smp.c

一个2.4.21版本的嵌入式linux内核

/* smp.c: Sparc64 SMP support.
 *
 * Copyright (C) 1997 David S. Miller ([e-mail])
 */

#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/mm.h>
#include <linux/pagemap.h>
#include <linux/threads.h>
#include <linux/smp.h>
#include <linux/smp_lock.h>
#include <linux/interrupt.h>
#include <linux/kernel_stat.h>
#include <linux/delay.h>
#include <linux/init.h>
#include <linux/spinlock.h>
#include <linux/fs.h>
#include <linux/seq_file.h>
#include <linux/cache.h>
#include <linux/timer.h>

#include <asm/head.h>
#include <asm/ptrace.h>
#include <asm/atomic.h>

#include <asm/irq.h>
#include <asm/page.h>
#include <asm/pgtable.h>
#include <asm/oplib.h>
#include <asm/hardirq.h>
#include <asm/softirq.h>
#include <asm/uaccess.h>
#include <asm/timer.h>
#include <asm/starfire.h>

#define __KERNEL_SYSCALLS__
#include <linux/unistd.h>

extern int linux_num_cpus;
extern void calibrate_delay(void);
extern unsigned prom_cpu_nodes[];

cpuinfo_sparc cpu_data[NR_CPUS];

volatile int __cpu_number_map[NR_CPUS]  __attribute__ ((aligned (SMP_CACHE_BYTES)));
volatile int __cpu_logical_map[NR_CPUS] __attribute__ ((aligned (SMP_CACHE_BYTES)));

/* Please don't make this stuff initdata!!!  --DaveM */
static unsigned char boot_cpu_id;
static int smp_activated;

/* Kernel spinlock */
spinlock_t kernel_flag __cacheline_aligned_in_smp = SPIN_LOCK_UNLOCKED;

volatile int smp_processors_ready = 0;
unsigned long cpu_present_map = 0;
int smp_num_cpus = 1;
int smp_threads_ready = 0;

void __init smp_setup(char *str, int *ints)
{
	/* XXX implement me XXX */
}

static int max_cpus = NR_CPUS;
static int __init maxcpus(char *str)
{
	get_option(&str, &max_cpus);
	return 1;
}

__setup("maxcpus=", maxcpus);

void smp_info(struct seq_file *m)
{
	int i;
	
	seq_printf(m, "State:\n");
	for (i = 0; i < NR_CPUS; i++) {
		if (cpu_present_map & (1UL << i))
			seq_printf(m,
				   "CPU%d:\t\tonline\n", i);
	}
}

void smp_bogo(struct seq_file *m)
{
	int i;
	
	for (i = 0; i < NR_CPUS; i++)
		if (cpu_present_map & (1UL << i))
			seq_printf(m,
				   "Cpu%dBogo\t: %lu.%02lu\n"
				   "Cpu%dClkTck\t: %016lx\n",
				   i, cpu_data[i].udelay_val / (500000/HZ),
				   (cpu_data[i].udelay_val / (5000/HZ)) % 100,
				   i, cpu_data[i].clock_tick);
}

void __init smp_store_cpu_info(int id)
{
	int i, no;

	/* multiplier and counter set by
	   smp_setup_percpu_timer()  */
	cpu_data[id].udelay_val			= loops_per_jiffy;

	for (no = 0; no < linux_num_cpus; no++)
		if (linux_cpus[no].mid == id)
			break;

	cpu_data[id].clock_tick = prom_getintdefault(linux_cpus[no].prom_node,
						     "clock-frequency", 0);

	cpu_data[id].pgcache_size		= 0;
	cpu_data[id].pte_cache[0]		= NULL;
	cpu_data[id].pte_cache[1]		= NULL;
	cpu_data[id].pgdcache_size		= 0;
	cpu_data[id].pgd_cache			= NULL;
	cpu_data[id].idle_volume		= 1;

	for (i = 0; i < 16; i++)
		cpu_data[id].irq_worklists[i] = 0;
}

void __init smp_commence(void)
{
}

static void smp_setup_percpu_timer(void);

static volatile unsigned long callin_flag = 0;

extern void inherit_locked_prom_mappings(int save_p);
extern void cpu_probe(void);

void __init smp_callin(void)
{
	int cpuid = hard_smp_processor_id();
	extern int bigkernel;
	extern unsigned long kern_locked_tte_data;

	if (bigkernel) {
		prom_dtlb_load(sparc64_highest_locked_tlbent()-1, 
			kern_locked_tte_data + 0x400000, KERNBASE + 0x400000);
		prom_itlb_load(sparc64_highest_locked_tlbent()-1, 
			kern_locked_tte_data + 0x400000, KERNBASE + 0x400000);
	}

	inherit_locked_prom_mappings(0);

	__flush_cache_all();
	__flush_tlb_all();

	cpu_probe();

	smp_setup_percpu_timer();

	__sti();

	calibrate_delay();
	smp_store_cpu_info(cpuid);
	callin_flag = 1;
	__asm__ __volatile__("membar #Sync\n\t"
			     "flush  %%g6" : : : "memory");

	/* Clear this or we will die instantly when we
	 * schedule back to this idler...
	 */
	current->thread.flags &= ~(SPARC_FLAG_NEWCHILD);

	/* Attach to the address space of init_task. */
	atomic_inc(&init_mm.mm_count);
	current->active_mm = &init_mm;

	while (!smp_threads_ready)
		membar("#LoadLoad");
}

extern int cpu_idle(void);
extern void init_IRQ(void);

int start_secondary(void *unused)
{
	trap_init();
	init_IRQ();
	return cpu_idle();
}

void cpu_panic(void)
{
	printk("CPU[%d]: Returns from cpu_idle!\n", smp_processor_id());
	panic("SMP bolixed\n");
}

static unsigned long current_tick_offset;

/* This tick register synchronization scheme is taken entirely from
 * the ia64 port, see arch/ia64/kernel/smpboot.c for details and credit.
 *
 * The only change I've made is to rework it so that the master
 * initiates the synchonization instead of the slave. -DaveM
 */

#define MASTER	0
#define SLAVE	(SMP_CACHE_BYTES/sizeof(unsigned long))

#define NUM_ROUNDS	64	/* magic value */
#define NUM_ITERS	5	/* likewise */

static spinlock_t itc_sync_lock = SPIN_LOCK_UNLOCKED;
static unsigned long go[SLAVE + 1];

#define DEBUG_TICK_SYNC	0

static inline long get_delta (long *rt, long *master)
{
	unsigned long best_t0 = 0, best_t1 = ~0UL, best_tm = 0;
	unsigned long tcenter, t0, t1, tm;
	unsigned long i;

	for (i = 0; i < NUM_ITERS; i++) {
		t0 = tick_ops->get_tick();
		go[MASTER] = 1;
		membar("#StoreLoad");
		while (!(tm = go[SLAVE]))
			membar("#LoadLoad");
		go[SLAVE] = 0;
		membar("#StoreStore");
		t1 = tick_ops->get_tick();

		if (t1 - t0 < best_t1 - best_t0)
			best_t0 = t0, best_t1 = t1, best_tm = tm;
	}

	*rt = best_t1 - best_t0;
	*master = best_tm - best_t0;

	/* average best_t0 and best_t1 without overflow: */
	tcenter = (best_t0/2 + best_t1/2);
	if (best_t0 % 2 + best_t1 % 2 == 2)
		tcenter++;
	return tcenter - best_tm;
}

void smp_synchronize_tick_client(void)
{
	long i, delta, adj, adjust_latency = 0, done = 0;
	unsigned long flags, rt, master_time_stamp, bound;
#if DEBUG_TICK_SYNC
	struct {
		long rt;	/* roundtrip time */
		long master;	/* master's timestamp */
		long diff;	/* difference between midpoint and master's timestamp */
		long lat;	/* estimate of itc adjustment latency */
	} t[NUM_ROUNDS];
#endif

	go[MASTER] = 1;

	while (go[MASTER])
		membar("#LoadLoad");

	local_irq_save(flags);
	{
		for (i = 0; i < NUM_ROUNDS; i++) {
			delta = get_delta(&rt, &master_time_stamp);
			if (delta == 0) {
				done = 1;	/* let's lock on to this... */
				bound = rt;
			}

			if (!done) {
				if (i > 0) {
					adjust_latency += -delta;
					adj = -delta + adjust_latency/4;
				} else
					adj = -delta;

				tick_ops->add_tick(adj, current_tick_offset);
			}
#if DEBUG_TICK_SYNC
			t[i].rt = rt;
			t[i].master = master_time_stamp;
			t[i].diff = delta;
			t[i].lat = adjust_latency/4;
#endif
		}
	}
	local_irq_restore(flags);

#if DEBUG_TICK_SYNC
	for (i = 0; i < NUM_ROUNDS; i++)
		printk("rt=%5ld master=%5ld diff=%5ld adjlat=%5ld\n",
		       t[i].rt, t[i].master, t[i].diff, t[i].lat);
#endif

	printk(KERN_INFO "CPU %d: synchronized TICK with master CPU (last diff %ld cycles,"
	       "maxerr %lu cycles)\n", smp_processor_id(), delta, rt);
}

static void smp_start_sync_tick_client(int cpu);

static void smp_synchronize_one_tick(int cpu)
{
	unsigned long flags, i;

	go[MASTER] = 0;

	smp_start_sync_tick_client(cpu);

	/* wait for client to be ready */
	while (!go[MASTER])
		membar("#LoadLoad");

	/* now let the client proceed into his loop */
	go[MASTER] = 0;
	membar("#StoreLoad");

	spin_lock_irqsave(&itc_sync_lock, flags);
	{
		for (i = 0; i < NUM_ROUNDS*NUM_ITERS; i++) {
			while (!go[MASTER])
				membar("#LoadLoad");
			go[MASTER] = 0;
			membar("#StoreStore");
			go[SLAVE] = tick_ops->get_tick();
			membar("#StoreLoad");
		}
	}
	spin_unlock_irqrestore(&itc_sync_lock, flags);
}

static void smp_synchronize_tick(void)
{
	int cpu = smp_processor_id();
	int i;

	for (i = 0; i < NR_CPUS; i++) {
		if (cpu_present_map & (1UL << i)) {
			if (i == cpu)
				continue;
			smp_synchronize_one_tick(i);
		}
	}
}

extern struct prom_cpuinfo linux_cpus[64];

extern unsigned long sparc64_cpu_startup;

/* The OBP cpu startup callback truncates the 3rd arg cookie to
 * 32-bits (I think) so to be safe we have it read the pointer
 * contained here so we work on >4GB machines. -DaveM
 */
static struct task_struct *cpu_new_task = NULL;

void __init smp_boot_cpus(void)
{
	int cpucount = 0, i;

	printk("Entering UltraSMPenguin Mode...\n");
	__sti();
	smp_store_cpu_info(boot_cpu_id);
	init_idle();

	if (linux_num_cpus == 1)
		return;

	for (i = 0; i < NR_CPUS; i++) {
		if (i == boot_cpu_id)
			continue;

		if ((cpucount + 1) == max_cpus)
			goto ignorecpu;
		if (cpu_present_map & (1UL << i)) {
			unsigned long entry = (unsigned long)(&sparc64_cpu_startup);
			unsigned long cookie = (unsigned long)(&cpu_new_task);
			struct task_struct *p;
			int timeout;
			int no;

			prom_printf("Starting CPU %d... ", i);
			kernel_thread(start_secondary, NULL, CLONE_PID);
			cpucount++;

			p = init_task.prev_task;
			init_tasks[cpucount] = p;

			p->processor = i;
			p->cpus_runnable = 1UL << i; /* we schedule the first task manually */

			del_from_runqueue(p);
			unhash_process(p);

			callin_flag = 0;
			for (no = 0; no < linux_num_cpus; no++)
				if (linux_cpus[no].mid == i)
					break;
			cpu_new_task = p;
			prom_startcpu(linux_cpus[no].prom_node,
				      entry, cookie);
			for (timeout = 0; timeout < 5000000; timeout++) {
				if (callin_flag)
					break;
				udelay(100);
			}
			if (callin_flag) {
				__cpu_number_map[i] = cpucount;
				__cpu_logical_map[cpucount] = i;
				prom_cpu_nodes[i] = linux_cpus[no].prom_node;
				prom_printf("OK\n");
			} else {
				cpucount--;
				printk("Processor %d is stuck.\n", i);
				prom_printf("FAILED\n");