[deliverable/linux.git] / arch / x86 / xen / smp.c

/*
 * Xen SMP support
 *
 * This file implements the Xen versions of smp_ops.  SMP under Xen is
 * very straightforward.  Bringing a CPU up is simply a matter of
 * loading its initial context and setting it running.
 *
 * IPIs are handled through the Xen event mechanism.
 *
 * Because virtual CPUs can be scheduled onto any real CPU, there's no
 * useful topology information for the kernel to make use of.  As a
 * result, all CPUs are treated as if they're single-core and
 * single-threaded.
 */
#include <linux/sched.h>
#include <linux/err.h>
#include <linux/slab.h>
#include <linux/smp.h>

#include <asm/paravirt.h>
#include <asm/desc.h>
#include <asm/pgtable.h>
#include <asm/cpu.h>

#include <xen/interface/xen.h>
#include <xen/interface/vcpu.h>

#include <asm/xen/interface.h>
#include <asm/xen/hypercall.h>

#include <xen/xen.h>
#include <xen/page.h>
#include <xen/events.h>

#include <xen/hvc-console.h>
#include "xen-ops.h"
#include "mmu.h"

cpumask_var_t xen_cpu_initialized_map;

static DEFINE_PER_CPU(int, xen_resched_irq);
static DEFINE_PER_CPU(int, xen_callfunc_irq);
static DEFINE_PER_CPU(int, xen_callfuncsingle_irq);
static DEFINE_PER_CPU(int, xen_debug_irq) = -1;

static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id);
static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id);

/*
 * Reschedule call back.
 */
static irqreturn_t xen_reschedule_interrupt(int irq, void *dev_id)
{
	inc_irq_stat(irq_resched_count);
	scheduler_ipi();

	return IRQ_HANDLED;
}

static void __cpuinit cpu_bringup(void)
{
	int cpu;

	cpu_init();
	touch_softlockup_watchdog();
	preempt_disable();

	xen_enable_sysenter();
	xen_enable_syscall();

	cpu = smp_processor_id();
	smp_store_cpu_info(cpu);
	cpu_data(cpu).x86_max_cores = 1;
	set_cpu_sibling_map(cpu);

	xen_setup_cpu_clockevents();

	notify_cpu_starting(cpu);

	ipi_call_lock();
	set_cpu_online(cpu, true);
	ipi_call_unlock();

	this_cpu_write(cpu_state, CPU_ONLINE);

	wmb();

	/* We can take interrupts now: we're officially "up". */
	local_irq_enable();

	wmb();			/* make sure everything is out */
}

static void __cpuinit cpu_bringup_and_idle(void)
{
	cpu_bringup();
	cpu_idle();
}

static int xen_smp_intr_init(unsigned int cpu)
{
	int rc;
	const char *resched_name, *callfunc_name, *debug_name;

	resched_name = kasprintf(GFP_KERNEL, "resched%d", cpu);
	rc = bind_ipi_to_irqhandler(XEN_RESCHEDULE_VECTOR,
				    cpu,
				    xen_reschedule_interrupt,
				    IRQF_DISABLED|IRQF_PERCPU|IRQF_NOBALANCING,
				    resched_name,
				    NULL);
	if (rc < 0)
		goto fail;
	per_cpu(xen_resched_irq, cpu) = rc;

	callfunc_name = kasprintf(GFP_KERNEL, "callfunc%d", cpu);
	rc = bind_ipi_to_irqhandler(XEN_CALL_FUNCTION_VECTOR,
				    cpu,
				    xen_call_function_interrupt,
				    IRQF_DISABLED|IRQF_PERCPU|IRQF_NOBALANCING,
				    callfunc_name,
				    NULL);
	if (rc < 0)
		goto fail;
	per_cpu(xen_callfunc_irq, cpu) = rc;

	debug_name = kasprintf(GFP_KERNEL, "debug%d", cpu);
	rc = bind_virq_to_irqhandler(VIRQ_DEBUG, cpu, xen_debug_interrupt,
				     IRQF_DISABLED | IRQF_PERCPU | IRQF_NOBALANCING,
				     debug_name, NULL);
	if (rc < 0)
		goto fail;
	per_cpu(xen_debug_irq, cpu) = rc;

	callfunc_name = kasprintf(GFP_KERNEL, "callfuncsingle%d", cpu);
	rc = bind_ipi_to_irqhandler(XEN_CALL_FUNCTION_SINGLE_VECTOR,
				    cpu,
				    xen_call_function_single_interrupt,
				    IRQF_DISABLED|IRQF_PERCPU|IRQF_NOBALANCING,
				    callfunc_name,
				    NULL);
	if (rc < 0)
		goto fail;
	per_cpu(xen_callfuncsingle_irq, cpu) = rc;

	return 0;

 fail:
	if (per_cpu(xen_resched_irq, cpu) >= 0)
		unbind_from_irqhandler(per_cpu(xen_resched_irq, cpu), NULL);
	if (per_cpu(xen_callfunc_irq, cpu) >= 0)
		unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
	if (per_cpu(xen_debug_irq, cpu) >= 0)
		unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
	if (per_cpu(xen_callfuncsingle_irq, cpu) >= 0)
		unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu),
				       NULL);

	return rc;
}

static void __init xen_fill_possible_map(void)
{
	int i, rc;

	if (xen_initial_domain())
		return;

	for (i = 0; i < nr_cpu_ids; i++) {
		rc = HYPERVISOR_vcpu_op(VCPUOP_is_up, i, NULL);
		if (rc >= 0) {
			num_processors++;
			set_cpu_possible(i, true);
		}
	}
}

static void __init xen_filter_cpu_maps(void)
{
	int i, rc;

	if (!xen_initial_domain())
		return;

	num_processors = 0;
	disabled_cpus = 0;
	for (i = 0; i < nr_cpu_ids; i++) {
		rc = HYPERVISOR_vcpu_op(VCPUOP_is_up, i, NULL);
		if (rc >= 0) {
			num_processors++;
			set_cpu_possible(i, true);
		} else {
			set_cpu_possible(i, false);
			set_cpu_present(i, false);
		}
	}
}

static void __init xen_smp_prepare_boot_cpu(void)
{
	BUG_ON(smp_processor_id() != 0);
	native_smp_prepare_boot_cpu();

	/* We've switched to the "real" per-cpu gdt, so make sure the
	   old memory can be recycled */
	make_lowmem_page_readwrite(xen_initial_gdt);

	xen_filter_cpu_maps();
	xen_setup_vcpu_info_placement();
}

static void __init xen_smp_prepare_cpus(unsigned int max_cpus)
{
	unsigned cpu;
	unsigned int i;

	if (skip_ioapic_setup) {
		char *m = (max_cpus == 0) ?
			"The nosmp parameter is incompatible with Xen; " \
			"use Xen dom0_max_vcpus=1 parameter" :
			"The noapic parameter is incompatible with Xen";

		xen_raw_printk(m);
		panic(m);
	}
	xen_init_lock_cpu(0);

	smp_store_cpu_info(0);
	cpu_data(0).x86_max_cores = 1;

	for_each_possible_cpu(i) {
		zalloc_cpumask_var(&per_cpu(cpu_sibling_map, i), GFP_KERNEL);
		zalloc_cpumask_var(&per_cpu(cpu_core_map, i), GFP_KERNEL);
		zalloc_cpumask_var(&per_cpu(cpu_llc_shared_map, i), GFP_KERNEL);
	}
	set_cpu_sibling_map(0);

	if (xen_smp_intr_init(0))
		BUG();

	if (!alloc_cpumask_var(&xen_cpu_initialized_map, GFP_KERNEL))
		panic("could not allocate xen_cpu_initialized_map\n");

	cpumask_copy(xen_cpu_initialized_map, cpumask_of(0));

	/* Restrict the possible_map according to max_cpus. */
	while ((num_possible_cpus() > 1) && (num_possible_cpus() > max_cpus)) {
		for (cpu = nr_cpu_ids - 1; !cpu_possible(cpu); cpu--)
			continue;
		set_cpu_possible(cpu, false);
	}

	for_each_possible_cpu (cpu) {
		struct task_struct *idle;

		if (cpu == 0)
			continue;

		idle = fork_idle(cpu);
		if (IS_ERR(idle))
			panic("failed fork for CPU %d", cpu);

		set_cpu_present(cpu, true);
	}
}

static int __cpuinit
cpu_initialize_context(unsigned int cpu, struct task_struct *idle)
{
	struct vcpu_guest_context *ctxt;
	struct desc_struct *gdt;
	unsigned long gdt_mfn;

	if (cpumask_test_and_set_cpu(cpu, xen_cpu_initialized_map))
		return 0;

	ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
	if (ctxt == NULL)
		return -ENOMEM;

	gdt = get_cpu_gdt_table(cpu);

	ctxt->flags = VGCF_IN_KERNEL;
	ctxt->user_regs.ds = __USER_DS;
	ctxt->user_regs.es = __USER_DS;
	ctxt->user_regs.ss = __KERNEL_DS;
#ifdef CONFIG_X86_32
	ctxt->user_regs.fs = __KERNEL_PERCPU;
	ctxt->user_regs.gs = __KERNEL_STACK_CANARY;
#else
	ctxt->gs_base_kernel = per_cpu_offset(cpu);
#endif
	ctxt->user_regs.eip = (unsigned long)cpu_bringup_and_idle;
	ctxt->user_regs.eflags = 0x1000; /* IOPL_RING1 */

	memset(&ctxt->fpu_ctxt, 0, sizeof(ctxt->fpu_ctxt));

	xen_copy_trap_info(ctxt->trap_ctxt);

	ctxt->ldt_ents = 0;

	BUG_ON((unsigned long)gdt & ~PAGE_MASK);

	gdt_mfn = arbitrary_virt_to_mfn(gdt);
	make_lowmem_page_readonly(gdt);
	make_lowmem_page_readonly(mfn_to_virt(gdt_mfn));

	ctxt->gdt_frames[0] = gdt_mfn;
	ctxt->gdt_ents      = GDT_ENTRIES;

	ctxt->user_regs.cs = __KERNEL_CS;
	ctxt->user_regs.esp = idle->thread.sp0 - sizeof(struct pt_regs);

	ctxt->kernel_ss = __KERNEL_DS;
	ctxt->kernel_sp = idle->thread.sp0;

#ifdef CONFIG_X86_32
	ctxt->event_callback_cs     = __KERNEL_CS;
	ctxt->failsafe_callback_cs  = __KERNEL_CS;
#endif
	ctxt->event_callback_eip    = (unsigned long)xen_hypervisor_callback;
	ctxt->failsafe_callback_eip = (unsigned long)xen_failsafe_callback;

	per_cpu(xen_cr3, cpu) = __pa(swapper_pg_dir);
	ctxt->ctrlreg[3] = xen_pfn_to_cr3(virt_to_mfn(swapper_pg_dir));

	if (HYPERVISOR_vcpu_op(VCPUOP_initialise, cpu, ctxt))
		BUG();

	kfree(ctxt);
	return 0;
}

static int __cpuinit xen_cpu_up(unsigned int cpu, struct task_struct *tidle)
{
	struct task_struct *idle = idle_task(cpu);
	int rc;

	per_cpu(current_task, cpu) = idle;
#ifdef CONFIG_X86_32
	irq_ctx_init(cpu);
#else
	clear_tsk_thread_flag(idle, TIF_FORK);
	per_cpu(kernel_stack, cpu) =
		(unsigned long)task_stack_page(idle) -
		KERNEL_STACK_OFFSET + THREAD_SIZE;
#endif
	xen_setup_runstate_info(cpu);
	xen_setup_timer(cpu);
	xen_init_lock_cpu(cpu);

	per_cpu(cpu_state, cpu) = CPU_UP_PREPARE;

	/* make sure interrupts start blocked */
	per_cpu(xen_vcpu, cpu)->evtchn_upcall_mask = 1;

	rc = cpu_initialize_context(cpu, idle);
	if (rc)
		return rc;

	if (num_online_cpus() == 1)
		alternatives_smp_switch(1);

	rc = xen_smp_intr_init(cpu);
	if (rc)
		return rc;

	rc = HYPERVISOR_vcpu_op(VCPUOP_up, cpu, NULL);
	BUG_ON(rc);

	while(per_cpu(cpu_state, cpu) != CPU_ONLINE) {
		HYPERVISOR_sched_op(SCHEDOP_yield, NULL);
		barrier();
	}

	return 0;
}

static void xen_smp_cpus_done(unsigned int max_cpus)
{
}

#ifdef CONFIG_HOTPLUG_CPU
static int xen_cpu_disable(void)
{
	unsigned int cpu = smp_processor_id();
	if (cpu == 0)
		return -EBUSY;

	cpu_disable_common();

	load_cr3(swapper_pg_dir);
	return 0;
}

static void xen_cpu_die(unsigned int cpu)
{
	while (HYPERVISOR_vcpu_op(VCPUOP_is_up, cpu, NULL)) {
		current->state = TASK_UNINTERRUPTIBLE;
		schedule_timeout(HZ/10);
	}
	unbind_from_irqhandler(per_cpu(xen_resched_irq, cpu), NULL);
	unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
	unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
	unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu), NULL);
	xen_uninit_lock_cpu(cpu);
	xen_teardown_timer(cpu);

	if (num_online_cpus() == 1)
		alternatives_smp_switch(0);
}

static void __cpuinit xen_play_dead(void) /* used only with HOTPLUG_CPU */
{
	play_dead_common();
	HYPERVISOR_vcpu_op(VCPUOP_down, smp_processor_id(), NULL);
	cpu_bringup();
	/*
	 * Balance out the preempt calls - as we are running in cpu_idle
	 * loop which has been called at bootup from cpu_bringup_and_idle.
	 * The cpucpu_bringup_and_idle called cpu_bringup which made a
	 * preempt_disable() So this preempt_enable will balance it out.
	 */
	preempt_enable();
}

#else /* !CONFIG_HOTPLUG_CPU */
static int xen_cpu_disable(void)
{
	return -ENOSYS;
}

static void xen_cpu_die(unsigned int cpu)
{
	BUG();
}

static void xen_play_dead(void)
{
	BUG();
}

#endif
static void stop_self(void *v)
{
	int cpu = smp_processor_id();

	/* make sure we're not pinning something down */
	load_cr3(swapper_pg_dir);
	/* should set up a minimal gdt */

	set_cpu_online(cpu, false);

	HYPERVISOR_vcpu_op(VCPUOP_down, cpu, NULL);
	BUG();
}

static void xen_stop_other_cpus(int wait)
{
	smp_call_function(stop_self, NULL, wait);
}

static void xen_smp_send_reschedule(int cpu)
{
	xen_send_IPI_one(cpu, XEN_RESCHEDULE_VECTOR);
}

static void xen_send_IPI_mask(const struct cpumask *mask,
			      enum ipi_vector vector)
{
	unsigned cpu;

	for_each_cpu_and(cpu, mask, cpu_online_mask)
		xen_send_IPI_one(cpu, vector);
}

static void xen_smp_send_call_function_ipi(const struct cpumask *mask)
{
	int cpu;

	xen_send_IPI_mask(mask, XEN_CALL_FUNCTION_VECTOR);

	/* Make sure other vcpus get a chance to run if they need to. */
	for_each_cpu(cpu, mask) {
		if (xen_vcpu_stolen(cpu)) {
			HYPERVISOR_sched_op(SCHEDOP_yield, NULL);
			break;
		}
	}
}

static void xen_smp_send_call_function_single_ipi(int cpu)
{
	xen_send_IPI_mask(cpumask_of(cpu),
			  XEN_CALL_FUNCTION_SINGLE_VECTOR);
}

static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id)
{
	irq_enter();
	generic_smp_call_function_interrupt();
	inc_irq_stat(irq_call_count);
	irq_exit();

	return IRQ_HANDLED;
}

static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id)
{
	irq_enter();
	generic_smp_call_function_single_interrupt();
	inc_irq_stat(irq_call_count);
	irq_exit();

	return IRQ_HANDLED;
}

static const struct smp_ops xen_smp_ops __initconst = {
	.smp_prepare_boot_cpu = xen_smp_prepare_boot_cpu,
	.smp_prepare_cpus = xen_smp_prepare_cpus,
	.smp_cpus_done = xen_smp_cpus_done,

	.cpu_up = xen_cpu_up,
	.cpu_die = xen_cpu_die,
	.cpu_disable = xen_cpu_disable,
	.play_dead = xen_play_dead,

	.stop_other_cpus = xen_stop_other_cpus,
	.smp_send_reschedule = xen_smp_send_reschedule,

	.send_call_func_ipi = xen_smp_send_call_function_ipi,
	.send_call_func_single_ipi = xen_smp_send_call_function_single_ipi,
};

void __init xen_smp_init(void)
{
	smp_ops = xen_smp_ops;
	xen_fill_possible_map();
	xen_init_spinlocks();
}

static void __init xen_hvm_smp_prepare_cpus(unsigned int max_cpus)
{
	native_smp_prepare_cpus(max_cpus);
	WARN_ON(xen_smp_intr_init(0));

	xen_init_lock_cpu(0);
}

static int __cpuinit xen_hvm_cpu_up(unsigned int cpu, struct task_struct *tidle)
{
	int rc;
	rc = native_cpu_up(cpu, tidle);
	WARN_ON (xen_smp_intr_init(cpu));
	return rc;
}

static void xen_hvm_cpu_die(unsigned int cpu)
{
	unbind_from_irqhandler(per_cpu(xen_resched_irq, cpu), NULL);
	unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
	unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
	unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu), NULL);
	native_cpu_die(cpu);
}

void __init xen_hvm_smp_init(void)
{
	if (!xen_have_vector_callback)
		return;
	smp_ops.smp_prepare_cpus = xen_hvm_smp_prepare_cpus;
	smp_ops.smp_send_reschedule = xen_smp_send_reschedule;
	smp_ops.cpu_up = xen_hvm_cpu_up;
	smp_ops.cpu_die = xen_hvm_cpu_die;
	smp_ops.send_call_func_ipi = xen_smp_send_call_function_ipi;
	smp_ops.send_call_func_single_ipi = xen_smp_send_call_function_single_ipi;
}
Commit	Line	Data
f87e4cac JF	1	/*
	2	* Xen SMP support
	3	*
	4	* This file implements the Xen versions of smp_ops. SMP under Xen is
	5	* very straightforward. Bringing a CPU up is simply a matter of
	6	* loading its initial context and setting it running.
	7	*
	8	* IPIs are handled through the Xen event mechanism.
	9	*
	10	* Because virtual CPUs can be scheduled onto any real CPU, there's no
	11	* useful topology information for the kernel to make use of. As a
	12	* result, all CPUs are treated as if they're single-core and
	13	* single-threaded.
f87e4cac JF	14	*/
	15	#include <linux/sched.h>
	16	#include <linux/err.h>
5a0e3ad6	17	#include <linux/slab.h>
f87e4cac JF	18	#include <linux/smp.h>
	19
	20	#include <asm/paravirt.h>
	21	#include <asm/desc.h>
	22	#include <asm/pgtable.h>
	23	#include <asm/cpu.h>
	24
	25	#include <xen/interface/xen.h>
	26	#include <xen/interface/vcpu.h>
	27
	28	#include <asm/xen/interface.h>
	29	#include <asm/xen/hypercall.h>
	30
ea5b8f73	31	#include <xen/xen.h>
f87e4cac JF	32	#include <xen/page.h>
	33	#include <xen/events.h>
	34
ed467e69	35	#include <xen/hvc-console.h>
f87e4cac JF	36	#include "xen-ops.h"
	37	#include "mmu.h"
	38
b78936e1	39	cpumask_var_t xen_cpu_initialized_map;
f87e4cac	40
c6e22f9e TH	41	static DEFINE_PER_CPU(int, xen_resched_irq);
	42	static DEFINE_PER_CPU(int, xen_callfunc_irq);
	43	static DEFINE_PER_CPU(int, xen_callfuncsingle_irq);
	44	static DEFINE_PER_CPU(int, xen_debug_irq) = -1;
f87e4cac JF	45
f87e4cac JF	46	static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id);
3b16cf87	47	static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id);
f87e4cac JF	48
f87e4cac JF	49	/*
184748cc	50	* Reschedule call back.
f87e4cac JF	51	*/
	52	static irqreturn_t xen_reschedule_interrupt(int irq, void *dev_id)
	53	{
1b437c8c	54	inc_irq_stat(irq_resched_count);
184748cc	55	scheduler_ipi();
38bb5ab4	56
f87e4cac JF	57	return IRQ_HANDLED;
	58	}
	59
b53cedeb	60	static void __cpuinit cpu_bringup(void)
f87e4cac	61	{
e8c9e788	62	int cpu;
f87e4cac JF	63
f87e4cac JF	64	cpu_init();
d68d82af	65	touch_softlockup_watchdog();
c7b75947 JF	66	preempt_disable();
c7b75947 JF	67
e2a81baf	68	xen_enable_sysenter();
6fcac6d3	69	xen_enable_syscall();
f87e4cac	70
c7b75947 JF	71	cpu = smp_processor_id();
	72	smp_store_cpu_info(cpu);
	73	cpu_data(cpu).x86_max_cores = 1;
	74	set_cpu_sibling_map(cpu);
f87e4cac JF	75
	76	xen_setup_cpu_clockevents();
	77
106b4438 KRW	78	notify_cpu_starting(cpu);
	79
	80	ipi_call_lock();
d7d3756c	81	set_cpu_online(cpu, true);
106b4438 KRW	82	ipi_call_unlock();
106b4438 KRW	83
2113f469	84	this_cpu_write(cpu_state, CPU_ONLINE);
106b4438	85
c7b75947 JF	86	wmb();
c7b75947 JF	87
f87e4cac JF	88	/* We can take interrupts now: we're officially "up". */
	89	local_irq_enable();
	90
	91	wmb(); /* make sure everything is out */
d68d82af AN	92	}
d68d82af AN	93
b53cedeb	94	static void __cpuinit cpu_bringup_and_idle(void)
d68d82af AN	95	{
d68d82af AN	96	cpu_bringup();
f87e4cac JF	97	cpu_idle();
	98	}
	99
	100	static int xen_smp_intr_init(unsigned int cpu)
	101	{
	102	int rc;
ee523ca1	103	const char resched_name, callfunc_name, *debug_name;
f87e4cac JF	104
	105	resched_name = kasprintf(GFP_KERNEL, "resched%d", cpu);
	106	rc = bind_ipi_to_irqhandler(XEN_RESCHEDULE_VECTOR,
	107	cpu,
	108	xen_reschedule_interrupt,
	109	IRQF_DISABLED\|IRQF_PERCPU\|IRQF_NOBALANCING,
	110	resched_name,
	111	NULL);
	112	if (rc < 0)
	113	goto fail;
c6e22f9e	114	per_cpu(xen_resched_irq, cpu) = rc;
f87e4cac JF	115
	116	callfunc_name = kasprintf(GFP_KERNEL, "callfunc%d", cpu);
	117	rc = bind_ipi_to_irqhandler(XEN_CALL_FUNCTION_VECTOR,
	118	cpu,
	119	xen_call_function_interrupt,
	120	IRQF_DISABLED\|IRQF_PERCPU\|IRQF_NOBALANCING,
	121	callfunc_name,
	122	NULL);
	123	if (rc < 0)
	124	goto fail;
c6e22f9e	125	per_cpu(xen_callfunc_irq, cpu) = rc;
f87e4cac	126
ee523ca1 JF	127	debug_name = kasprintf(GFP_KERNEL, "debug%d", cpu);
	128	rc = bind_virq_to_irqhandler(VIRQ_DEBUG, cpu, xen_debug_interrupt,
	129	IRQF_DISABLED \| IRQF_PERCPU \| IRQF_NOBALANCING,
	130	debug_name, NULL);
	131	if (rc < 0)
	132	goto fail;
c6e22f9e	133	per_cpu(xen_debug_irq, cpu) = rc;
ee523ca1	134
3b16cf87 JA	135	callfunc_name = kasprintf(GFP_KERNEL, "callfuncsingle%d", cpu);
	136	rc = bind_ipi_to_irqhandler(XEN_CALL_FUNCTION_SINGLE_VECTOR,
	137	cpu,
	138	xen_call_function_single_interrupt,
	139	IRQF_DISABLED\|IRQF_PERCPU\|IRQF_NOBALANCING,
	140	callfunc_name,
	141	NULL);
	142	if (rc < 0)
	143	goto fail;
c6e22f9e	144	per_cpu(xen_callfuncsingle_irq, cpu) = rc;
3b16cf87	145
f87e4cac JF	146	return 0;
	147
	148	fail:
c6e22f9e TH	149	if (per_cpu(xen_resched_irq, cpu) >= 0)
	150	unbind_from_irqhandler(per_cpu(xen_resched_irq, cpu), NULL);
	151	if (per_cpu(xen_callfunc_irq, cpu) >= 0)
	152	unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
	153	if (per_cpu(xen_debug_irq, cpu) >= 0)
	154	unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
	155	if (per_cpu(xen_callfuncsingle_irq, cpu) >= 0)
	156	unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu),
	157	NULL);
3b16cf87	158
f87e4cac JF	159	return rc;
	160	}
	161
c7b75947	162	static void __init xen_fill_possible_map(void)
f87e4cac JF	163	{
	164	int i, rc;
	165
ea5b8f73 SS	166	if (xen_initial_domain())
	167	return;
	168
	169	for (i = 0; i < nr_cpu_ids; i++) {
	170	rc = HYPERVISOR_vcpu_op(VCPUOP_is_up, i, NULL);
	171	if (rc >= 0) {
	172	num_processors++;
	173	set_cpu_possible(i, true);
	174	}
	175	}
	176	}
	177
	178	static void __init xen_filter_cpu_maps(void)
	179	{
	180	int i, rc;
	181
	182	if (!xen_initial_domain())
	183	return;
	184
801fd14a SS	185	num_processors = 0;
801fd14a SS	186	disabled_cpus = 0;
e7986739	187	for (i = 0; i < nr_cpu_ids; i++) {
f87e4cac	188	rc = HYPERVISOR_vcpu_op(VCPUOP_is_up, i, NULL);
4560a294 JF	189	if (rc >= 0) {
4560a294 JF	190	num_processors++;
4f062896	191	set_cpu_possible(i, true);
801fd14a SS	192	} else {
	193	set_cpu_possible(i, false);
	194	set_cpu_present(i, false);
4560a294	195	}
f87e4cac JF	196	}
	197	}
	198
a9e7062d	199	static void __init xen_smp_prepare_boot_cpu(void)
f87e4cac	200	{
f87e4cac JF	201	BUG_ON(smp_processor_id() != 0);
	202	native_smp_prepare_boot_cpu();
	203
f87e4cac JF	204	/* We've switched to the "real" per-cpu gdt, so make sure the
f87e4cac JF	205	old memory can be recycled */
38341432	206	make_lowmem_page_readwrite(xen_initial_gdt);
60223a32	207
ea5b8f73	208	xen_filter_cpu_maps();
60223a32	209	xen_setup_vcpu_info_placement();
f87e4cac JF	210	}
f87e4cac JF	211
a9e7062d	212	static void __init xen_smp_prepare_cpus(unsigned int max_cpus)
f87e4cac JF	213	{
f87e4cac JF	214	unsigned cpu;
900cba88	215	unsigned int i;
f87e4cac	216
ed467e69 KRW	217	if (skip_ioapic_setup) {
	218	char *m = (max_cpus == 0) ?
	219	"The nosmp parameter is incompatible with Xen; " \
	220	"use Xen dom0_max_vcpus=1 parameter" :
	221	"The noapic parameter is incompatible with Xen";
	222
	223	xen_raw_printk(m);
	224	panic(m);
	225	}
2d9e1e2f JF	226	xen_init_lock_cpu(0);
2d9e1e2f JF	227
f87e4cac	228	smp_store_cpu_info(0);
c7b75947	229	cpu_data(0).x86_max_cores = 1;
900cba88 AJ	230
	231	for_each_possible_cpu(i) {
	232	zalloc_cpumask_var(&per_cpu(cpu_sibling_map, i), GFP_KERNEL);
	233	zalloc_cpumask_var(&per_cpu(cpu_core_map, i), GFP_KERNEL);
	234	zalloc_cpumask_var(&per_cpu(cpu_llc_shared_map, i), GFP_KERNEL);
	235	}
f87e4cac JF	236	set_cpu_sibling_map(0);
	237
	238	if (xen_smp_intr_init(0))
	239	BUG();
	240
b78936e1 MT	241	if (!alloc_cpumask_var(&xen_cpu_initialized_map, GFP_KERNEL))
	242	panic("could not allocate xen_cpu_initialized_map\n");
	243
	244	cpumask_copy(xen_cpu_initialized_map, cpumask_of(0));
f87e4cac JF	245
	246	/* Restrict the possible_map according to max_cpus. */
	247	while ((num_possible_cpus() > 1) && (num_possible_cpus() > max_cpus)) {
e7986739	248	for (cpu = nr_cpu_ids - 1; !cpu_possible(cpu); cpu--)
f87e4cac	249	continue;
4f062896	250	set_cpu_possible(cpu, false);
f87e4cac JF	251	}
	252
	253	for_each_possible_cpu (cpu) {
	254	struct task_struct *idle;
	255
	256	if (cpu == 0)
	257	continue;
	258
	259	idle = fork_idle(cpu);
	260	if (IS_ERR(idle))
	261	panic("failed fork for CPU %d", cpu);
	262
4f062896	263	set_cpu_present(cpu, true);
f87e4cac	264	}
f87e4cac JF	265	}
f87e4cac JF	266
b53cedeb	267	static int __cpuinit
f87e4cac JF	268	cpu_initialize_context(unsigned int cpu, struct task_struct *idle)
	269	{
	270	struct vcpu_guest_context *ctxt;
c7b75947	271	struct desc_struct *gdt;
9976b39b	272	unsigned long gdt_mfn;
f87e4cac	273
b78936e1	274	if (cpumask_test_and_set_cpu(cpu, xen_cpu_initialized_map))
f87e4cac JF	275	return 0;
	276
	277	ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
	278	if (ctxt == NULL)
	279	return -ENOMEM;
	280
c7b75947 JF	281	gdt = get_cpu_gdt_table(cpu);
c7b75947 JF	282
f87e4cac JF	283	ctxt->flags = VGCF_IN_KERNEL;
	284	ctxt->user_regs.ds = __USER_DS;
	285	ctxt->user_regs.es = __USER_DS;
f87e4cac	286	ctxt->user_regs.ss = __KERNEL_DS;
c7b75947 JF	287	#ifdef CONFIG_X86_32
c7b75947 JF	288	ctxt->user_regs.fs = __KERNEL_PERCPU;
577eebea	289	ctxt->user_regs.gs = __KERNEL_STACK_CANARY;
795f99b6 JF	290	#else
795f99b6 JF	291	ctxt->gs_base_kernel = per_cpu_offset(cpu);
c7b75947	292	#endif
f87e4cac JF	293	ctxt->user_regs.eip = (unsigned long)cpu_bringup_and_idle;
	294	ctxt->user_regs.eflags = 0x1000; /* IOPL_RING1 */
	295
	296	memset(&ctxt->fpu_ctxt, 0, sizeof(ctxt->fpu_ctxt));
	297
	298	xen_copy_trap_info(ctxt->trap_ctxt);
	299
	300	ctxt->ldt_ents = 0;
	301
c7b75947	302	BUG_ON((unsigned long)gdt & ~PAGE_MASK);
9976b39b JF	303
9976b39b JF	304	gdt_mfn = arbitrary_virt_to_mfn(gdt);
c7b75947	305	make_lowmem_page_readonly(gdt);
9976b39b	306	make_lowmem_page_readonly(mfn_to_virt(gdt_mfn));
f87e4cac	307
9976b39b	308	ctxt->gdt_frames[0] = gdt_mfn;
c7b75947	309	ctxt->gdt_ents = GDT_ENTRIES;
f87e4cac JF	310
f87e4cac JF	311	ctxt->user_regs.cs = __KERNEL_CS;
faca6227	312	ctxt->user_regs.esp = idle->thread.sp0 - sizeof(struct pt_regs);
f87e4cac JF	313
f87e4cac JF	314	ctxt->kernel_ss = __KERNEL_DS;
faca6227	315	ctxt->kernel_sp = idle->thread.sp0;
f87e4cac	316
c7b75947	317	#ifdef CONFIG_X86_32
f87e4cac	318	ctxt->event_callback_cs = __KERNEL_CS;
f87e4cac	319	ctxt->failsafe_callback_cs = __KERNEL_CS;
c7b75947 JF	320	#endif
c7b75947 JF	321	ctxt->event_callback_eip = (unsigned long)xen_hypervisor_callback;
f87e4cac JF	322	ctxt->failsafe_callback_eip = (unsigned long)xen_failsafe_callback;
	323
	324	per_cpu(xen_cr3, cpu) = __pa(swapper_pg_dir);
	325	ctxt->ctrlreg[3] = xen_pfn_to_cr3(virt_to_mfn(swapper_pg_dir));
	326
	327	if (HYPERVISOR_vcpu_op(VCPUOP_initialise, cpu, ctxt))
	328	BUG();
	329
	330	kfree(ctxt);
	331	return 0;
	332	}
	333
5cdaf183	334	static int __cpuinit xen_cpu_up(unsigned int cpu, struct task_struct *tidle)
f87e4cac JF	335	{
	336	struct task_struct *idle = idle_task(cpu);
	337	int rc;
	338
c6f5e0ac	339	per_cpu(current_task, cpu) = idle;
c7b75947	340	#ifdef CONFIG_X86_32
f87e4cac	341	irq_ctx_init(cpu);
c7b75947	342	#else
c7b75947	343	clear_tsk_thread_flag(idle, TIF_FORK);
38341432 JF	344	per_cpu(kernel_stack, cpu) =
	345	(unsigned long)task_stack_page(idle) -
	346	KERNEL_STACK_OFFSET + THREAD_SIZE;
c7b75947	347	#endif
02889672	348	xen_setup_runstate_info(cpu);
f87e4cac	349	xen_setup_timer(cpu);
2d9e1e2f	350	xen_init_lock_cpu(cpu);
f87e4cac	351
c7b75947 JF	352	per_cpu(cpu_state, cpu) = CPU_UP_PREPARE;
c7b75947 JF	353
f87e4cac JF	354	/* make sure interrupts start blocked */
	355	per_cpu(xen_vcpu, cpu)->evtchn_upcall_mask = 1;
	356
	357	rc = cpu_initialize_context(cpu, idle);
	358	if (rc)
	359	return rc;
	360
	361	if (num_online_cpus() == 1)
	362	alternatives_smp_switch(1);
	363
	364	rc = xen_smp_intr_init(cpu);
	365	if (rc)
	366	return rc;
	367
f87e4cac JF	368	rc = HYPERVISOR_vcpu_op(VCPUOP_up, cpu, NULL);
	369	BUG_ON(rc);
	370
c7b75947	371	while(per_cpu(cpu_state, cpu) != CPU_ONLINE) {
1207cf8e	372	HYPERVISOR_sched_op(SCHEDOP_yield, NULL);
c7b75947 JF	373	barrier();
	374	}
	375
f87e4cac JF	376	return 0;
	377	}
	378
a9e7062d	379	static void xen_smp_cpus_done(unsigned int max_cpus)
f87e4cac JF	380	{
	381	}
	382
2737146b	383	#ifdef CONFIG_HOTPLUG_CPU
26fd1051	384	static int xen_cpu_disable(void)
d68d82af AN	385	{
	386	unsigned int cpu = smp_processor_id();
	387	if (cpu == 0)
	388	return -EBUSY;
	389
	390	cpu_disable_common();
	391
	392	load_cr3(swapper_pg_dir);
	393	return 0;
	394	}
	395
26fd1051	396	static void xen_cpu_die(unsigned int cpu)
d68d82af AN	397	{
	398	while (HYPERVISOR_vcpu_op(VCPUOP_is_up, cpu, NULL)) {
	399	current->state = TASK_UNINTERRUPTIBLE;
	400	schedule_timeout(HZ/10);
	401	}
c6e22f9e TH	402	unbind_from_irqhandler(per_cpu(xen_resched_irq, cpu), NULL);
	403	unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
	404	unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
	405	unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu), NULL);
d68d82af AN	406	xen_uninit_lock_cpu(cpu);
	407	xen_teardown_timer(cpu);
	408
	409	if (num_online_cpus() == 1)
	410	alternatives_smp_switch(0);
	411	}
	412
71709247	413	static void __cpuinit xen_play_dead(void) /* used only with HOTPLUG_CPU */
d68d82af AN	414	{
	415	play_dead_common();
	416	HYPERVISOR_vcpu_op(VCPUOP_down, smp_processor_id(), NULL);
	417	cpu_bringup();
41bd956d KRW	418	/*
	419	* Balance out the preempt calls - as we are running in cpu_idle
	420	* loop which has been called at bootup from cpu_bringup_and_idle.
	421	* The cpucpu_bringup_and_idle called cpu_bringup which made a
	422	* preempt_disable() So this preempt_enable will balance it out.
	423	*/
	424	preempt_enable();
d68d82af AN	425	}
d68d82af AN	426
2737146b	427	#else /* !CONFIG_HOTPLUG_CPU */
26fd1051	428	static int xen_cpu_disable(void)
2737146b AN	429	{
	430	return -ENOSYS;
	431	}
	432
26fd1051	433	static void xen_cpu_die(unsigned int cpu)
2737146b AN	434	{
	435	BUG();
	436	}
	437
26fd1051	438	static void xen_play_dead(void)
2737146b AN	439	{
	440	BUG();
	441	}
	442
	443	#endif
f87e4cac JF	444	static void stop_self(void *v)
	445	{
	446	int cpu = smp_processor_id();
	447
	448	/* make sure we're not pinning something down */
	449	load_cr3(swapper_pg_dir);
	450	/* should set up a minimal gdt */
	451
086748e5 IC	452	set_cpu_online(cpu, false);
086748e5 IC	453
f87e4cac JF	454	HYPERVISOR_vcpu_op(VCPUOP_down, cpu, NULL);
	455	BUG();
	456	}
	457
76fac077	458	static void xen_stop_other_cpus(int wait)
f87e4cac	459	{
76fac077	460	smp_call_function(stop_self, NULL, wait);
f87e4cac JF	461	}
f87e4cac JF	462
a9e7062d	463	static void xen_smp_send_reschedule(int cpu)
f87e4cac JF	464	{
	465	xen_send_IPI_one(cpu, XEN_RESCHEDULE_VECTOR);
	466	}
	467
bcda016e MT	468	static void xen_send_IPI_mask(const struct cpumask *mask,
bcda016e MT	469	enum ipi_vector vector)
f87e4cac JF	470	{
	471	unsigned cpu;
	472
bcda016e	473	for_each_cpu_and(cpu, mask, cpu_online_mask)
f87e4cac JF	474	xen_send_IPI_one(cpu, vector);
	475	}
	476
bcda016e	477	static void xen_smp_send_call_function_ipi(const struct cpumask *mask)
3b16cf87 JA	478	{
	479	int cpu;
	480
	481	xen_send_IPI_mask(mask, XEN_CALL_FUNCTION_VECTOR);
	482
	483	/* Make sure other vcpus get a chance to run if they need to. */
bcda016e	484	for_each_cpu(cpu, mask) {
3b16cf87	485	if (xen_vcpu_stolen(cpu)) {
1207cf8e	486	HYPERVISOR_sched_op(SCHEDOP_yield, NULL);
3b16cf87 JA	487	break;
	488	}
	489	}
	490	}
	491
a9e7062d	492	static void xen_smp_send_call_function_single_ipi(int cpu)
3b16cf87	493	{
bcda016e	494	xen_send_IPI_mask(cpumask_of(cpu),
e7986739	495	XEN_CALL_FUNCTION_SINGLE_VECTOR);
3b16cf87 JA	496	}
3b16cf87 JA	497
f87e4cac JF	498	static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id)
f87e4cac JF	499	{
f87e4cac	500	irq_enter();
3b16cf87	501	generic_smp_call_function_interrupt();
1b437c8c	502	inc_irq_stat(irq_call_count);
f87e4cac JF	503	irq_exit();
f87e4cac JF	504
f87e4cac JF	505	return IRQ_HANDLED;
	506	}
	507
3b16cf87	508	static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id)
f87e4cac	509	{
3b16cf87 JA	510	irq_enter();
3b16cf87 JA	511	generic_smp_call_function_single_interrupt();
1b437c8c	512	inc_irq_stat(irq_call_count);
3b16cf87	513	irq_exit();
f87e4cac	514
3b16cf87	515	return IRQ_HANDLED;
f87e4cac	516	}
a9e7062d	517
b53cedeb	518	static const struct smp_ops xen_smp_ops __initconst = {
a9e7062d JF	519	.smp_prepare_boot_cpu = xen_smp_prepare_boot_cpu,
a9e7062d JF	520	.smp_prepare_cpus = xen_smp_prepare_cpus,
a9e7062d JF	521	.smp_cpus_done = xen_smp_cpus_done,
a9e7062d JF	522
d68d82af AN	523	.cpu_up = xen_cpu_up,
	524	.cpu_die = xen_cpu_die,
	525	.cpu_disable = xen_cpu_disable,
	526	.play_dead = xen_play_dead,
	527
76fac077	528	.stop_other_cpus = xen_stop_other_cpus,
a9e7062d JF	529	.smp_send_reschedule = xen_smp_send_reschedule,
	530
	531	.send_call_func_ipi = xen_smp_send_call_function_ipi,
	532	.send_call_func_single_ipi = xen_smp_send_call_function_single_ipi,
	533	};
	534
	535	void __init xen_smp_init(void)
	536	{
	537	smp_ops = xen_smp_ops;
c7b75947	538	xen_fill_possible_map();
2d9e1e2f	539	xen_init_spinlocks();
a9e7062d	540	}
99bbb3a8 SS	541
	542	static void __init xen_hvm_smp_prepare_cpus(unsigned int max_cpus)
	543	{
	544	native_smp_prepare_cpus(max_cpus);
	545	WARN_ON(xen_smp_intr_init(0));
	546
99bbb3a8	547	xen_init_lock_cpu(0);
99bbb3a8 SS	548	}
99bbb3a8 SS	549
5cdaf183	550	static int __cpuinit xen_hvm_cpu_up(unsigned int cpu, struct task_struct *tidle)
99bbb3a8 SS	551	{
99bbb3a8 SS	552	int rc;
5cdaf183	553	rc = native_cpu_up(cpu, tidle);
99bbb3a8 SS	554	WARN_ON (xen_smp_intr_init(cpu));
	555	return rc;
	556	}
	557
	558	static void xen_hvm_cpu_die(unsigned int cpu)
	559	{
	560	unbind_from_irqhandler(per_cpu(xen_resched_irq, cpu), NULL);
	561	unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
	562	unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
	563	unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu), NULL);
	564	native_cpu_die(cpu);
	565	}
	566
	567	void __init xen_hvm_smp_init(void)
	568	{
3c05c4be SS	569	if (!xen_have_vector_callback)
3c05c4be SS	570	return;
99bbb3a8 SS	571	smp_ops.smp_prepare_cpus = xen_hvm_smp_prepare_cpus;
	572	smp_ops.smp_send_reschedule = xen_smp_send_reschedule;
	573	smp_ops.cpu_up = xen_hvm_cpu_up;
	574	smp_ops.cpu_die = xen_hvm_cpu_die;
	575	smp_ops.send_call_func_ipi = xen_smp_send_call_function_ipi;
	576	smp_ops.send_call_func_single_ipi = xen_smp_send_call_function_single_ipi;
	577	}