[mirror_ubuntu-bionic-kernel.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/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;

	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_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_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 */

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

static void xen_smp_send_stop(void)
{
	smp_call_function(stop_self, NULL, 0);
}

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,

	.smp_send_stop = xen_smp_send_stop,
	.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();
}
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
	31	#include <xen/page.h>
	32	#include <xen/events.h>
	33
	34	#include "xen-ops.h"
	35	#include "mmu.h"
	36
b78936e1	37	cpumask_var_t xen_cpu_initialized_map;
f87e4cac	38
c6e22f9e TH	39	static DEFINE_PER_CPU(int, xen_resched_irq);
	40	static DEFINE_PER_CPU(int, xen_callfunc_irq);
	41	static DEFINE_PER_CPU(int, xen_callfuncsingle_irq);
	42	static DEFINE_PER_CPU(int, xen_debug_irq) = -1;
f87e4cac JF	43
f87e4cac JF	44	static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id);
3b16cf87	45	static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id);
f87e4cac JF	46
	47	/*
	48	* Reschedule call back. Nothing to do,
	49	* all the work is done automatically when
	50	* we return from the interrupt.
	51	*/
	52	static irqreturn_t xen_reschedule_interrupt(int irq, void *dev_id)
	53	{
1b437c8c	54	inc_irq_stat(irq_resched_count);
38bb5ab4	55
f87e4cac JF	56	return IRQ_HANDLED;
	57	}
	58
d68d82af	59	static __cpuinit void cpu_bringup(void)
f87e4cac JF	60	{
	61	int cpu = smp_processor_id();
	62
	63	cpu_init();
d68d82af	64	touch_softlockup_watchdog();
c7b75947 JF	65	preempt_disable();
c7b75947 JF	66
e2a81baf	67	xen_enable_sysenter();
6fcac6d3	68	xen_enable_syscall();
f87e4cac	69
c7b75947 JF	70	cpu = smp_processor_id();
	71	smp_store_cpu_info(cpu);
	72	cpu_data(cpu).x86_max_cores = 1;
	73	set_cpu_sibling_map(cpu);
f87e4cac JF	74
	75	xen_setup_cpu_clockevents();
	76
d7d3756c	77	set_cpu_online(cpu, true);
6dbde353	78	percpu_write(cpu_state, CPU_ONLINE);
c7b75947 JF	79	wmb();
c7b75947 JF	80
f87e4cac JF	81	/* We can take interrupts now: we're officially "up". */
	82	local_irq_enable();
	83
	84	wmb(); /* make sure everything is out */
d68d82af AN	85	}
	86
	87	static __cpuinit void cpu_bringup_and_idle(void)
	88	{
	89	cpu_bringup();
f87e4cac JF	90	cpu_idle();
	91	}
	92
	93	static int xen_smp_intr_init(unsigned int cpu)
	94	{
	95	int rc;
ee523ca1	96	const char resched_name, callfunc_name, *debug_name;
f87e4cac JF	97
	98	resched_name = kasprintf(GFP_KERNEL, "resched%d", cpu);
	99	rc = bind_ipi_to_irqhandler(XEN_RESCHEDULE_VECTOR,
	100	cpu,
	101	xen_reschedule_interrupt,
	102	IRQF_DISABLED\|IRQF_PERCPU\|IRQF_NOBALANCING,
	103	resched_name,
	104	NULL);
	105	if (rc < 0)
	106	goto fail;
c6e22f9e	107	per_cpu(xen_resched_irq, cpu) = rc;
f87e4cac JF	108
	109	callfunc_name = kasprintf(GFP_KERNEL, "callfunc%d", cpu);
	110	rc = bind_ipi_to_irqhandler(XEN_CALL_FUNCTION_VECTOR,
	111	cpu,
	112	xen_call_function_interrupt,
	113	IRQF_DISABLED\|IRQF_PERCPU\|IRQF_NOBALANCING,
	114	callfunc_name,
	115	NULL);
	116	if (rc < 0)
	117	goto fail;
c6e22f9e	118	per_cpu(xen_callfunc_irq, cpu) = rc;
f87e4cac	119
ee523ca1 JF	120	debug_name = kasprintf(GFP_KERNEL, "debug%d", cpu);
	121	rc = bind_virq_to_irqhandler(VIRQ_DEBUG, cpu, xen_debug_interrupt,
	122	IRQF_DISABLED \| IRQF_PERCPU \| IRQF_NOBALANCING,
	123	debug_name, NULL);
	124	if (rc < 0)
	125	goto fail;
c6e22f9e	126	per_cpu(xen_debug_irq, cpu) = rc;
ee523ca1	127
3b16cf87 JA	128	callfunc_name = kasprintf(GFP_KERNEL, "callfuncsingle%d", cpu);
	129	rc = bind_ipi_to_irqhandler(XEN_CALL_FUNCTION_SINGLE_VECTOR,
	130	cpu,
	131	xen_call_function_single_interrupt,
	132	IRQF_DISABLED\|IRQF_PERCPU\|IRQF_NOBALANCING,
	133	callfunc_name,
	134	NULL);
	135	if (rc < 0)
	136	goto fail;
c6e22f9e	137	per_cpu(xen_callfuncsingle_irq, cpu) = rc;
3b16cf87	138
f87e4cac JF	139	return 0;
	140
	141	fail:
c6e22f9e TH	142	if (per_cpu(xen_resched_irq, cpu) >= 0)
	143	unbind_from_irqhandler(per_cpu(xen_resched_irq, cpu), NULL);
	144	if (per_cpu(xen_callfunc_irq, cpu) >= 0)
	145	unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
	146	if (per_cpu(xen_debug_irq, cpu) >= 0)
	147	unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
	148	if (per_cpu(xen_callfuncsingle_irq, cpu) >= 0)
	149	unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu),
	150	NULL);
3b16cf87	151
f87e4cac JF	152	return rc;
	153	}
	154
c7b75947	155	static void __init xen_fill_possible_map(void)
f87e4cac JF	156	{
	157	int i, rc;
	158
e7986739	159	for (i = 0; i < nr_cpu_ids; i++) {
f87e4cac	160	rc = HYPERVISOR_vcpu_op(VCPUOP_is_up, i, NULL);
4560a294 JF	161	if (rc >= 0) {
4560a294 JF	162	num_processors++;
4f062896	163	set_cpu_possible(i, true);
4560a294	164	}
f87e4cac JF	165	}
	166	}
	167
a9e7062d	168	static void __init xen_smp_prepare_boot_cpu(void)
f87e4cac	169	{
f87e4cac JF	170	BUG_ON(smp_processor_id() != 0);
	171	native_smp_prepare_boot_cpu();
	172
f87e4cac JF	173	/* We've switched to the "real" per-cpu gdt, so make sure the
f87e4cac JF	174	old memory can be recycled */
38341432	175	make_lowmem_page_readwrite(xen_initial_gdt);
60223a32 JF	176
60223a32 JF	177	xen_setup_vcpu_info_placement();
f87e4cac JF	178	}
f87e4cac JF	179
a9e7062d	180	static void __init xen_smp_prepare_cpus(unsigned int max_cpus)
f87e4cac JF	181	{
	182	unsigned cpu;
	183
2d9e1e2f JF	184	xen_init_lock_cpu(0);
2d9e1e2f JF	185
f87e4cac	186	smp_store_cpu_info(0);
c7b75947	187	cpu_data(0).x86_max_cores = 1;
f87e4cac JF	188	set_cpu_sibling_map(0);
	189
	190	if (xen_smp_intr_init(0))
	191	BUG();
	192
b78936e1 MT	193	if (!alloc_cpumask_var(&xen_cpu_initialized_map, GFP_KERNEL))
	194	panic("could not allocate xen_cpu_initialized_map\n");
	195
	196	cpumask_copy(xen_cpu_initialized_map, cpumask_of(0));
f87e4cac JF	197
	198	/* Restrict the possible_map according to max_cpus. */
	199	while ((num_possible_cpus() > 1) && (num_possible_cpus() > max_cpus)) {
e7986739	200	for (cpu = nr_cpu_ids - 1; !cpu_possible(cpu); cpu--)
f87e4cac	201	continue;
4f062896	202	set_cpu_possible(cpu, false);
f87e4cac JF	203	}
	204
	205	for_each_possible_cpu (cpu) {
	206	struct task_struct *idle;
	207
	208	if (cpu == 0)
	209	continue;
	210
	211	idle = fork_idle(cpu);
	212	if (IS_ERR(idle))
	213	panic("failed fork for CPU %d", cpu);
	214
4f062896	215	set_cpu_present(cpu, true);
f87e4cac	216	}
f87e4cac JF	217	}
	218
	219	static __cpuinit int
	220	cpu_initialize_context(unsigned int cpu, struct task_struct *idle)
	221	{
	222	struct vcpu_guest_context *ctxt;
c7b75947	223	struct desc_struct *gdt;
9976b39b	224	unsigned long gdt_mfn;
f87e4cac	225
b78936e1	226	if (cpumask_test_and_set_cpu(cpu, xen_cpu_initialized_map))
f87e4cac JF	227	return 0;
	228
	229	ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
	230	if (ctxt == NULL)
	231	return -ENOMEM;
	232
c7b75947 JF	233	gdt = get_cpu_gdt_table(cpu);
c7b75947 JF	234
f87e4cac JF	235	ctxt->flags = VGCF_IN_KERNEL;
	236	ctxt->user_regs.ds = __USER_DS;
	237	ctxt->user_regs.es = __USER_DS;
f87e4cac	238	ctxt->user_regs.ss = __KERNEL_DS;
c7b75947 JF	239	#ifdef CONFIG_X86_32
c7b75947 JF	240	ctxt->user_regs.fs = __KERNEL_PERCPU;
577eebea	241	ctxt->user_regs.gs = __KERNEL_STACK_CANARY;
795f99b6 JF	242	#else
795f99b6 JF	243	ctxt->gs_base_kernel = per_cpu_offset(cpu);
c7b75947	244	#endif
f87e4cac JF	245	ctxt->user_regs.eip = (unsigned long)cpu_bringup_and_idle;
	246	ctxt->user_regs.eflags = 0x1000; /* IOPL_RING1 */
	247
	248	memset(&ctxt->fpu_ctxt, 0, sizeof(ctxt->fpu_ctxt));
	249
	250	xen_copy_trap_info(ctxt->trap_ctxt);
	251
	252	ctxt->ldt_ents = 0;
	253
c7b75947	254	BUG_ON((unsigned long)gdt & ~PAGE_MASK);
9976b39b JF	255
9976b39b JF	256	gdt_mfn = arbitrary_virt_to_mfn(gdt);
c7b75947	257	make_lowmem_page_readonly(gdt);
9976b39b	258	make_lowmem_page_readonly(mfn_to_virt(gdt_mfn));
f87e4cac	259
9976b39b	260	ctxt->gdt_frames[0] = gdt_mfn;
c7b75947	261	ctxt->gdt_ents = GDT_ENTRIES;
f87e4cac JF	262
f87e4cac JF	263	ctxt->user_regs.cs = __KERNEL_CS;
faca6227	264	ctxt->user_regs.esp = idle->thread.sp0 - sizeof(struct pt_regs);
f87e4cac JF	265
f87e4cac JF	266	ctxt->kernel_ss = __KERNEL_DS;
faca6227	267	ctxt->kernel_sp = idle->thread.sp0;
f87e4cac	268
c7b75947	269	#ifdef CONFIG_X86_32
f87e4cac	270	ctxt->event_callback_cs = __KERNEL_CS;
f87e4cac	271	ctxt->failsafe_callback_cs = __KERNEL_CS;
c7b75947 JF	272	#endif
c7b75947 JF	273	ctxt->event_callback_eip = (unsigned long)xen_hypervisor_callback;
f87e4cac JF	274	ctxt->failsafe_callback_eip = (unsigned long)xen_failsafe_callback;
	275
	276	per_cpu(xen_cr3, cpu) = __pa(swapper_pg_dir);
	277	ctxt->ctrlreg[3] = xen_pfn_to_cr3(virt_to_mfn(swapper_pg_dir));
	278
	279	if (HYPERVISOR_vcpu_op(VCPUOP_initialise, cpu, ctxt))
	280	BUG();
	281
	282	kfree(ctxt);
	283	return 0;
	284	}
	285
a9e7062d	286	static int __cpuinit xen_cpu_up(unsigned int cpu)
f87e4cac JF	287	{
	288	struct task_struct *idle = idle_task(cpu);
	289	int rc;
	290
c6f5e0ac	291	per_cpu(current_task, cpu) = idle;
c7b75947	292	#ifdef CONFIG_X86_32
f87e4cac	293	irq_ctx_init(cpu);
c7b75947	294	#else
c7b75947	295	clear_tsk_thread_flag(idle, TIF_FORK);
38341432 JF	296	per_cpu(kernel_stack, cpu) =
	297	(unsigned long)task_stack_page(idle) -
	298	KERNEL_STACK_OFFSET + THREAD_SIZE;
c7b75947	299	#endif
02889672	300	xen_setup_runstate_info(cpu);
f87e4cac	301	xen_setup_timer(cpu);
2d9e1e2f	302	xen_init_lock_cpu(cpu);
f87e4cac	303
c7b75947 JF	304	per_cpu(cpu_state, cpu) = CPU_UP_PREPARE;
c7b75947 JF	305
f87e4cac JF	306	/* make sure interrupts start blocked */
	307	per_cpu(xen_vcpu, cpu)->evtchn_upcall_mask = 1;
	308
	309	rc = cpu_initialize_context(cpu, idle);
	310	if (rc)
	311	return rc;
	312
	313	if (num_online_cpus() == 1)
	314	alternatives_smp_switch(1);
	315
	316	rc = xen_smp_intr_init(cpu);
	317	if (rc)
	318	return rc;
	319
f87e4cac JF	320	rc = HYPERVISOR_vcpu_op(VCPUOP_up, cpu, NULL);
	321	BUG_ON(rc);
	322
c7b75947	323	while(per_cpu(cpu_state, cpu) != CPU_ONLINE) {
1207cf8e	324	HYPERVISOR_sched_op(SCHEDOP_yield, NULL);
c7b75947 JF	325	barrier();
	326	}
	327
f87e4cac JF	328	return 0;
	329	}
	330
a9e7062d	331	static void xen_smp_cpus_done(unsigned int max_cpus)
f87e4cac JF	332	{
	333	}
	334
2737146b	335	#ifdef CONFIG_HOTPLUG_CPU
26fd1051	336	static int xen_cpu_disable(void)
d68d82af AN	337	{
	338	unsigned int cpu = smp_processor_id();
	339	if (cpu == 0)
	340	return -EBUSY;
	341
	342	cpu_disable_common();
	343
	344	load_cr3(swapper_pg_dir);
	345	return 0;
	346	}
	347
26fd1051	348	static void xen_cpu_die(unsigned int cpu)
d68d82af AN	349	{
	350	while (HYPERVISOR_vcpu_op(VCPUOP_is_up, cpu, NULL)) {
	351	current->state = TASK_UNINTERRUPTIBLE;
	352	schedule_timeout(HZ/10);
	353	}
c6e22f9e TH	354	unbind_from_irqhandler(per_cpu(xen_resched_irq, cpu), NULL);
	355	unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
	356	unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
	357	unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu), NULL);
d68d82af AN	358	xen_uninit_lock_cpu(cpu);
	359	xen_teardown_timer(cpu);
	360
	361	if (num_online_cpus() == 1)
	362	alternatives_smp_switch(0);
	363	}
	364
71709247	365	static void __cpuinit xen_play_dead(void) /* used only with HOTPLUG_CPU */
d68d82af AN	366	{
	367	play_dead_common();
	368	HYPERVISOR_vcpu_op(VCPUOP_down, smp_processor_id(), NULL);
	369	cpu_bringup();
	370	}
	371
2737146b	372	#else /* !CONFIG_HOTPLUG_CPU */
26fd1051	373	static int xen_cpu_disable(void)
2737146b AN	374	{
	375	return -ENOSYS;
	376	}
	377
26fd1051	378	static void xen_cpu_die(unsigned int cpu)
2737146b AN	379	{
	380	BUG();
	381	}
	382
26fd1051	383	static void xen_play_dead(void)
2737146b AN	384	{
	385	BUG();
	386	}
	387
	388	#endif
f87e4cac JF	389	static void stop_self(void *v)
	390	{
	391	int cpu = smp_processor_id();
	392
	393	/* make sure we're not pinning something down */
	394	load_cr3(swapper_pg_dir);
	395	/* should set up a minimal gdt */
	396
	397	HYPERVISOR_vcpu_op(VCPUOP_down, cpu, NULL);
	398	BUG();
	399	}
	400
a9e7062d	401	static void xen_smp_send_stop(void)
f87e4cac	402	{
8691e5a8	403	smp_call_function(stop_self, NULL, 0);
f87e4cac JF	404	}
f87e4cac JF	405
a9e7062d	406	static void xen_smp_send_reschedule(int cpu)
f87e4cac JF	407	{
	408	xen_send_IPI_one(cpu, XEN_RESCHEDULE_VECTOR);
	409	}
	410
bcda016e MT	411	static void xen_send_IPI_mask(const struct cpumask *mask,
bcda016e MT	412	enum ipi_vector vector)
f87e4cac JF	413	{
	414	unsigned cpu;
	415
bcda016e	416	for_each_cpu_and(cpu, mask, cpu_online_mask)
f87e4cac JF	417	xen_send_IPI_one(cpu, vector);
	418	}
	419
bcda016e	420	static void xen_smp_send_call_function_ipi(const struct cpumask *mask)
3b16cf87 JA	421	{
	422	int cpu;
	423
	424	xen_send_IPI_mask(mask, XEN_CALL_FUNCTION_VECTOR);
	425
	426	/* Make sure other vcpus get a chance to run if they need to. */
bcda016e	427	for_each_cpu(cpu, mask) {
3b16cf87	428	if (xen_vcpu_stolen(cpu)) {
1207cf8e	429	HYPERVISOR_sched_op(SCHEDOP_yield, NULL);
3b16cf87 JA	430	break;
	431	}
	432	}
	433	}
	434
a9e7062d	435	static void xen_smp_send_call_function_single_ipi(int cpu)
3b16cf87	436	{
bcda016e	437	xen_send_IPI_mask(cpumask_of(cpu),
e7986739	438	XEN_CALL_FUNCTION_SINGLE_VECTOR);
3b16cf87 JA	439	}
3b16cf87 JA	440
f87e4cac JF	441	static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id)
f87e4cac JF	442	{
f87e4cac	443	irq_enter();
3b16cf87	444	generic_smp_call_function_interrupt();
1b437c8c	445	inc_irq_stat(irq_call_count);
f87e4cac JF	446	irq_exit();
f87e4cac JF	447
f87e4cac JF	448	return IRQ_HANDLED;
	449	}
	450
3b16cf87	451	static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id)
f87e4cac	452	{
3b16cf87 JA	453	irq_enter();
3b16cf87 JA	454	generic_smp_call_function_single_interrupt();
1b437c8c	455	inc_irq_stat(irq_call_count);
3b16cf87	456	irq_exit();
f87e4cac	457
3b16cf87	458	return IRQ_HANDLED;
f87e4cac	459	}
a9e7062d JF	460
	461	static const struct smp_ops xen_smp_ops __initdata = {
	462	.smp_prepare_boot_cpu = xen_smp_prepare_boot_cpu,
	463	.smp_prepare_cpus = xen_smp_prepare_cpus,
a9e7062d JF	464	.smp_cpus_done = xen_smp_cpus_done,
a9e7062d JF	465
d68d82af AN	466	.cpu_up = xen_cpu_up,
	467	.cpu_die = xen_cpu_die,
	468	.cpu_disable = xen_cpu_disable,
	469	.play_dead = xen_play_dead,
	470
a9e7062d JF	471	.smp_send_stop = xen_smp_send_stop,
	472	.smp_send_reschedule = xen_smp_send_reschedule,
	473
	474	.send_call_func_ipi = xen_smp_send_call_function_ipi,
	475	.send_call_func_single_ipi = xen_smp_send_call_function_single_ipi,
	476	};
	477
	478	void __init xen_smp_init(void)
	479	{
	480	smp_ops = xen_smp_ops;
c7b75947	481	xen_fill_possible_map();
2d9e1e2f	482	xen_init_spinlocks();
a9e7062d	483	}