mirror of
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git
synced 2024-09-12 13:55:32 +00:00
![Ralf Baechle](/assets/img/avatar_default.png)
A recent change requires cpu_possible_map to be initialized before smp_sched_init() but most MIPS platforms were initializing their processors in the prom_prepare_cpus callback of smp_prepare_cpus. The simple fix of calling prom_prepare_cpus from one of the earlier SMP initialization hooks doesn't work well either since IPIs may require init_IRQ() to have completed, so bit the bullet and split prom_prepare_cpus into two initialization functions, plat_smp_setup which is called early from setup_arch and plat_prepare_cpus called where prom_prepare_cpus used to be called. Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
450 lines
11 KiB
C
450 lines
11 KiB
C
/*
|
|
* This program is free software; you can redistribute it and/or
|
|
* modify it under the terms of the GNU General Public License
|
|
* as published by the Free Software Foundation; either version 2
|
|
* of the License, or (at your option) any later version.
|
|
*
|
|
* This program is distributed in the hope that it will be useful,
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
* GNU General Public License for more details.
|
|
*
|
|
* You should have received a copy of the GNU General Public License
|
|
* along with this program; if not, write to the Free Software
|
|
* Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
|
|
*
|
|
* Copyright (C) 2000, 2001 Kanoj Sarcar
|
|
* Copyright (C) 2000, 2001 Ralf Baechle
|
|
* Copyright (C) 2000, 2001 Silicon Graphics, Inc.
|
|
* Copyright (C) 2000, 2001, 2003 Broadcom Corporation
|
|
*/
|
|
#include <linux/cache.h>
|
|
#include <linux/delay.h>
|
|
#include <linux/init.h>
|
|
#include <linux/interrupt.h>
|
|
#include <linux/spinlock.h>
|
|
#include <linux/threads.h>
|
|
#include <linux/module.h>
|
|
#include <linux/time.h>
|
|
#include <linux/timex.h>
|
|
#include <linux/sched.h>
|
|
#include <linux/cpumask.h>
|
|
#include <linux/cpu.h>
|
|
|
|
#include <asm/atomic.h>
|
|
#include <asm/cpu.h>
|
|
#include <asm/processor.h>
|
|
#include <asm/system.h>
|
|
#include <asm/mmu_context.h>
|
|
#include <asm/smp.h>
|
|
|
|
cpumask_t phys_cpu_present_map; /* Bitmask of available CPUs */
|
|
volatile cpumask_t cpu_callin_map; /* Bitmask of started secondaries */
|
|
cpumask_t cpu_online_map; /* Bitmask of currently online CPUs */
|
|
int __cpu_number_map[NR_CPUS]; /* Map physical to logical */
|
|
int __cpu_logical_map[NR_CPUS]; /* Map logical to physical */
|
|
|
|
EXPORT_SYMBOL(phys_cpu_present_map);
|
|
EXPORT_SYMBOL(cpu_online_map);
|
|
|
|
static void smp_tune_scheduling (void)
|
|
{
|
|
struct cache_desc *cd = ¤t_cpu_data.scache;
|
|
unsigned long cachesize; /* kB */
|
|
unsigned long cpu_khz;
|
|
|
|
/*
|
|
* Crude estimate until we actually meassure ...
|
|
*/
|
|
cpu_khz = loops_per_jiffy * 2 * HZ / 1000;
|
|
|
|
/*
|
|
* Rough estimation for SMP scheduling, this is the number of
|
|
* cycles it takes for a fully memory-limited process to flush
|
|
* the SMP-local cache.
|
|
*
|
|
* (For a P5 this pretty much means we will choose another idle
|
|
* CPU almost always at wakeup time (this is due to the small
|
|
* L1 cache), on PIIs it's around 50-100 usecs, depending on
|
|
* the cache size)
|
|
*/
|
|
if (!cpu_khz)
|
|
return;
|
|
|
|
cachesize = cd->linesz * cd->sets * cd->ways;
|
|
}
|
|
|
|
extern void __init calibrate_delay(void);
|
|
extern ATTRIB_NORET void cpu_idle(void);
|
|
|
|
/*
|
|
* First C code run on the secondary CPUs after being started up by
|
|
* the master.
|
|
*/
|
|
asmlinkage void start_secondary(void)
|
|
{
|
|
unsigned int cpu;
|
|
|
|
cpu_probe();
|
|
cpu_report();
|
|
per_cpu_trap_init();
|
|
prom_init_secondary();
|
|
|
|
/*
|
|
* XXX parity protection should be folded in here when it's converted
|
|
* to an option instead of something based on .cputype
|
|
*/
|
|
|
|
calibrate_delay();
|
|
preempt_disable();
|
|
cpu = smp_processor_id();
|
|
cpu_data[cpu].udelay_val = loops_per_jiffy;
|
|
|
|
prom_smp_finish();
|
|
|
|
cpu_set(cpu, cpu_callin_map);
|
|
|
|
cpu_idle();
|
|
}
|
|
|
|
DEFINE_SPINLOCK(smp_call_lock);
|
|
|
|
struct call_data_struct *call_data;
|
|
|
|
/*
|
|
* Run a function on all other CPUs.
|
|
* <func> The function to run. This must be fast and non-blocking.
|
|
* <info> An arbitrary pointer to pass to the function.
|
|
* <retry> If true, keep retrying until ready.
|
|
* <wait> If true, wait until function has completed on other CPUs.
|
|
* [RETURNS] 0 on success, else a negative status code.
|
|
*
|
|
* Does not return until remote CPUs are nearly ready to execute <func>
|
|
* or are or have executed.
|
|
*
|
|
* You must not call this function with disabled interrupts or from a
|
|
* hardware interrupt handler or from a bottom half handler:
|
|
*
|
|
* CPU A CPU B
|
|
* Disable interrupts
|
|
* smp_call_function()
|
|
* Take call_lock
|
|
* Send IPIs
|
|
* Wait for all cpus to acknowledge IPI
|
|
* CPU A has not responded, spin waiting
|
|
* for cpu A to respond, holding call_lock
|
|
* smp_call_function()
|
|
* Spin waiting for call_lock
|
|
* Deadlock Deadlock
|
|
*/
|
|
int smp_call_function (void (*func) (void *info), void *info, int retry,
|
|
int wait)
|
|
{
|
|
struct call_data_struct data;
|
|
int i, cpus = num_online_cpus() - 1;
|
|
int cpu = smp_processor_id();
|
|
|
|
/*
|
|
* Can die spectacularly if this CPU isn't yet marked online
|
|
*/
|
|
BUG_ON(!cpu_online(cpu));
|
|
|
|
if (!cpus)
|
|
return 0;
|
|
|
|
/* Can deadlock when called with interrupts disabled */
|
|
WARN_ON(irqs_disabled());
|
|
|
|
data.func = func;
|
|
data.info = info;
|
|
atomic_set(&data.started, 0);
|
|
data.wait = wait;
|
|
if (wait)
|
|
atomic_set(&data.finished, 0);
|
|
|
|
spin_lock(&smp_call_lock);
|
|
call_data = &data;
|
|
mb();
|
|
|
|
/* Send a message to all other CPUs and wait for them to respond */
|
|
for (i = 0; i < NR_CPUS; i++)
|
|
if (cpu_online(i) && i != cpu)
|
|
core_send_ipi(i, SMP_CALL_FUNCTION);
|
|
|
|
/* Wait for response */
|
|
/* FIXME: lock-up detection, backtrace on lock-up */
|
|
while (atomic_read(&data.started) != cpus)
|
|
barrier();
|
|
|
|
if (wait)
|
|
while (atomic_read(&data.finished) != cpus)
|
|
barrier();
|
|
spin_unlock(&smp_call_lock);
|
|
|
|
return 0;
|
|
}
|
|
|
|
void smp_call_function_interrupt(void)
|
|
{
|
|
void (*func) (void *info) = call_data->func;
|
|
void *info = call_data->info;
|
|
int wait = call_data->wait;
|
|
|
|
/*
|
|
* Notify initiating CPU that I've grabbed the data and am
|
|
* about to execute the function.
|
|
*/
|
|
mb();
|
|
atomic_inc(&call_data->started);
|
|
|
|
/*
|
|
* At this point the info structure may be out of scope unless wait==1.
|
|
*/
|
|
irq_enter();
|
|
(*func)(info);
|
|
irq_exit();
|
|
|
|
if (wait) {
|
|
mb();
|
|
atomic_inc(&call_data->finished);
|
|
}
|
|
}
|
|
|
|
static void stop_this_cpu(void *dummy)
|
|
{
|
|
/*
|
|
* Remove this CPU:
|
|
*/
|
|
cpu_clear(smp_processor_id(), cpu_online_map);
|
|
local_irq_enable(); /* May need to service _machine_restart IPI */
|
|
for (;;); /* Wait if available. */
|
|
}
|
|
|
|
void smp_send_stop(void)
|
|
{
|
|
smp_call_function(stop_this_cpu, NULL, 1, 0);
|
|
}
|
|
|
|
void __init smp_cpus_done(unsigned int max_cpus)
|
|
{
|
|
prom_cpus_done();
|
|
}
|
|
|
|
/* called from main before smp_init() */
|
|
void __init smp_prepare_cpus(unsigned int max_cpus)
|
|
{
|
|
init_new_context(current, &init_mm);
|
|
current_thread_info()->cpu = 0;
|
|
smp_tune_scheduling();
|
|
plat_prepare_cpus(max_cpus);
|
|
}
|
|
|
|
/* preload SMP state for boot cpu */
|
|
void __devinit smp_prepare_boot_cpu(void)
|
|
{
|
|
/*
|
|
* This assumes that bootup is always handled by the processor
|
|
* with the logic and physical number 0.
|
|
*/
|
|
__cpu_number_map[0] = 0;
|
|
__cpu_logical_map[0] = 0;
|
|
cpu_set(0, phys_cpu_present_map);
|
|
cpu_set(0, cpu_online_map);
|
|
cpu_set(0, cpu_callin_map);
|
|
}
|
|
|
|
/*
|
|
* Called once for each "cpu_possible(cpu)". Needs to spin up the cpu
|
|
* and keep control until "cpu_online(cpu)" is set. Note: cpu is
|
|
* physical, not logical.
|
|
*/
|
|
int __devinit __cpu_up(unsigned int cpu)
|
|
{
|
|
struct task_struct *idle;
|
|
|
|
/*
|
|
* Processor goes to start_secondary(), sets online flag
|
|
* The following code is purely to make sure
|
|
* Linux can schedule processes on this slave.
|
|
*/
|
|
idle = fork_idle(cpu);
|
|
if (IS_ERR(idle))
|
|
panic(KERN_ERR "Fork failed for CPU %d", cpu);
|
|
|
|
prom_boot_secondary(cpu, idle);
|
|
|
|
/*
|
|
* Trust is futile. We should really have timeouts ...
|
|
*/
|
|
while (!cpu_isset(cpu, cpu_callin_map))
|
|
udelay(100);
|
|
|
|
cpu_set(cpu, cpu_online_map);
|
|
|
|
return 0;
|
|
}
|
|
|
|
/* Not really SMP stuff ... */
|
|
int setup_profiling_timer(unsigned int multiplier)
|
|
{
|
|
return 0;
|
|
}
|
|
|
|
static void flush_tlb_all_ipi(void *info)
|
|
{
|
|
local_flush_tlb_all();
|
|
}
|
|
|
|
void flush_tlb_all(void)
|
|
{
|
|
on_each_cpu(flush_tlb_all_ipi, 0, 1, 1);
|
|
}
|
|
|
|
static void flush_tlb_mm_ipi(void *mm)
|
|
{
|
|
local_flush_tlb_mm((struct mm_struct *)mm);
|
|
}
|
|
|
|
/*
|
|
* The following tlb flush calls are invoked when old translations are
|
|
* being torn down, or pte attributes are changing. For single threaded
|
|
* address spaces, a new context is obtained on the current cpu, and tlb
|
|
* context on other cpus are invalidated to force a new context allocation
|
|
* at switch_mm time, should the mm ever be used on other cpus. For
|
|
* multithreaded address spaces, intercpu interrupts have to be sent.
|
|
* Another case where intercpu interrupts are required is when the target
|
|
* mm might be active on another cpu (eg debuggers doing the flushes on
|
|
* behalf of debugees, kswapd stealing pages from another process etc).
|
|
* Kanoj 07/00.
|
|
*/
|
|
|
|
void flush_tlb_mm(struct mm_struct *mm)
|
|
{
|
|
preempt_disable();
|
|
|
|
if ((atomic_read(&mm->mm_users) != 1) || (current->mm != mm)) {
|
|
smp_call_function(flush_tlb_mm_ipi, (void *)mm, 1, 1);
|
|
} else {
|
|
int i;
|
|
for (i = 0; i < num_online_cpus(); i++)
|
|
if (smp_processor_id() != i)
|
|
cpu_context(i, mm) = 0;
|
|
}
|
|
local_flush_tlb_mm(mm);
|
|
|
|
preempt_enable();
|
|
}
|
|
|
|
struct flush_tlb_data {
|
|
struct vm_area_struct *vma;
|
|
unsigned long addr1;
|
|
unsigned long addr2;
|
|
};
|
|
|
|
static void flush_tlb_range_ipi(void *info)
|
|
{
|
|
struct flush_tlb_data *fd = (struct flush_tlb_data *)info;
|
|
|
|
local_flush_tlb_range(fd->vma, fd->addr1, fd->addr2);
|
|
}
|
|
|
|
void flush_tlb_range(struct vm_area_struct *vma, unsigned long start, unsigned long end)
|
|
{
|
|
struct mm_struct *mm = vma->vm_mm;
|
|
|
|
preempt_disable();
|
|
if ((atomic_read(&mm->mm_users) != 1) || (current->mm != mm)) {
|
|
struct flush_tlb_data fd;
|
|
|
|
fd.vma = vma;
|
|
fd.addr1 = start;
|
|
fd.addr2 = end;
|
|
smp_call_function(flush_tlb_range_ipi, (void *)&fd, 1, 1);
|
|
} else {
|
|
int i;
|
|
for (i = 0; i < num_online_cpus(); i++)
|
|
if (smp_processor_id() != i)
|
|
cpu_context(i, mm) = 0;
|
|
}
|
|
local_flush_tlb_range(vma, start, end);
|
|
preempt_enable();
|
|
}
|
|
|
|
static void flush_tlb_kernel_range_ipi(void *info)
|
|
{
|
|
struct flush_tlb_data *fd = (struct flush_tlb_data *)info;
|
|
|
|
local_flush_tlb_kernel_range(fd->addr1, fd->addr2);
|
|
}
|
|
|
|
void flush_tlb_kernel_range(unsigned long start, unsigned long end)
|
|
{
|
|
struct flush_tlb_data fd;
|
|
|
|
fd.addr1 = start;
|
|
fd.addr2 = end;
|
|
on_each_cpu(flush_tlb_kernel_range_ipi, (void *)&fd, 1, 1);
|
|
}
|
|
|
|
static void flush_tlb_page_ipi(void *info)
|
|
{
|
|
struct flush_tlb_data *fd = (struct flush_tlb_data *)info;
|
|
|
|
local_flush_tlb_page(fd->vma, fd->addr1);
|
|
}
|
|
|
|
void flush_tlb_page(struct vm_area_struct *vma, unsigned long page)
|
|
{
|
|
preempt_disable();
|
|
if ((atomic_read(&vma->vm_mm->mm_users) != 1) || (current->mm != vma->vm_mm)) {
|
|
struct flush_tlb_data fd;
|
|
|
|
fd.vma = vma;
|
|
fd.addr1 = page;
|
|
smp_call_function(flush_tlb_page_ipi, (void *)&fd, 1, 1);
|
|
} else {
|
|
int i;
|
|
for (i = 0; i < num_online_cpus(); i++)
|
|
if (smp_processor_id() != i)
|
|
cpu_context(i, vma->vm_mm) = 0;
|
|
}
|
|
local_flush_tlb_page(vma, page);
|
|
preempt_enable();
|
|
}
|
|
|
|
static void flush_tlb_one_ipi(void *info)
|
|
{
|
|
unsigned long vaddr = (unsigned long) info;
|
|
|
|
local_flush_tlb_one(vaddr);
|
|
}
|
|
|
|
void flush_tlb_one(unsigned long vaddr)
|
|
{
|
|
smp_call_function(flush_tlb_one_ipi, (void *) vaddr, 1, 1);
|
|
local_flush_tlb_one(vaddr);
|
|
}
|
|
|
|
static DEFINE_PER_CPU(struct cpu, cpu_devices);
|
|
|
|
static int __init topology_init(void)
|
|
{
|
|
int cpu;
|
|
int ret;
|
|
|
|
for_each_cpu(cpu) {
|
|
ret = register_cpu(&per_cpu(cpu_devices, cpu), cpu, NULL);
|
|
if (ret)
|
|
printk(KERN_WARNING "topology_init: register_cpu %d "
|
|
"failed (%d)\n", cpu, ret);
|
|
}
|
|
|
|
return 0;
|
|
}
|
|
|
|
subsys_initcall(topology_init);
|
|
|
|
EXPORT_SYMBOL(flush_tlb_page);
|
|
EXPORT_SYMBOL(flush_tlb_one);
|
|
EXPORT_SYMBOL(cpu_data);
|
|
EXPORT_SYMBOL(synchronize_irq);
|