smpboot.c

来自「优龙2410linux2.6.8内核源代码」· C语言 代码 · 共 973 行 · 第 1/2 页

/*
 *	x86 SMP booting functions
 *
 *	(c) 1995 Alan Cox, Building #3 <[e-mail]>
 *	(c) 1998, 1999, 2000 Ingo Molnar <[e-mail]>
 *	Copyright 2001 Andi Kleen, SuSE Labs.
 *
 *	Much of the core SMP work is based on previous work by Thomas Radke, to
 *	whom a great many thanks are extended.
 *
 *	Thanks to Intel for making available several different Pentium,
 *	Pentium Pro and Pentium-II/Xeon MP machines.
 *	Original development of Linux SMP code supported by Caldera.
 *
 *	This code is released under the GNU General Public License version 2 or
 *	later.
 *
 *	Fixes
 *		Felix Koop	:	NR_CPUS used properly
 *		Jose Renau	:	Handle single CPU case.
 *		Alan Cox	:	By repeated request 8) - Total BogoMIP report.
 *		Greg Wright	:	Fix for kernel stacks panic.
 *		Erich Boleyn	:	MP v1.4 and additional changes.
 *	Matthias Sattler	:	Changes for 2.1 kernel map.
 *	Michel Lespinasse	:	Changes for 2.1 kernel map.
 *	Michael Chastain	:	Change trampoline.S to gnu as.
 *		Alan Cox	:	Dumb bug: 'B' step PPro's are fine
 *		Ingo Molnar	:	Added APIC timers, based on code
 *					from Jose Renau
 *		Ingo Molnar	:	various cleanups and rewrites
 *		Tigran Aivazian	:	fixed "0.00 in /proc/uptime on SMP" bug.
 *	Maciej W. Rozycki	:	Bits for genuine 82489DX APICs
 *	Andi Kleen		:	Changed for SMP boot into long mode.
 *		Rusty Russell	:	Hacked into shape for new "hotplug" boot process. 
 */

#include <linux/config.h>
#include <linux/init.h>

#include <linux/mm.h>
#include <linux/kernel_stat.h>
#include <linux/smp_lock.h>
#include <linux/irq.h>
#include <linux/bootmem.h>
#include <linux/thread_info.h>

#include <linux/delay.h>
#include <linux/mc146818rtc.h>
#include <asm/mtrr.h>
#include <asm/pgalloc.h>
#include <asm/desc.h>
#include <asm/kdebug.h>
#include <asm/tlbflush.h>
#include <asm/proto.h>

/* Number of siblings per CPU package */
int smp_num_siblings = 1;
char phys_proc_id[NR_CPUS]; /* Package ID of each logical CPU */

/* Bitmask of currently online CPUs */
cpumask_t cpu_online_map;

/* which logical CPU number maps to which CPU (physical APIC ID) */
volatile char x86_cpu_to_apicid[NR_CPUS];

static cpumask_t cpu_callin_map;
cpumask_t cpu_callout_map;
static cpumask_t smp_commenced_mask;

/* Per CPU bogomips and other parameters */
struct cpuinfo_x86 cpu_data[NR_CPUS] __cacheline_aligned;

/* Set when the idlers are all forked */
int smp_threads_ready;

cpumask_t cpu_sibling_map[NR_CPUS] __cacheline_aligned;

/*
 * Trampoline 80x86 program as an array.
 */

extern unsigned char trampoline_data [];
extern unsigned char trampoline_end  [];

/*
 * Currently trivial. Write the real->protected mode
 * bootstrap into the page concerned. The caller
 * has made sure it's suitably aligned.
 */

static unsigned long __init setup_trampoline(void)
{
	void *tramp = __va(SMP_TRAMPOLINE_BASE); 
	extern volatile __u32 tramp_gdt_ptr; 
	tramp_gdt_ptr = __pa_symbol(&cpu_gdt_table); 
	memcpy(tramp, trampoline_data, trampoline_end - trampoline_data);
	return virt_to_phys(tramp);
}

/*
 * The bootstrap kernel entry code has set these up. Save them for
 * a given CPU
 */

static void __init smp_store_cpu_info(int id)
{
	struct cpuinfo_x86 *c = cpu_data + id;

	*c = boot_cpu_data;
	identify_cpu(c);
}

/*
 * TSC synchronization.
 *
 * We first check whether all CPUs have their TSC's synchronized,
 * then we print a warning if not, and always resync.
 */

static atomic_t tsc_start_flag = ATOMIC_INIT(0);
static atomic_t tsc_count_start = ATOMIC_INIT(0);
static atomic_t tsc_count_stop = ATOMIC_INIT(0);
static unsigned long long tsc_values[NR_CPUS];

#define NR_LOOPS 5

extern unsigned int fast_gettimeoffset_quotient;

static void __init synchronize_tsc_bp (void)
{
	int i;
	unsigned long long t0;
	unsigned long long sum, avg;
	long long delta;
	long one_usec;
	int buggy = 0;

	printk(KERN_INFO "checking TSC synchronization across %u CPUs: ",num_booting_cpus());

	one_usec = cpu_khz; 

	atomic_set(&tsc_start_flag, 1);
	wmb();

	/*
	 * We loop a few times to get a primed instruction cache,
	 * then the last pass is more or less synchronized and
	 * the BP and APs set their cycle counters to zero all at
	 * once. This reduces the chance of having random offsets
	 * between the processors, and guarantees that the maximum
	 * delay between the cycle counters is never bigger than
	 * the latency of information-passing (cachelines) between
	 * two CPUs.
	 */
	for (i = 0; i < NR_LOOPS; i++) {
		/*
		 * all APs synchronize but they loop on '== num_cpus'
		 */
		while (atomic_read(&tsc_count_start) != num_booting_cpus()-1) mb();
		atomic_set(&tsc_count_stop, 0);
		wmb();
		/*
		 * this lets the APs save their current TSC:
		 */
		atomic_inc(&tsc_count_start);

		sync_core();
		rdtscll(tsc_values[smp_processor_id()]);
		/*
		 * We clear the TSC in the last loop:
		 */
		if (i == NR_LOOPS-1)
			write_tsc(0, 0);

		/*
		 * Wait for all APs to leave the synchronization point:
		 */
		while (atomic_read(&tsc_count_stop) != num_booting_cpus()-1) mb();
		atomic_set(&tsc_count_start, 0);
		wmb();
		atomic_inc(&tsc_count_stop);
	}

	sum = 0;
	for (i = 0; i < NR_CPUS; i++) {
		if (cpu_isset(i, cpu_callout_map)) {
		t0 = tsc_values[i];
		sum += t0;
	}
	}
	avg = sum / num_booting_cpus();

	sum = 0;
	for (i = 0; i < NR_CPUS; i++) {
		if (!cpu_isset(i, cpu_callout_map))
			continue;

		delta = tsc_values[i] - avg;
		if (delta < 0)
			delta = -delta;
		/*
		 * We report bigger than 2 microseconds clock differences.
		 */
		if (delta > 2*one_usec) {
			long realdelta;
			if (!buggy) {
				buggy = 1;
				printk("\n");
			}
			realdelta = delta / one_usec;
			if (tsc_values[i] < avg)
				realdelta = -realdelta;

			printk("BIOS BUG: CPU#%d improperly initialized, has %ld usecs TSC skew! FIXED.\n",
				i, realdelta);
		}

		sum += delta;
	}
	if (!buggy)
		printk("passed.\n");
}

static void __init synchronize_tsc_ap (void)
{
	int i;

	/*
	 * Not every cpu is online at the time
	 * this gets called, so we first wait for the BP to
	 * finish SMP initialization:
	 */
	while (!atomic_read(&tsc_start_flag)) mb();

	for (i = 0; i < NR_LOOPS; i++) {
		atomic_inc(&tsc_count_start);
		while (atomic_read(&tsc_count_start) != num_booting_cpus()) mb();

		sync_core();
		rdtscll(tsc_values[smp_processor_id()]);
		if (i == NR_LOOPS-1)
			write_tsc(0, 0);

		atomic_inc(&tsc_count_stop);
		while (atomic_read(&tsc_count_stop) != num_booting_cpus()) mb();
	}
}
#undef NR_LOOPS

static atomic_t init_deasserted;

void __init smp_callin(void)
{
	int cpuid, phys_id;
	unsigned long timeout;

	/*
	 * If waken up by an INIT in an 82489DX configuration
	 * we may get here before an INIT-deassert IPI reaches
	 * our local APIC.  We have to wait for the IPI or we'll
	 * lock up on an APIC access.
	 */
	while (!atomic_read(&init_deasserted));

	/*
	 * (This works even if the APIC is not enabled.)
	 */
	phys_id = GET_APIC_ID(apic_read(APIC_ID));
	cpuid = smp_processor_id();
	if (cpu_isset(cpuid, cpu_callin_map)) {
		panic("smp_callin: phys CPU#%d, CPU#%d already present??\n",
					phys_id, cpuid);
	}
	Dprintk("CPU#%d (phys ID: %d) waiting for CALLOUT\n", cpuid, phys_id);

	/*
	 * STARTUP IPIs are fragile beasts as they might sometimes
	 * trigger some glue motherboard logic. Complete APIC bus
	 * silence for 1 second, this overestimates the time the
	 * boot CPU is spending to send the up to 2 STARTUP IPIs
	 * by a factor of two. This should be enough.
	 */

	/*
	 * Waiting 2s total for startup (udelay is not yet working)
	 */
	timeout = jiffies + 2*HZ;
	while (time_before(jiffies, timeout)) {
		/*
		 * Has the boot CPU finished it's STARTUP sequence?
		 */
		if (cpu_isset(cpuid, cpu_callout_map))
			break;
		rep_nop();
	}

	if (!time_before(jiffies, timeout)) {
		panic("smp_callin: CPU%d started up but did not get a callout!\n",
			cpuid);
	}

	/*
	 * the boot CPU has finished the init stage and is spinning
	 * on callin_map until we finish. We are free to set up this
	 * CPU, first the APIC. (this is probably redundant on most
	 * boards)
	 */

	Dprintk("CALLIN, before setup_local_APIC().\n");
	setup_local_APIC();

	local_irq_enable();

	/*
	 * Get our bogomips.
	 */
	calibrate_delay();
	Dprintk("Stack at about %p\n",&cpuid);

	disable_APIC_timer();

	/*
	 * Save our processor parameters
	 */
 	smp_store_cpu_info(cpuid);

	local_irq_disable();

	/*
	 * Allow the master to continue.
	 */
	cpu_set(cpuid, cpu_callin_map);

	/*
	 *      Synchronize the TSC with the BP
	 */
	if (cpu_has_tsc)
		synchronize_tsc_ap();
}

int cpucount;

/*
 * Activate a secondary processor.
 */
void __init start_secondary(void)
{
	/*
	 * Dont put anything before smp_callin(), SMP
	 * booting is too fragile that we want to limit the
	 * things done here to the most necessary things.
	 */
	cpu_init();
	smp_callin();

	/* otherwise gcc will move up the smp_processor_id before the cpu_init */
	barrier();

	Dprintk("cpu %d: waiting for commence\n", smp_processor_id()); 
	while (!cpu_isset(smp_processor_id(), smp_commenced_mask))
		rep_nop();

	Dprintk("cpu %d: setting up apic clock\n", smp_processor_id()); 	
	setup_secondary_APIC_clock();

	Dprintk("cpu %d: enabling apic timer\n", smp_processor_id()); 

	if (nmi_watchdog == NMI_IO_APIC) {
		disable_8259A_irq(0);
		enable_NMI_through_LVT0(NULL);
		enable_8259A_irq(0);
	}


	enable_APIC_timer(); 

	/*
	 * low-memory mappings have been cleared, flush them from
	 * the local TLBs too.
	 */
	local_flush_tlb();

	Dprintk("cpu %d eSetting cpu_online_map\n", smp_processor_id()); 
	cpu_set(smp_processor_id(), cpu_online_map);
	wmb();
	
	cpu_idle();
}

extern volatile unsigned long init_rsp; 
extern void (*initial_code)(void);

static struct task_struct * __init fork_by_hand(void)
{
	struct pt_regs regs;
	/*
	 * don't care about the eip and regs settings since
	 * we'll never reschedule the forked task.
	 */
	return copy_process(CLONE_VM|CLONE_IDLETASK, 0, &regs, 0, NULL, NULL);
}

#if APIC_DEBUG
static inline void inquire_remote_apic(int apicid)
{
	unsigned i, regs[] = { APIC_ID >> 4, APIC_LVR >> 4, APIC_SPIV >> 4 };
	char *names[] = { "ID", "VERSION", "SPIV" };
	int timeout, status;

	printk(KERN_INFO "Inquiring remote APIC #%d...\n", apicid);

	for (i = 0; i < sizeof(regs) / sizeof(*regs); i++) {
		printk("... APIC #%d %s: ", apicid, names[i]);

		/*
		 * Wait for idle.
		 */
		apic_wait_icr_idle();

		apic_write_around(APIC_ICR2, SET_APIC_DEST_FIELD(apicid));
		apic_write_around(APIC_ICR, APIC_DM_REMRD | regs[i]);

		timeout = 0;
		do {
			udelay(100);
			status = apic_read(APIC_ICR) & APIC_ICR_RR_MASK;
		} while (status == APIC_ICR_RR_INPROG && timeout++ < 1000);

		switch (status) {
		case APIC_ICR_RR_VALID:
			status = apic_read(APIC_RRR);
			printk("%08x\n", status);
			break;
		default:
			printk("failed\n");
		}
	}
}
#endif

static int __init wakeup_secondary_via_INIT(int phys_apicid, unsigned int start_rip)
{
	unsigned long send_status = 0, accept_status = 0;
	int maxlvt, timeout, num_starts, j;

	Dprintk("Asserting INIT.\n");

	/*
	 * Turn INIT on target chip
	 */
	apic_write_around(APIC_ICR2, SET_APIC_DEST_FIELD(phys_apicid));

	/*
	 * Send IPI
	 */
	apic_write_around(APIC_ICR, APIC_INT_LEVELTRIG | APIC_INT_ASSERT
				| APIC_DM_INIT);

	Dprintk("Waiting for send to finish...\n");
	timeout = 0;
	do {
		Dprintk("+");
		udelay(100);
		send_status = apic_read(APIC_ICR) & APIC_ICR_BUSY;
	} while (send_status && (timeout++ < 1000));

	mdelay(10);

	Dprintk("Deasserting INIT.\n");

	/* Target chip */
	apic_write_around(APIC_ICR2, SET_APIC_DEST_FIELD(phys_apicid));

	/* Send IPI */
	apic_write_around(APIC_ICR, APIC_INT_LEVELTRIG | APIC_DM_INIT);

	Dprintk("Waiting for send to finish...\n");
	timeout = 0;
	do {
		Dprintk("+");
		udelay(100);
		send_status = apic_read(APIC_ICR) & APIC_ICR_BUSY;
	} while (send_status && (timeout++ < 1000));

	atomic_set(&init_deasserted, 1);

	/*