#include <asm/acpi.h>
#include <asm/k8.h>
-#ifndef Dprintk
-#define Dprintk(x...)
+#ifdef CONFIG_DEBUG_PER_CPU_MAPS
+# define DBG(x...) printk(KERN_DEBUG x)
+#else
+# define DBG(x...)
#endif
struct pglist_data *node_data[MAX_NUMNODES] __read_mostly;
EXPORT_SYMBOL(node_data);
-bootmem_data_t plat_node_bdata[MAX_NUMNODES];
-
struct memnode memnode;
-int x86_cpu_to_node_map_init[NR_CPUS] = {
- [0 ... NR_CPUS-1] = NUMA_NO_NODE
-};
-void *x86_cpu_to_node_map_early_ptr;
-DEFINE_PER_CPU(int, x86_cpu_to_node_map) = NUMA_NO_NODE;
-EXPORT_PER_CPU_SYMBOL(x86_cpu_to_node_map);
-EXPORT_SYMBOL(x86_cpu_to_node_map_early_ptr);
-
s16 apicid_to_node[MAX_LOCAL_APIC] __cpuinitdata = {
[0 ... MAX_LOCAL_APIC-1] = NUMA_NO_NODE
};
-cpumask_t node_to_cpumask_map[MAX_NUMNODES] __read_mostly;
-EXPORT_SYMBOL(node_to_cpumask_map);
-
int numa_off __initdata;
-unsigned long __initdata nodemap_addr;
-unsigned long __initdata nodemap_size;
+static unsigned long __initdata nodemap_addr;
+static unsigned long __initdata nodemap_size;
+
+DEFINE_PER_CPU(int, node_number) = 0;
+EXPORT_PER_CPU_SYMBOL(node_number);
+
+/*
+ * Map cpu index to node index
+ */
+DEFINE_EARLY_PER_CPU(int, x86_cpu_to_node_map, NUMA_NO_NODE);
+EXPORT_EARLY_PER_CPU_SYMBOL(x86_cpu_to_node_map);
+
+/*
+ * Which logical CPUs are on which nodes
+ */
+cpumask_t *node_to_cpumask_map;
+EXPORT_SYMBOL(node_to_cpumask_map);
/*
* Given a shift value, try to populate memnodemap[]
* -1 if node overlap or lost ram (shift too big)
*/
static int __init populate_memnodemap(const struct bootnode *nodes,
- int numnodes, int shift)
+ int numnodes, int shift, int *nodeids)
{
unsigned long addr, end;
int i, res = -1;
do {
if (memnodemap[addr >> shift] != NUMA_NO_NODE)
return -1;
- memnodemap[addr >> shift] = i;
+
+ if (!nodeids)
+ memnodemap[addr >> shift] = i;
+ else
+ memnodemap[addr >> shift] = nodeids[i];
+
addr += (1UL << shift);
} while (addr < end);
res = 1;
static int __init allocate_cachealigned_memnodemap(void)
{
- unsigned long pad, pad_addr;
+ unsigned long addr;
memnodemap = memnode.embedded_map;
if (memnodemapsize <= ARRAY_SIZE(memnode.embedded_map))
return 0;
- pad = L1_CACHE_BYTES - 1;
- pad_addr = 0x8000;
- nodemap_size = pad + sizeof(s16) * memnodemapsize;
- nodemap_addr = find_e820_area(pad_addr, end_pfn<<PAGE_SHIFT,
- nodemap_size);
+ addr = 0x8000;
+ nodemap_size = roundup(sizeof(s16) * memnodemapsize, L1_CACHE_BYTES);
+ nodemap_addr = find_e820_area(addr, max_pfn<<PAGE_SHIFT,
+ nodemap_size, L1_CACHE_BYTES);
if (nodemap_addr == -1UL) {
printk(KERN_ERR
"NUMA: Unable to allocate Memory to Node hash map\n");
nodemap_addr = nodemap_size = 0;
return -1;
}
- pad_addr = (nodemap_addr + pad) & ~pad;
- memnodemap = phys_to_virt(pad_addr);
- reserve_early(nodemap_addr, nodemap_addr + nodemap_size);
+ memnodemap = phys_to_virt(nodemap_addr);
+ reserve_early(nodemap_addr, nodemap_addr + nodemap_size, "MEMNODEMAP");
printk(KERN_DEBUG "NUMA: Allocated memnodemap from %lx - %lx\n",
nodemap_addr, nodemap_addr + nodemap_size);
return i;
}
-int __init compute_hash_shift(struct bootnode *nodes, int numnodes)
+int __init compute_hash_shift(struct bootnode *nodes, int numnodes,
+ int *nodeids)
{
int shift;
printk(KERN_DEBUG "NUMA: Using %d for the hash shift.\n",
shift);
- if (populate_memnodemap(nodes, numnodes, shift) != 1) {
+ if (populate_memnodemap(nodes, numnodes, shift, nodeids) != 1) {
printk(KERN_INFO "Your memory is not aligned you need to "
"rebuild your kernel with a bigger NODEMAPSIZE "
"shift=%d\n", shift);
}
static void * __init early_node_mem(int nodeid, unsigned long start,
- unsigned long end, unsigned long size)
+ unsigned long end, unsigned long size,
+ unsigned long align)
{
- unsigned long mem = find_e820_area(start, end, size);
+ unsigned long mem = find_e820_area(start, end, size, align);
void *ptr;
if (mem != -1L)
return __va(mem);
- ptr = __alloc_bootmem_nopanic(size,
- SMP_CACHE_BYTES, __pa(MAX_DMA_ADDRESS));
+
+ ptr = __alloc_bootmem_nopanic(size, align, __pa(MAX_DMA_ADDRESS));
if (ptr == NULL) {
printk(KERN_ERR "Cannot find %lu bytes in node %d\n",
size, nodeid);
void __init setup_node_bootmem(int nodeid, unsigned long start,
unsigned long end)
{
- unsigned long start_pfn, end_pfn, bootmap_pages, bootmap_size;
+ unsigned long start_pfn, last_pfn, bootmap_pages, bootmap_size;
unsigned long bootmap_start, nodedata_phys;
void *bootmap;
- const int pgdat_size = round_up(sizeof(pg_data_t), PAGE_SIZE);
+ const int pgdat_size = roundup(sizeof(pg_data_t), PAGE_SIZE);
+ int nid;
- start = round_up(start, ZONE_ALIGN);
+ start = roundup(start, ZONE_ALIGN);
printk(KERN_INFO "Bootmem setup node %d %016lx-%016lx\n", nodeid,
start, end);
start_pfn = start >> PAGE_SHIFT;
- end_pfn = end >> PAGE_SHIFT;
+ last_pfn = end >> PAGE_SHIFT;
- node_data[nodeid] = early_node_mem(nodeid, start, end, pgdat_size);
+ node_data[nodeid] = early_node_mem(nodeid, start, end, pgdat_size,
+ SMP_CACHE_BYTES);
if (node_data[nodeid] == NULL)
return;
nodedata_phys = __pa(node_data[nodeid]);
+ printk(KERN_INFO " NODE_DATA [%016lx - %016lx]\n", nodedata_phys,
+ nodedata_phys + pgdat_size - 1);
memset(NODE_DATA(nodeid), 0, sizeof(pg_data_t));
- NODE_DATA(nodeid)->bdata = &plat_node_bdata[nodeid];
+ NODE_DATA(nodeid)->bdata = &bootmem_node_data[nodeid];
NODE_DATA(nodeid)->node_start_pfn = start_pfn;
- NODE_DATA(nodeid)->node_spanned_pages = end_pfn - start_pfn;
+ NODE_DATA(nodeid)->node_spanned_pages = last_pfn - start_pfn;
- /* Find a place for the bootmem map */
- bootmap_pages = bootmem_bootmap_pages(end_pfn - start_pfn);
- bootmap_start = round_up(nodedata_phys + pgdat_size, PAGE_SIZE);
+ /*
+ * Find a place for the bootmem map
+ * nodedata_phys could be on other nodes by alloc_bootmem,
+ * so need to sure bootmap_start not to be small, otherwise
+ * early_node_mem will get that with find_e820_area instead
+ * of alloc_bootmem, that could clash with reserved range
+ */
+ bootmap_pages = bootmem_bootmap_pages(last_pfn - start_pfn);
+ nid = phys_to_nid(nodedata_phys);
+ if (nid == nodeid)
+ bootmap_start = roundup(nodedata_phys + pgdat_size, PAGE_SIZE);
+ else
+ bootmap_start = roundup(start, PAGE_SIZE);
+ /*
+ * SMP_CACHE_BYTES could be enough, but init_bootmem_node like
+ * to use that to align to PAGE_SIZE
+ */
bootmap = early_node_mem(nodeid, bootmap_start, end,
- bootmap_pages<<PAGE_SHIFT);
+ bootmap_pages<<PAGE_SHIFT, PAGE_SIZE);
if (bootmap == NULL) {
if (nodedata_phys < start || nodedata_phys >= end)
- free_bootmem((unsigned long)node_data[nodeid],
- pgdat_size);
+ free_bootmem(nodedata_phys, pgdat_size);
node_data[nodeid] = NULL;
return;
}
bootmap_start = __pa(bootmap);
- Dprintk("bootmap start %lu pages %lu\n", bootmap_start, bootmap_pages);
bootmap_size = init_bootmem_node(NODE_DATA(nodeid),
bootmap_start >> PAGE_SHIFT,
- start_pfn, end_pfn);
+ start_pfn, last_pfn);
+
+ printk(KERN_INFO " bootmap [%016lx - %016lx] pages %lx\n",
+ bootmap_start, bootmap_start + bootmap_size - 1,
+ bootmap_pages);
free_bootmem_with_active_regions(nodeid, end);
- reserve_bootmem_node(NODE_DATA(nodeid), nodedata_phys, pgdat_size);
- reserve_bootmem_node(NODE_DATA(nodeid), bootmap_start,
- bootmap_pages<<PAGE_SHIFT);
+ /*
+ * convert early reserve to bootmem reserve earlier
+ * otherwise early_node_mem could use early reserved mem
+ * on previous node
+ */
+ early_res_to_bootmem(start, end);
+
+ /*
+ * in some case early_node_mem could use alloc_bootmem
+ * to get range on other node, don't reserve that again
+ */
+ if (nid != nodeid)
+ printk(KERN_INFO " NODE_DATA(%d) on node %d\n", nodeid, nid);
+ else
+ reserve_bootmem_node(NODE_DATA(nodeid), nodedata_phys,
+ pgdat_size, BOOTMEM_DEFAULT);
+ nid = phys_to_nid(bootmap_start);
+ if (nid != nodeid)
+ printk(KERN_INFO " bootmap(%d) on node %d\n", nodeid, nid);
+ else
+ reserve_bootmem_node(NODE_DATA(nodeid), bootmap_start,
+ bootmap_pages<<PAGE_SHIFT, BOOTMEM_DEFAULT);
+
#ifdef CONFIG_ACPI_NUMA
srat_reserve_add_area(nodeid);
#endif
int rr, i;
rr = first_node(node_online_map);
- for (i = 0; i < NR_CPUS; i++) {
+ for (i = 0; i < nr_cpu_ids; i++) {
if (early_cpu_to_node(i) != NUMA_NO_NODE)
continue;
numa_set_node(i, rr);
#ifdef CONFIG_NUMA_EMU
/* Numa emulation */
-char *cmdline __initdata;
+static char *cmdline __initdata;
/*
* Setups up nid to range from addr to addr + size. If the end
}
/*
- * Sets up the system RAM area from start_pfn to end_pfn according to the
+ * Sets up the system RAM area from start_pfn to last_pfn according to the
* numa=fake command-line option.
*/
-static int __init numa_emulation(unsigned long start_pfn, unsigned long end_pfn)
+static struct bootnode nodes[MAX_NUMNODES] __initdata;
+
+static int __init numa_emulation(unsigned long start_pfn, unsigned long last_pfn)
{
- struct bootnode nodes[MAX_NUMNODES];
u64 size, addr = start_pfn << PAGE_SHIFT;
- u64 max_addr = end_pfn << PAGE_SHIFT;
+ u64 max_addr = last_pfn << PAGE_SHIFT;
int num_nodes = 0, num = 0, coeff_flag, coeff = -1, i;
memset(&nodes, 0, sizeof(nodes));
}
}
out:
- memnode_shift = compute_hash_shift(nodes, num_nodes);
+ memnode_shift = compute_hash_shift(nodes, num_nodes, NULL);
if (memnode_shift < 0) {
memnode_shift = 0;
printk(KERN_ERR "No NUMA hash function found. NUMA emulation "
}
#endif /* CONFIG_NUMA_EMU */
-void __init numa_initmem_init(unsigned long start_pfn, unsigned long end_pfn)
+void __init initmem_init(unsigned long start_pfn, unsigned long last_pfn)
{
int i;
nodes_clear(node_possible_map);
+ nodes_clear(node_online_map);
#ifdef CONFIG_NUMA_EMU
- if (cmdline && !numa_emulation(start_pfn, end_pfn))
+ if (cmdline && !numa_emulation(start_pfn, last_pfn))
return;
nodes_clear(node_possible_map);
+ nodes_clear(node_online_map);
#endif
#ifdef CONFIG_ACPI_NUMA
if (!numa_off && !acpi_scan_nodes(start_pfn << PAGE_SHIFT,
- end_pfn << PAGE_SHIFT))
+ last_pfn << PAGE_SHIFT))
return;
nodes_clear(node_possible_map);
+ nodes_clear(node_online_map);
#endif
#ifdef CONFIG_K8_NUMA
if (!numa_off && !k8_scan_nodes(start_pfn<<PAGE_SHIFT,
- end_pfn<<PAGE_SHIFT))
+ last_pfn<<PAGE_SHIFT))
return;
nodes_clear(node_possible_map);
+ nodes_clear(node_online_map);
#endif
printk(KERN_INFO "%s\n",
numa_off ? "NUMA turned off" : "No NUMA configuration found");
printk(KERN_INFO "Faking a node at %016lx-%016lx\n",
start_pfn << PAGE_SHIFT,
- end_pfn << PAGE_SHIFT);
+ last_pfn << PAGE_SHIFT);
/* setup dummy node covering all memory */
memnode_shift = 63;
memnodemap = memnode.embedded_map;
memnodemap[0] = 0;
- nodes_clear(node_online_map);
node_set_online(0);
node_set(0, node_possible_map);
- for (i = 0; i < NR_CPUS; i++)
+ for (i = 0; i < nr_cpu_ids; i++)
numa_set_node(i, 0);
- /* cpumask_of_cpu() may not be available during early startup */
- memset(&node_to_cpumask_map[0], 0, sizeof(node_to_cpumask_map[0]));
- cpu_set(0, node_to_cpumask_map[0]);
- e820_register_active_regions(0, start_pfn, end_pfn);
- setup_node_bootmem(0, start_pfn << PAGE_SHIFT, end_pfn << PAGE_SHIFT);
-}
-
-__cpuinit void numa_add_cpu(int cpu)
-{
- set_bit(cpu,
- (unsigned long *)&node_to_cpumask_map[early_cpu_to_node(cpu)]);
-}
-
-void __cpuinit numa_set_node(int cpu, int node)
-{
- int *cpu_to_node_map = x86_cpu_to_node_map_early_ptr;
-
- cpu_pda(cpu)->nodenumber = node;
-
- if(cpu_to_node_map)
- cpu_to_node_map[cpu] = node;
- else if(per_cpu_offset(cpu))
- per_cpu(x86_cpu_to_node_map, cpu) = node;
- else
- Dprintk(KERN_INFO "Setting node for non-present cpu %d\n", cpu);
+ e820_register_active_regions(0, start_pfn, last_pfn);
+ setup_node_bootmem(0, start_pfn << PAGE_SHIFT, last_pfn << PAGE_SHIFT);
}
unsigned long __init numa_free_all_bootmem(void)
memset(max_zone_pfns, 0, sizeof(max_zone_pfns));
max_zone_pfns[ZONE_DMA] = MAX_DMA_PFN;
max_zone_pfns[ZONE_DMA32] = MAX_DMA32_PFN;
- max_zone_pfns[ZONE_NORMAL] = end_pfn;
+ max_zone_pfns[ZONE_NORMAL] = max_pfn;
sparse_memory_present_with_active_regions(MAX_NUMNODES);
sparse_init();
}
early_param("numa", numa_setup);
+#ifdef CONFIG_NUMA
/*
* Setup early cpu_to_node.
*
* is already initialized in a round robin manner at numa_init_array,
* prior to this call, and this initialization is good enough
* for the fake NUMA cases.
+ *
+ * Called before the per_cpu areas are setup.
*/
void __init init_cpu_to_node(void)
{
- int i;
+ int cpu;
+ u16 *cpu_to_apicid = early_per_cpu_ptr(x86_cpu_to_apicid);
- for (i = 0; i < NR_CPUS; i++) {
- u16 apicid = x86_cpu_to_apicid_init[i];
+ BUG_ON(cpu_to_apicid == NULL);
+
+ for_each_possible_cpu(cpu) {
+ int node;
+ u16 apicid = cpu_to_apicid[cpu];
if (apicid == BAD_APICID)
continue;
- if (apicid_to_node[apicid] == NUMA_NO_NODE)
+ node = apicid_to_node[apicid];
+ if (node == NUMA_NO_NODE)
+ continue;
+ if (!node_online(node))
continue;
- numa_set_node(i, apicid_to_node[apicid]);
+ numa_set_node(cpu, node);
}
}
+#endif
+
+
+/*
+ * Allocate node_to_cpumask_map based on number of available nodes
+ * Requires node_possible_map to be valid.
+ *
+ * Note: node_to_cpumask() is not valid until after this is done.
+ * (Use CONFIG_DEBUG_PER_CPU_MAPS to check this.)
+ */
+void __init setup_node_to_cpumask_map(void)
+{
+ unsigned int node, num = 0;
+ cpumask_t *map;
+
+ /* setup nr_node_ids if not done yet */
+ if (nr_node_ids == MAX_NUMNODES) {
+ for_each_node_mask(node, node_possible_map)
+ num = node;
+ nr_node_ids = num + 1;
+ }
+ /* allocate the map */
+ map = alloc_bootmem_low(nr_node_ids * sizeof(cpumask_t));
+ DBG("node_to_cpumask_map at %p for %d nodes\n", map, nr_node_ids);
+
+ pr_debug("Node to cpumask map at %p for %d nodes\n",
+ map, nr_node_ids);
+
+ /* node_to_cpumask() will now work */
+ node_to_cpumask_map = map;
+}
+
+void __cpuinit numa_set_node(int cpu, int node)
+{
+ int *cpu_to_node_map = early_per_cpu_ptr(x86_cpu_to_node_map);
+
+ /* early setting, no percpu area yet */
+ if (cpu_to_node_map) {
+ cpu_to_node_map[cpu] = node;
+ return;
+ }
+
+#ifdef CONFIG_DEBUG_PER_CPU_MAPS
+ if (cpu >= nr_cpu_ids || !cpu_possible(cpu)) {
+ printk(KERN_ERR "numa_set_node: invalid cpu# (%d)\n", cpu);
+ dump_stack();
+ return;
+ }
+#endif
+ per_cpu(x86_cpu_to_node_map, cpu) = node;
+
+ if (node != NUMA_NO_NODE)
+ per_cpu(node_number, cpu) = node;
+}
+
+void __cpuinit numa_clear_node(int cpu)
+{
+ numa_set_node(cpu, NUMA_NO_NODE);
+}
+
+#ifndef CONFIG_DEBUG_PER_CPU_MAPS
+
+void __cpuinit numa_add_cpu(int cpu)
+{
+ cpu_set(cpu, node_to_cpumask_map[early_cpu_to_node(cpu)]);
+}
+
+void __cpuinit numa_remove_cpu(int cpu)
+{
+ cpu_clear(cpu, node_to_cpumask_map[early_cpu_to_node(cpu)]);
+}
+
+#else /* CONFIG_DEBUG_PER_CPU_MAPS */
+
+/*
+ * --------- debug versions of the numa functions ---------
+ */
+static void __cpuinit numa_set_cpumask(int cpu, int enable)
+{
+ int node = early_cpu_to_node(cpu);
+ cpumask_t *mask;
+ char buf[64];
+
+ if (node_to_cpumask_map == NULL) {
+ printk(KERN_ERR "node_to_cpumask_map NULL\n");
+ dump_stack();
+ return;
+ }
+
+ mask = &node_to_cpumask_map[node];
+ if (enable)
+ cpu_set(cpu, *mask);
+ else
+ cpu_clear(cpu, *mask);
+
+ cpulist_scnprintf(buf, sizeof(buf), mask);
+ printk(KERN_DEBUG "%s cpu %d node %d: mask now %s\n",
+ enable ? "numa_add_cpu" : "numa_remove_cpu", cpu, node, buf);
+}
+
+void __cpuinit numa_add_cpu(int cpu)
+{
+ numa_set_cpumask(cpu, 1);
+}
+
+void __cpuinit numa_remove_cpu(int cpu)
+{
+ numa_set_cpumask(cpu, 0);
+}
+
+int cpu_to_node(int cpu)
+{
+ if (early_per_cpu_ptr(x86_cpu_to_node_map)) {
+ printk(KERN_WARNING
+ "cpu_to_node(%d): usage too early!\n", cpu);
+ dump_stack();
+ return early_per_cpu_ptr(x86_cpu_to_node_map)[cpu];
+ }
+ return per_cpu(x86_cpu_to_node_map, cpu);
+}
+EXPORT_SYMBOL(cpu_to_node);
+
+/*
+ * Same function as cpu_to_node() but used if called before the
+ * per_cpu areas are setup.
+ */
+int early_cpu_to_node(int cpu)
+{
+ if (early_per_cpu_ptr(x86_cpu_to_node_map))
+ return early_per_cpu_ptr(x86_cpu_to_node_map)[cpu];
+
+ if (!cpu_possible(cpu)) {
+ printk(KERN_WARNING
+ "early_cpu_to_node(%d): no per_cpu area!\n", cpu);
+ dump_stack();
+ return NUMA_NO_NODE;
+ }
+ return per_cpu(x86_cpu_to_node_map, cpu);
+}
+
+
+/* empty cpumask */
+static const cpumask_t cpu_mask_none;
+
+/*
+ * Returns a pointer to the bitmask of CPUs on Node 'node'.
+ */
+const cpumask_t *cpumask_of_node(int node)
+{
+ if (node_to_cpumask_map == NULL) {
+ printk(KERN_WARNING
+ "cpumask_of_node(%d): no node_to_cpumask_map!\n",
+ node);
+ dump_stack();
+ return (const cpumask_t *)&cpu_online_map;
+ }
+ if (node >= nr_node_ids) {
+ printk(KERN_WARNING
+ "cpumask_of_node(%d): node > nr_node_ids(%d)\n",
+ node, nr_node_ids);
+ dump_stack();
+ return &cpu_mask_none;
+ }
+ return &node_to_cpumask_map[node];
+}
+EXPORT_SYMBOL(cpumask_of_node);
+
+/*
+ * Returns a bitmask of CPUs on Node 'node'.
+ *
+ * Side note: this function creates the returned cpumask on the stack
+ * so with a high NR_CPUS count, excessive stack space is used. The
+ * node_to_cpumask_ptr function should be used whenever possible.
+ */
+cpumask_t node_to_cpumask(int node)
+{
+ if (node_to_cpumask_map == NULL) {
+ printk(KERN_WARNING
+ "node_to_cpumask(%d): no node_to_cpumask_map!\n", node);
+ dump_stack();
+ return cpu_online_map;
+ }
+ if (node >= nr_node_ids) {
+ printk(KERN_WARNING
+ "node_to_cpumask(%d): node > nr_node_ids(%d)\n",
+ node, nr_node_ids);
+ dump_stack();
+ return cpu_mask_none;
+ }
+ return node_to_cpumask_map[node];
+}
+EXPORT_SYMBOL(node_to_cpumask);
+
+/*
+ * --------- end of debug versions of the numa functions ---------
+ */
+#endif /* CONFIG_DEBUG_PER_CPU_MAPS */