#include "qemu/bitmap.h"
#include "qom/cpu.h"
#include "qemu/error-report.h"
-#include "include/exec/cpu-common.h" /* for RAM_ADDR_FMT */
-#include "qapi-visit.h"
+#include "qapi/error.h"
#include "qapi/opts-visitor.h"
+#include "qapi/qapi-commands-misc.h"
+#include "qapi/qapi-visit-misc.h"
#include "hw/boards.h"
#include "sysemu/hostmem.h"
-#include "qmp-commands.h"
#include "hw/mem/pc-dimm.h"
+#include "hw/mem/memory-device.h"
#include "qemu/option.h"
#include "qemu/config-file.h"
+#include "qemu/cutils.h"
QemuOptsList qemu_numa_opts = {
.name = "numa",
* For all nodes, nodeid < max_numa_nodeid
*/
int nb_numa_nodes;
+bool have_numa_distance;
NodeInfo numa_info[MAX_NODES];
-void numa_set_mem_node_id(ram_addr_t addr, uint64_t size, uint32_t node)
-{
- struct numa_addr_range *range;
-
- /*
- * Memory-less nodes can come here with 0 size in which case,
- * there is nothing to do.
- */
- if (!size) {
- return;
- }
-
- range = g_malloc0(sizeof(*range));
- range->mem_start = addr;
- range->mem_end = addr + size - 1;
- QLIST_INSERT_HEAD(&numa_info[node].addr, range, entry);
-}
-
-void numa_unset_mem_node_id(ram_addr_t addr, uint64_t size, uint32_t node)
-{
- struct numa_addr_range *range, *next;
-
- QLIST_FOREACH_SAFE(range, &numa_info[node].addr, entry, next) {
- if (addr == range->mem_start && (addr + size - 1) == range->mem_end) {
- QLIST_REMOVE(range, entry);
- g_free(range);
- return;
- }
- }
-}
-
-static void numa_set_mem_ranges(void)
-{
- int i;
- ram_addr_t mem_start = 0;
-
- /*
- * Deduce start address of each node and use it to store
- * the address range info in numa_info address range list
- */
- for (i = 0; i < nb_numa_nodes; i++) {
- numa_set_mem_node_id(mem_start, numa_info[i].node_mem, i);
- mem_start += numa_info[i].node_mem;
- }
-}
-
-/*
- * Check if @addr falls under NUMA @node.
- */
-static bool numa_addr_belongs_to_node(ram_addr_t addr, uint32_t node)
-{
- struct numa_addr_range *range;
-
- QLIST_FOREACH(range, &numa_info[node].addr, entry) {
- if (addr >= range->mem_start && addr <= range->mem_end) {
- return true;
- }
- }
- return false;
-}
-
-/*
- * Given an address, return the index of the NUMA node to which the
- * address belongs to.
- */
-uint32_t numa_get_node(ram_addr_t addr, Error **errp)
-{
- uint32_t i;
-
- /* For non NUMA configurations, check if the addr falls under node 0 */
- if (!nb_numa_nodes) {
- if (numa_addr_belongs_to_node(addr, 0)) {
- return 0;
- }
- }
-
- for (i = 0; i < nb_numa_nodes; i++) {
- if (numa_addr_belongs_to_node(addr, i)) {
- return i;
- }
- }
-
- error_setg(errp, "Address 0x" RAM_ADDR_FMT " doesn't belong to any "
- "NUMA node", addr);
- return -1;
-}
-static void numa_node_parse(NumaNodeOptions *node, QemuOpts *opts, Error **errp)
+static void parse_numa_node(MachineState *ms, NumaNodeOptions *node,
+ Error **errp)
{
+ Error *err = NULL;
uint16_t nodenr;
uint16List *cpus = NULL;
+ MachineClass *mc = MACHINE_GET_CLASS(ms);
if (node->has_nodeid) {
nodenr = node->nodeid;
return;
}
+ if (!mc->cpu_index_to_instance_props || !mc->get_default_cpu_node_id) {
+ error_setg(errp, "NUMA is not supported by this machine-type");
+ return;
+ }
for (cpus = node->cpus; cpus; cpus = cpus->next) {
+ CpuInstanceProperties props;
if (cpus->value >= max_cpus) {
error_setg(errp,
"CPU index (%" PRIu16 ")"
cpus->value, max_cpus);
return;
}
- bitmap_set(numa_info[nodenr].node_cpu, cpus->value, 1);
+ props = mc->cpu_index_to_instance_props(ms, cpus->value);
+ props.node_id = nodenr;
+ props.has_node_id = true;
+ machine_set_cpu_numa_node(ms, &props, &err);
+ if (err) {
+ error_propagate(errp, err);
+ return;
+ }
}
if (node->has_mem && node->has_memdev) {
- error_setg(errp, "qemu: cannot specify both mem= and memdev=");
+ error_setg(errp, "cannot specify both mem= and memdev=");
return;
}
have_memdevs = node->has_memdev;
}
if (node->has_memdev != have_memdevs) {
- error_setg(errp, "qemu: memdev option must be specified for either "
+ error_setg(errp, "memdev option must be specified for either "
"all or no nodes");
return;
}
if (node->has_mem) {
- uint64_t mem_size = node->mem;
- const char *mem_str = qemu_opt_get(opts, "mem");
- /* Fix up legacy suffix-less format */
- if (g_ascii_isdigit(mem_str[strlen(mem_str) - 1])) {
- mem_size <<= 20;
- }
- numa_info[nodenr].node_mem = mem_size;
+ numa_info[nodenr].node_mem = node->mem;
}
if (node->has_memdev) {
Object *o;
}
object_ref(o);
- numa_info[nodenr].node_mem = object_property_get_int(o, "size", NULL);
+ numa_info[nodenr].node_mem = object_property_get_uint(o, "size", NULL);
numa_info[nodenr].node_memdev = MEMORY_BACKEND(o);
}
numa_info[nodenr].present = true;
max_numa_nodeid = MAX(max_numa_nodeid, nodenr + 1);
+ nb_numa_nodes++;
}
-static int parse_numa(void *opaque, QemuOpts *opts, Error **errp)
+static void parse_numa_distance(NumaDistOptions *dist, Error **errp)
{
- NumaOptions *object = NULL;
- Error *err = NULL;
+ uint16_t src = dist->src;
+ uint16_t dst = dist->dst;
+ uint8_t val = dist->val;
- {
- Visitor *v = opts_visitor_new(opts);
- visit_type_NumaOptions(v, NULL, &object, &err);
- visit_free(v);
+ if (src >= MAX_NODES || dst >= MAX_NODES) {
+ error_setg(errp, "Parameter '%s' expects an integer between 0 and %d",
+ src >= MAX_NODES ? "src" : "dst", MAX_NODES - 1);
+ return;
}
- if (err) {
- goto end;
+ if (!numa_info[src].present || !numa_info[dst].present) {
+ error_setg(errp, "Source/Destination NUMA node is missing. "
+ "Please use '-numa node' option to declare it first.");
+ return;
+ }
+
+ if (val < NUMA_DISTANCE_MIN) {
+ error_setg(errp, "NUMA distance (%" PRIu8 ") is invalid, "
+ "it shouldn't be less than %d.",
+ val, NUMA_DISTANCE_MIN);
+ return;
}
+ if (src == dst && val != NUMA_DISTANCE_MIN) {
+ error_setg(errp, "Local distance of node %d should be %d.",
+ src, NUMA_DISTANCE_MIN);
+ return;
+ }
+
+ numa_info[src].distance[dst] = val;
+ have_numa_distance = true;
+}
+
+static
+void set_numa_options(MachineState *ms, NumaOptions *object, Error **errp)
+{
+ Error *err = NULL;
+
switch (object->type) {
case NUMA_OPTIONS_TYPE_NODE:
- numa_node_parse(&object->u.node, opts, &err);
+ parse_numa_node(ms, &object->u.node, &err);
if (err) {
goto end;
}
- nb_numa_nodes++;
+ break;
+ case NUMA_OPTIONS_TYPE_DIST:
+ parse_numa_distance(&object->u.dist, &err);
+ if (err) {
+ goto end;
+ }
+ break;
+ case NUMA_OPTIONS_TYPE_CPU:
+ if (!object->u.cpu.has_node_id) {
+ error_setg(&err, "Missing mandatory node-id property");
+ goto end;
+ }
+ if (!numa_info[object->u.cpu.node_id].present) {
+ error_setg(&err, "Invalid node-id=%" PRId64 ", NUMA node must be "
+ "defined with -numa node,nodeid=ID before it's used with "
+ "-numa cpu,node-id=ID", object->u.cpu.node_id);
+ goto end;
+ }
+
+ machine_set_cpu_numa_node(ms, qapi_NumaCpuOptions_base(&object->u.cpu),
+ &err);
break;
default:
abort();
}
+end:
+ error_propagate(errp, err);
+}
+
+static int parse_numa(void *opaque, QemuOpts *opts, Error **errp)
+{
+ NumaOptions *object = NULL;
+ MachineState *ms = MACHINE(opaque);
+ Error *err = NULL;
+ Visitor *v = opts_visitor_new(opts);
+
+ visit_type_NumaOptions(v, NULL, &object, &err);
+ visit_free(v);
+ if (err) {
+ goto end;
+ }
+
+ /* Fix up legacy suffix-less format */
+ if ((object->type == NUMA_OPTIONS_TYPE_NODE) && object->u.node.has_mem) {
+ const char *mem_str = qemu_opt_get(opts, "mem");
+ qemu_strtosz_MiB(mem_str, NULL, &object->u.node.mem);
+ }
+
+ set_numa_options(ms, object, &err);
+
end:
qapi_free_NumaOptions(object);
if (err) {
- error_report_err(err);
+ error_propagate(errp, err);
return -1;
}
return 0;
}
-static char *enumerate_cpus(unsigned long *cpus, int max_cpus)
+/* If all node pair distances are symmetric, then only distances
+ * in one direction are enough. If there is even one asymmetric
+ * pair, though, then all distances must be provided. The
+ * distance from a node to itself is always NUMA_DISTANCE_MIN,
+ * so providing it is never necessary.
+ */
+static void validate_numa_distance(void)
{
- int cpu;
- bool first = true;
- GString *s = g_string_new(NULL);
+ int src, dst;
+ bool is_asymmetrical = false;
+
+ for (src = 0; src < nb_numa_nodes; src++) {
+ for (dst = src; dst < nb_numa_nodes; dst++) {
+ if (numa_info[src].distance[dst] == 0 &&
+ numa_info[dst].distance[src] == 0) {
+ if (src != dst) {
+ error_report("The distance between node %d and %d is "
+ "missing, at least one distance value "
+ "between each nodes should be provided.",
+ src, dst);
+ exit(EXIT_FAILURE);
+ }
+ }
- for (cpu = find_first_bit(cpus, max_cpus);
- cpu < max_cpus;
- cpu = find_next_bit(cpus, max_cpus, cpu + 1)) {
- g_string_append_printf(s, "%s%d", first ? "" : " ", cpu);
- first = false;
+ if (numa_info[src].distance[dst] != 0 &&
+ numa_info[dst].distance[src] != 0 &&
+ numa_info[src].distance[dst] !=
+ numa_info[dst].distance[src]) {
+ is_asymmetrical = true;
+ }
+ }
+ }
+
+ if (is_asymmetrical) {
+ for (src = 0; src < nb_numa_nodes; src++) {
+ for (dst = 0; dst < nb_numa_nodes; dst++) {
+ if (src != dst && numa_info[src].distance[dst] == 0) {
+ error_report("At least one asymmetrical pair of "
+ "distances is given, please provide distances "
+ "for both directions of all node pairs.");
+ exit(EXIT_FAILURE);
+ }
+ }
+ }
}
- return g_string_free(s, FALSE);
}
-static void validate_numa_cpus(void)
+static void complete_init_numa_distance(void)
{
- int i;
- unsigned long *seen_cpus = bitmap_new(max_cpus);
+ int src, dst;
- for (i = 0; i < nb_numa_nodes; i++) {
- if (bitmap_intersects(seen_cpus, numa_info[i].node_cpu, max_cpus)) {
- bitmap_and(seen_cpus, seen_cpus,
- numa_info[i].node_cpu, max_cpus);
- error_report("CPU(s) present in multiple NUMA nodes: %s",
- enumerate_cpus(seen_cpus, max_cpus));
- g_free(seen_cpus);
- exit(EXIT_FAILURE);
+ /* Fixup NUMA distance by symmetric policy because if it is an
+ * asymmetric distance table, it should be a complete table and
+ * there would not be any missing distance except local node, which
+ * is verified by validate_numa_distance above.
+ */
+ for (src = 0; src < nb_numa_nodes; src++) {
+ for (dst = 0; dst < nb_numa_nodes; dst++) {
+ if (numa_info[src].distance[dst] == 0) {
+ if (src == dst) {
+ numa_info[src].distance[dst] = NUMA_DISTANCE_MIN;
+ } else {
+ numa_info[src].distance[dst] = numa_info[dst].distance[src];
+ }
+ }
}
- bitmap_or(seen_cpus, seen_cpus,
- numa_info[i].node_cpu, max_cpus);
}
+}
+
+void numa_legacy_auto_assign_ram(MachineClass *mc, NodeInfo *nodes,
+ int nb_nodes, ram_addr_t size)
+{
+ int i;
+ uint64_t usedmem = 0;
- if (!bitmap_full(seen_cpus, max_cpus)) {
- char *msg;
- bitmap_complement(seen_cpus, seen_cpus, max_cpus);
- msg = enumerate_cpus(seen_cpus, max_cpus);
- error_report("warning: CPU(s) not present in any NUMA nodes: %s", msg);
- error_report("warning: All CPU(s) up to maxcpus should be described "
- "in NUMA config");
- g_free(msg);
+ /* Align each node according to the alignment
+ * requirements of the machine class
+ */
+
+ for (i = 0; i < nb_nodes - 1; i++) {
+ nodes[i].node_mem = (size / nb_nodes) &
+ ~((1 << mc->numa_mem_align_shift) - 1);
+ usedmem += nodes[i].node_mem;
}
- g_free(seen_cpus);
+ nodes[i].node_mem = size - usedmem;
}
-void parse_numa_opts(MachineClass *mc)
+void numa_default_auto_assign_ram(MachineClass *mc, NodeInfo *nodes,
+ int nb_nodes, ram_addr_t size)
{
int i;
+ uint64_t usedmem = 0, node_mem;
+ uint64_t granularity = size / nb_nodes;
+ uint64_t propagate = 0;
+
+ for (i = 0; i < nb_nodes - 1; i++) {
+ node_mem = (granularity + propagate) &
+ ~((1 << mc->numa_mem_align_shift) - 1);
+ propagate = granularity + propagate - node_mem;
+ nodes[i].node_mem = node_mem;
+ usedmem += node_mem;
+ }
+ nodes[i].node_mem = size - usedmem;
+}
- for (i = 0; i < MAX_NODES; i++) {
- numa_info[i].node_cpu = bitmap_new(max_cpus);
- }
+void numa_complete_configuration(MachineState *ms)
+{
+ int i;
+ MachineClass *mc = MACHINE_GET_CLASS(ms);
- if (qemu_opts_foreach(qemu_find_opts("numa"), parse_numa, NULL, NULL)) {
- exit(1);
+ /*
+ * If memory hotplug is enabled (slots > 0) but without '-numa'
+ * options explicitly on CLI, guestes will break.
+ *
+ * Windows: won't enable memory hotplug without SRAT table at all
+ *
+ * Linux: if QEMU is started with initial memory all below 4Gb
+ * and no SRAT table present, guest kernel will use nommu DMA ops,
+ * which breaks 32bit hw drivers when memory is hotplugged and
+ * guest tries to use it with that drivers.
+ *
+ * Enable NUMA implicitly by adding a new NUMA node automatically.
+ */
+ if (ms->ram_slots > 0 && nb_numa_nodes == 0 &&
+ mc->auto_enable_numa_with_memhp) {
+ NumaNodeOptions node = { };
+ parse_numa_node(ms, &node, &error_abort);
}
assert(max_numa_nodeid <= MAX_NODES);
}
}
if (i == nb_numa_nodes) {
- uint64_t usedmem = 0;
-
- /* Align each node according to the alignment
- * requirements of the machine class
- */
- for (i = 0; i < nb_numa_nodes - 1; i++) {
- numa_info[i].node_mem = (ram_size / nb_numa_nodes) &
- ~((1 << mc->numa_mem_align_shift) - 1);
- usedmem += numa_info[i].node_mem;
- }
- numa_info[i].node_mem = ram_size - usedmem;
+ assert(mc->numa_auto_assign_ram);
+ mc->numa_auto_assign_ram(mc, numa_info, nb_numa_nodes, ram_size);
}
numa_total = 0;
exit(1);
}
- for (i = 0; i < nb_numa_nodes; i++) {
- QLIST_INIT(&numa_info[i].addr);
- }
-
- numa_set_mem_ranges();
-
- for (i = 0; i < nb_numa_nodes; i++) {
- if (!bitmap_empty(numa_info[i].node_cpu, max_cpus)) {
- break;
- }
- }
- /* Historically VCPUs were assigned in round-robin order to NUMA
- * nodes. However it causes issues with guest not handling it nice
- * in case where cores/threads from a multicore CPU appear on
- * different nodes. So allow boards to override default distribution
- * rule grouping VCPUs by socket so that VCPUs from the same socket
- * would be on the same node.
+ /* QEMU needs at least all unique node pair distances to build
+ * the whole NUMA distance table. QEMU treats the distance table
+ * as symmetric by default, i.e. distance A->B == distance B->A.
+ * Thus, QEMU is able to complete the distance table
+ * initialization even though only distance A->B is provided and
+ * distance B->A is not. QEMU knows the distance of a node to
+ * itself is always 10, so A->A distances may be omitted. When
+ * the distances of two nodes of a pair differ, i.e. distance
+ * A->B != distance B->A, then that means the distance table is
+ * asymmetric. In this case, the distances for both directions
+ * of all node pairs are required.
*/
- if (i == nb_numa_nodes) {
- for (i = 0; i < max_cpus; i++) {
- unsigned node_id = i % nb_numa_nodes;
- if (mc->cpu_index_to_socket_id) {
- node_id = mc->cpu_index_to_socket_id(i) % nb_numa_nodes;
- }
+ if (have_numa_distance) {
+ /* Validate enough NUMA distance information was provided. */
+ validate_numa_distance();
- set_bit(i, numa_info[node_id].node_cpu);
- }
+ /* Validation succeeded, now fill in any missing distances. */
+ complete_init_numa_distance();
}
-
- validate_numa_cpus();
- } else {
- numa_set_mem_node_id(0, ram_size, 0);
}
}
-void numa_post_machine_init(void)
+void parse_numa_opts(MachineState *ms)
{
- CPUState *cpu;
- int i;
+ qemu_opts_foreach(qemu_find_opts("numa"), parse_numa, ms, &error_fatal);
+}
- CPU_FOREACH(cpu) {
- for (i = 0; i < nb_numa_nodes; i++) {
- assert(cpu->cpu_index < max_cpus);
- if (test_bit(cpu->cpu_index, numa_info[i].node_cpu)) {
- cpu->numa_node = i;
- }
+void qmp_set_numa_node(NumaOptions *cmd, Error **errp)
+{
+ if (!runstate_check(RUN_STATE_PRECONFIG)) {
+ error_setg(errp, "The command is permitted only in '%s' state",
+ RunState_str(RUN_STATE_PRECONFIG));
+ return;
+ }
+
+ set_numa_options(MACHINE(qdev_get_machine()), cmd, errp);
+}
+
+void numa_cpu_pre_plug(const CPUArchId *slot, DeviceState *dev, Error **errp)
+{
+ int node_id = object_property_get_int(OBJECT(dev), "node-id", &error_abort);
+
+ if (node_id == CPU_UNSET_NUMA_NODE_ID) {
+ /* due to bug in libvirt, it doesn't pass node-id from props on
+ * device_add as expected, so we have to fix it up here */
+ if (slot->props.has_node_id) {
+ object_property_set_int(OBJECT(dev), slot->props.node_id,
+ "node-id", errp);
}
+ } else if (node_id != slot->props.node_id) {
+ error_setg(errp, "node-id=%d must match numa node specified "
+ "with -numa option", node_id);
}
}
if (mem_path) {
#ifdef __linux__
Error *err = NULL;
- memory_region_init_ram_from_file(mr, owner, name, ram_size, false,
+ memory_region_init_ram_from_file(mr, owner, name, ram_size, 0, 0,
mem_path, &err);
if (err) {
error_report_err(err);
if (mem_prealloc) {
exit(1);
}
+ error_report("falling back to regular RAM allocation.");
/* Legacy behavior: if allocation failed, fall back to
* regular RAM allocation.
*/
- memory_region_init_ram(mr, owner, name, ram_size, &error_fatal);
+ mem_path = NULL;
+ memory_region_init_ram_nomigrate(mr, owner, name, ram_size, &error_fatal);
}
#else
fprintf(stderr, "-mem-path not supported on this host\n");
exit(1);
#endif
} else {
- memory_region_init_ram(mr, owner, name, ram_size, &error_fatal);
+ memory_region_init_ram_nomigrate(mr, owner, name, ram_size, &error_fatal);
}
vmstate_register_ram_global(mr);
}
}
memory_region_init(mr, owner, name, ram_size);
- for (i = 0; i < MAX_NODES; i++) {
+ for (i = 0; i < nb_numa_nodes; i++) {
uint64_t size = numa_info[i].node_mem;
HostMemoryBackend *backend = numa_info[i].node_memdev;
if (!backend) {
continue;
}
- MemoryRegion *seg = host_memory_backend_get_memory(backend,
- &error_fatal);
+ MemoryRegion *seg = host_memory_backend_get_memory(backend);
if (memory_region_is_mapped(seg)) {
char *path = object_get_canonical_path_component(OBJECT(backend));
error_report("memory backend %s is used multiple times. Each "
"-numa option must use a different memdev value.",
path);
+ g_free(path);
exit(1);
}
}
}
-static void numa_stat_memory_devices(uint64_t node_mem[])
+static void numa_stat_memory_devices(NumaNodeMem node_mem[])
{
- MemoryDeviceInfoList *info_list = NULL;
- MemoryDeviceInfoList **prev = &info_list;
+ MemoryDeviceInfoList *info_list = qmp_memory_device_list();
MemoryDeviceInfoList *info;
+ PCDIMMDeviceInfo *pcdimm_info;
- qmp_pc_dimm_device_list(qdev_get_machine(), &prev);
for (info = info_list; info; info = info->next) {
MemoryDeviceInfo *value = info->value;
if (value) {
switch (value->type) {
case MEMORY_DEVICE_INFO_KIND_DIMM:
- node_mem[value->u.dimm.data->node] += value->u.dimm.data->size;
+ pcdimm_info = value->u.dimm.data;
break;
+
+ case MEMORY_DEVICE_INFO_KIND_NVDIMM:
+ pcdimm_info = value->u.nvdimm.data;
+ break;
+
default:
+ pcdimm_info = NULL;
break;
}
+
+ if (pcdimm_info) {
+ node_mem[pcdimm_info->node].node_mem += pcdimm_info->size;
+ node_mem[pcdimm_info->node].node_plugged_mem +=
+ pcdimm_info->size;
+ }
}
}
qapi_free_MemoryDeviceInfoList(info_list);
}
-void query_numa_node_mem(uint64_t node_mem[])
+void query_numa_node_mem(NumaNodeMem node_mem[])
{
int i;
numa_stat_memory_devices(node_mem);
for (i = 0; i < nb_numa_nodes; i++) {
- node_mem[i] += numa_info[i].node_mem;
+ node_mem[i].node_mem += numa_info[i].node_mem;
}
}
m->value = g_malloc0(sizeof(*m->value));
- m->value->id = object_property_get_str(obj, "id", NULL);
+ m->value->id = object_get_canonical_path_component(obj);
m->value->has_id = !!m->value->id;
- m->value->size = object_property_get_int(obj, "size",
- &error_abort);
+ m->value->size = object_property_get_uint(obj, "size",
+ &error_abort);
m->value->merge = object_property_get_bool(obj, "merge",
&error_abort);
m->value->dump = object_property_get_bool(obj, "dump",
return list;
}
-int numa_get_node_for_cpu(int idx)
-{
- int i;
-
- assert(idx < max_cpus);
-
- for (i = 0; i < nb_numa_nodes; i++) {
- if (test_bit(idx, numa_info[i].node_cpu)) {
- break;
- }
- }
- return i;
-}
-
void ram_block_notifier_add(RAMBlockNotifier *n)
{
QLIST_INSERT_HEAD(&ram_list.ramblock_notifiers, n, next);
projects / mirror_qemu.git / blobdiff