* [PATCH 1/3] x86: Change size of node ids from u8 to u16 fixup
2008-01-17 22:35 [PATCH 0/3] x86: Reduce memory and intra-node effects with large count NR_CPUs fixup travis
@ 2008-01-17 22:35 ` travis
2008-01-17 22:35 ` [PATCH 2/3] x86: Change NR_CPUS arrays in numa_64 fixup travis
` (2 subsequent siblings)
3 siblings, 0 replies; 8+ messages in thread
From: travis @ 2008-01-17 22:35 UTC (permalink / raw)
To: Andrew Morton, Andi Kleen, mingo
Cc: Christoph Lameter, linux-mm, linux-kernel, Eric Dumazet
[-- Attachment #1: big_nodeids-fixup --]
[-- Type: text/plain, Size: 2124 bytes --]
Change the size of node ids from 8 bits to 16 bits to
accomodate more than 256 nodes.
Cc: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: Mike Travis <travis@sgi.com>
Reviewed-by: Christoph Lameter <clameter@sgi.com>
---
Fixup:
Size of memnode.embedded_map needs to be changed to
accomodate 16-bit node ids as suggested by Eric.
V2->V3:
- changed memnode.embedded_map from [64-16] to [64-8]
(and size comment to 128 bytes)
V1->V2:
- changed pxm_to_node_map to u16
- changed memnode map entries to u16
---
arch/x86/mm/numa_64.c | 2 +-
drivers/acpi/numa.c | 2 +-
include/asm-x86/mmzone_64.h | 6 +++---
3 files changed, 5 insertions(+), 5 deletions(-)
--- a/arch/x86/mm/numa_64.c
+++ b/arch/x86/mm/numa_64.c
@@ -88,7 +88,7 @@ static int __init allocate_cachealigned_
unsigned long pad, pad_addr;
memnodemap = memnode.embedded_map;
- if (memnodemapsize <= 48)
+ if (memnodemapsize <= ARRAY_SIZE(memnode.embedded_map))
return 0;
pad = L1_CACHE_BYTES - 1;
--- a/drivers/acpi/numa.c
+++ b/drivers/acpi/numa.c
@@ -38,7 +38,7 @@ ACPI_MODULE_NAME("numa");
static nodemask_t nodes_found_map = NODE_MASK_NONE;
/* maps to convert between proximity domain and logical node ID */
-static int pxm_to_node_map[MAX_PXM_DOMAINS]
+static u16 pxm_to_node_map[MAX_PXM_DOMAINS]
= { [0 ... MAX_PXM_DOMAINS - 1] = NID_INVAL };
static int node_to_pxm_map[MAX_NUMNODES]
= { [0 ... MAX_NUMNODES - 1] = PXM_INVAL };
--- a/include/asm-x86/mmzone_64.h
+++ b/include/asm-x86/mmzone_64.h
@@ -15,9 +15,9 @@
struct memnode {
int shift;
unsigned int mapsize;
- u8 *map;
- u8 embedded_map[64-16];
-} ____cacheline_aligned; /* total size = 64 bytes */
+ u16 *map;
+ u16 embedded_map[64-8];
+} ____cacheline_aligned; /* total size = 128 bytes */
extern struct memnode memnode;
#define memnode_shift memnode.shift
#define memnodemap memnode.map
--
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 8+ messages in thread* [PATCH 2/3] x86: Change NR_CPUS arrays in numa_64 fixup
2008-01-17 22:35 [PATCH 0/3] x86: Reduce memory and intra-node effects with large count NR_CPUs fixup travis
2008-01-17 22:35 ` [PATCH 1/3] x86: Change size of node ids from u8 to u16 fixup travis
@ 2008-01-17 22:35 ` travis
2008-01-17 22:35 ` [PATCH 3/3] x86: Change bios_cpu_apicid to percpu data variable fixup travis
2008-01-17 22:42 ` [PATCH 0/3] x86: Reduce memory and intra-node effects with large count NR_CPUs fixup Mike Travis
3 siblings, 0 replies; 8+ messages in thread
From: travis @ 2008-01-17 22:35 UTC (permalink / raw)
To: Andrew Morton, Andi Kleen, mingo
Cc: Christoph Lameter, linux-mm, linux-kernel
[-- Attachment #1: NR_CPUS-arrays-in-numa_64-fixup --]
[-- Type: text/plain, Size: 7800 bytes --]
Change the following static arrays sized by NR_CPUS to
per_cpu data variables:
char cpu_to_node_map[NR_CPUS];
Signed-off-by: Mike Travis <travis@sgi.com>
Reviewed-by: Christoph Lameter <clameter@sgi.com>
---
fixup:
Split cpu_to_node function into "early" and "late" versions
so that x86_cpu_to_node_map_{init,early_ptr} are not EXPORT'ed.
This also involves setting up the percpu maps as early as possible.
V2->V3:
- add early_cpu_to_node function to keep cpu_to_node efficient
- move and rename smp_set_apicids() to setup_percpu_maps()
- call setup_percpu_maps() as early as possible
V1->V2:
- Removed extraneous casts
- Fix !NUMA builds with '#ifdef CONFIG_NUMA"
---
arch/x86/kernel/setup64.c | 37 +++++++++++++++++++++++++++++++++++--
arch/x86/kernel/smpboot_64.c | 38 ++------------------------------------
arch/x86/mm/numa_64.c | 2 --
arch/x86/mm/srat_64.c | 5 +++--
include/asm-x86/numa_64.h | 7 -------
include/asm-x86/topology.h | 10 +++++++++-
mm/page_alloc.c | 2 +-
7 files changed, 50 insertions(+), 51 deletions(-)
--- a/arch/x86/kernel/setup64.c
+++ b/arch/x86/kernel/setup64.c
@@ -84,6 +84,36 @@ static int __init nonx32_setup(char *str
__setup("noexec32=", nonx32_setup);
/*
+ * Copy data used in early init routines from the initial arrays to the
+ * per cpu data areas. These arrays then become expendable and the *_ptrs
+ * are zeroed indicating that the static arrays are gone.
+ */
+void __init setup_percpu_maps(void)
+{
+ int cpu;
+
+ for_each_possible_cpu(cpu) {
+ if (per_cpu_offset(cpu)) {
+ per_cpu(x86_cpu_to_apicid, cpu) =
+ x86_cpu_to_apicid_init[cpu];
+#ifdef CONFIG_NUMA
+ per_cpu(x86_cpu_to_node_map, cpu) =
+ x86_cpu_to_node_map_init[cpu];
+#endif
+ }
+ else
+ printk(KERN_NOTICE "per_cpu_offset zero for cpu %d\n",
+ cpu);
+ }
+
+ /* indicate the early static arrays are gone */
+ x86_cpu_to_apicid_early_ptr = NULL;
+#ifdef CONFIG_NUMA
+ x86_cpu_to_node_map_early_ptr = NULL;
+#endif
+}
+
+/*
* Great future plan:
* Declare PDA itself and support (irqstack,tss,pgd) as per cpu data.
* Always point %gs to its beginning
@@ -104,18 +134,21 @@ void __init setup_per_cpu_areas(void)
for_each_cpu_mask (i, cpu_possible_map) {
char *ptr;
- if (!NODE_DATA(cpu_to_node(i))) {
+ if (!NODE_DATA(early_cpu_to_node(i))) {
printk("cpu with no node %d, num_online_nodes %d\n",
i, num_online_nodes());
ptr = alloc_bootmem_pages(size);
} else {
- ptr = alloc_bootmem_pages_node(NODE_DATA(cpu_to_node(i)), size);
+ ptr = alloc_bootmem_pages_node(NODE_DATA(early_cpu_to_node(i)), size);
}
if (!ptr)
panic("Cannot allocate cpu data for CPU %d\n", i);
cpu_pda(i)->data_offset = ptr - __per_cpu_start;
memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
}
+
+ /* setup percpu data maps early */
+ setup_percpu_maps();
}
void pda_init(int cpu)
--- a/arch/x86/kernel/smpboot_64.c
+++ b/arch/x86/kernel/smpboot_64.c
@@ -702,7 +702,7 @@ do_rest:
if (boot_error) {
cpu_clear(cpu, cpu_callout_map); /* was set here (do_boot_cpu()) */
clear_bit(cpu, (unsigned long *)&cpu_initialized); /* was set by cpu_init() */
- clear_node_cpumask(cpu); /* was set by numa_add_cpu */
+ clear_bit(cpu, (unsigned long *)&node_to_cpumask_map[cpu_to_node(cpu)]);
cpu_clear(cpu, cpu_present_map);
cpu_clear(cpu, cpu_possible_map);
per_cpu(x86_cpu_to_apicid, cpu) = BAD_APICID;
@@ -851,39 +851,6 @@ static int __init smp_sanity_check(unsig
return 0;
}
-/*
- * Copy data used in early init routines from the initial arrays to the
- * per cpu data areas. These arrays then become expendable and the
- * *_ptrs are zeroed indicating that the static arrays are gone.
- */
-void __init smp_set_apicids(void)
-{
- int cpu;
-
- for_each_possible_cpu(cpu) {
- if (per_cpu_offset(cpu)) {
- per_cpu(x86_cpu_to_apicid, cpu) =
- x86_cpu_to_apicid_init[cpu];
-#ifdef CONFIG_NUMA
- per_cpu(x86_cpu_to_node_map, cpu) =
- x86_cpu_to_node_map_init[cpu];
-#endif
- per_cpu(x86_bios_cpu_apicid, cpu) =
- x86_bios_cpu_apicid_init[cpu];
- }
- else
- printk(KERN_NOTICE "per_cpu_offset zero for cpu %d\n",
- cpu);
- }
-
- /* indicate the early static arrays are gone */
- x86_cpu_to_apicid_early_ptr = NULL;
-#ifdef CONFIG_NUMA
- x86_cpu_to_node_map_early_ptr = NULL;
-#endif
- x86_bios_cpu_apicid_early_ptr = NULL;
-}
-
static void __init smp_cpu_index_default(void)
{
int i;
@@ -906,7 +873,6 @@ void __init smp_prepare_cpus(unsigned in
smp_cpu_index_default();
current_cpu_data = boot_cpu_data;
current_thread_info()->cpu = 0; /* needed? */
- smp_set_apicids();
set_cpu_sibling_map(0);
if (smp_sanity_check(max_cpus) < 0) {
@@ -1060,7 +1026,7 @@ void remove_cpu_from_maps(void)
cpu_clear(cpu, cpu_callout_map);
cpu_clear(cpu, cpu_callin_map);
clear_bit(cpu, (unsigned long *)&cpu_initialized); /* was set by cpu_init() */
- clear_node_cpumask(cpu);
+ clear_bit(cpu, (unsigned long *)&node_to_cpumask_map[cpu_to_node(cpu)]);
}
int __cpu_disable(void)
--- a/arch/x86/mm/numa_64.c
+++ b/arch/x86/mm/numa_64.c
@@ -35,8 +35,6 @@ u16 x86_cpu_to_node_map_init[NR_CPUS] =
[0 ... NR_CPUS-1] = NUMA_NO_NODE
};
void *x86_cpu_to_node_map_early_ptr;
-EXPORT_SYMBOL(x86_cpu_to_node_map_init);
-EXPORT_SYMBOL(x86_cpu_to_node_map_early_ptr);
DEFINE_PER_CPU(u16, x86_cpu_to_node_map) = NUMA_NO_NODE;
EXPORT_PER_CPU_SYMBOL(x86_cpu_to_node_map);
--- a/arch/x86/mm/srat_64.c
+++ b/arch/x86/mm/srat_64.c
@@ -382,9 +382,10 @@ int __init acpi_scan_nodes(unsigned long
setup_node_bootmem(i, nodes[i].start, nodes[i].end);
for (i = 0; i < NR_CPUS; i++) {
- if (cpu_to_node(i) == NUMA_NO_NODE)
+ int node = cpu_to_node(i);
+ if (node == NUMA_NO_NODE)
continue;
- if (!node_isset(cpu_to_node(i), node_possible_map))
+ if (!node_isset(node, node_possible_map))
numa_set_node(i, NUMA_NO_NODE);
}
numa_init_array();
--- a/include/asm-x86/numa_64.h
+++ b/include/asm-x86/numa_64.h
@@ -29,15 +29,8 @@ extern void setup_node_bootmem(int nodei
#ifdef CONFIG_NUMA
extern void __init init_cpu_to_node(void);
-
-static inline void clear_node_cpumask(int cpu)
-{
- clear_bit(cpu, (unsigned long *)&node_to_cpumask_map[cpu_to_node(cpu)]);
-}
-
#else
#define init_cpu_to_node() do {} while (0)
-#define clear_node_cpumask(cpu) do {} while (0)
#endif
#endif
--- a/include/asm-x86/topology.h
+++ b/include/asm-x86/topology.h
@@ -38,7 +38,7 @@ extern cpumask_t node_to_cpumask_map[];
#define NUMA_NO_NODE ((u16)(~0))
/* Returns the number of the node containing CPU 'cpu' */
-static inline int cpu_to_node(int cpu)
+static inline int early_cpu_to_node(int cpu)
{
u16 *cpu_to_node_map = x86_cpu_to_node_map_early_ptr;
@@ -50,6 +50,14 @@ static inline int cpu_to_node(int cpu)
return NUMA_NO_NODE;
}
+static inline int cpu_to_node(int cpu)
+{
+ if(per_cpu_offset(cpu))
+ return per_cpu(x86_cpu_to_node_map, cpu);
+ else
+ return NUMA_NO_NODE;
+}
+
/*
* Returns the number of the node containing Node 'node'. This
* architecture is flat, so it is a pretty simple function!
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -1783,7 +1783,7 @@ EXPORT_SYMBOL(free_pages);
static unsigned int nr_free_zone_pages(int offset)
{
/* Just pick one node, since fallback list is circular */
- pg_data_t *pgdat = NODE_DATA(numa_node_id());
+ pg_data_t *pgdat = NODE_DATA(cpu_to_node(raw_smp_processor_id()));
unsigned int sum = 0;
struct zonelist *zonelist = pgdat->node_zonelists + offset;
--
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 8+ messages in thread* [PATCH 3/3] x86: Change bios_cpu_apicid to percpu data variable fixup
2008-01-17 22:35 [PATCH 0/3] x86: Reduce memory and intra-node effects with large count NR_CPUs fixup travis
2008-01-17 22:35 ` [PATCH 1/3] x86: Change size of node ids from u8 to u16 fixup travis
2008-01-17 22:35 ` [PATCH 2/3] x86: Change NR_CPUS arrays in numa_64 fixup travis
@ 2008-01-17 22:35 ` travis
2008-01-17 22:42 ` [PATCH 0/3] x86: Reduce memory and intra-node effects with large count NR_CPUs fixup Mike Travis
3 siblings, 0 replies; 8+ messages in thread
From: travis @ 2008-01-17 22:35 UTC (permalink / raw)
To: Andrew Morton, Andi Kleen, mingo
Cc: Christoph Lameter, linux-mm, linux-kernel
[-- Attachment #1: change-bios_cpu_apicid-to-percpu-fixup --]
[-- Type: text/plain, Size: 2790 bytes --]
Change static bios_cpu_apicid array to a per_cpu data variable.
This includes using a static array used during initialization
similar to the way x86_cpu_to_apicid[] is handled.
There is one early use of bios_cpu_apicid in apic_is_clustered_box().
The other reference in cpu_present_to_apicid() is called after
smp_set_apicids() has setup the percpu version of bios_cpu_apicid.
Signed-off-by: Mike Travis <travis@sgi.com>
Reviewed-by: Christoph Lameter <clameter@sgi.com>
---
V1->V2:
- Removed extraneous casts
- Add slight optimization to apic_is_clustered_box()
[don't reference x86_bios_cpu_apicid_early_ptr each pass.]
---
arch/x86/kernel/apic_64.c | 6 +++---
arch/x86/kernel/setup64.c | 3 +++
arch/x86/kernel/setup_64.c | 1 +
3 files changed, 7 insertions(+), 3 deletions(-)
--- a/arch/x86/kernel/apic_64.c
+++ b/arch/x86/kernel/apic_64.c
@@ -1191,9 +1191,9 @@ __cpuinit int apic_is_clustered_box(void
/* Problem: Partially populated chassis may not have CPUs in some of
* the APIC clusters they have been allocated. Only present CPUs have
- * x86_bios_cpu_apicid entries, thus causing zeroes in the bitmap. Since
- * clusters are allocated sequentially, count zeros only if they are
- * bounded by ones.
+ * x86_bios_cpu_apicid entries, thus causing zeroes in the bitmap.
+ * Since clusters are allocated sequentially, count zeros only if
+ * they are bounded by ones.
*/
clusters = 0;
zeros = 0;
--- a/arch/x86/kernel/setup64.c
+++ b/arch/x86/kernel/setup64.c
@@ -96,6 +96,8 @@ void __init setup_percpu_maps(void)
if (per_cpu_offset(cpu)) {
per_cpu(x86_cpu_to_apicid, cpu) =
x86_cpu_to_apicid_init[cpu];
+ per_cpu(x86_bios_cpu_apicid, cpu) =
+ x86_bios_cpu_apicid_init[cpu];
#ifdef CONFIG_NUMA
per_cpu(x86_cpu_to_node_map, cpu) =
x86_cpu_to_node_map_init[cpu];
@@ -108,6 +110,7 @@ void __init setup_percpu_maps(void)
/* indicate the early static arrays are gone */
x86_cpu_to_apicid_early_ptr = NULL;
+ x86_bios_cpu_apicid_early_ptr = NULL;
#ifdef CONFIG_NUMA
x86_cpu_to_node_map_early_ptr = NULL;
#endif
--- a/arch/x86/kernel/setup_64.c
+++ b/arch/x86/kernel/setup_64.c
@@ -390,6 +390,7 @@ void __init setup_arch(char **cmdline_p)
#ifdef CONFIG_SMP
/* setup to use the early static init tables during kernel startup */
x86_cpu_to_apicid_early_ptr = (void *)&x86_cpu_to_apicid_init;
+ x86_bios_cpu_apicid_early_ptr = (void *)&x86_bios_cpu_apicid_init;
#ifdef CONFIG_NUMA
x86_cpu_to_node_map_early_ptr = (void *)&x86_cpu_to_node_map_init;
#endif
--
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 8+ messages in thread