Commit e3cfe529 authored by Thomas Gleixner's avatar Thomas Gleixner Committed by Ingo Molnar

x86: cleanup numa_64.c

Clean it up before applying more patches.
Signed-off-by: default avatarThomas Gleixner <tglx@linutronix.de>
Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
parent e1d91978
...@@ -24,6 +24,8 @@ ...@@ -24,6 +24,8 @@
#endif #endif
struct pglist_data *node_data[MAX_NUMNODES] __read_mostly; struct pglist_data *node_data[MAX_NUMNODES] __read_mostly;
EXPORT_SYMBOL(node_data);
bootmem_data_t plat_node_bdata[MAX_NUMNODES]; bootmem_data_t plat_node_bdata[MAX_NUMNODES];
struct memnode memnode; struct memnode memnode;
...@@ -31,16 +33,19 @@ struct memnode memnode; ...@@ -31,16 +33,19 @@ struct memnode memnode;
unsigned char cpu_to_node[NR_CPUS] __read_mostly = { unsigned char cpu_to_node[NR_CPUS] __read_mostly = {
[0 ... NR_CPUS-1] = NUMA_NO_NODE [0 ... NR_CPUS-1] = NUMA_NO_NODE
}; };
EXPORT_SYMBOL(cpu_to_node);
unsigned char apicid_to_node[MAX_LOCAL_APIC] __cpuinitdata = { unsigned char apicid_to_node[MAX_LOCAL_APIC] __cpuinitdata = {
[0 ... MAX_LOCAL_APIC-1] = NUMA_NO_NODE [0 ... MAX_LOCAL_APIC-1] = NUMA_NO_NODE
}; };
cpumask_t node_to_cpumask[MAX_NUMNODES] __read_mostly; cpumask_t node_to_cpumask[MAX_NUMNODES] __read_mostly;
EXPORT_SYMBOL(node_to_cpumask);
int numa_off __initdata; int numa_off __initdata;
unsigned long __initdata nodemap_addr; unsigned long __initdata nodemap_addr;
unsigned long __initdata nodemap_size; unsigned long __initdata nodemap_size;
/* /*
* Given a shift value, try to populate memnodemap[] * Given a shift value, try to populate memnodemap[]
* Returns : * Returns :
...@@ -48,12 +53,11 @@ unsigned long __initdata nodemap_size; ...@@ -48,12 +53,11 @@ unsigned long __initdata nodemap_size;
* 0 if memnodmap[] too small (of shift too small) * 0 if memnodmap[] too small (of shift too small)
* -1 if node overlap or lost ram (shift too big) * -1 if node overlap or lost ram (shift too big)
*/ */
static int __init static int __init populate_memnodemap(const struct bootnode *nodes,
populate_memnodemap(const struct bootnode *nodes, int numnodes, int shift) int numnodes, int shift)
{ {
int i;
int res = -1;
unsigned long addr, end; unsigned long addr, end;
int i, res = -1;
memset(memnodemap, 0xff, memnodemapsize); memset(memnodemap, 0xff, memnodemapsize);
for (i = 0; i < numnodes; i++) { for (i = 0; i < numnodes; i++) {
...@@ -105,8 +109,8 @@ static int __init allocate_cachealigned_memnodemap(void) ...@@ -105,8 +109,8 @@ static int __init allocate_cachealigned_memnodemap(void)
* The LSB of all start and end addresses in the node map is the value of the * The LSB of all start and end addresses in the node map is the value of the
* maximum possible shift. * maximum possible shift.
*/ */
static int __init static int __init extract_lsb_from_nodes(const struct bootnode *nodes,
extract_lsb_from_nodes (const struct bootnode *nodes, int numnodes) int numnodes)
{ {
int i, nodes_used = 0; int i, nodes_used = 0;
unsigned long start, end; unsigned long start, end;
...@@ -141,10 +145,9 @@ int __init compute_hash_shift(struct bootnode *nodes, int numnodes) ...@@ -141,10 +145,9 @@ int __init compute_hash_shift(struct bootnode *nodes, int numnodes)
shift); shift);
if (populate_memnodemap(nodes, numnodes, shift) != 1) { if (populate_memnodemap(nodes, numnodes, shift) != 1) {
printk(KERN_INFO printk(KERN_INFO "Your memory is not aligned you need to "
"Your memory is not aligned you need to rebuild your kernel " "rebuild your kernel with a bigger NODEMAPSIZE "
"with a bigger NODEMAPSIZE shift=%d\n", "shift=%d\n", shift);
shift);
return -1; return -1;
} }
return shift; return shift;
...@@ -157,12 +160,12 @@ int early_pfn_to_nid(unsigned long pfn) ...@@ -157,12 +160,12 @@ int early_pfn_to_nid(unsigned long pfn)
} }
#endif #endif
static void * __init static void * __init early_node_mem(int nodeid, unsigned long start,
early_node_mem(int nodeid, unsigned long start, unsigned long end, unsigned long end, unsigned long size)
unsigned long size)
{ {
unsigned long mem = find_e820_area(start, end, size); unsigned long mem = find_e820_area(start, end, size);
void *ptr; void *ptr;
if (mem != -1L) if (mem != -1L)
return __va(mem); return __va(mem);
ptr = __alloc_bootmem_nopanic(size, ptr = __alloc_bootmem_nopanic(size,
...@@ -176,16 +179,18 @@ early_node_mem(int nodeid, unsigned long start, unsigned long end, ...@@ -176,16 +179,18 @@ early_node_mem(int nodeid, unsigned long start, unsigned long end,
} }
/* Initialize bootmem allocator for a node */ /* Initialize bootmem allocator for a node */
void __init setup_node_bootmem(int nodeid, unsigned long start, unsigned long end) void __init setup_node_bootmem(int nodeid, unsigned long start,
unsigned long end)
{ {
unsigned long start_pfn, end_pfn, bootmap_pages, bootmap_size, bootmap_start; unsigned long start_pfn, end_pfn, bootmap_pages, bootmap_size;
unsigned long nodedata_phys; unsigned long bootmap_start, nodedata_phys;
void *bootmap; void *bootmap;
const int pgdat_size = round_up(sizeof(pg_data_t), PAGE_SIZE); const int pgdat_size = round_up(sizeof(pg_data_t), PAGE_SIZE);
start = round_up(start, ZONE_ALIGN); start = round_up(start, ZONE_ALIGN);
printk(KERN_INFO "Bootmem setup node %d %016lx-%016lx\n", nodeid, start, end); printk(KERN_INFO "Bootmem setup node %d %016lx-%016lx\n", nodeid,
start, end);
start_pfn = start >> PAGE_SHIFT; start_pfn = start >> PAGE_SHIFT;
end_pfn = end >> PAGE_SHIFT; end_pfn = end >> PAGE_SHIFT;
...@@ -207,7 +212,8 @@ void __init setup_node_bootmem(int nodeid, unsigned long start, unsigned long en ...@@ -207,7 +212,8 @@ void __init setup_node_bootmem(int nodeid, unsigned long start, unsigned long en
bootmap_pages<<PAGE_SHIFT); bootmap_pages<<PAGE_SHIFT);
if (bootmap == NULL) { if (bootmap == NULL) {
if (nodedata_phys < start || nodedata_phys >= end) if (nodedata_phys < start || nodedata_phys >= end)
free_bootmem((unsigned long)node_data[nodeid],pgdat_size); free_bootmem((unsigned long)node_data[nodeid],
pgdat_size);
node_data[nodeid] = NULL; node_data[nodeid] = NULL;
return; return;
} }
...@@ -221,7 +227,8 @@ void __init setup_node_bootmem(int nodeid, unsigned long start, unsigned long en ...@@ -221,7 +227,8 @@ void __init setup_node_bootmem(int nodeid, unsigned long start, unsigned long en
free_bootmem_with_active_regions(nodeid, end); free_bootmem_with_active_regions(nodeid, end);
reserve_bootmem_node(NODE_DATA(nodeid), nodedata_phys, pgdat_size); reserve_bootmem_node(NODE_DATA(nodeid), nodedata_phys, pgdat_size);
reserve_bootmem_node(NODE_DATA(nodeid), bootmap_start, bootmap_pages<<PAGE_SHIFT); reserve_bootmem_node(NODE_DATA(nodeid), bootmap_start,
bootmap_pages<<PAGE_SHIFT);
#ifdef CONFIG_ACPI_NUMA #ifdef CONFIG_ACPI_NUMA
srat_reserve_add_area(nodeid); srat_reserve_add_area(nodeid);
#endif #endif
...@@ -239,8 +246,10 @@ void __init setup_node_zones(int nodeid) ...@@ -239,8 +246,10 @@ void __init setup_node_zones(int nodeid)
Dprintk(KERN_INFO "Setting up memmap for node %d %lx-%lx\n", Dprintk(KERN_INFO "Setting up memmap for node %d %lx-%lx\n",
nodeid, start_pfn, end_pfn); nodeid, start_pfn, end_pfn);
/* Try to allocate mem_map at end to not fill up precious <4GB /*
memory. */ * Try to allocate mem_map at end to not fill up precious <4GB
* memory.
*/
memmapsize = sizeof(struct page) * (end_pfn-start_pfn); memmapsize = sizeof(struct page) * (end_pfn-start_pfn);
limit = end_pfn << PAGE_SHIFT; limit = end_pfn << PAGE_SHIFT;
#ifdef CONFIG_FLAT_NODE_MEM_MAP #ifdef CONFIG_FLAT_NODE_MEM_MAP
...@@ -252,14 +261,17 @@ void __init setup_node_zones(int nodeid) ...@@ -252,14 +261,17 @@ void __init setup_node_zones(int nodeid)
#endif #endif
} }
/*
* There are unfortunately some poorly designed mainboards around that
* only connect memory to a single CPU. This breaks the 1:1 cpu->node
* mapping. To avoid this fill in the mapping for all possible CPUs,
* as the number of CPUs is not known yet. We round robin the existing
* nodes.
*/
void __init numa_init_array(void) void __init numa_init_array(void)
{ {
int rr, i; int rr, i;
/* There are unfortunately some poorly designed mainboards around
that only connect memory to a single CPU. This breaks the 1:1 cpu->node
mapping. To avoid this fill in the mapping for all possible
CPUs, as the number of CPUs is not known yet.
We round robin the existing nodes. */
rr = first_node(node_online_map); rr = first_node(node_online_map);
for (i = 0; i < NR_CPUS; i++) { for (i = 0; i < NR_CPUS; i++) {
if (cpu_to_node(i) != NUMA_NO_NODE) if (cpu_to_node(i) != NUMA_NO_NODE)
...@@ -269,7 +281,6 @@ void __init numa_init_array(void) ...@@ -269,7 +281,6 @@ void __init numa_init_array(void)
if (rr == MAX_NUMNODES) if (rr == MAX_NUMNODES)
rr = first_node(node_online_map); rr = first_node(node_online_map);
} }
} }
#ifdef CONFIG_NUMA_EMU #ifdef CONFIG_NUMA_EMU
...@@ -277,15 +288,17 @@ void __init numa_init_array(void) ...@@ -277,15 +288,17 @@ void __init numa_init_array(void)
char *cmdline __initdata; char *cmdline __initdata;
/* /*
* Setups up nid to range from addr to addr + size. If the end boundary is * Setups up nid to range from addr to addr + size. If the end
* greater than max_addr, then max_addr is used instead. The return value is 0 * boundary is greater than max_addr, then max_addr is used instead.
* if there is additional memory left for allocation past addr and -1 otherwise. * The return value is 0 if there is additional memory left for
* addr is adjusted to be at the end of the node. * allocation past addr and -1 otherwise. addr is adjusted to be at
* the end of the node.
*/ */
static int __init setup_node_range(int nid, struct bootnode *nodes, u64 *addr, static int __init setup_node_range(int nid, struct bootnode *nodes, u64 *addr,
u64 size, u64 max_addr) u64 size, u64 max_addr)
{ {
int ret = 0; int ret = 0;
nodes[nid].start = *addr; nodes[nid].start = *addr;
*addr += size; *addr += size;
if (*addr >= max_addr) { if (*addr >= max_addr) {
...@@ -336,6 +349,7 @@ static int __init split_nodes_equally(struct bootnode *nodes, u64 *addr, ...@@ -336,6 +349,7 @@ static int __init split_nodes_equally(struct bootnode *nodes, u64 *addr,
for (i = node_start; i < num_nodes + node_start; i++) { for (i = node_start; i < num_nodes + node_start; i++) {
u64 end = *addr + size; u64 end = *addr + size;
if (i < big) if (i < big)
end += FAKE_NODE_MIN_SIZE; end += FAKE_NODE_MIN_SIZE;
/* /*
...@@ -381,14 +395,9 @@ static int __init split_nodes_by_size(struct bootnode *nodes, u64 *addr, ...@@ -381,14 +395,9 @@ static int __init split_nodes_by_size(struct bootnode *nodes, u64 *addr,
static int __init numa_emulation(unsigned long start_pfn, unsigned long end_pfn) static int __init numa_emulation(unsigned long start_pfn, unsigned long end_pfn)
{ {
struct bootnode nodes[MAX_NUMNODES]; struct bootnode nodes[MAX_NUMNODES];
u64 addr = start_pfn << PAGE_SHIFT; u64 size, addr = start_pfn << PAGE_SHIFT;
u64 max_addr = end_pfn << PAGE_SHIFT; u64 max_addr = end_pfn << PAGE_SHIFT;
int num_nodes = 0; int num_nodes = 0, num = 0, coeff_flag, coeff = -1, i;
int coeff_flag;
int coeff = -1;
int num = 0;
u64 size;
int i;
memset(&nodes, 0, sizeof(nodes)); memset(&nodes, 0, sizeof(nodes));
/* /*
...@@ -396,8 +405,9 @@ static int __init numa_emulation(unsigned long start_pfn, unsigned long end_pfn) ...@@ -396,8 +405,9 @@ static int __init numa_emulation(unsigned long start_pfn, unsigned long end_pfn)
* system RAM into N fake nodes. * system RAM into N fake nodes.
*/ */
if (!strchr(cmdline, '*') && !strchr(cmdline, ',')) { if (!strchr(cmdline, '*') && !strchr(cmdline, ',')) {
num_nodes = split_nodes_equally(nodes, &addr, max_addr, 0, long n = simple_strtol(cmdline, NULL, 0);
simple_strtol(cmdline, NULL, 0));
num_nodes = split_nodes_equally(nodes, &addr, max_addr, 0, n);
if (num_nodes < 0) if (num_nodes < 0)
return num_nodes; return num_nodes;
goto out; goto out;
...@@ -512,7 +522,8 @@ void __init numa_initmem_init(unsigned long start_pfn, unsigned long end_pfn) ...@@ -512,7 +522,8 @@ void __init numa_initmem_init(unsigned long start_pfn, unsigned long end_pfn)
#endif #endif
#ifdef CONFIG_K8_NUMA #ifdef CONFIG_K8_NUMA
if (!numa_off && !k8_scan_nodes(start_pfn<<PAGE_SHIFT, end_pfn<<PAGE_SHIFT)) if (!numa_off && !k8_scan_nodes(start_pfn<<PAGE_SHIFT,
end_pfn<<PAGE_SHIFT))
return; return;
nodes_clear(node_possible_map); nodes_clear(node_possible_map);
#endif #endif
...@@ -549,18 +560,20 @@ void __cpuinit numa_set_node(int cpu, int node) ...@@ -549,18 +560,20 @@ void __cpuinit numa_set_node(int cpu, int node)
unsigned long __init numa_free_all_bootmem(void) unsigned long __init numa_free_all_bootmem(void)
{ {
int i;
unsigned long pages = 0; unsigned long pages = 0;
for_each_online_node(i) { int i;
for_each_online_node(i)
pages += free_all_bootmem_node(NODE_DATA(i)); pages += free_all_bootmem_node(NODE_DATA(i));
}
return pages; return pages;
} }
void __init paging_init(void) void __init paging_init(void)
{ {
int i;
unsigned long max_zone_pfns[MAX_NR_ZONES]; unsigned long max_zone_pfns[MAX_NR_ZONES];
int i;
memset(max_zone_pfns, 0, sizeof(max_zone_pfns)); memset(max_zone_pfns, 0, sizeof(max_zone_pfns));
max_zone_pfns[ZONE_DMA] = MAX_DMA_PFN; max_zone_pfns[ZONE_DMA] = MAX_DMA_PFN;
max_zone_pfns[ZONE_DMA32] = MAX_DMA32_PFN; max_zone_pfns[ZONE_DMA32] = MAX_DMA32_PFN;
...@@ -569,9 +582,8 @@ void __init paging_init(void) ...@@ -569,9 +582,8 @@ void __init paging_init(void)
sparse_memory_present_with_active_regions(MAX_NUMNODES); sparse_memory_present_with_active_regions(MAX_NUMNODES);
sparse_init(); sparse_init();
for_each_online_node(i) { for_each_online_node(i)
setup_node_zones(i); setup_node_zones(i);
}
free_area_init_nodes(max_zone_pfns); free_area_init_nodes(max_zone_pfns);
} }
...@@ -580,21 +592,20 @@ static __init int numa_setup(char *opt) ...@@ -580,21 +592,20 @@ static __init int numa_setup(char *opt)
{ {
if (!opt) if (!opt)
return -EINVAL; return -EINVAL;
if (!strncmp(opt,"off",3)) if (!strncmp(opt, "off", 3))
numa_off = 1; numa_off = 1;
#ifdef CONFIG_NUMA_EMU #ifdef CONFIG_NUMA_EMU
if (!strncmp(opt, "fake=", 5)) if (!strncmp(opt, "fake=", 5))
cmdline = opt + 5; cmdline = opt + 5;
#endif #endif
#ifdef CONFIG_ACPI_NUMA #ifdef CONFIG_ACPI_NUMA
if (!strncmp(opt,"noacpi",6)) if (!strncmp(opt, "noacpi", 6))
acpi_numa = -1; acpi_numa = -1;
if (!strncmp(opt,"hotadd=", 7)) if (!strncmp(opt, "hotadd=", 7))
hotadd_percent = simple_strtoul(opt+7, NULL, 10); hotadd_percent = simple_strtoul(opt+7, NULL, 10);
#endif #endif
return 0; return 0;
} }
early_param("numa", numa_setup); early_param("numa", numa_setup);
/* /*
...@@ -612,20 +623,18 @@ early_param("numa", numa_setup); ...@@ -612,20 +623,18 @@ early_param("numa", numa_setup);
void __init init_cpu_to_node(void) void __init init_cpu_to_node(void)
{ {
int i; int i;
for (i = 0; i < NR_CPUS; i++) { for (i = 0; i < NR_CPUS; i++) {
u8 apicid = x86_cpu_to_apicid_init[i]; u8 apicid = x86_cpu_to_apicid_init[i];
if (apicid == BAD_APICID) if (apicid == BAD_APICID)
continue; continue;
if (apicid_to_node[apicid] == NUMA_NO_NODE) if (apicid_to_node[apicid] == NUMA_NO_NODE)
continue; continue;
numa_set_node(i,apicid_to_node[apicid]); numa_set_node(i, apicid_to_node[apicid]);
} }
} }
EXPORT_SYMBOL(cpu_to_node);
EXPORT_SYMBOL(node_to_cpumask);
EXPORT_SYMBOL(node_data);
#ifdef CONFIG_DISCONTIGMEM #ifdef CONFIG_DISCONTIGMEM
/* /*
* Functions to convert PFNs from/to per node page addresses. * Functions to convert PFNs from/to per node page addresses.
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment