Allow setting NUMA distance for different NUMA nodes

This patch is going to add SLIT table support in QEMU, and provides additional option dist for command -numa to allow user set vNUMA distance by QEMU command.

With this patch, when a user wants to create a guest that contains several vNUMA nodes and also wants to set distance among those nodes, the QEMU command would like:

```

(quoted diff) +0 -10

-numa node,nodeid=0,cpus=0 \
-numa node,nodeid=1,cpus=1 \
-numa node,nodeid=2,cpus=2 \
-numa node,nodeid=3,cpus=3 \
-numa dist,src=0,dst=1,val=21 \
-numa dist,src=0,dst=2,val=31 \
-numa dist,src=0,dst=3,val=41 \
-numa dist,src=1,dst=2,val=21 \
-numa dist,src=1,dst=3,val=31 \
-numa dist,src=2,dst=3,val=21 \

```

Signed-off-by He Chen <[email protected]>

---
 hw/acpi/aml-build.c         |  25 +++++++++
 hw/i386/acpi-build.c        |   2 +
 include/hw/acpi/aml-build.h |   1 +
 include/sysemu/numa.h       |   1 +
 include/sysemu/sysemu.h     |   4 ++
 numa.c                      | 121 ++++++++++++++++++++++++++++++++++++++++++++
 qapi-schema.json            |  30 ++++++++++-
 qemu-options.hx             |  17 ++++++-
 8 files changed, 198 insertions(+), 3 deletions(-)

hw/acpi/aml-build.c +25 -0

index c6f2032..2c6ab07 100644
@@ -24,6 +24,7 @@
 #include "hw/acpi/aml-build.h"
 #include "qemu/bswap.h"
 #include "qemu/bitops.h"
+#include "sysemu/numa.h"
 
 static GArray *build_alloc_array(void)
 {

@@ -1609,3 +1610,27 @@ void build_srat_memory(AcpiSratMemoryAffinity *numamem, uint64_t base,
     numamem->base_addr = cpu_to_le64(base);
     numamem->range_length = cpu_to_le64(len);
 }
+
+/*
+ * ACPI spec 5.2.17 System Locality Distance Information Table
+ * (Revision 2.0 or later)
+ */
+void build_slit(GArray *table_data, BIOSLinker *linker)
+{
+    int slit_start, i, j;
+    slit_start = table_data->len;
+
+    acpi_data_push(table_data, sizeof(AcpiTableHeader));
+
+    build_append_int_noprefix(table_data, nb_numa_nodes, 8);
+    for (i = 0; i < nb_numa_nodes; i++) {
+        for (j = 0; j < nb_numa_nodes; j++) {
+            build_append_int_noprefix(table_data, numa_info[i].distance[j], 1);
+        }
+    }
+
+    build_header(linker, table_data,
+                 (void *)(table_data->data + slit_start),
+                 "SLIT",
+                 table_data->len - slit_start, 1, NULL, NULL);
+}

hw/i386/acpi-build.c +0 -0

index 2073108..12730ea 100644

hw/i386/acpi-build.c +0 -0

     }
     if (acpi_get_mcfg(&mcfg)) {
         acpi_add_table(table_offsets, tables_blob);

include/hw/acpi/aml-build.h +1 -0

index 00c21f1..329a0d0 100644
@@ -389,4 +389,5 @@ GCC_FMT_ATTR(2, 3);
 void build_srat_memory(AcpiSratMemoryAffinity *numamem, uint64_t base,
                        uint64_t len, int node, MemoryAffinityFlags flags);
 
+void build_slit(GArray *table_data, BIOSLinker *linker);
 #endif

include/sysemu/numa.h +1 -0

index 8f09dcf..2f7a941 100644
@@ -21,6 +21,7 @@ typedef struct node_info {
     struct HostMemoryBackend *node_memdev;
     bool present;
     QLIST_HEAD(, numa_addr_range) addr; /* List to store address ranges */
+    uint8_t distance[MAX_NODES];
 } NodeInfo;
 
 extern NodeInfo numa_info[MAX_NODES];

include/sysemu/sysemu.h +4 -0

index 576c7ce..6999545 100644
@@ -169,6 +169,10 @@ extern int mem_prealloc;
 
 #define MAX_NODES 128
 #define NUMA_NODE_UNASSIGNED MAX_NODES
+#define NUMA_DISTANCE_MIN         10
+#define NUMA_DISTANCE_DEFAULT     20
+#define NUMA_DISTANCE_MAX         254
+#define NUMA_DISTANCE_UNREACHABLE 255
 
 #define MAX_OPTION_ROMS 16
 typedef struct QEMUOptionRom {

numa.c +6 -0

index 6fc2393..838e45a 100644
@@ -52,6 +52,7 @@ static int max_numa_nodeid; /* Highest specified NUMA node ID, plus one.
                              */
 int nb_numa_nodes;
 NodeInfo numa_info[MAX_NODES];
+static bool have_numa_distance;
 
 void numa_set_mem_node_id(ram_addr_t addr, uint64_t size, uint32_t node)
 {

@@ -212,6 +213,41 @@ static void numa_node_parse(NumaNodeOptions *node, QemuOpts *opts, Error **errp)
     max_numa_nodeid = MAX(max_numa_nodeid, nodenr + 1);
 }
 
+static void numa_distance_parse(NumaDistOptions *dist, QemuOpts *opts, Error **errp)
+{
+    uint16_t src = dist->src;
+    uint16_t dst = dist->dst;
+    uint8_t val = dist->val;

numa.c +26 -0

+        return;
+    }
+
+    if (val < NUMA_DISTANCE_MIN) {
+        error_setg(errp, "NUMA distance (%" PRIu8 ") is invalid, "
+                   "it should be larger 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 int parse_numa(void *opaque, QemuOpts *opts, Error **errp)
 {
     NumaOptions *object = NULL;

@@ -235,6 +271,12 @@ static int parse_numa(void *opaque, QemuOpts *opts, Error **errp)
         }
         nb_numa_nodes++;
         break;
+    case NUMA_OPTIONS_TYPE_DIST:
+        numa_distance_parse(&object->u.dist, opts, &err);
+        if (err) {
+            goto end;
+        }
+        break;
     default:
         abort();
     }

numa.c +6 -0

+
+    if (!have_numa_distance) {
+        for (src = 0; src < nb_numa_nodes; src++) {
+            for (dst = 0; dst < nb_numa_nodes; dst++) {
+                if (numa_info[src].present && numa_info[dst].present) {
+                    if (src == dst) {

numa.c +5 -0

+
+    for (src = 0; src < nb_numa_nodes; src++) {
+        for (dst = src; dst < nb_numa_nodes; dst++) {
+            s = src;
+            d = dst;

numa.c +8 -0

+
+                if (numa_info[s].distance[d] == 0) {
+                    s = dst;
+                    d = src;
+                }
+
+                if (numa_info[d].distance[s] == 0) {
+                    opposite_miss = true;

numa.c +3 -0

+}
+
 void parse_numa_opts(MachineClass *mc)
 {
     int i;

@@ -390,6 +510,7 @@ void parse_numa_opts(MachineClass *mc)
         }
 
         validate_numa_cpus();
+        validate_numa_distance();
     } else {
         numa_set_mem_node_id(0, ram_size, 0);
     }

qapi-schema.json +8 -2

index 250e4dc..7552777 100644
@@ -5673,10 +5673,14 @@
 ##
 # @NumaOptionsType:
 #
+# @node: NUMA nodes configuration
+#
+# @dist: NUMA distance configuration (since 2.10)
+#
 # Since: 2.1
 ##
 { 'enum': 'NumaOptionsType',
-  'data': [ 'node' ] }
+  'data': [ 'node', 'dist' ] }
 
 ##
 # @NumaOptions:

@@ -5689,7 +5693,8 @@
   'base': { 'type': 'NumaOptionsType' },
   'discriminator': 'type',
   'data': {
-    'node': 'NumaNodeOptions' }}
+    'node': 'NumaNodeOptions',
+    'dist': 'NumaDistOptions' }}
 
 ##
 # @NumaNodeOptions:

@@ -5718,6 +5723,27 @@
    '*memdev': 'str' }}
 
 ##
+# @NumaDistOptions:

qapi-schema.json +10 -0

+#
+# Since: 2.10
+##
+{ 'struct': 'NumaDistOptions',
+  'data': {
+   'src': 'uint16',
+   'dst': 'uint16',
+   'val': 'uint8' }}
+
+##
 # @HostMemPolicy:
 #
 # Host memory policy types

qemu-options.hx +5 -1

index 99af8ed..2318d85 100644
@@ -139,12 +139,15 @@ ETEXI
 
 DEF("numa", HAS_ARG, QEMU_OPTION_numa,
     "-numa node[,mem=size][,cpus=firstcpu[-lastcpu]][,nodeid=node]\n"
-    "-numa node[,memdev=id][,cpus=firstcpu[-lastcpu]][,nodeid=node]\n", QEMU_ARCH_ALL)
+    "-numa node[,memdev=id][,cpus=firstcpu[-lastcpu]][,nodeid=node]\n"
+    "-numa dist,src=source,dst=destination,val=distance\n", QEMU_ARCH_ALL)
 STEXI
 @item -numa node[,mem=@var{size}][,cpus=@var{firstcpu}[-@var{lastcpu}]][,nodeid=@var{node}]
 @itemx -numa node[,memdev=@var{id}][,cpus=@var{firstcpu}[-@var{lastcpu}]][,nodeid=@var{node}]
+@itemx -numa dist,src=@var{source},dst=@var{destination},val=@var{distance}
 @findex -numa
 Define a NUMA node and assign RAM and VCPUs to it.
+Set the NUMA distance from a source node to a destination node.
 
 @var{firstcpu} and @var{lastcpu} are CPU indexes. Each
 @samp{cpus} option represent a contiguous range of CPU indexes

@@ -167,6 +170,18 @@ split equally between them.
 @samp{mem} and @samp{memdev} are mutually exclusive. Furthermore,
 if one node uses @samp{memdev}, all of them have to use it.
 
+@var{source} and @var{destination} are NUMA node IDs.

qemu-options.hx +1 -0

+
 Note that the -@option{numa} option doesn't allocate any of the
 specified resources, it just assigns existing resources to NUMA
 nodes. This means that one still has to use the @option{-m},

Signature · 1 line

2.7.4