob(dom);
cleanup:
virDomainObjEndAPI(&dom);
return ret;
}
2017-01-31 07:44:13 +00:00
static int
vzDomainReset(virDomainPtr domain, unsigned int flags)
{
virDomainObjPtr dom = NULL;
int ret = -1;
bool job = false;
virCheckFlags(0, -1);
if (!(dom = vzDomObjFromDomain(domain)))
2017-01-31 07:44:13 +00:00
return -1;
if (virDomainResetEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
ret = prlsdkReset(dom);
cleanup:
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static int vzDomainSetVcpusFlags(virDomainPtr domain, unsigned int nvcpus,
unsigned int flags)
{
virDomainObjPtr dom = NULL;
int ret = -1;
bool job = false;
virCheckFlags(VIR_DOMAIN_AFFECT_LIVE |
VIR_DOMAIN_AFFECT_CONFIG, -1);
if (!(dom = vzDomObjFromDomain(domain)))
goto cleanup;
if (vzCheckConfigUpdateFlags(dom, &flags) < 0)
goto cleanup;
if (virDomainSetVcpusFlagsEnsureACL(domain->conn, dom->def, flags) < 0)
goto cleanup;
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
ret = prlsdkSetCpuCount(dom, nvcpus);
cleanup:
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static int vzDomainSetVcpus(virDomainPtr dom, unsigned int nvcpus)
{
return vzDomainSetVcpusFlags(dom, nvcpus,
VIR_DOMAIN_AFFECT_LIVE | VIR_DOMAIN_AFFECT_CONFIG);
}
static int
vzDomainBlockResize(virDomainPtr domain,
const char *path,
unsigned long long size,
unsigned int flags)
{
virDomainObjPtr dom = NULL;
virDomainDiskDefPtr disk = NULL;
int ret = -1;
bool job = false;
virCheckFlags(VIR_DOMAIN_BLOCK_RESIZE_BYTES, -1);
if (!(dom = vzDomObjFromDomain(domain)))
goto cleanup;
if (virDomainBlockResizeEnsureACL(domain->conn, dom->def) < 0)
goto cleanup;
if (path[0] == '\0') {
virReportError(VIR_ERR_INVALID_ARG,
"%s", _("empty path"));
goto cleanup;
}
/* sdk wants Mb */
if (flags & VIR_DOMAIN_BLOCK_RESIZE_BYTES)
size /= 1024;
size /= 1024;
if (vzDomainObjBeginJob(dom) < 0)
goto cleanup;
job = true;
if (vzEnsureDomainExists(dom) < 0)
goto cleanup;
if (virDomainObjCheckActive(dom) < 0)
goto cleanup;
if (!(disk = virDomainDiskByName(dom->def, path, false))) {
virReportError(VIR_ERR_INVALID_ARG,
_("invalid path: %s"), path);
goto cleanup;
}
ret = prlsdkResizeImage(dom, disk, size);
cleanup:
if (job)
vzDomainObjEndJob(dom);
virDomainObjEndAPI(&dom);
return ret;
}
static virHypervisorDriver vzHypervisorDriver = {
.name = "vz",
.connectOpen = vzConnectOpen, /* 0.10.0 */
.connectClose = vzConnectClose, /* 0.10.0 */
.connectGetVersion = vzConnectGetVersion, /* 0.10.0 */
.connectGetHostname = vzConnectGetHostname, /* 0.10.0 */
.connectGetSysinfo = vzConnectGetSysinfo, /* 1.3.4 */
.connectGetMaxVcpus = vzConnectGetMaxVcpus, /* 1.2.21 */
.nodeGetInfo = vzNodeGetInfo, /* 0.10.0 */
.nodeGetCPUStats = vzNodeGetCPUStats, /* 1.2.21 */
.nodeGetMemoryStats = vzNodeGetMemoryStats, /* 1.2.21 */
.nodeGetCellsFreeMemory = vzNodeGetCellsFreeMemory, /* 1.2.21 */
.nodeGetFreeMemory = vzNodeGetFreeMemory, /* 1.2.21 */
.connectGetCapabilities = vzConnectGetCapabilities, /* 0.10.0 */
.connectBaselineCPU = vzConnectBaselineCPU, /* 1.2.6 */
.connectListDomains = vzConnectListDomains, /* 0.10.0 */
.connectNumOfDomains = vzConnectNumOfDomains, /* 0.10.0 */
.connectListDefinedDomains = vzConnectListDefinedDomains, /* 0.10.0 */
.connectNumOfDefinedDomains = vzConnectNumOfDefinedDomains, /* 0.10.0 */
.connectListAllDomains = vzConnectListAllDomains, /* 0.10.0 */
.domainLookupByID = vzDomainLookupByID, /* 0.10.0 */
.domainLookupByUUID = vzDomainLookupByUUID, /* 0.10.0 */
.domainLookupByName = vzDomainLookupByName, /* 0.10.0 */
.domainGetOSType = vzDomainGetOSType, /* 0.10.0 */
.domainGetInfo = vzDomainGetInfo, /* 0.10.0 */
.domainGetState = vzDomainGetState, /* 0.10.0 */
.domainGetXMLDesc = vzDomainGetXMLDesc, /* 0.10.0 */
.domainIsPersistent = vzDomainIsPersistent, /* 0.10.0 */
.domainGetAutostart = vzDomainGetAutostart, /* 0.10.0 */
.domainGetVcpus = vzDomainGetVcpus, /* 1.2.6 */
.domainSuspend = vzDomainSuspend, /* 0.10.0 */
.domainResume = vzDomainResume, /* 0.10.0 */
.domainDestroy = vzDomainDestroy, /* 0.10.0 */
.domainDestroyFlags = vzDomainDestroyFlags, /* 2.2.0 */
.domainShutdown = vzDomainShutdown, /* 0.10.0 */
.domainShutdownFlags = vzDomainShutdownFlags, /* 2.2.0 */
.domainCreate = vzDomainCreate, /* 0.10.0 */
.domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */
.domainReboot = vzDomainReboot, /* 1.3.0 */
.domainDefineXML = vzDomainDefineXML, /* 0.10.0 */
.domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */
.domainUndefine = vzDomainUndefine, /* 1.2.10 */
.domainUndefineFlags = vzDomainUndefineFlags, /* 1.2.10 */
.domainAttachDevice = vzDomainAttachDevice, /* 1.2.15 */
.domainAttachDeviceFlags = vzDomainAttachDeviceFlags, /* 1.2.15 */
.domainDetachDevice = vzDomainDetachDevice, /* 1.2.15 */
.domainDetachDeviceFlags = vzDomainDetachDeviceFlags, /* 1.2.15 */
.domainIsActive = vzDomainIsActive, /* 1.2.10 */
.domainIsUpdated = vzDomainIsUpdated, /* 1.2.21 */
.domainSetVcpus = vzDomainSetVcpus, /* 3.3.0 */
.domainSetVcpusFlags = vzDomainSetVcpusFlags, /* 3.3.0 */
.domainGetVcpusFlags = vzDomainGetVcpusFlags, /* 1.2.21 */
.domainGetMaxVcpus = vzDomainGetMaxVcpus, /* 1.2.21 */
.domainSetUserPassword = vzDomainSetUserPassword, /* 2.0.0 */
.connectDomainEventRegisterAny = vzConnectDomainEventRegisterAny, /* 1.2.10 */
.connectDomainEventDeregisterAny = vzConnectDomainEventDeregisterAny, /* 1.2.10 */
.nodeGetCPUMap = vzNodeGetCPUMap, /* 1.2.8 */
.connectIsEncrypted = vzConnectIsEncrypted, /* 1.2.5 */
.connectIsSecure = vzConnectIsSecure, /* 1.2.5 */
.connectIsAlive = vzConnectIsAlive, /* 1.2.5 */
.domainHasManagedSaveImage = vzDomainHasManagedSaveImage, /* 1.2.13 */
.domainManagedSave = vzDomainManagedSave, /* 1.2.14 */
.domainManagedSaveRemove = vzDomainManagedSaveRemove, /* 1.2.14 */
.domainGetMaxMemory = vzDomainGetMaxMemory, /* 1.2.15 */
.domainBlockStats = vzDomainBlockStats, /* 1.2.17 */
.domainBlockStatsFlags = vzDomainBlockStatsFlags, /* 1.2.17 */
.domainInterfaceStats = vzDomainInterfaceStats, /* 1.2.17 */
.domainMemoryStats = vzDomainMemoryStats, /* 1.2.17 */
.connectRegisterCloseCallback = vzConnectRegisterCloseCallback, /* 1.3.2 */
.connectUnregisterCloseCallback = vzConnectUnregisterCloseCallback, /* 1.3.2 */
.domainSetMemoryFlags = vzDomainSetMemoryFlags, /* 1.3.4 */
.domainSetMemory = vzDomainSetMemory, /* 1.3.4 */
.domainSnapshotNum = vzDomainSnapshotNum, /* 1.3.5 */
.domainSnapshotListNames = vzDomainSnapshotListNames, /* 1.3.5 */
.domainListAllSnapshots = vzDomainListAllSnapshots, /* 1.3.5 */
.domainSnapshotGetXMLDesc = vzDomainSnapshotGetXMLDesc, /* 1.3.5 */
.domainSnapshotNumChildren = vzDomainSnapshotNumChildren, /* 1.3.5 */
.domainSnapshotListChildrenNames = vzDomainSnapshotListChildrenNames, /* 1.3.5 */
.domainSnapshotListAllChildren = vzDomainSnapshotListAllChildren, /* 1.3.5 */
.domainSnapshotLookupByName = vzDomainSnapshotLookupByName, /* 1.3.5 */
.domainHasCurrentSnapshot = vzDomainHasCurrentSnapshot, /* 1.3.5 */
.domainSnapshotGetParent = vzDomainSnapshotGetParent, /* 1.3.5 */
.domainSnapshotCurrent = vzDomainSnapshotCurrent, /* 1.3.5 */
.domainSnapshotIsCurrent = vzDomainSnapshotIsCurrent, /* 1.3.5 */
.domainSnapshotHasMetadata = vzDomainSnapshotHasMetadata, /* 1.3.5 */
.domainSnapshotCreateXML = vzDomainSnapshotCreateXML, /* 1.3.5 */
.domainSnapshotDelete = vzDomainSnapshotDelete, /* 1.3.5 */
.domainRevertToSnapshot = vzDomainRevertToSnapshot, /* 1.3.5 */
.connectSupportsFeature = vzConnectSupportsFeature, /* 1.3.5 */
.domainMigrateBegin3Params = vzDomainMigrateBegin3Params, /* 1.3.5 */
.domainMigratePrepare3Params = vzDomainMigratePrepare3Params, /* 1.3.5 */
.domainMigratePerform3Params = vzDomainMigratePerform3Params, /* 1.3.5 */
.domainMigrateFinish3Params = vzDomainMigrateFinish3Params, /* 1.3.5 */
.domainMigrateConfirm3Params = vzDomainMigrateConfirm3Params, /* 1.3.5 */
.domainUpdateDeviceFlags = vzDomainUpdateDeviceFlags, /* 2.0.0 */
.domainGetJobInfo = vzDomainGetJobInfo, /* 2.2.0 */
.domainGetJobStats = vzDomainGetJobStats, /* 2.2.0 */
.connectGetAllDomainStats = vzConnectGetAllDomainStats, /* 3.1.0 */
2017-01-31 07:44:12 +00:00
.domainAbortJob = vzDomainAbortJob, /* 3.1.0 */
2017-01-31 07:44:13 +00:00
.domainReset = vzDomainReset, /* 3.1.0 */
.domainBlockResize = vzDomainBlockResize, /* 3.3.0 */
};
static virConnectDriver vzConnectDriver = {
.localOnly = true,
.uriSchemes = (const char *[]){ "vz", NULL },
.hypervisorDriver = &vzHypervisorDriver,
};
static int
vzStateCleanup(void)
{
if (vz_driver_privileged) {
virObjectUnref(vz_driver);
vz_driver = NULL;
if (vz_driver_lock_fd != -1)
virPidFileRelease(VZ_STATEDIR, "driver", vz_driver_lock_fd);
virMutexDestroy(&vz_driver_lock);
prlsdkDeinit();
}
return 0;
}
static int
vzStateInitialize(bool privileged,
virStateInhibitCallback callback G_GNUC_UNUSED,
void *opaque G_GNUC_UNUSED)
{
if (!privileged)
return VIR_DRV_STATE_INIT_SKIPPED;
vz_driver_privileged = privileged;
if (virFileMakePathWithMode(VZ_STATEDIR, S_IRWXU) < 0) {
virReportSystemError(errno, _("cannot create state directory '%s'"),
VZ_STATEDIR);
return VIR_DRV_STATE_INIT_ERROR;
}
if ((vz_driver_lock_fd =
virPidFileAcquire(VZ_STATEDIR, "driver", false, getpid())) < 0)
return VIR_DRV_STATE_INIT_ERROR;
if (prlsdkInit() < 0) {
VIR_DEBUG("%s", _("Can't initialize Parallels SDK"));
return VIR_DRV_STATE_INIT_ERROR;
}
if (virMutexInit(&vz_driver_lock) < 0)
goto error;
/* Failing to create driver here is not fatal and only means
* that next driver client will try once more when connecting */
vz_driver = vzDriverObjNew();
return VIR_DRV_STATE_INIT_COMPLETE;
error:
vzStateCleanup();
return VIR_DRV_STATE_INIT_ERROR;
}
static virStateDriver vzStateDriver = {
.name = "vz",
.stateInitialize = vzStateInitialize,
.stateCleanup = vzStateCleanup,
};
/* Parallels domain type backward compatibility*/
static virHypervisorDriver parallelsHypervisorDriver;
static virConnectDriver parallelsConnectDriver = {
.localOnly = true,
.uriSchemes = (const char *[]){ "parallels", NULL },
.hypervisorDriver = &parallelsHypervisorDriver,
};
/**
* vzRegister:
*
* Registers the vz driver
*/
int
vzRegister(void)
{
char *prlctl_path;
prlctl_path = virFindFileInPath(PRLCTL);
if (!prlctl_path) {
VIR_DEBUG("%s", _("Can't find prlctl command in the PATH env"));
return 0;
}
VIR_FREE(prlctl_path);
/* Backward compatibility with Parallels domain type */
parallelsHypervisorDriver = vzHypervisorDriver;
parallelsHypervisorDriver.name = "Parallels";
if (virRegisterConnectDriver(&parallelsConnectDriver, true) < 0)
return -1;
if (virRegisterConnectDriver(&vzConnectDriver, true) < 0)
return -1;
if (virRegisterStateDriver(&vzStateDriver) < 0)
return -1;
return 0;
}