[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-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. Nothing to do,
 * all the work is done automatically when
 * we return from the interrupt.
 */
static irqreturn_t xen_reschedule_interrupt(int irq, void *dev_id)
{
	inc_irq_stat(irq_resched_count);

	return IRQ_HANDLED;
}

static __cpuinit void cpu_bringup(void)
{
	int cpu = smp_processor_id();

	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();

	set_cpu_online(cpu, true);
	percpu_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 __cpuinit void 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;

	xen_init_lock_cpu(0);

	smp_store_cpu_info(0);
	cpu_data(0).x86_max_cores = 1;
	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 __cpuinit int
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 *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();
}

#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 __initdata = {
	.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));

	if (!xen_have_vector_callback)
		return;
	xen_init_lock_cpu(0);
	xen_init_spinlocks();
}

static int __cpuinit xen_hvm_cpu_up(unsigned int cpu)
{
	int rc;
	rc = native_cpu_up(cpu);
	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)
{
	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
	35	#include "xen-ops.h"
	36	#include "mmu.h"
	37
b78936e1	38	cpumask_var_t xen_cpu_initialized_map;
f87e4cac	39
c6e22f9e TH	40	static DEFINE_PER_CPU(int, xen_resched_irq);
	41	static DEFINE_PER_CPU(int, xen_callfunc_irq);
	42	static DEFINE_PER_CPU(int, xen_callfuncsingle_irq);
	43	static DEFINE_PER_CPU(int, xen_debug_irq) = -1;
f87e4cac JF	44
f87e4cac JF	45	static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id);
3b16cf87	46	static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id);
f87e4cac JF	47
	48	/*
	49	* Reschedule call back. Nothing to do,
	50	* all the work is done automatically when
	51	* we return from the interrupt.
	52	*/
	53	static irqreturn_t xen_reschedule_interrupt(int irq, void *dev_id)
	54	{
1b437c8c	55	inc_irq_stat(irq_resched_count);
38bb5ab4	56
f87e4cac JF	57	return IRQ_HANDLED;
	58	}
	59
d68d82af	60	static __cpuinit void cpu_bringup(void)
f87e4cac JF	61	{
	62	int cpu = smp_processor_id();
	63
	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
d7d3756c	78	set_cpu_online(cpu, true);
6dbde353	79	percpu_write(cpu_state, CPU_ONLINE);
c7b75947 JF	80	wmb();
c7b75947 JF	81
f87e4cac JF	82	/* We can take interrupts now: we're officially "up". */
	83	local_irq_enable();
	84
	85	wmb(); /* make sure everything is out */
d68d82af AN	86	}
	87
	88	static __cpuinit void cpu_bringup_and_idle(void)
	89	{
	90	cpu_bringup();
f87e4cac JF	91	cpu_idle();
	92	}
	93
	94	static int xen_smp_intr_init(unsigned int cpu)
	95	{
	96	int rc;
ee523ca1	97	const char resched_name, callfunc_name, *debug_name;
f87e4cac JF	98
	99	resched_name = kasprintf(GFP_KERNEL, "resched%d", cpu);
	100	rc = bind_ipi_to_irqhandler(XEN_RESCHEDULE_VECTOR,
	101	cpu,
	102	xen_reschedule_interrupt,
	103	IRQF_DISABLED\|IRQF_PERCPU\|IRQF_NOBALANCING,
	104	resched_name,
	105	NULL);
	106	if (rc < 0)
	107	goto fail;
c6e22f9e	108	per_cpu(xen_resched_irq, cpu) = rc;
f87e4cac JF	109
	110	callfunc_name = kasprintf(GFP_KERNEL, "callfunc%d", cpu);
	111	rc = bind_ipi_to_irqhandler(XEN_CALL_FUNCTION_VECTOR,
	112	cpu,
	113	xen_call_function_interrupt,
	114	IRQF_DISABLED\|IRQF_PERCPU\|IRQF_NOBALANCING,
	115	callfunc_name,
	116	NULL);
	117	if (rc < 0)
	118	goto fail;
c6e22f9e	119	per_cpu(xen_callfunc_irq, cpu) = rc;
f87e4cac	120
ee523ca1 JF	121	debug_name = kasprintf(GFP_KERNEL, "debug%d", cpu);
	122	rc = bind_virq_to_irqhandler(VIRQ_DEBUG, cpu, xen_debug_interrupt,
	123	IRQF_DISABLED \| IRQF_PERCPU \| IRQF_NOBALANCING,
	124	debug_name, NULL);
	125	if (rc < 0)
	126	goto fail;
c6e22f9e	127	per_cpu(xen_debug_irq, cpu) = rc;
ee523ca1	128
3b16cf87 JA	129	callfunc_name = kasprintf(GFP_KERNEL, "callfuncsingle%d", cpu);
	130	rc = bind_ipi_to_irqhandler(XEN_CALL_FUNCTION_SINGLE_VECTOR,
	131	cpu,
	132	xen_call_function_single_interrupt,
	133	IRQF_DISABLED\|IRQF_PERCPU\|IRQF_NOBALANCING,
	134	callfunc_name,
	135	NULL);
	136	if (rc < 0)
	137	goto fail;
c6e22f9e	138	per_cpu(xen_callfuncsingle_irq, cpu) = rc;
3b16cf87	139
f87e4cac JF	140	return 0;
	141
	142	fail:
c6e22f9e TH	143	if (per_cpu(xen_resched_irq, cpu) >= 0)
	144	unbind_from_irqhandler(per_cpu(xen_resched_irq, cpu), NULL);
	145	if (per_cpu(xen_callfunc_irq, cpu) >= 0)
	146	unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
	147	if (per_cpu(xen_debug_irq, cpu) >= 0)
	148	unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
	149	if (per_cpu(xen_callfuncsingle_irq, cpu) >= 0)
	150	unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu),
	151	NULL);
3b16cf87	152
f87e4cac JF	153	return rc;
	154	}
	155
c7b75947	156	static void __init xen_fill_possible_map(void)
f87e4cac JF	157	{
	158	int i, rc;
	159
ea5b8f73 SS	160	if (xen_initial_domain())
	161	return;
	162
	163	for (i = 0; i < nr_cpu_ids; i++) {
	164	rc = HYPERVISOR_vcpu_op(VCPUOP_is_up, i, NULL);
	165	if (rc >= 0) {
	166	num_processors++;
	167	set_cpu_possible(i, true);
	168	}
	169	}
	170	}
	171
	172	static void __init xen_filter_cpu_maps(void)
	173	{
	174	int i, rc;
	175
	176	if (!xen_initial_domain())
	177	return;
	178
801fd14a SS	179	num_processors = 0;
801fd14a SS	180	disabled_cpus = 0;
e7986739	181	for (i = 0; i < nr_cpu_ids; i++) {
f87e4cac	182	rc = HYPERVISOR_vcpu_op(VCPUOP_is_up, i, NULL);
4560a294 JF	183	if (rc >= 0) {
4560a294 JF	184	num_processors++;
4f062896	185	set_cpu_possible(i, true);
801fd14a SS	186	} else {
	187	set_cpu_possible(i, false);
	188	set_cpu_present(i, false);
4560a294	189	}
f87e4cac JF	190	}
	191	}
	192
a9e7062d	193	static void __init xen_smp_prepare_boot_cpu(void)
f87e4cac	194	{
f87e4cac JF	195	BUG_ON(smp_processor_id() != 0);
	196	native_smp_prepare_boot_cpu();
	197
f87e4cac JF	198	/* We've switched to the "real" per-cpu gdt, so make sure the
f87e4cac JF	199	old memory can be recycled */
38341432	200	make_lowmem_page_readwrite(xen_initial_gdt);
60223a32	201
ea5b8f73	202	xen_filter_cpu_maps();
60223a32	203	xen_setup_vcpu_info_placement();
f87e4cac JF	204	}
f87e4cac JF	205
a9e7062d	206	static void __init xen_smp_prepare_cpus(unsigned int max_cpus)
f87e4cac JF	207	{
	208	unsigned cpu;
	209
2d9e1e2f JF	210	xen_init_lock_cpu(0);
2d9e1e2f JF	211
f87e4cac	212	smp_store_cpu_info(0);
c7b75947	213	cpu_data(0).x86_max_cores = 1;
f87e4cac JF	214	set_cpu_sibling_map(0);
	215
	216	if (xen_smp_intr_init(0))
	217	BUG();
	218
b78936e1 MT	219	if (!alloc_cpumask_var(&xen_cpu_initialized_map, GFP_KERNEL))
	220	panic("could not allocate xen_cpu_initialized_map\n");
	221
	222	cpumask_copy(xen_cpu_initialized_map, cpumask_of(0));
f87e4cac JF	223
	224	/* Restrict the possible_map according to max_cpus. */
	225	while ((num_possible_cpus() > 1) && (num_possible_cpus() > max_cpus)) {
e7986739	226	for (cpu = nr_cpu_ids - 1; !cpu_possible(cpu); cpu--)
f87e4cac	227	continue;
4f062896	228	set_cpu_possible(cpu, false);
f87e4cac JF	229	}
	230
	231	for_each_possible_cpu (cpu) {
	232	struct task_struct *idle;
	233
	234	if (cpu == 0)
	235	continue;
	236
	237	idle = fork_idle(cpu);
	238	if (IS_ERR(idle))
	239	panic("failed fork for CPU %d", cpu);
	240
4f062896	241	set_cpu_present(cpu, true);
f87e4cac	242	}
f87e4cac JF	243	}
	244
	245	static __cpuinit int
	246	cpu_initialize_context(unsigned int cpu, struct task_struct *idle)
	247	{
	248	struct vcpu_guest_context *ctxt;
c7b75947	249	struct desc_struct *gdt;
9976b39b	250	unsigned long gdt_mfn;
f87e4cac	251
b78936e1	252	if (cpumask_test_and_set_cpu(cpu, xen_cpu_initialized_map))
f87e4cac JF	253	return 0;
	254
	255	ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
	256	if (ctxt == NULL)
	257	return -ENOMEM;
	258
c7b75947 JF	259	gdt = get_cpu_gdt_table(cpu);
c7b75947 JF	260
f87e4cac JF	261	ctxt->flags = VGCF_IN_KERNEL;
	262	ctxt->user_regs.ds = __USER_DS;
	263	ctxt->user_regs.es = __USER_DS;
f87e4cac	264	ctxt->user_regs.ss = __KERNEL_DS;
c7b75947 JF	265	#ifdef CONFIG_X86_32
c7b75947 JF	266	ctxt->user_regs.fs = __KERNEL_PERCPU;
577eebea	267	ctxt->user_regs.gs = __KERNEL_STACK_CANARY;
795f99b6 JF	268	#else
795f99b6 JF	269	ctxt->gs_base_kernel = per_cpu_offset(cpu);
c7b75947	270	#endif
f87e4cac JF	271	ctxt->user_regs.eip = (unsigned long)cpu_bringup_and_idle;
	272	ctxt->user_regs.eflags = 0x1000; /* IOPL_RING1 */
	273
	274	memset(&ctxt->fpu_ctxt, 0, sizeof(ctxt->fpu_ctxt));
	275
	276	xen_copy_trap_info(ctxt->trap_ctxt);
	277
	278	ctxt->ldt_ents = 0;
	279
c7b75947	280	BUG_ON((unsigned long)gdt & ~PAGE_MASK);
9976b39b JF	281
9976b39b JF	282	gdt_mfn = arbitrary_virt_to_mfn(gdt);
c7b75947	283	make_lowmem_page_readonly(gdt);
9976b39b	284	make_lowmem_page_readonly(mfn_to_virt(gdt_mfn));
f87e4cac	285
9976b39b	286	ctxt->gdt_frames[0] = gdt_mfn;
c7b75947	287	ctxt->gdt_ents = GDT_ENTRIES;
f87e4cac JF	288
f87e4cac JF	289	ctxt->user_regs.cs = __KERNEL_CS;
faca6227	290	ctxt->user_regs.esp = idle->thread.sp0 - sizeof(struct pt_regs);
f87e4cac JF	291
f87e4cac JF	292	ctxt->kernel_ss = __KERNEL_DS;
faca6227	293	ctxt->kernel_sp = idle->thread.sp0;
f87e4cac	294
c7b75947	295	#ifdef CONFIG_X86_32
f87e4cac	296	ctxt->event_callback_cs = __KERNEL_CS;
f87e4cac	297	ctxt->failsafe_callback_cs = __KERNEL_CS;
c7b75947 JF	298	#endif
c7b75947 JF	299	ctxt->event_callback_eip = (unsigned long)xen_hypervisor_callback;
f87e4cac JF	300	ctxt->failsafe_callback_eip = (unsigned long)xen_failsafe_callback;
	301
	302	per_cpu(xen_cr3, cpu) = __pa(swapper_pg_dir);
	303	ctxt->ctrlreg[3] = xen_pfn_to_cr3(virt_to_mfn(swapper_pg_dir));
	304
	305	if (HYPERVISOR_vcpu_op(VCPUOP_initialise, cpu, ctxt))
	306	BUG();
	307
	308	kfree(ctxt);
	309	return 0;
	310	}
	311
a9e7062d	312	static int __cpuinit xen_cpu_up(unsigned int cpu)
f87e4cac JF	313	{
	314	struct task_struct *idle = idle_task(cpu);
	315	int rc;
	316
c6f5e0ac	317	per_cpu(current_task, cpu) = idle;
c7b75947	318	#ifdef CONFIG_X86_32
f87e4cac	319	irq_ctx_init(cpu);
c7b75947	320	#else
c7b75947	321	clear_tsk_thread_flag(idle, TIF_FORK);
38341432 JF	322	per_cpu(kernel_stack, cpu) =
	323	(unsigned long)task_stack_page(idle) -
	324	KERNEL_STACK_OFFSET + THREAD_SIZE;
c7b75947	325	#endif
02889672	326	xen_setup_runstate_info(cpu);
f87e4cac	327	xen_setup_timer(cpu);
2d9e1e2f	328	xen_init_lock_cpu(cpu);
f87e4cac	329
c7b75947 JF	330	per_cpu(cpu_state, cpu) = CPU_UP_PREPARE;
c7b75947 JF	331
f87e4cac JF	332	/* make sure interrupts start blocked */
	333	per_cpu(xen_vcpu, cpu)->evtchn_upcall_mask = 1;
	334
	335	rc = cpu_initialize_context(cpu, idle);
	336	if (rc)
	337	return rc;
	338
	339	if (num_online_cpus() == 1)
	340	alternatives_smp_switch(1);
	341
	342	rc = xen_smp_intr_init(cpu);
	343	if (rc)
	344	return rc;
	345
f87e4cac JF	346	rc = HYPERVISOR_vcpu_op(VCPUOP_up, cpu, NULL);
	347	BUG_ON(rc);
	348
c7b75947	349	while(per_cpu(cpu_state, cpu) != CPU_ONLINE) {
1207cf8e	350	HYPERVISOR_sched_op(SCHEDOP_yield, NULL);
c7b75947 JF	351	barrier();
	352	}
	353
f87e4cac JF	354	return 0;
	355	}
	356
a9e7062d	357	static void xen_smp_cpus_done(unsigned int max_cpus)
f87e4cac JF	358	{
	359	}
	360
2737146b	361	#ifdef CONFIG_HOTPLUG_CPU
26fd1051	362	static int xen_cpu_disable(void)
d68d82af AN	363	{
	364	unsigned int cpu = smp_processor_id();
	365	if (cpu == 0)
	366	return -EBUSY;
	367
	368	cpu_disable_common();
	369
	370	load_cr3(swapper_pg_dir);
	371	return 0;
	372	}
	373
26fd1051	374	static void xen_cpu_die(unsigned int cpu)
d68d82af AN	375	{
	376	while (HYPERVISOR_vcpu_op(VCPUOP_is_up, cpu, NULL)) {
	377	current->state = TASK_UNINTERRUPTIBLE;
	378	schedule_timeout(HZ/10);
	379	}
c6e22f9e TH	380	unbind_from_irqhandler(per_cpu(xen_resched_irq, cpu), NULL);
	381	unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
	382	unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
	383	unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu), NULL);
d68d82af AN	384	xen_uninit_lock_cpu(cpu);
	385	xen_teardown_timer(cpu);
	386
	387	if (num_online_cpus() == 1)
	388	alternatives_smp_switch(0);
	389	}
	390
71709247	391	static void __cpuinit xen_play_dead(void) /* used only with HOTPLUG_CPU */
d68d82af AN	392	{
	393	play_dead_common();
	394	HYPERVISOR_vcpu_op(VCPUOP_down, smp_processor_id(), NULL);
	395	cpu_bringup();
	396	}
	397
2737146b	398	#else /* !CONFIG_HOTPLUG_CPU */
26fd1051	399	static int xen_cpu_disable(void)
2737146b AN	400	{
	401	return -ENOSYS;
	402	}
	403
26fd1051	404	static void xen_cpu_die(unsigned int cpu)
2737146b AN	405	{
	406	BUG();
	407	}
	408
26fd1051	409	static void xen_play_dead(void)
2737146b AN	410	{
	411	BUG();
	412	}
	413
	414	#endif
f87e4cac JF	415	static void stop_self(void *v)
	416	{
	417	int cpu = smp_processor_id();
	418
	419	/* make sure we're not pinning something down */
	420	load_cr3(swapper_pg_dir);
	421	/* should set up a minimal gdt */
	422
086748e5 IC	423	set_cpu_online(cpu, false);
086748e5 IC	424
f87e4cac JF	425	HYPERVISOR_vcpu_op(VCPUOP_down, cpu, NULL);
	426	BUG();
	427	}
	428
76fac077	429	static void xen_stop_other_cpus(int wait)
f87e4cac	430	{
76fac077	431	smp_call_function(stop_self, NULL, wait);
f87e4cac JF	432	}
f87e4cac JF	433
a9e7062d	434	static void xen_smp_send_reschedule(int cpu)
f87e4cac JF	435	{
	436	xen_send_IPI_one(cpu, XEN_RESCHEDULE_VECTOR);
	437	}
	438
bcda016e MT	439	static void xen_send_IPI_mask(const struct cpumask *mask,
bcda016e MT	440	enum ipi_vector vector)
f87e4cac JF	441	{
	442	unsigned cpu;
	443
bcda016e	444	for_each_cpu_and(cpu, mask, cpu_online_mask)
f87e4cac JF	445	xen_send_IPI_one(cpu, vector);
	446	}
	447
bcda016e	448	static void xen_smp_send_call_function_ipi(const struct cpumask *mask)
3b16cf87 JA	449	{
	450	int cpu;
	451
	452	xen_send_IPI_mask(mask, XEN_CALL_FUNCTION_VECTOR);
	453
	454	/* Make sure other vcpus get a chance to run if they need to. */
bcda016e	455	for_each_cpu(cpu, mask) {
3b16cf87	456	if (xen_vcpu_stolen(cpu)) {
1207cf8e	457	HYPERVISOR_sched_op(SCHEDOP_yield, NULL);
3b16cf87 JA	458	break;
	459	}
	460	}
	461	}
	462
a9e7062d	463	static void xen_smp_send_call_function_single_ipi(int cpu)
3b16cf87	464	{
bcda016e	465	xen_send_IPI_mask(cpumask_of(cpu),
e7986739	466	XEN_CALL_FUNCTION_SINGLE_VECTOR);
3b16cf87 JA	467	}
3b16cf87 JA	468
f87e4cac JF	469	static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id)
f87e4cac JF	470	{
f87e4cac	471	irq_enter();
3b16cf87	472	generic_smp_call_function_interrupt();
1b437c8c	473	inc_irq_stat(irq_call_count);
f87e4cac JF	474	irq_exit();
f87e4cac JF	475
f87e4cac JF	476	return IRQ_HANDLED;
	477	}
	478
3b16cf87	479	static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id)
f87e4cac	480	{
3b16cf87 JA	481	irq_enter();
3b16cf87 JA	482	generic_smp_call_function_single_interrupt();
1b437c8c	483	inc_irq_stat(irq_call_count);
3b16cf87	484	irq_exit();
f87e4cac	485
3b16cf87	486	return IRQ_HANDLED;
f87e4cac	487	}
a9e7062d JF	488
	489	static const struct smp_ops xen_smp_ops __initdata = {
	490	.smp_prepare_boot_cpu = xen_smp_prepare_boot_cpu,
	491	.smp_prepare_cpus = xen_smp_prepare_cpus,
a9e7062d JF	492	.smp_cpus_done = xen_smp_cpus_done,
a9e7062d JF	493
d68d82af AN	494	.cpu_up = xen_cpu_up,
	495	.cpu_die = xen_cpu_die,
	496	.cpu_disable = xen_cpu_disable,
	497	.play_dead = xen_play_dead,
	498
76fac077	499	.stop_other_cpus = xen_stop_other_cpus,
a9e7062d JF	500	.smp_send_reschedule = xen_smp_send_reschedule,
	501
	502	.send_call_func_ipi = xen_smp_send_call_function_ipi,
	503	.send_call_func_single_ipi = xen_smp_send_call_function_single_ipi,
	504	};
	505
	506	void __init xen_smp_init(void)
	507	{
	508	smp_ops = xen_smp_ops;
c7b75947	509	xen_fill_possible_map();
2d9e1e2f	510	xen_init_spinlocks();
a9e7062d	511	}
99bbb3a8 SS	512
	513	static void __init xen_hvm_smp_prepare_cpus(unsigned int max_cpus)
	514	{
	515	native_smp_prepare_cpus(max_cpus);
	516	WARN_ON(xen_smp_intr_init(0));
	517
	518	if (!xen_have_vector_callback)
	519	return;
	520	xen_init_lock_cpu(0);
	521	xen_init_spinlocks();
	522	}
	523
	524	static int __cpuinit xen_hvm_cpu_up(unsigned int cpu)
	525	{
	526	int rc;
	527	rc = native_cpu_up(cpu);
	528	WARN_ON (xen_smp_intr_init(cpu));
	529	return rc;
	530	}
	531
	532	static void xen_hvm_cpu_die(unsigned int cpu)
	533	{
	534	unbind_from_irqhandler(per_cpu(xen_resched_irq, cpu), NULL);
	535	unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
	536	unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
	537	unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu), NULL);
	538	native_cpu_die(cpu);
	539	}
	540
	541	void __init xen_hvm_smp_init(void)
	542	{
	543	smp_ops.smp_prepare_cpus = xen_hvm_smp_prepare_cpus;
	544	smp_ops.smp_send_reschedule = xen_smp_send_reschedule;
	545	smp_ops.cpu_up = xen_hvm_cpu_up;
	546	smp_ops.cpu_die = xen_hvm_cpu_die;
	547	smp_ops.send_call_func_ipi = xen_smp_send_call_function_ipi;
	548	smp_ops.send_call_func_single_ipi = xen_smp_send_call_function_single_ipi;
	549	}