static int
lxcSetCpusetTune(virDomainDefPtr def, virConfPtr properties)
{
VIR_AUTOFREE(char *) value = NULL;
virBitmapPtr nodeset = NULL;
if (virConfGetValueString(properties, "lxc.cgroup.cpuset.cpus",
&value) > 0) {
if (virBitmapParse(value, &def->cpumask, VIR_DOMAIN_CPUMASK_LEN) < 0)
return -1;
def->placement_mode = VIR_DOMAIN_CPU_PLACEMENT_MODE_STATIC;
VIR_FREE(value);
}
if (virConfGetValueString(properties, "lxc.cgroup.cpuset.mems",
&value) > 0) {
if (virBitmapParse(value, &nodeset, VIR_DOMAIN_CPUMASK_LEN) < 0)
return -1;
if (virDomainNumatuneSet(def->numa,
def->placement_mode ==
VIR_DOMAIN_CPU_PLACEMENT_MODE_STATIC,
VIR_DOMAIN_NUMATUNE_PLACEMENT_STATIC,
VIR_DOMAIN_NUMATUNE_MEM_STRICT,
nodeset) < 0) {
virBitmapFree(nodeset);
return -1;
}
virBitmapFree(nodeset);
}
return 0;
}
static int
lxcSetCpusetTune(virDomainDefPtr def, virConfPtr properties)
{
virConfValuePtr value;
virBitmapPtr nodeset = NULL;
if ((value = virConfGetValue(properties, "lxc.cgroup.cpuset.cpus")) &&
value->str) {
if (virBitmapParse(value->str, 0, &def->cpumask,
VIR_DOMAIN_CPUMASK_LEN) < 0)
return -1;
def->placement_mode = VIR_DOMAIN_CPU_PLACEMENT_MODE_STATIC;
}
if ((value = virConfGetValue(properties, "lxc.cgroup.cpuset.mems")) &&
value->str) {
if (virBitmapParse(value->str, 0, &nodeset, VIR_DOMAIN_CPUMASK_LEN) < 0)
return -1;
if (virDomainNumatuneSet(def->numa,
def->placement_mode ==
VIR_DOMAIN_CPU_PLACEMENT_MODE_STATIC,
VIR_DOMAIN_NUMATUNE_PLACEMENT_STATIC,
VIR_DOMAIN_NUMATUNE_MEM_STRICT,
nodeset) < 0) {
virBitmapFree(nodeset);
return -1;
}
virBitmapFree(nodeset);
}
return 0;
}
void
virDomainNumatuneFree(virDomainNumatunePtr numatune)
{
size_t i = 0;
if (!numatune)
return;
virBitmapFree(numatune->memory.nodeset);
for (i = 0; i < numatune->nmem_nodes; i++)
virBitmapFree(numatune->mem_nodes[i].nodeset);
VIR_FREE(numatune->mem_nodes);
VIR_FREE(numatune);
}
static void
virPortAllocatorDispose(void *obj)
{
virPortAllocatorPtr pa = obj;
virBitmapFree(pa->bitmap);
}
/*
* Linux maintains cpu bit map. For example, if cpuid=5's flag is not set
* and max cpu is 7. The map file shows 0-4,6-7. This function parses
* it and returns cpumap.
*/
static virBitmapPtr
linuxParseCPUmap(int *max_cpuid, const char *path)
{
virBitmapPtr map = NULL;
char *str = NULL;
int max_id = 0, i;
if (virFileReadAll(path, 5 * VIR_DOMAIN_CPUMASK_LEN, &str) < 0) {
virReportOOMError();
goto error;
}
if (virBitmapParse(str, 0, &map,
VIR_DOMAIN_CPUMASK_LEN) < 0) {
goto error;
}
i = -1;
while ((i = virBitmapNextSetBit(map, i)) >= 0) {
max_id = i;
}
*max_cpuid = max_id;
VIR_FREE(str);
return map;
error:
VIR_FREE(str);
virBitmapFree(map);
return NULL;
}
int
virQEMUBuildCommandLineJSONArrayBitmap(const char *key,
virJSONValuePtr array,
virBufferPtr buf)
{
ssize_t pos = -1;
ssize_t end;
virBitmapPtr bitmap = NULL;
if (virJSONValueGetArrayAsBitmap(array, &bitmap) < 0)
return -1;
while ((pos = virBitmapNextSetBit(bitmap, pos)) > -1) {
if ((end = virBitmapNextClearBit(bitmap, pos)) < 0)
end = virBitmapLastSetBit(bitmap) + 1;
if (end - 1 > pos) {
virBufferAsprintf(buf, "%s=%zd-%zd,", key, pos, end - 1);
pos = end;
} else {
virBufferAsprintf(buf, "%s=%zd,", key, pos);
}
}
virBitmapFree(bitmap);
return 0;
}
/* Check whether the host subcore configuration is valid.
*
* A valid configuration is one where no secondary thread is online;
* the primary thread in a subcore is always the first one */
static bool
virHostCPUHasValidSubcoreConfiguration(int threads_per_subcore)
{
virBitmapPtr online_cpus = NULL;
int cpu = -1;
bool ret = false;
/* No point in checking if subcores are not in use */
if (threads_per_subcore <= 0)
goto cleanup;
if (!(online_cpus = virHostCPUGetOnlineBitmap()))
goto cleanup;
while ((cpu = virBitmapNextSetBit(online_cpus, cpu)) >= 0) {
/* A single online secondary thread is enough to
* make the configuration invalid */
if (cpu % threads_per_subcore != 0)
goto cleanup;
}
ret = true;
cleanup:
virBitmapFree(online_cpus);
return ret;
}
static void
dnsmasqCapsDispose(void *obj)
{
dnsmasqCapsPtr caps = obj;
virBitmapFree(caps->flags);
VIR_FREE(caps->binaryPath);
}
void
virDomainNumaFree(virDomainNumaPtr numa)
{
size_t i = 0;
if (!numa)
return;
virBitmapFree(numa->memory.nodeset);
for (i = 0; i < numa->nmem_nodes; i++) {
virBitmapFree(numa->mem_nodes[i].cpumask);
virBitmapFree(numa->mem_nodes[i].nodeset);
}
VIR_FREE(numa->mem_nodes);
VIR_FREE(numa);
}
static void
virDomainVirtioSerialControllerFree(virDomainVirtioSerialControllerPtr cont)
{
if (cont) {
virBitmapFree(cont->ports);
VIR_FREE(cont);
}
}
int
virNumaGetNodeCPUs(int node,
virBitmapPtr *cpus)
{
unsigned long *mask = NULL;
unsigned long *allonesmask = NULL;
virBitmapPtr cpumap = NULL;
int ncpus = 0;
int max_n_cpus = virNumaGetMaxCPUs();
int mask_n_bytes = max_n_cpus / 8;
size_t i;
int ret = -1;
*cpus = NULL;
if (VIR_ALLOC_N(mask, mask_n_bytes / sizeof(*mask)) < 0)
goto cleanup;
if (VIR_ALLOC_N(allonesmask, mask_n_bytes / sizeof(*mask)) < 0)
goto cleanup;
memset(allonesmask, 0xff, mask_n_bytes);
/* The first time this returns -1, ENOENT if node doesn't exist... */
if (numa_node_to_cpus(node, mask, mask_n_bytes) < 0) {
VIR_WARN("NUMA topology for cell %d is not available, ignoring", node);
ret = -2;
goto cleanup;
}
/* second, third... times it returns an all-1's mask */
if (memcmp(mask, allonesmask, mask_n_bytes) == 0) {
VIR_DEBUG("NUMA topology for cell %d is invalid, ignoring", node);
ret = -2;
goto cleanup;
}
if (!(cpumap = virBitmapNew(max_n_cpus)))
goto cleanup;
for (i = 0; i < max_n_cpus; i++) {
if (MASK_CPU_ISSET(mask, i)) {
ignore_value(virBitmapSetBit(cpumap, i));
ncpus++;
}
}
*cpus = cpumap;
cpumap = NULL;
ret = ncpus;
cleanup:
VIR_FREE(mask);
VIR_FREE(allonesmask);
virBitmapFree(cpumap);
return ret;
}
/*
* To be run while still single threaded
*/
static int virLXCControllerSetupCpuAffinity(virLXCControllerPtr ctrl)
{
int hostcpus, maxcpu = CPU_SETSIZE;
virNodeInfo nodeinfo;
virBitmapPtr cpumap, cpumapToSet;
VIR_DEBUG("Setting CPU affinity");
if (nodeGetInfo(NULL, &nodeinfo) < 0)
return -1;
/* setaffinity fails if you set bits for CPUs which
* aren't present, so we have to limit ourselves */
hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
if (maxcpu > hostcpus)
maxcpu = hostcpus;
cpumap = virBitmapNew(maxcpu);
if (!cpumap)
return -1;
cpumapToSet = cpumap;
if (ctrl->def->cpumask) {
cpumapToSet = ctrl->def->cpumask;
} else {
/* You may think this is redundant, but we can't assume libvirtd
* itself is running on all pCPUs, so we need to explicitly set
* the spawned LXC instance to all pCPUs if no map is given in
* its config file */
virBitmapSetAll(cpumap);
}
/* We are presuming we are running between fork/exec of LXC
* so use '0' to indicate our own process ID. No threads are
* running at this point
*/
if (virProcessInfoSetAffinity(0 /* Self */, cpumapToSet) < 0) {
virBitmapFree(cpumap);
return -1;
}
virBitmapFree(cpumap);
return 0;
}
bool
virNumaNodeIsAvailable(int node)
{
bool ret = false;
virBitmapPtr map = NULL;
if (virFileReadValueBitmap(&map, "%s/node/online", SYSFS_SYSTEM_PATH) < 0)
return false;
ret = virBitmapIsBitSet(map, node);
virBitmapFree(map);
return ret;
}
int
virNumaGetMaxNode(void)
{
int ret = -1;
virBitmapPtr map = NULL;
if (virFileReadValueBitmap(&map, "%s/node/online", SYSFS_SYSTEM_PATH) < 0)
return -1;
ret = virBitmapLastSetBit(map);
virBitmapFree(map);
return ret;
}
/**
* virBitmapNewCopy:
* @src: the source bitmap.
*
* Makes a copy of bitmap @src.
*
* returns the copied bitmap on success, or NULL otherwise. Caller
* should call virBitmapFree to free the returned bitmap.
*/
virBitmapPtr virBitmapNewCopy(virBitmapPtr src)
{
virBitmapPtr dst;
if ((dst = virBitmapNew(src->max_bit)) == NULL)
return NULL;
if (virBitmapCopy(dst, src) != 0) {
virBitmapFree(dst);
return NULL;
}
return dst;
}
void
virCapabilitiesClearHostNUMACellCPUTopology(virCapsHostNUMACellCPUPtr cpus,
size_t ncpus)
{
size_t i;
if (!cpus)
return;
for (i = 0; i < ncpus; i++) {
virBitmapFree(cpus[i].siblings);
cpus[i].siblings = NULL;
}
}
static int
nodeDeviceUpdateCaps(virNodeDeviceDefPtr def)
{
virNodeDevCapsDefPtr cap = def->caps;
while (cap) {
switch (cap->data.type) {
case VIR_NODE_DEV_CAP_SCSI_HOST:
nodeDeviceSysfsGetSCSIHostCaps(&cap->data.scsi_host);
break;
case VIR_NODE_DEV_CAP_SCSI_TARGET:
nodeDeviceSysfsGetSCSITargetCaps(def->sysfs_path,
&cap->data.scsi_target);
break;
case VIR_NODE_DEV_CAP_NET:
if (virNetDevGetLinkInfo(cap->data.net.ifname, &cap->data.net.lnk) < 0)
return -1;
virBitmapFree(cap->data.net.features);
if (virNetDevGetFeatures(cap->data.net.ifname, &cap->data.net.features) < 0)
return -1;
break;
case VIR_NODE_DEV_CAP_PCI_DEV:
if (nodeDeviceSysfsGetPCIRelatedDevCaps(def->sysfs_path,
&cap->data.pci_dev) < 0)
return -1;
break;
/* all types that (supposedly) don't require any updates
* relative to what's in the cache.
*/
case VIR_NODE_DEV_CAP_DRM:
case VIR_NODE_DEV_CAP_SYSTEM:
case VIR_NODE_DEV_CAP_USB_DEV:
case VIR_NODE_DEV_CAP_USB_INTERFACE:
case VIR_NODE_DEV_CAP_SCSI:
case VIR_NODE_DEV_CAP_STORAGE:
case VIR_NODE_DEV_CAP_FC_HOST:
case VIR_NODE_DEV_CAP_VPORTS:
case VIR_NODE_DEV_CAP_SCSI_GENERIC:
case VIR_NODE_DEV_CAP_MDEV_TYPES:
case VIR_NODE_DEV_CAP_MDEV:
case VIR_NODE_DEV_CAP_CCW_DEV:
case VIR_NODE_DEV_CAP_LAST:
break;
}
cap = cap->next;
}
return 0;
}
void
virCPUDefFree(virCPUDefPtr def)
{
unsigned int i;
if (!def)
return;
VIR_FREE(def->arch);
virCPUDefFreeModel(def);
for (i = 0 ; i < def->ncells ; i++) {
virBitmapFree(def->cells[i].cpumask);
VIR_FREE(def->cells[i].cpustr);
}
VIR_FREE(def->cells);
VIR_FREE(def);
}
void
virCPUDefFree(virCPUDefPtr def)
{
size_t i;
if (!def)
return;
virCPUDefFreeModel(def);
for (i = 0; i < def->ncells; i++) {
virBitmapFree(def->cells[i].cpumask);
VIR_FREE(def->cells[i].cpustr);
}
VIR_FREE(def->cells);
VIR_FREE(def->vendor_id);
VIR_FREE(def);
}
/*
* Linux maintains cpu bit map under cpu/online. For example, if
* cpuid=5's flag is not set and max cpu is 7, the map file shows
* 0-4,6-7. This function parses it and returns cpumap.
*/
static virBitmapPtr
linuxParseCPUmap(int max_cpuid, const char *path)
{
virBitmapPtr map = NULL;
char *str = NULL;
if (virFileReadAll(path, 5 * VIR_DOMAIN_CPUMASK_LEN, &str) < 0)
goto error;
if (virBitmapParse(str, 0, &map, max_cpuid) < 0)
goto error;
VIR_FREE(str);
return map;
error:
VIR_FREE(str);
virBitmapFree(map);
return NULL;
}
/*
* Linux maintains cpu bit map under cpu/online. For example, if
* cpuid=5's flag is not set and max cpu is 7, the map file shows
* 0-4,6-7. This function parses it and returns cpumap.
*/
static virBitmapPtr
virHostCPUParseMapLinux(int max_cpuid, const char *path)
{
virBitmapPtr map = NULL;
char *str = NULL;
if (virFileReadAll(path, 5 * VIR_HOST_CPU_MASK_LEN, &str) < 0)
goto error;
if (virBitmapParse(str, 0, &map, max_cpuid) < 0)
goto error;
VIR_FREE(str);
return map;
error:
VIR_FREE(str);
virBitmapFree(map);
return NULL;
}
virBitmapPtr
virCapabilitiesGetCpusForNodemask(virCapsPtr caps,
virBitmapPtr nodemask)
{
virBitmapPtr ret = NULL;
unsigned int maxcpu = virCapabilitiesGetHostMaxcpu(caps);
ssize_t node = -1;
if (!(ret = virBitmapNew(maxcpu + 1)))
return NULL;
while ((node = virBitmapNextSetBit(nodemask, node)) >= 0) {
if (virCapabilitiesGetCpusForNode(caps, node, ret) < 0) {
virBitmapFree(ret);
return NULL;
}
}
return ret;
}
/**
* qemuCgroupEmulatorAllNodesAllow:
* @cgroup: domain cgroup pointer
* @retData: filled with structure used to roll back the operation
*
* Allows all NUMA nodes for the qemu emulator thread temporarily. This is
* necessary when hotplugging cpus since it requires memory allocated in the
* DMA region. Afterwards the operation can be reverted by
* qemuCgroupEmulatorAllNodesRestore.
*
* Returns 0 on success -1 on error
*/
int
qemuCgroupEmulatorAllNodesAllow(virCgroupPtr cgroup,
qemuCgroupEmulatorAllNodesDataPtr *retData)
{
qemuCgroupEmulatorAllNodesDataPtr data = NULL;
char *all_nodes_str = NULL;
virBitmapPtr all_nodes = NULL;
int ret = -1;
if (!virNumaIsAvailable() ||
!virCgroupHasController(cgroup, VIR_CGROUP_CONTROLLER_CPUSET))
return 0;
if (!(all_nodes = virNumaGetHostMemoryNodeset()))
goto cleanup;
if (!(all_nodes_str = virBitmapFormat(all_nodes)))
goto cleanup;
if (VIR_ALLOC(data) < 0)
goto cleanup;
if (virCgroupNewThread(cgroup, VIR_CGROUP_THREAD_EMULATOR, 0,
false, &data->emulatorCgroup) < 0)
goto cleanup;
if (virCgroupGetCpusetMems(data->emulatorCgroup, &data->emulatorMemMask) < 0 ||
virCgroupSetCpusetMems(data->emulatorCgroup, all_nodes_str) < 0)
goto cleanup;
VIR_STEAL_PTR(*retData, data);
ret = 0;
cleanup:
VIR_FREE(all_nodes_str);
virBitmapFree(all_nodes);
qemuCgroupEmulatorAllNodesDataFree(data);
return ret;
}
/**
* virBitmapParse:
* @str: points to a string representing a human-readable bitmap
* @terminator: character separating the bitmap to parse
* @bitmap: a bitmap created from @str
* @bitmapSize: the upper limit of num of bits in created bitmap
*
* This function is the counterpart of virBitmapFormat. This function creates
* a bitmap, in which bits are set according to the content of @str.
*
* @str is a comma separated string of fields N, which means a number of bit
* to set, and ^N, which means to unset the bit, and N-M for ranges of bits
* to set.
*
* To allow parsing of bitmaps within larger strings it is possible to set
* a termination character in the argument @terminator. When the character
* in @terminator is encountered in @str, the parsing of the bitmap stops.
* Pass 0 as @terminator if it is not needed. Whitespace characters may not
* be used as terminators.
*
* Returns the number of bits set in @bitmap, or -1 in case of error.
*/
int
virBitmapParse(const char *str,
char terminator,
virBitmapPtr *bitmap,
size_t bitmapSize)
{
bool neg = false;
const char *cur = str;
char *tmp;
size_t i;
int start, last;
if (!(*bitmap = virBitmapNew(bitmapSize)))
return -1;
if (!str)
goto error;
virSkipSpaces(&cur);
if (*cur == '\0')
goto error;
while (*cur != 0 && *cur != terminator) {
/*
* 3 constructs are allowed:
* - N : a single CPU number
* - N-M : a range of CPU numbers with N < M
* - ^N : remove a single CPU number from the current set
*/
if (*cur == '^') {
cur++;
neg = true;
}
if (!c_isdigit(*cur))
goto error;
if (virStrToLong_i(cur, &tmp, 10, &start) < 0)
goto error;
if (start < 0)
goto error;
cur = tmp;
virSkipSpaces(&cur);
if (*cur == ',' || *cur == 0 || *cur == terminator) {
if (neg) {
if (virBitmapClearBit(*bitmap, start) < 0)
goto error;
} else {
if (virBitmapSetBit(*bitmap, start) < 0)
goto error;
}
} else if (*cur == '-') {
if (neg)
goto error;
cur++;
virSkipSpaces(&cur);
if (virStrToLong_i(cur, &tmp, 10, &last) < 0)
goto error;
if (last < start)
goto error;
cur = tmp;
for (i = start; i <= last; i++) {
if (virBitmapSetBit(*bitmap, i) < 0)
goto error;
}
virSkipSpaces(&cur);
}
if (*cur == ',') {
cur++;
virSkipSpaces(&cur);
neg = false;
} else if (*cur == 0 || *cur == terminator) {
break;
} else {
goto error;
}
}
if (virBitmapIsAllClear(*bitmap))
goto error;
return virBitmapCountBits(*bitmap);
error:
virReportError(VIR_ERR_INVALID_ARG,
_("Failed to parse bitmap '%s'"), str);
virBitmapFree(*bitmap);
*bitmap = NULL;
return -1;
}
int
virDomainNumatuneSet(virDomainNumaPtr numa,
bool placement_static,
int placement,
int mode,
virBitmapPtr nodeset)
{
int ret = -1;
/* No need to do anything in this case */
if (mode == -1 && placement == -1 && !nodeset)
return 0;
if (!numa->memory.specified) {
if (mode == -1)
mode = VIR_DOMAIN_NUMATUNE_MEM_STRICT;
if (placement == -1)
placement = VIR_DOMAIN_NUMATUNE_PLACEMENT_DEFAULT;
}
/* Range checks */
if (mode != -1 &&
(mode < 0 || mode >= VIR_DOMAIN_NUMATUNE_MEM_LAST)) {
virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
_("Unsupported numatune mode '%d'"),
mode);
goto cleanup;
}
if (placement != -1 &&
(placement < 0 || placement >= VIR_DOMAIN_NUMATUNE_PLACEMENT_LAST)) {
virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
_("Unsupported numatune placement '%d'"),
mode);
goto cleanup;
}
if (mode != -1)
numa->memory.mode = mode;
if (nodeset) {
virBitmapFree(numa->memory.nodeset);
if (!(numa->memory.nodeset = virBitmapNewCopy(nodeset)))
goto cleanup;
if (placement == -1)
placement = VIR_DOMAIN_NUMATUNE_PLACEMENT_STATIC;
}
if (placement == VIR_DOMAIN_NUMATUNE_PLACEMENT_DEFAULT) {
if (numa->memory.nodeset || placement_static)
placement = VIR_DOMAIN_NUMATUNE_PLACEMENT_STATIC;
else
placement = VIR_DOMAIN_NUMATUNE_PLACEMENT_AUTO;
}
if (placement == VIR_DOMAIN_NUMATUNE_PLACEMENT_STATIC &&
!numa->memory.nodeset) {
virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
_("nodeset for NUMA memory tuning must be set "
"if 'placement' is 'static'"));
goto cleanup;
}
/* setting nodeset when placement auto is invalid */
if (placement == VIR_DOMAIN_NUMATUNE_PLACEMENT_AUTO &&
numa->memory.nodeset) {
virBitmapFree(numa->memory.nodeset);
numa->memory.nodeset = NULL;
}
if (placement != -1)
numa->memory.placement = placement;
numa->memory.specified = true;
ret = 0;
cleanup:
return ret;
}
/* Align def->disks to def->domain. Sort the list of def->disks,
* filling in any missing disks or snapshot state defaults given by
* the domain, with a fallback to a passed in default. Convert paths
* to disk targets for uniformity. Issue an error and return -1 if
* any def->disks[n]->name appears more than once or does not map to
* dom->disks. If require_match, also ensure that there is no
* conflicting requests for both internal and external snapshots. */
int
virDomainSnapshotAlignDisks(virDomainSnapshotDefPtr def,
int default_snapshot,
bool require_match)
{
int ret = -1;
virBitmapPtr map = NULL;
int i;
int ndisks;
bool inuse;
if (!def->dom) {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("missing domain in snapshot"));
goto cleanup;
}
if (def->ndisks > def->dom->ndisks) {
virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
_("too many disk snapshot requests for domain"));
goto cleanup;
}
/* Unlikely to have a guest without disks but technically possible. */
if (!def->dom->ndisks) {
ret = 0;
goto cleanup;
}
if (!(map = virBitmapNew(def->dom->ndisks))) {
virReportOOMError();
goto cleanup;
}
/* Double check requested disks. */
for (i = 0; i < def->ndisks; i++) {
virDomainSnapshotDiskDefPtr disk = &def->disks[i];
int idx = virDomainDiskIndexByName(def->dom, disk->name, false);
int disk_snapshot;
if (idx < 0) {
virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
_("no disk named '%s'"), disk->name);
goto cleanup;
}
if (virBitmapGetBit(map, idx, &inuse) < 0 || inuse) {
virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
_("disk '%s' specified twice"),
disk->name);
goto cleanup;
}
ignore_value(virBitmapSetBit(map, idx));
disk->index = idx;
disk_snapshot = def->dom->disks[idx]->snapshot;
if (!disk->snapshot) {
if (disk_snapshot &&
(!require_match ||
disk_snapshot == VIR_DOMAIN_SNAPSHOT_LOCATION_NONE))
disk->snapshot = disk_snapshot;
else
disk->snapshot = default_snapshot;
} else if (require_match &&
disk->snapshot != default_snapshot &&
!(disk->snapshot == VIR_DOMAIN_SNAPSHOT_LOCATION_NONE &&
disk_snapshot == VIR_DOMAIN_SNAPSHOT_LOCATION_NONE)) {
const char *tmp;
tmp = virDomainSnapshotLocationTypeToString(default_snapshot);
virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
_("disk '%s' must use snapshot mode '%s'"),
disk->name, tmp);
goto cleanup;
}
if (disk->file &&
disk->snapshot != VIR_DOMAIN_SNAPSHOT_LOCATION_EXTERNAL) {
virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
_("file '%s' for disk '%s' requires "
"use of external snapshot mode"),
disk->file, disk->name);
goto cleanup;
}
if (STRNEQ(disk->name, def->dom->disks[idx]->dst)) {
VIR_FREE(disk->name);
if (!(disk->name = strdup(def->dom->disks[idx]->dst))) {
virReportOOMError();
goto cleanup;
}
}
}
/* Provide defaults for all remaining disks. */
ndisks = def->ndisks;
if (VIR_EXPAND_N(def->disks, def->ndisks,
def->dom->ndisks - def->ndisks) < 0) {
virReportOOMError();
goto cleanup;
}
for (i = 0; i < def->dom->ndisks; i++) {
virDomainSnapshotDiskDefPtr disk;
ignore_value(virBitmapGetBit(map, i, &inuse));
if (inuse)
continue;
disk = &def->disks[ndisks++];
if (!(disk->name = strdup(def->dom->disks[i]->dst))) {
virReportOOMError();
goto cleanup;
}
disk->index = i;
disk->snapshot = def->dom->disks[i]->snapshot;
if (!disk->snapshot)
disk->snapshot = default_snapshot;
}
qsort(&def->disks[0], def->ndisks, sizeof(def->disks[0]), disksorter);
/* Generate any default external file names, but only if the
* backing file is a regular file. */
for (i = 0; i < def->ndisks; i++) {
virDomainSnapshotDiskDefPtr disk = &def->disks[i];
if (disk->snapshot == VIR_DOMAIN_SNAPSHOT_LOCATION_EXTERNAL &&
!disk->file) {
const char *original = def->dom->disks[i]->src;
const char *tmp;
struct stat sb;
if (!original) {
virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
_("cannot generate external snapshot name "
"for disk '%s' without source"),
disk->name);
goto cleanup;
}
if (stat(original, &sb) < 0 || !S_ISREG(sb.st_mode)) {
virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
_("source for disk '%s' is not a regular "
"file; refusing to generate external "
"snapshot name"),
disk->name);
goto cleanup;
}
tmp = strrchr(original, '.');
if (!tmp || strchr(tmp, '/')) {
ignore_value(virAsprintf(&disk->file, "%s.%s",
original, def->name));
} else {
if ((tmp - original) > INT_MAX) {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("integer overflow"));
goto cleanup;
}
ignore_value(virAsprintf(&disk->file, "%.*s.%s",
(int) (tmp - original), original,
def->name));
}
if (!disk->file) {
virReportOOMError();
goto cleanup;
}
}
}
ret = 0;
cleanup:
virBitmapFree(map);
return ret;
}
int virBitmapParse(const char *str,
char sep,
virBitmapPtr *bitmap,
size_t bitmapSize)
{
int ret = 0;
bool neg = false;
const char *cur;
char *tmp;
int i, start, last;
if (!str)
return -1;
cur = str;
virSkipSpaces(&cur);
if (*cur == 0)
return -1;
*bitmap = virBitmapNew(bitmapSize);
if (!*bitmap)
return -1;
while (*cur != 0 && *cur != sep) {
/*
* 3 constructs are allowed:
* - N : a single CPU number
* - N-M : a range of CPU numbers with N < M
* - ^N : remove a single CPU number from the current set
*/
if (*cur == '^') {
cur++;
neg = true;
}
if (!c_isdigit(*cur))
goto parse_error;
if (virStrToLong_i(cur, &tmp, 10, &start) < 0)
goto parse_error;
if (start < 0)
goto parse_error;
cur = tmp;
virSkipSpaces(&cur);
if (*cur == ',' || *cur == 0 || *cur == sep) {
if (neg) {
if (virBitmapIsSet(*bitmap, start)) {
ignore_value(virBitmapClearBit(*bitmap, start));
ret--;
}
} else {
if (!virBitmapIsSet(*bitmap, start)) {
ignore_value(virBitmapSetBit(*bitmap, start));
ret++;
}
}
} else if (*cur == '-') {
if (neg)
goto parse_error;
cur++;
virSkipSpaces(&cur);
if (virStrToLong_i(cur, &tmp, 10, &last) < 0)
goto parse_error;
if (last < start)
goto parse_error;
cur = tmp;
for (i = start; i <= last; i++) {
if (!virBitmapIsSet(*bitmap, i)) {
ignore_value(virBitmapSetBit(*bitmap, i));
ret++;
}
}
virSkipSpaces(&cur);
}
if (*cur == ',') {
cur++;
virSkipSpaces(&cur);
neg = false;
} else if(*cur == 0 || *cur == sep) {
break;
} else {
goto parse_error;
}
}
return ret;
parse_error:
virBitmapFree(*bitmap);
*bitmap = NULL;
return -1;
}
int
virHostCPUGetInfoPopulateLinux(FILE *cpuinfo,
virArch arch,
unsigned int *cpus,
unsigned int *mhz,
unsigned int *nodes,
unsigned int *sockets,
unsigned int *cores,
unsigned int *threads)
{
virBitmapPtr present_cpus_map = NULL;
virBitmapPtr online_cpus_map = NULL;
char line[1024];
DIR *nodedir = NULL;
struct dirent *nodedirent = NULL;
int nodecpus, nodecores, nodesockets, nodethreads, offline = 0;
int threads_per_subcore = 0;
unsigned int node;
int ret = -1;
char *sysfs_nodedir = NULL;
char *sysfs_cpudir = NULL;
int direrr;
*mhz = 0;
*cpus = *nodes = *sockets = *cores = *threads = 0;
/* Start with parsing CPU clock speed from /proc/cpuinfo */
while (fgets(line, sizeof(line), cpuinfo) != NULL) {
if (ARCH_IS_X86(arch)) {
char *buf = line;
if (STRPREFIX(buf, "cpu MHz")) {
char *p;
unsigned int ui;
buf += 7;
while (*buf && c_isspace(*buf))
buf++;
if (*buf != ':' || !buf[1]) {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("parsing cpu MHz from cpuinfo"));
goto cleanup;
}
if (virStrToLong_ui(buf+1, &p, 10, &ui) == 0 &&
/* Accept trailing fractional part. */
(*p == '\0' || *p == '.' || c_isspace(*p)))
*mhz = ui;
}
} else if (ARCH_IS_PPC(arch)) {
char *buf = line;
if (STRPREFIX(buf, "clock")) {
char *p;
unsigned int ui;
buf += 5;
while (*buf && c_isspace(*buf))
buf++;
if (*buf != ':' || !buf[1]) {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("parsing cpu MHz from cpuinfo"));
goto cleanup;
}
if (virStrToLong_ui(buf+1, &p, 10, &ui) == 0 &&
/* Accept trailing fractional part. */
(*p == '\0' || *p == '.' || c_isspace(*p)))
*mhz = ui;
/* No other interesting infos are available in /proc/cpuinfo.
* However, there is a line identifying processor's version,
* identification and machine, but we don't want it to be caught
* and parsed in next iteration, because it is not in expected
* format and thus lead to error. */
}
} else if (ARCH_IS_ARM(arch)) {
char *buf = line;
if (STRPREFIX(buf, "BogoMIPS")) {
char *p;
unsigned int ui;
buf += 8;
while (*buf && c_isspace(*buf))
buf++;
if (*buf != ':' || !buf[1]) {
virReportError(VIR_ERR_INTERNAL_ERROR,
"%s", _("parsing cpu MHz from cpuinfo"));
goto cleanup;
}
if (virStrToLong_ui(buf+1, &p, 10, &ui) == 0
/* Accept trailing fractional part. */
&& (*p == '\0' || *p == '.' || c_isspace(*p)))
*mhz = ui;
}
} else if (ARCH_IS_S390(arch)) {
/* s390x has no realistic value for CPU speed,
* assign a value of zero to signify this */
*mhz = 0;
} else {
VIR_WARN("Parser for /proc/cpuinfo needs to be adapted for your architecture");
break;
}
}
/* Get information about what CPUs are present in the host and what
* CPUs are online, so that we don't have to so for each node */
present_cpus_map = virHostCPUGetPresentBitmap();
if (!present_cpus_map)
goto cleanup;
online_cpus_map = virHostCPUGetOnlineBitmap();
if (!online_cpus_map)
goto cleanup;
/* OK, we've parsed clock speed out of /proc/cpuinfo. Get the
* core, node, socket, thread and topology information from /sys
*/
if (virAsprintf(&sysfs_nodedir, "%s/node", sysfs_system_path) < 0)
goto cleanup;
if (!(nodedir = opendir(sysfs_nodedir))) {
/* the host isn't probably running a NUMA architecture */
goto fallback;
}
/* PPC-KVM needs the secondary threads of a core to be offline on the
* host. The kvm scheduler brings the secondary threads online in the
* guest context. Moreover, P8 processor has split-core capability
* where, there can be 1,2 or 4 subcores per core. The primaries of the
* subcores alone will be online on the host for a subcore in the
* host. Even though the actual threads per core for P8 processor is 8,
* depending on the subcores_per_core = 1, 2 or 4, the threads per
* subcore will vary accordingly to 8, 4 and 2 repectively.
* So, On host threads_per_core what is arrived at from sysfs in the
* current logic is actually the subcores_per_core. Threads per subcore
* can only be obtained from the kvm device. For example, on P8 wih 1
* core having 8 threads, sub_cores_percore=4, the threads 0,2,4 & 6
* will be online. The sysfs reflects this and in the current logic
* variable 'threads' will be 4 which is nothing but subcores_per_core.
* If the user tampers the cpu online/offline states using chcpu or other
* means, then it is an unsupported configuration for kvm.
* The code below tries to keep in mind
* - when the libvirtd is run inside a KVM guest or Phyp based guest.
* - Or on the kvm host where user manually tampers the cpu states to
* offline/online randomly.
* On hosts other than POWER this will be 0, in which case a simpler
* thread-counting logic will be used */
if ((threads_per_subcore = virHostCPUGetThreadsPerSubcore(arch)) < 0)
goto cleanup;
/* If the subcore configuration is not valid, just pretend subcores
* are not in use and count threads one by one */
if (!virHostCPUHasValidSubcoreConfiguration(threads_per_subcore))
threads_per_subcore = 0;
while ((direrr = virDirRead(nodedir, &nodedirent, sysfs_nodedir)) > 0) {
if (sscanf(nodedirent->d_name, "node%u", &node) != 1)
continue;
(*nodes)++;
if (virAsprintf(&sysfs_cpudir, "%s/node/%s",
sysfs_system_path, nodedirent->d_name) < 0)
goto cleanup;
if ((nodecpus = virHostCPUParseNode(sysfs_cpudir, arch,
present_cpus_map,
online_cpus_map,
threads_per_subcore,
&nodesockets, &nodecores,
&nodethreads, &offline)) < 0)
goto cleanup;
VIR_FREE(sysfs_cpudir);
*cpus += nodecpus;
if (nodesockets > *sockets)
*sockets = nodesockets;
if (nodecores > *cores)
*cores = nodecores;
if (nodethreads > *threads)
*threads = nodethreads;
}
if (direrr < 0)
goto cleanup;
if (*cpus && *nodes)
goto done;
fallback:
VIR_FREE(sysfs_cpudir);
if (virAsprintf(&sysfs_cpudir, "%s/cpu", sysfs_system_path) < 0)
goto cleanup;
if ((nodecpus = virHostCPUParseNode(sysfs_cpudir, arch,
present_cpus_map,
online_cpus_map,
threads_per_subcore,
&nodesockets, &nodecores,
&nodethreads, &offline)) < 0)
goto cleanup;
*nodes = 1;
*cpus = nodecpus;
*sockets = nodesockets;
*cores = nodecores;
*threads = nodethreads;
done:
/* There should always be at least one cpu, socket, node, and thread. */
if (*cpus == 0) {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("no CPUs found"));
goto cleanup;
}
if (*sockets == 0) {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("no sockets found"));
goto cleanup;
}
if (*threads == 0) {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("no threads found"));
goto cleanup;
}
/* Now check if the topology makes sense. There are machines that don't
* expose their real number of nodes or for example the AMD Bulldozer
* architecture that exposes their Clustered integer core modules as both
* threads and cores. This approach throws off our detection. Unfortunately
* the nodeinfo structure isn't designed to carry the full topology so
* we're going to lie about the detected topology to notify the user
* to check the host capabilities for the actual topology. */
if ((*nodes *
*sockets *
*cores *
*threads) != (*cpus + offline)) {
*nodes = 1;
*sockets = 1;
*cores = *cpus + offline;
*threads = 1;
}
ret = 0;
cleanup:
/* don't shadow a more serious error */
if (nodedir && closedir(nodedir) < 0 && ret >= 0) {
virReportSystemError(errno, _("problem closing %s"), sysfs_nodedir);
ret = -1;
}
virBitmapFree(present_cpus_map);
virBitmapFree(online_cpus_map);
VIR_FREE(sysfs_nodedir);
VIR_FREE(sysfs_cpudir);
return ret;
}
virStorageBackendProbeTarget(virStorageSourcePtr target,
char **backingStore,
int *backingStoreFormat,
virStorageEncryptionPtr *encryption)
{
int fd = -1;
int ret = -1;
virStorageFileMetadata *meta = NULL;
struct stat sb;
char *header = NULL;
ssize_t len = VIR_STORAGE_MAX_HEADER;
*backingStore = NULL;
*backingStoreFormat = VIR_STORAGE_FILE_AUTO;
if (encryption)
*encryption = NULL;
if ((ret = virStorageBackendVolOpen(target->path, &sb,
VIR_STORAGE_VOL_FS_PROBE_FLAGS)) < 0)
goto error; /* Take care to propagate ret, it is not always -1 */
fd = ret;
if ((ret = virStorageBackendUpdateVolTargetInfoFD(target, fd, &sb)) < 0) {
goto error;
}
if (S_ISDIR(sb.st_mode)) {
target->format = VIR_STORAGE_FILE_DIR;
} else {
if ((len = virFileReadHeaderFD(fd, len, &header)) < 0) {
virReportSystemError(errno, _("cannot read header '%s'"),
target->path);
goto error;
}
target->format = virStorageFileProbeFormatFromBuf(target->path,
header, len);
if (target->format < 0) {
ret = -1;
goto error;
}
if (!(meta = virStorageFileGetMetadataFromBuf(target->path,
header, len,
target->format,
backingStore,
backingStoreFormat))) {
ret = -1;
goto error;
}
}
VIR_FORCE_CLOSE(fd);
if (meta && *backingStore &&
*backingStoreFormat == VIR_STORAGE_FILE_AUTO &&
virStorageIsFile(*backingStore)) {
if ((ret = virStorageFileProbeFormat(*backingStore, -1, -1)) < 0) {
/* If the backing file is currently unavailable, only log an error,
* but continue. Returning -1 here would disable the whole storage
* pool, making it unavailable for even maintenance. */
virReportError(VIR_ERR_INTERNAL_ERROR,
_("cannot probe backing volume format: %s"),
*backingStore);
ret = -3;
} else {
*backingStoreFormat = ret;
ret = 0;
}
} else {
ret = 0;
}
if (meta && meta->capacity)
target->capacity = meta->capacity;
if (encryption && meta && meta->encryption) {
*encryption = meta->encryption;
meta->encryption = NULL;
switch (target->format) {
case VIR_STORAGE_FILE_QCOW:
case VIR_STORAGE_FILE_QCOW2:
(*encryption)->format = VIR_STORAGE_ENCRYPTION_FORMAT_QCOW;
break;
default:
break;
}
/* XXX ideally we'd fill in secret UUID here
* but we cannot guarantee 'conn' is non-NULL
* at this point in time :-( So we only fill
* in secrets when someone first queries a vol
*/
}
virBitmapFree(target->features);
if (meta) {
target->features = meta->features;
meta->features = NULL;
}
if (meta && meta->compat) {
VIR_FREE(target->compat);
target->compat = meta->compat;
meta->compat = NULL;
}
goto cleanup;
error:
VIR_FORCE_CLOSE(fd);
cleanup:
virStorageFileFreeMetadata(meta);
VIR_FREE(header);
return ret;
}
virHostCPUParseNode(const char *node,
virArch arch,
virBitmapPtr present_cpus_map,
virBitmapPtr online_cpus_map,
int threads_per_subcore,
int *sockets,
int *cores,
int *threads,
int *offline)
{
/* Biggest value we can expect to be used as either socket id
* or core id. Bitmaps will need to be sized accordingly */
const int ID_MAX = 4095;
int ret = -1;
int processors = 0;
DIR *cpudir = NULL;
struct dirent *cpudirent = NULL;
virBitmapPtr node_cpus_map = NULL;
virBitmapPtr sockets_map = NULL;
virBitmapPtr *cores_maps = NULL;
int npresent_cpus = virBitmapSize(present_cpus_map);
int sock_max = 0;
int sock;
int core;
size_t i;
int siblings;
unsigned int cpu;
int direrr;
*threads = 0;
*cores = 0;
*sockets = 0;
if (!(cpudir = opendir(node))) {
virReportSystemError(errno, _("cannot opendir %s"), node);
goto cleanup;
}
/* Keep track of the CPUs that belong to the current node */
if (!(node_cpus_map = virBitmapNew(npresent_cpus)))
goto cleanup;
/* enumerate sockets in the node */
if (!(sockets_map = virBitmapNew(ID_MAX + 1)))
goto cleanup;
while ((direrr = virDirRead(cpudir, &cpudirent, node)) > 0) {
if (sscanf(cpudirent->d_name, "cpu%u", &cpu) != 1)
continue;
if (!virBitmapIsBitSet(present_cpus_map, cpu))
continue;
/* Mark this CPU as part of the current node */
if (virBitmapSetBit(node_cpus_map, cpu) < 0)
goto cleanup;
if (!virBitmapIsBitSet(online_cpus_map, cpu))
continue;
/* Parse socket */
if ((sock = virHostCPUParseSocket(node, arch, cpu)) < 0)
goto cleanup;
if (sock > ID_MAX) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Socket %d can't be handled (max socket is %d)"),
sock, ID_MAX);
goto cleanup;
}
if (virBitmapSetBit(sockets_map, sock) < 0)
goto cleanup;
if (sock > sock_max)
sock_max = sock;
}
if (direrr < 0)
goto cleanup;
sock_max++;
/* allocate cores maps for each socket */
if (VIR_ALLOC_N(cores_maps, sock_max) < 0)
goto cleanup;
for (i = 0; i < sock_max; i++)
if (!(cores_maps[i] = virBitmapNew(ID_MAX + 1)))
goto cleanup;
/* Iterate over all CPUs in the node, in ascending order */
for (cpu = 0; cpu < npresent_cpus; cpu++) {
/* Skip CPUs that are not part of the current node */
if (!virBitmapIsBitSet(node_cpus_map, cpu))
continue;
if (!virBitmapIsBitSet(online_cpus_map, cpu)) {
if (threads_per_subcore > 0 &&
cpu % threads_per_subcore != 0 &&
virBitmapIsBitSet(online_cpus_map,
cpu - (cpu % threads_per_subcore))) {
/* Secondary offline threads are counted as online when
* subcores are in use and the corresponding primary
* thread is online */
processors++;
} else {
/* But they are counted as offline otherwise */
(*offline)++;
}
continue;
}
processors++;
/* Parse socket */
if ((sock = virHostCPUParseSocket(node, arch, cpu)) < 0)
goto cleanup;
if (!virBitmapIsBitSet(sockets_map, sock)) {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("CPU socket topology has changed"));
goto cleanup;
}
/* Parse core */
if (ARCH_IS_S390(arch)) {
/* logical cpu is equivalent to a core on s390 */
core = cpu;
} else {
if ((core = virHostCPUGetValue(node, cpu,
"topology/core_id", 0)) < 0)
goto cleanup;
}
if (core > ID_MAX) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Core %d can't be handled (max core is %d)"),
core, ID_MAX);
goto cleanup;
}
if (virBitmapSetBit(cores_maps[sock], core) < 0)
goto cleanup;
if (!(siblings = virHostCPUCountThreadSiblings(node, cpu)))
goto cleanup;
if (siblings > *threads)
*threads = siblings;
}
/* finalize the returned data */
*sockets = virBitmapCountBits(sockets_map);
for (i = 0; i < sock_max; i++) {
if (!virBitmapIsBitSet(sockets_map, i))
continue;
core = virBitmapCountBits(cores_maps[i]);
if (core > *cores)
*cores = core;
}
if (threads_per_subcore > 0) {
/* The thread count ignores offline threads, which means that only
* only primary threads have been considered so far. If subcores
* are in use, we need to also account for secondary threads */
*threads *= threads_per_subcore;
}
ret = processors;
cleanup:
/* don't shadow a more serious error */
if (cpudir && closedir(cpudir) < 0 && ret >= 0) {
virReportSystemError(errno, _("problem closing %s"), node);
ret = -1;
}
if (cores_maps)
for (i = 0; i < sock_max; i++)
virBitmapFree(cores_maps[i]);
VIR_FREE(cores_maps);
virBitmapFree(sockets_map);
virBitmapFree(node_cpus_map);
return ret;
}
