smpboot.c

linux-2.6.15.6

		}

		if (cpu_isset(cpu, cpu_callin_map)) {
			/* number CPUs logically, starting from 1 (BSP is 0) */
			Dprintk("OK.\n");
			printk("CPU%d: ", cpu);
			print_cpu_info(&cpu_data[cpu]);
			Dprintk("CPU has booted.\n");
		} else {
			boot_error= 1;
			if (*((volatile unsigned char *)trampoline_base)
					== 0xA5)
				/* trampoline started but...? */
				printk("Stuck ??\n");
			else
				/* trampoline code not run */
				printk("Not responding.\n");
			inquire_remote_apic(apicid);
		}
	}

	if (boot_error) {
		/* Try to put things back the way they were before ... */
		unmap_cpu_to_logical_apicid(cpu);
		cpu_clear(cpu, cpu_callout_map); /* was set here (do_boot_cpu()) */
		cpu_clear(cpu, cpu_initialized); /* was set by cpu_init() */
		cpucount--;
	} else {
		x86_cpu_to_apicid[cpu] = apicid;
		cpu_set(cpu, cpu_present_map);
	}

	/* mark "stuck" area as not stuck */
	*((volatile unsigned long *)trampoline_base) = 0;

	return boot_error;
}

#ifdef CONFIG_HOTPLUG_CPU
void cpu_exit_clear(void)
{
	int cpu = raw_smp_processor_id();

	idle_task_exit();

	cpucount --;
	cpu_uninit();
	irq_ctx_exit(cpu);

	cpu_clear(cpu, cpu_callout_map);
	cpu_clear(cpu, cpu_callin_map);
	cpu_clear(cpu, cpu_present_map);

	cpu_clear(cpu, smp_commenced_mask);
	unmap_cpu_to_logical_apicid(cpu);
}

struct warm_boot_cpu_info {
	struct completion *complete;
	int apicid;
	int cpu;
};

static void __devinit do_warm_boot_cpu(void *p)
{
	struct warm_boot_cpu_info *info = p;
	do_boot_cpu(info->apicid, info->cpu);
	complete(info->complete);
}

int __devinit smp_prepare_cpu(int cpu)
{
	DECLARE_COMPLETION(done);
	struct warm_boot_cpu_info info;
	struct work_struct task;
	int	apicid, ret;

	lock_cpu_hotplug();
	apicid = x86_cpu_to_apicid[cpu];
	if (apicid == BAD_APICID) {
		ret = -ENODEV;
		goto exit;
	}

	info.complete = &done;
	info.apicid = apicid;
	info.cpu = cpu;
	INIT_WORK(&task, do_warm_boot_cpu, &info);

	tsc_sync_disabled = 1;

	/* init low mem mapping */
	clone_pgd_range(swapper_pg_dir, swapper_pg_dir + USER_PGD_PTRS,
			KERNEL_PGD_PTRS);
	flush_tlb_all();
	schedule_work(&task);
	wait_for_completion(&done);

	tsc_sync_disabled = 0;
	zap_low_mappings();
	ret = 0;
exit:
	unlock_cpu_hotplug();
	return ret;
}
#endif

static void smp_tune_scheduling (void)
{
	unsigned long cachesize;       /* kB   */
	unsigned long bandwidth = 350; /* MB/s */
	/*
	 * 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) {
		/*
		 * this basically disables processor-affinity
		 * scheduling on SMP without a TSC.
		 */
		return;
	} else {
		cachesize = boot_cpu_data.x86_cache_size;
		if (cachesize == -1) {
			cachesize = 16; /* Pentiums, 2x8kB cache */
			bandwidth = 100;
		}
	}
}

/*
 * Cycle through the processors sending APIC IPIs to boot each.
 */

static int boot_cpu_logical_apicid;
/* Where the IO area was mapped on multiquad, always 0 otherwise */
void *xquad_portio;
#ifdef CONFIG_X86_NUMAQ
EXPORT_SYMBOL(xquad_portio);
#endif

static void __init smp_boot_cpus(unsigned int max_cpus)
{
	int apicid, cpu, bit, kicked;
	unsigned long bogosum = 0;

	/*
	 * Setup boot CPU information
	 */
	smp_store_cpu_info(0); /* Final full version of the data */
	printk("CPU%d: ", 0);
	print_cpu_info(&cpu_data[0]);

	boot_cpu_physical_apicid = GET_APIC_ID(apic_read(APIC_ID));
	boot_cpu_logical_apicid = logical_smp_processor_id();
	x86_cpu_to_apicid[0] = boot_cpu_physical_apicid;

	current_thread_info()->cpu = 0;
	smp_tune_scheduling();

	set_cpu_sibling_map(0);

	/*
	 * If we couldn't find an SMP configuration at boot time,
	 * get out of here now!
	 */
	if (!smp_found_config && !acpi_lapic) {
		printk(KERN_NOTICE "SMP motherboard not detected.\n");
		smpboot_clear_io_apic_irqs();
		phys_cpu_present_map = physid_mask_of_physid(0);
		if (APIC_init_uniprocessor())
			printk(KERN_NOTICE "Local APIC not detected."
					   " Using dummy APIC emulation.\n");
		map_cpu_to_logical_apicid();
		cpu_set(0, cpu_sibling_map[0]);
		cpu_set(0, cpu_core_map[0]);
		return;
	}

	/*
	 * Should not be necessary because the MP table should list the boot
	 * CPU too, but we do it for the sake of robustness anyway.
	 * Makes no sense to do this check in clustered apic mode, so skip it
	 */
	if (!check_phys_apicid_present(boot_cpu_physical_apicid)) {
		printk("weird, boot CPU (#%d) not listed by the BIOS.\n",
				boot_cpu_physical_apicid);
		physid_set(hard_smp_processor_id(), phys_cpu_present_map);
	}

	/*
	 * If we couldn't find a local APIC, then get out of here now!
	 */
	if (APIC_INTEGRATED(apic_version[boot_cpu_physical_apicid]) && !cpu_has_apic) {
		printk(KERN_ERR "BIOS bug, local APIC #%d not detected!...\n",
			boot_cpu_physical_apicid);
		printk(KERN_ERR "... forcing use of dummy APIC emulation. (tell your hw vendor)\n");
		smpboot_clear_io_apic_irqs();
		phys_cpu_present_map = physid_mask_of_physid(0);
		cpu_set(0, cpu_sibling_map[0]);
		cpu_set(0, cpu_core_map[0]);
		return;
	}

	verify_local_APIC();

	/*
	 * If SMP should be disabled, then really disable it!
	 */
	if (!max_cpus) {
		smp_found_config = 0;
		printk(KERN_INFO "SMP mode deactivated, forcing use of dummy APIC emulation.\n");
		smpboot_clear_io_apic_irqs();
		phys_cpu_present_map = physid_mask_of_physid(0);
		cpu_set(0, cpu_sibling_map[0]);
		cpu_set(0, cpu_core_map[0]);
		return;
	}

	connect_bsp_APIC();
	setup_local_APIC();
	map_cpu_to_logical_apicid();


	setup_portio_remap();

	/*
	 * Scan the CPU present map and fire up the other CPUs via do_boot_cpu
	 *
	 * In clustered apic mode, phys_cpu_present_map is a constructed thus:
	 * bits 0-3 are quad0, 4-7 are quad1, etc. A perverse twist on the 
	 * clustered apic ID.
	 */
	Dprintk("CPU present map: %lx\n", physids_coerce(phys_cpu_present_map));

	kicked = 1;
	for (bit = 0; kicked < NR_CPUS && bit < MAX_APICS; bit++) {
		apicid = cpu_present_to_apicid(bit);
		/*
		 * Don't even attempt to start the boot CPU!
		 */
		if ((apicid == boot_cpu_apicid) || (apicid == BAD_APICID))
			continue;

		if (!check_apicid_present(bit))
			continue;
		if (max_cpus <= cpucount+1)
			continue;

		if (((cpu = alloc_cpu_id()) <= 0) || do_boot_cpu(apicid, cpu))
			printk("CPU #%d not responding - cannot use it.\n",
								apicid);
		else
			++kicked;
	}

	/*
	 * Cleanup possible dangling ends...
	 */
	smpboot_restore_warm_reset_vector();

	/*
	 * Allow the user to impress friends.
	 */
	Dprintk("Before bogomips.\n");
	for (cpu = 0; cpu < NR_CPUS; cpu++)
		if (cpu_isset(cpu, cpu_callout_map))
			bogosum += cpu_data[cpu].loops_per_jiffy;
	printk(KERN_INFO
		"Total of %d processors activated (%lu.%02lu BogoMIPS).\n",
		cpucount+1,
		bogosum/(500000/HZ),
		(bogosum/(5000/HZ))%100);
	
	Dprintk("Before bogocount - setting activated=1.\n");

	if (smp_b_stepping)
		printk(KERN_WARNING "WARNING: SMP operation may be unreliable with B stepping processors.\n");

	/*
	 * Don't taint if we are running SMP kernel on a single non-MP
	 * approved Athlon
	 */
	if (tainted & TAINT_UNSAFE_SMP) {
		if (cpucount)
			printk (KERN_INFO "WARNING: This combination of AMD processors is not suitable for SMP.\n");
		else
			tainted &= ~TAINT_UNSAFE_SMP;
	}

	Dprintk("Boot done.\n");

	/*
	 * construct cpu_sibling_map[], so that we can tell sibling CPUs
	 * efficiently.
	 */
	for (cpu = 0; cpu < NR_CPUS; cpu++) {
		cpus_clear(cpu_sibling_map[cpu]);
		cpus_clear(cpu_core_map[cpu]);
	}

	cpu_set(0, cpu_sibling_map[0]);
	cpu_set(0, cpu_core_map[0]);

	smpboot_setup_io_apic();

	setup_boot_APIC_clock();

	/*
	 * Synchronize the TSC with the AP
	 */
	if (cpu_has_tsc && cpucount && cpu_khz)
		synchronize_tsc_bp();
}

/* These are wrappers to interface to the new boot process.  Someone
   who understands all this stuff should rewrite it properly. --RR 15/Jul/02 */
void __init smp_prepare_cpus(unsigned int max_cpus)
{
	smp_commenced_mask = cpumask_of_cpu(0);
	cpu_callin_map = cpumask_of_cpu(0);
	mb();
	smp_boot_cpus(max_cpus);
}

void __devinit smp_prepare_boot_cpu(void)
{
	cpu_set(smp_processor_id(), cpu_online_map);
	cpu_set(smp_processor_id(), cpu_callout_map);
	cpu_set(smp_processor_id(), cpu_present_map);
	cpu_set(smp_processor_id(), cpu_possible_map);
	per_cpu(cpu_state, smp_processor_id()) = CPU_ONLINE;
}

#ifdef CONFIG_HOTPLUG_CPU
static void
remove_siblinginfo(int cpu)
{
	int sibling;
	struct cpuinfo_x86 *c = cpu_data;

	for_each_cpu_mask(sibling, cpu_core_map[cpu]) {
		cpu_clear(cpu, cpu_core_map[sibling]);
		/*
		 * last thread sibling in this cpu core going down
		 */
		if (cpus_weight(cpu_sibling_map[cpu]) == 1)
			c[sibling].booted_cores--;
	}
			
	for_each_cpu_mask(sibling, cpu_sibling_map[cpu])
		cpu_clear(cpu, cpu_sibling_map[sibling]);
	cpus_clear(cpu_sibling_map[cpu]);
	cpus_clear(cpu_core_map[cpu]);
	phys_proc_id[cpu] = BAD_APICID;
	cpu_core_id[cpu] = BAD_APICID;
	cpu_clear(cpu, cpu_sibling_setup_map);
}

int __cpu_disable(void)
{
	cpumask_t map = cpu_online_map;
	int cpu = smp_processor_id();

	/*
	 * Perhaps use cpufreq to drop frequency, but that could go
	 * into generic code.
 	 *
	 * We won't take down the boot processor on i386 due to some
	 * interrupts only being able to be serviced by the BSP.
	 * Especially so if we're not using an IOAPIC	-zwane
	 */
	if (cpu == 0)
		return -EBUSY;

	clear_local_APIC();
	/* Allow any queued timer interrupts to get serviced */
	local_irq_enable();
	mdelay(1);
	local_irq_disable();

	remove_siblinginfo(cpu);

	cpu_clear(cpu, map);
	fixup_irqs(map);
	/* It's now safe to remove this processor from the online map */
	cpu_clear(cpu, cpu_online_map);
	return 0;
}

void __cpu_die(unsigned int cpu)
{
	/* We don't do anything here: idle task is faking death itself. */
	unsigned int i;

	for (i = 0; i < 10; i++) {
		/* They ack this in play_dead by setting CPU_DEAD */
		if (per_cpu(cpu_state, cpu) == CPU_DEAD) {
			printk ("CPU %d is now offline\n", cpu);
			return;
		}
		msleep(100);
	}
 	printk(KERN_ERR "CPU %u didn't die...\n", cpu);
}
#else /* ... !CONFIG_HOTPLUG_CPU */
int __cpu_disable(void)
{
	return -ENOSYS;
}

void __cpu_die(unsigned int cpu)
{
	/* We said "no" in __cpu_disable */
	BUG();
}
#endif /* CONFIG_HOTPLUG_CPU */

int __devinit __cpu_up(unsigned int cpu)
{
	/* In case one didn't come up */
	if (!cpu_isset(cpu, cpu_callin_map)) {
		printk(KERN_DEBUG "skipping cpu%d, didn't come online\n", cpu);
		local_irq_enable();
		return -EIO;
	}

	local_irq_enable();
	per_cpu(cpu_state, cpu) = CPU_UP_PREPARE;
	/* Unleash the CPU! */
	cpu_set(cpu, smp_commenced_mask);
	while (!cpu_isset(cpu, cpu_online_map))
		mb();
	return 0;
}

void __init smp_cpus_done(unsigned int max_cpus)
{
#ifdef CONFIG_X86_IO_APIC
	setup_ioapic_dest();
#endif
	zap_low_mappings();
#ifndef CONFIG_HOTPLUG_CPU
	/*
	 * Disable executability of the SMP trampoline:
	 */
	set_kernel_exec((unsigned long)trampoline_base, trampoline_exec);
#endif
}

void __init smp_intr_init(void)
{
	/*
	 * IRQ0 must be given a fixed assignment and initialized,
	 * because it's used before the IO-APIC is set up.
	 */
	set_intr_gate(FIRST_DEVICE_VECTOR, interrupt[0]);

	/*
	 * The reschedule interrupt is a CPU-to-CPU reschedule-helper
	 * IPI, driven by wakeup.
	 */
	set_intr_gate(RESCHEDULE_VECTOR, reschedule_interrupt);

	/* IPI for invalidation */
	set_intr_gate(INVALIDATE_TLB_VECTOR, invalidate_interrupt);

	/* IPI for generic function call */
	set_intr_gate(CALL_FUNCTION_VECTOR, call_function_interrupt);
}