forked from Minki/linux
9b130ad5bb
First, number of CPUs can't be negative number. Second, different signnnedness leads to suboptimal code in the following cases: 1) kmalloc(nr_cpu_ids * sizeof(X)); "int" has to be sign extended to size_t. 2) while (loff_t *pos < nr_cpu_ids) MOVSXD is 1 byte longed than the same MOV. Other cases exist as well. Basically compiler is told that nr_cpu_ids can't be negative which can't be deduced if it is "int". Code savings on allyesconfig kernel: -3KB add/remove: 0/0 grow/shrink: 25/264 up/down: 261/-3631 (-3370) function old new delta coretemp_cpu_online 450 512 +62 rcu_init_one 1234 1272 +38 pci_device_probe 374 399 +25 ... pgdat_reclaimable_pages 628 556 -72 select_fallback_rq 446 369 -77 task_numa_find_cpu 1923 1807 -116 Link: http://lkml.kernel.org/r/20170819114959.GA30580@avx2 Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
310 lines
8.7 KiB
C
310 lines
8.7 KiB
C
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
|
|
|
|
#include <linux/kernel.h>
|
|
#include <linux/export.h>
|
|
#include <linux/init.h>
|
|
#include <linux/bootmem.h>
|
|
#include <linux/percpu.h>
|
|
#include <linux/kexec.h>
|
|
#include <linux/crash_dump.h>
|
|
#include <linux/smp.h>
|
|
#include <linux/topology.h>
|
|
#include <linux/pfn.h>
|
|
#include <asm/sections.h>
|
|
#include <asm/processor.h>
|
|
#include <asm/desc.h>
|
|
#include <asm/setup.h>
|
|
#include <asm/mpspec.h>
|
|
#include <asm/apicdef.h>
|
|
#include <asm/highmem.h>
|
|
#include <asm/proto.h>
|
|
#include <asm/cpumask.h>
|
|
#include <asm/cpu.h>
|
|
#include <asm/stackprotector.h>
|
|
|
|
DEFINE_PER_CPU_READ_MOSTLY(int, cpu_number);
|
|
EXPORT_PER_CPU_SYMBOL(cpu_number);
|
|
|
|
#ifdef CONFIG_X86_64
|
|
#define BOOT_PERCPU_OFFSET ((unsigned long)__per_cpu_load)
|
|
#else
|
|
#define BOOT_PERCPU_OFFSET 0
|
|
#endif
|
|
|
|
DEFINE_PER_CPU_READ_MOSTLY(unsigned long, this_cpu_off) = BOOT_PERCPU_OFFSET;
|
|
EXPORT_PER_CPU_SYMBOL(this_cpu_off);
|
|
|
|
unsigned long __per_cpu_offset[NR_CPUS] __ro_after_init = {
|
|
[0 ... NR_CPUS-1] = BOOT_PERCPU_OFFSET,
|
|
};
|
|
EXPORT_SYMBOL(__per_cpu_offset);
|
|
|
|
/*
|
|
* On x86_64 symbols referenced from code should be reachable using
|
|
* 32bit relocations. Reserve space for static percpu variables in
|
|
* modules so that they are always served from the first chunk which
|
|
* is located at the percpu segment base. On x86_32, anything can
|
|
* address anywhere. No need to reserve space in the first chunk.
|
|
*/
|
|
#ifdef CONFIG_X86_64
|
|
#define PERCPU_FIRST_CHUNK_RESERVE PERCPU_MODULE_RESERVE
|
|
#else
|
|
#define PERCPU_FIRST_CHUNK_RESERVE 0
|
|
#endif
|
|
|
|
#ifdef CONFIG_X86_32
|
|
/**
|
|
* pcpu_need_numa - determine percpu allocation needs to consider NUMA
|
|
*
|
|
* If NUMA is not configured or there is only one NUMA node available,
|
|
* there is no reason to consider NUMA. This function determines
|
|
* whether percpu allocation should consider NUMA or not.
|
|
*
|
|
* RETURNS:
|
|
* true if NUMA should be considered; otherwise, false.
|
|
*/
|
|
static bool __init pcpu_need_numa(void)
|
|
{
|
|
#ifdef CONFIG_NEED_MULTIPLE_NODES
|
|
pg_data_t *last = NULL;
|
|
unsigned int cpu;
|
|
|
|
for_each_possible_cpu(cpu) {
|
|
int node = early_cpu_to_node(cpu);
|
|
|
|
if (node_online(node) && NODE_DATA(node) &&
|
|
last && last != NODE_DATA(node))
|
|
return true;
|
|
|
|
last = NODE_DATA(node);
|
|
}
|
|
#endif
|
|
return false;
|
|
}
|
|
#endif
|
|
|
|
/**
|
|
* pcpu_alloc_bootmem - NUMA friendly alloc_bootmem wrapper for percpu
|
|
* @cpu: cpu to allocate for
|
|
* @size: size allocation in bytes
|
|
* @align: alignment
|
|
*
|
|
* Allocate @size bytes aligned at @align for cpu @cpu. This wrapper
|
|
* does the right thing for NUMA regardless of the current
|
|
* configuration.
|
|
*
|
|
* RETURNS:
|
|
* Pointer to the allocated area on success, NULL on failure.
|
|
*/
|
|
static void * __init pcpu_alloc_bootmem(unsigned int cpu, unsigned long size,
|
|
unsigned long align)
|
|
{
|
|
const unsigned long goal = __pa(MAX_DMA_ADDRESS);
|
|
#ifdef CONFIG_NEED_MULTIPLE_NODES
|
|
int node = early_cpu_to_node(cpu);
|
|
void *ptr;
|
|
|
|
if (!node_online(node) || !NODE_DATA(node)) {
|
|
ptr = __alloc_bootmem_nopanic(size, align, goal);
|
|
pr_info("cpu %d has no node %d or node-local memory\n",
|
|
cpu, node);
|
|
pr_debug("per cpu data for cpu%d %lu bytes at %016lx\n",
|
|
cpu, size, __pa(ptr));
|
|
} else {
|
|
ptr = __alloc_bootmem_node_nopanic(NODE_DATA(node),
|
|
size, align, goal);
|
|
pr_debug("per cpu data for cpu%d %lu bytes on node%d at %016lx\n",
|
|
cpu, size, node, __pa(ptr));
|
|
}
|
|
return ptr;
|
|
#else
|
|
return __alloc_bootmem_nopanic(size, align, goal);
|
|
#endif
|
|
}
|
|
|
|
/*
|
|
* Helpers for first chunk memory allocation
|
|
*/
|
|
static void * __init pcpu_fc_alloc(unsigned int cpu, size_t size, size_t align)
|
|
{
|
|
return pcpu_alloc_bootmem(cpu, size, align);
|
|
}
|
|
|
|
static void __init pcpu_fc_free(void *ptr, size_t size)
|
|
{
|
|
free_bootmem(__pa(ptr), size);
|
|
}
|
|
|
|
static int __init pcpu_cpu_distance(unsigned int from, unsigned int to)
|
|
{
|
|
#ifdef CONFIG_NEED_MULTIPLE_NODES
|
|
if (early_cpu_to_node(from) == early_cpu_to_node(to))
|
|
return LOCAL_DISTANCE;
|
|
else
|
|
return REMOTE_DISTANCE;
|
|
#else
|
|
return LOCAL_DISTANCE;
|
|
#endif
|
|
}
|
|
|
|
static void __init pcpup_populate_pte(unsigned long addr)
|
|
{
|
|
populate_extra_pte(addr);
|
|
}
|
|
|
|
static inline void setup_percpu_segment(int cpu)
|
|
{
|
|
#ifdef CONFIG_X86_32
|
|
struct desc_struct d = GDT_ENTRY_INIT(0x8092, per_cpu_offset(cpu),
|
|
0xFFFFF);
|
|
|
|
write_gdt_entry(get_cpu_gdt_rw(cpu), GDT_ENTRY_PERCPU, &d, DESCTYPE_S);
|
|
#endif
|
|
}
|
|
|
|
void __init setup_per_cpu_areas(void)
|
|
{
|
|
unsigned int cpu;
|
|
unsigned long delta;
|
|
int rc;
|
|
|
|
pr_info("NR_CPUS:%d nr_cpumask_bits:%d nr_cpu_ids:%u nr_node_ids:%d\n",
|
|
NR_CPUS, nr_cpumask_bits, nr_cpu_ids, nr_node_ids);
|
|
|
|
/*
|
|
* Allocate percpu area. Embedding allocator is our favorite;
|
|
* however, on NUMA configurations, it can result in very
|
|
* sparse unit mapping and vmalloc area isn't spacious enough
|
|
* on 32bit. Use page in that case.
|
|
*/
|
|
#ifdef CONFIG_X86_32
|
|
if (pcpu_chosen_fc == PCPU_FC_AUTO && pcpu_need_numa())
|
|
pcpu_chosen_fc = PCPU_FC_PAGE;
|
|
#endif
|
|
rc = -EINVAL;
|
|
if (pcpu_chosen_fc != PCPU_FC_PAGE) {
|
|
const size_t dyn_size = PERCPU_MODULE_RESERVE +
|
|
PERCPU_DYNAMIC_RESERVE - PERCPU_FIRST_CHUNK_RESERVE;
|
|
size_t atom_size;
|
|
|
|
/*
|
|
* On 64bit, use PMD_SIZE for atom_size so that embedded
|
|
* percpu areas are aligned to PMD. This, in the future,
|
|
* can also allow using PMD mappings in vmalloc area. Use
|
|
* PAGE_SIZE on 32bit as vmalloc space is highly contended
|
|
* and large vmalloc area allocs can easily fail.
|
|
*/
|
|
#ifdef CONFIG_X86_64
|
|
atom_size = PMD_SIZE;
|
|
#else
|
|
atom_size = PAGE_SIZE;
|
|
#endif
|
|
rc = pcpu_embed_first_chunk(PERCPU_FIRST_CHUNK_RESERVE,
|
|
dyn_size, atom_size,
|
|
pcpu_cpu_distance,
|
|
pcpu_fc_alloc, pcpu_fc_free);
|
|
if (rc < 0)
|
|
pr_warning("%s allocator failed (%d), falling back to page size\n",
|
|
pcpu_fc_names[pcpu_chosen_fc], rc);
|
|
}
|
|
if (rc < 0)
|
|
rc = pcpu_page_first_chunk(PERCPU_FIRST_CHUNK_RESERVE,
|
|
pcpu_fc_alloc, pcpu_fc_free,
|
|
pcpup_populate_pte);
|
|
if (rc < 0)
|
|
panic("cannot initialize percpu area (err=%d)", rc);
|
|
|
|
/* alrighty, percpu areas up and running */
|
|
delta = (unsigned long)pcpu_base_addr - (unsigned long)__per_cpu_start;
|
|
for_each_possible_cpu(cpu) {
|
|
per_cpu_offset(cpu) = delta + pcpu_unit_offsets[cpu];
|
|
per_cpu(this_cpu_off, cpu) = per_cpu_offset(cpu);
|
|
per_cpu(cpu_number, cpu) = cpu;
|
|
setup_percpu_segment(cpu);
|
|
setup_stack_canary_segment(cpu);
|
|
/*
|
|
* Copy data used in early init routines from the
|
|
* initial arrays to the per cpu data areas. These
|
|
* arrays then become expendable and the *_early_ptr's
|
|
* are zeroed indicating that the static arrays are
|
|
* gone.
|
|
*/
|
|
#ifdef CONFIG_X86_LOCAL_APIC
|
|
per_cpu(x86_cpu_to_apicid, cpu) =
|
|
early_per_cpu_map(x86_cpu_to_apicid, cpu);
|
|
per_cpu(x86_bios_cpu_apicid, cpu) =
|
|
early_per_cpu_map(x86_bios_cpu_apicid, cpu);
|
|
per_cpu(x86_cpu_to_acpiid, cpu) =
|
|
early_per_cpu_map(x86_cpu_to_acpiid, cpu);
|
|
#endif
|
|
#ifdef CONFIG_X86_32
|
|
per_cpu(x86_cpu_to_logical_apicid, cpu) =
|
|
early_per_cpu_map(x86_cpu_to_logical_apicid, cpu);
|
|
#endif
|
|
#ifdef CONFIG_X86_64
|
|
per_cpu(irq_stack_ptr, cpu) =
|
|
per_cpu(irq_stack_union.irq_stack, cpu) +
|
|
IRQ_STACK_SIZE;
|
|
#endif
|
|
#ifdef CONFIG_NUMA
|
|
per_cpu(x86_cpu_to_node_map, cpu) =
|
|
early_per_cpu_map(x86_cpu_to_node_map, cpu);
|
|
/*
|
|
* Ensure that the boot cpu numa_node is correct when the boot
|
|
* cpu is on a node that doesn't have memory installed.
|
|
* Also cpu_up() will call cpu_to_node() for APs when
|
|
* MEMORY_HOTPLUG is defined, before per_cpu(numa_node) is set
|
|
* up later with c_init aka intel_init/amd_init.
|
|
* So set them all (boot cpu and all APs).
|
|
*/
|
|
set_cpu_numa_node(cpu, early_cpu_to_node(cpu));
|
|
#endif
|
|
/*
|
|
* Up to this point, the boot CPU has been using .init.data
|
|
* area. Reload any changed state for the boot CPU.
|
|
*/
|
|
if (!cpu)
|
|
switch_to_new_gdt(cpu);
|
|
}
|
|
|
|
/* indicate the early static arrays will soon be gone */
|
|
#ifdef CONFIG_X86_LOCAL_APIC
|
|
early_per_cpu_ptr(x86_cpu_to_apicid) = NULL;
|
|
early_per_cpu_ptr(x86_bios_cpu_apicid) = NULL;
|
|
early_per_cpu_ptr(x86_cpu_to_acpiid) = NULL;
|
|
#endif
|
|
#ifdef CONFIG_X86_32
|
|
early_per_cpu_ptr(x86_cpu_to_logical_apicid) = NULL;
|
|
#endif
|
|
#ifdef CONFIG_NUMA
|
|
early_per_cpu_ptr(x86_cpu_to_node_map) = NULL;
|
|
#endif
|
|
|
|
/* Setup node to cpumask map */
|
|
setup_node_to_cpumask_map();
|
|
|
|
/* Setup cpu initialized, callin, callout masks */
|
|
setup_cpu_local_masks();
|
|
|
|
#ifdef CONFIG_X86_32
|
|
/*
|
|
* Sync back kernel address range again. We already did this in
|
|
* setup_arch(), but percpu data also needs to be available in
|
|
* the smpboot asm. We can't reliably pick up percpu mappings
|
|
* using vmalloc_fault(), because exception dispatch needs
|
|
* percpu data.
|
|
*/
|
|
clone_pgd_range(initial_page_table + KERNEL_PGD_BOUNDARY,
|
|
swapper_pg_dir + KERNEL_PGD_BOUNDARY,
|
|
KERNEL_PGD_PTRS);
|
|
|
|
/*
|
|
* sync back low identity map too. It is used for example
|
|
* in the 32-bit EFI stub.
|
|
*/
|
|
clone_pgd_range(initial_page_table,
|
|
swapper_pg_dir + KERNEL_PGD_BOUNDARY,
|
|
min(KERNEL_PGD_PTRS, KERNEL_PGD_BOUNDARY));
|
|
#endif
|
|
}
|