[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/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, resched_irq);
static DEFINE_PER_CPU(int, callfunc_irq);
static DEFINE_PER_CPU(int, callfuncsingle_irq);
static DEFINE_PER_CPU(int, 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();

	cpu_set(cpu, cpu_online_map);
	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(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(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(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(callfuncsingle_irq, cpu) = rc;

	return 0;

 fail:
	if (per_cpu(resched_irq, cpu) >= 0)
		unbind_from_irqhandler(per_cpu(resched_irq, cpu), NULL);
	if (per_cpu(callfunc_irq, cpu) >= 0)
		unbind_from_irqhandler(per_cpu(callfunc_irq, cpu), NULL);
	if (per_cpu(debug_irq, cpu) >= 0)
		unbind_from_irqhandler(per_cpu(debug_irq, cpu), NULL);
	if (per_cpu(callfuncsingle_irq, cpu) >= 0)
		unbind_from_irqhandler(per_cpu(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_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(resched_irq, cpu), NULL);
	unbind_from_irqhandler(per_cpu(callfunc_irq, cpu), NULL);
	unbind_from_irqhandler(per_cpu(debug_irq, cpu), NULL);
	unbind_from_irqhandler(per_cpu(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 CPU_HOTPLUG */
{
	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>
	17	#include <linux/smp.h>
	18
	19	#include <asm/paravirt.h>
	20	#include <asm/desc.h>
	21	#include <asm/pgtable.h>
	22	#include <asm/cpu.h>
	23
	24	#include <xen/interface/xen.h>
	25	#include <xen/interface/vcpu.h>
	26
	27	#include <asm/xen/interface.h>
	28	#include <asm/xen/hypercall.h>
	29
	30	#include <xen/page.h>
	31	#include <xen/events.h>
	32
	33	#include "xen-ops.h"
	34	#include "mmu.h"
	35
b78936e1	36	cpumask_var_t xen_cpu_initialized_map;
f87e4cac	37
3b16cf87 JA	38	static DEFINE_PER_CPU(int, resched_irq);
	39	static DEFINE_PER_CPU(int, callfunc_irq);
	40	static DEFINE_PER_CPU(int, callfuncsingle_irq);
	41	static DEFINE_PER_CPU(int, debug_irq) = -1;
f87e4cac JF	42
f87e4cac JF	43	static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id);
3b16cf87	44	static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id);
f87e4cac JF	45
	46	/*
	47	* Reschedule call back. Nothing to do,
	48	* all the work is done automatically when
	49	* we return from the interrupt.
	50	*/
	51	static irqreturn_t xen_reschedule_interrupt(int irq, void *dev_id)
	52	{
1b437c8c	53	inc_irq_stat(irq_resched_count);
38bb5ab4	54
f87e4cac JF	55	return IRQ_HANDLED;
	56	}
	57
d68d82af	58	static __cpuinit void cpu_bringup(void)
f87e4cac JF	59	{
	60	int cpu = smp_processor_id();
	61
	62	cpu_init();
d68d82af	63	touch_softlockup_watchdog();
c7b75947 JF	64	preempt_disable();
c7b75947 JF	65
e2a81baf	66	xen_enable_sysenter();
6fcac6d3	67	xen_enable_syscall();
f87e4cac	68
c7b75947 JF	69	cpu = smp_processor_id();
	70	smp_store_cpu_info(cpu);
	71	cpu_data(cpu).x86_max_cores = 1;
	72	set_cpu_sibling_map(cpu);
f87e4cac JF	73
	74	xen_setup_cpu_clockevents();
	75
c7b75947	76	cpu_set(cpu, cpu_online_map);
6dbde353	77	percpu_write(cpu_state, CPU_ONLINE);
c7b75947 JF	78	wmb();
c7b75947 JF	79
f87e4cac JF	80	/* We can take interrupts now: we're officially "up". */
	81	local_irq_enable();
	82
	83	wmb(); /* make sure everything is out */
d68d82af AN	84	}
	85
	86	static __cpuinit void cpu_bringup_and_idle(void)
	87	{
	88	cpu_bringup();
f87e4cac JF	89	cpu_idle();
	90	}
	91
	92	static int xen_smp_intr_init(unsigned int cpu)
	93	{
	94	int rc;
ee523ca1	95	const char resched_name, callfunc_name, *debug_name;
f87e4cac JF	96
	97	resched_name = kasprintf(GFP_KERNEL, "resched%d", cpu);
	98	rc = bind_ipi_to_irqhandler(XEN_RESCHEDULE_VECTOR,
	99	cpu,
	100	xen_reschedule_interrupt,
	101	IRQF_DISABLED\|IRQF_PERCPU\|IRQF_NOBALANCING,
	102	resched_name,
	103	NULL);
	104	if (rc < 0)
	105	goto fail;
	106	per_cpu(resched_irq, cpu) = rc;
	107
	108	callfunc_name = kasprintf(GFP_KERNEL, "callfunc%d", cpu);
	109	rc = bind_ipi_to_irqhandler(XEN_CALL_FUNCTION_VECTOR,
	110	cpu,
	111	xen_call_function_interrupt,
	112	IRQF_DISABLED\|IRQF_PERCPU\|IRQF_NOBALANCING,
	113	callfunc_name,
	114	NULL);
	115	if (rc < 0)
	116	goto fail;
	117	per_cpu(callfunc_irq, cpu) = rc;
	118
ee523ca1 JF	119	debug_name = kasprintf(GFP_KERNEL, "debug%d", cpu);
	120	rc = bind_virq_to_irqhandler(VIRQ_DEBUG, cpu, xen_debug_interrupt,
	121	IRQF_DISABLED \| IRQF_PERCPU \| IRQF_NOBALANCING,
	122	debug_name, NULL);
	123	if (rc < 0)
	124	goto fail;
	125	per_cpu(debug_irq, cpu) = rc;
	126
3b16cf87 JA	127	callfunc_name = kasprintf(GFP_KERNEL, "callfuncsingle%d", cpu);
	128	rc = bind_ipi_to_irqhandler(XEN_CALL_FUNCTION_SINGLE_VECTOR,
	129	cpu,
	130	xen_call_function_single_interrupt,
	131	IRQF_DISABLED\|IRQF_PERCPU\|IRQF_NOBALANCING,
	132	callfunc_name,
	133	NULL);
	134	if (rc < 0)
	135	goto fail;
	136	per_cpu(callfuncsingle_irq, cpu) = rc;
	137
f87e4cac JF	138	return 0;
	139
	140	fail:
	141	if (per_cpu(resched_irq, cpu) >= 0)
	142	unbind_from_irqhandler(per_cpu(resched_irq, cpu), NULL);
	143	if (per_cpu(callfunc_irq, cpu) >= 0)
	144	unbind_from_irqhandler(per_cpu(callfunc_irq, cpu), NULL);
ee523ca1 JF	145	if (per_cpu(debug_irq, cpu) >= 0)
ee523ca1 JF	146	unbind_from_irqhandler(per_cpu(debug_irq, cpu), NULL);
3b16cf87 JA	147	if (per_cpu(callfuncsingle_irq, cpu) >= 0)
	148	unbind_from_irqhandler(per_cpu(callfuncsingle_irq, cpu), NULL);
	149
f87e4cac JF	150	return rc;
	151	}
	152
c7b75947	153	static void __init xen_fill_possible_map(void)
f87e4cac JF	154	{
	155	int i, rc;
	156
e7986739	157	for (i = 0; i < nr_cpu_ids; i++) {
f87e4cac	158	rc = HYPERVISOR_vcpu_op(VCPUOP_is_up, i, NULL);
4560a294 JF	159	if (rc >= 0) {
4560a294 JF	160	num_processors++;
4f062896	161	set_cpu_possible(i, true);
4560a294	162	}
f87e4cac JF	163	}
	164	}
	165
a9e7062d	166	static void __init xen_smp_prepare_boot_cpu(void)
f87e4cac	167	{
f87e4cac JF	168	BUG_ON(smp_processor_id() != 0);
	169	native_smp_prepare_boot_cpu();
	170
f87e4cac JF	171	/* We've switched to the "real" per-cpu gdt, so make sure the
f87e4cac JF	172	old memory can be recycled */
38341432	173	make_lowmem_page_readwrite(xen_initial_gdt);
60223a32 JF	174
60223a32 JF	175	xen_setup_vcpu_info_placement();
f87e4cac JF	176	}
f87e4cac JF	177
a9e7062d	178	static void __init xen_smp_prepare_cpus(unsigned int max_cpus)
f87e4cac JF	179	{
	180	unsigned cpu;
	181
2d9e1e2f JF	182	xen_init_lock_cpu(0);
2d9e1e2f JF	183
f87e4cac	184	smp_store_cpu_info(0);
c7b75947	185	cpu_data(0).x86_max_cores = 1;
f87e4cac JF	186	set_cpu_sibling_map(0);
	187
	188	if (xen_smp_intr_init(0))
	189	BUG();
	190
b78936e1 MT	191	if (!alloc_cpumask_var(&xen_cpu_initialized_map, GFP_KERNEL))
	192	panic("could not allocate xen_cpu_initialized_map\n");
	193
	194	cpumask_copy(xen_cpu_initialized_map, cpumask_of(0));
f87e4cac JF	195
	196	/* Restrict the possible_map according to max_cpus. */
	197	while ((num_possible_cpus() > 1) && (num_possible_cpus() > max_cpus)) {
e7986739	198	for (cpu = nr_cpu_ids - 1; !cpu_possible(cpu); cpu--)
f87e4cac	199	continue;
4f062896	200	set_cpu_possible(cpu, false);
f87e4cac JF	201	}
	202
	203	for_each_possible_cpu (cpu) {
	204	struct task_struct *idle;
	205
	206	if (cpu == 0)
	207	continue;
	208
	209	idle = fork_idle(cpu);
	210	if (IS_ERR(idle))
	211	panic("failed fork for CPU %d", cpu);
	212
4f062896	213	set_cpu_present(cpu, true);
f87e4cac	214	}
f87e4cac JF	215	}
	216
	217	static __cpuinit int
	218	cpu_initialize_context(unsigned int cpu, struct task_struct *idle)
	219	{
	220	struct vcpu_guest_context *ctxt;
c7b75947	221	struct desc_struct *gdt;
9976b39b	222	unsigned long gdt_mfn;
f87e4cac	223
b78936e1	224	if (cpumask_test_and_set_cpu(cpu, xen_cpu_initialized_map))
f87e4cac JF	225	return 0;
	226
	227	ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
	228	if (ctxt == NULL)
	229	return -ENOMEM;
	230
c7b75947 JF	231	gdt = get_cpu_gdt_table(cpu);
c7b75947 JF	232
f87e4cac JF	233	ctxt->flags = VGCF_IN_KERNEL;
	234	ctxt->user_regs.ds = __USER_DS;
	235	ctxt->user_regs.es = __USER_DS;
f87e4cac	236	ctxt->user_regs.ss = __KERNEL_DS;
c7b75947 JF	237	#ifdef CONFIG_X86_32
c7b75947 JF	238	ctxt->user_regs.fs = __KERNEL_PERCPU;
577eebea	239	ctxt->user_regs.gs = __KERNEL_STACK_CANARY;
795f99b6 JF	240	#else
795f99b6 JF	241	ctxt->gs_base_kernel = per_cpu_offset(cpu);
c7b75947	242	#endif
f87e4cac JF	243	ctxt->user_regs.eip = (unsigned long)cpu_bringup_and_idle;
	244	ctxt->user_regs.eflags = 0x1000; /* IOPL_RING1 */
	245
	246	memset(&ctxt->fpu_ctxt, 0, sizeof(ctxt->fpu_ctxt));
	247
	248	xen_copy_trap_info(ctxt->trap_ctxt);
	249
	250	ctxt->ldt_ents = 0;
	251
c7b75947	252	BUG_ON((unsigned long)gdt & ~PAGE_MASK);
9976b39b JF	253
9976b39b JF	254	gdt_mfn = arbitrary_virt_to_mfn(gdt);
c7b75947	255	make_lowmem_page_readonly(gdt);
9976b39b	256	make_lowmem_page_readonly(mfn_to_virt(gdt_mfn));
f87e4cac	257
9976b39b	258	ctxt->gdt_frames[0] = gdt_mfn;
c7b75947	259	ctxt->gdt_ents = GDT_ENTRIES;
f87e4cac JF	260
f87e4cac JF	261	ctxt->user_regs.cs = __KERNEL_CS;
faca6227	262	ctxt->user_regs.esp = idle->thread.sp0 - sizeof(struct pt_regs);
f87e4cac JF	263
f87e4cac JF	264	ctxt->kernel_ss = __KERNEL_DS;
faca6227	265	ctxt->kernel_sp = idle->thread.sp0;
f87e4cac	266
c7b75947	267	#ifdef CONFIG_X86_32
f87e4cac	268	ctxt->event_callback_cs = __KERNEL_CS;
f87e4cac	269	ctxt->failsafe_callback_cs = __KERNEL_CS;
c7b75947 JF	270	#endif
c7b75947 JF	271	ctxt->event_callback_eip = (unsigned long)xen_hypervisor_callback;
f87e4cac JF	272	ctxt->failsafe_callback_eip = (unsigned long)xen_failsafe_callback;
	273
	274	per_cpu(xen_cr3, cpu) = __pa(swapper_pg_dir);
	275	ctxt->ctrlreg[3] = xen_pfn_to_cr3(virt_to_mfn(swapper_pg_dir));
	276
	277	if (HYPERVISOR_vcpu_op(VCPUOP_initialise, cpu, ctxt))
	278	BUG();
	279
	280	kfree(ctxt);
	281	return 0;
	282	}
	283
a9e7062d	284	static int __cpuinit xen_cpu_up(unsigned int cpu)
f87e4cac JF	285	{
	286	struct task_struct *idle = idle_task(cpu);
	287	int rc;
	288
c6f5e0ac	289	per_cpu(current_task, cpu) = idle;
c7b75947	290	#ifdef CONFIG_X86_32
f87e4cac	291	irq_ctx_init(cpu);
c7b75947	292	#else
c7b75947	293	clear_tsk_thread_flag(idle, TIF_FORK);
38341432 JF	294	per_cpu(kernel_stack, cpu) =
	295	(unsigned long)task_stack_page(idle) -
	296	KERNEL_STACK_OFFSET + THREAD_SIZE;
c7b75947	297	#endif
f87e4cac	298	xen_setup_timer(cpu);
2d9e1e2f	299	xen_init_lock_cpu(cpu);
f87e4cac	300
c7b75947 JF	301	per_cpu(cpu_state, cpu) = CPU_UP_PREPARE;
c7b75947 JF	302
f87e4cac JF	303	/* make sure interrupts start blocked */
	304	per_cpu(xen_vcpu, cpu)->evtchn_upcall_mask = 1;
	305
	306	rc = cpu_initialize_context(cpu, idle);
	307	if (rc)
	308	return rc;
	309
	310	if (num_online_cpus() == 1)
	311	alternatives_smp_switch(1);
	312
	313	rc = xen_smp_intr_init(cpu);
	314	if (rc)
	315	return rc;
	316
f87e4cac JF	317	rc = HYPERVISOR_vcpu_op(VCPUOP_up, cpu, NULL);
	318	BUG_ON(rc);
	319
c7b75947	320	while(per_cpu(cpu_state, cpu) != CPU_ONLINE) {
1207cf8e	321	HYPERVISOR_sched_op(SCHEDOP_yield, NULL);
c7b75947 JF	322	barrier();
	323	}
	324
f87e4cac JF	325	return 0;
	326	}
	327
a9e7062d	328	static void xen_smp_cpus_done(unsigned int max_cpus)
f87e4cac JF	329	{
	330	}
	331
2737146b	332	#ifdef CONFIG_HOTPLUG_CPU
26fd1051	333	static int xen_cpu_disable(void)
d68d82af AN	334	{
	335	unsigned int cpu = smp_processor_id();
	336	if (cpu == 0)
	337	return -EBUSY;
	338
	339	cpu_disable_common();
	340
	341	load_cr3(swapper_pg_dir);
	342	return 0;
	343	}
	344
26fd1051	345	static void xen_cpu_die(unsigned int cpu)
d68d82af AN	346	{
	347	while (HYPERVISOR_vcpu_op(VCPUOP_is_up, cpu, NULL)) {
	348	current->state = TASK_UNINTERRUPTIBLE;
	349	schedule_timeout(HZ/10);
	350	}
	351	unbind_from_irqhandler(per_cpu(resched_irq, cpu), NULL);
	352	unbind_from_irqhandler(per_cpu(callfunc_irq, cpu), NULL);
	353	unbind_from_irqhandler(per_cpu(debug_irq, cpu), NULL);
	354	unbind_from_irqhandler(per_cpu(callfuncsingle_irq, cpu), NULL);
	355	xen_uninit_lock_cpu(cpu);
	356	xen_teardown_timer(cpu);
	357
	358	if (num_online_cpus() == 1)
	359	alternatives_smp_switch(0);
	360	}
	361
df6b0794	362	static void __cpuinit xen_play_dead(void) /* used only with CPU_HOTPLUG */
d68d82af AN	363	{
	364	play_dead_common();
	365	HYPERVISOR_vcpu_op(VCPUOP_down, smp_processor_id(), NULL);
	366	cpu_bringup();
	367	}
	368
2737146b	369	#else /* !CONFIG_HOTPLUG_CPU */
26fd1051	370	static int xen_cpu_disable(void)
2737146b AN	371	{
	372	return -ENOSYS;
	373	}
	374
26fd1051	375	static void xen_cpu_die(unsigned int cpu)
2737146b AN	376	{
	377	BUG();
	378	}
	379
26fd1051	380	static void xen_play_dead(void)
2737146b AN	381	{
	382	BUG();
	383	}
	384
	385	#endif
f87e4cac JF	386	static void stop_self(void *v)
	387	{
	388	int cpu = smp_processor_id();
	389
	390	/* make sure we're not pinning something down */
	391	load_cr3(swapper_pg_dir);
	392	/* should set up a minimal gdt */
	393
	394	HYPERVISOR_vcpu_op(VCPUOP_down, cpu, NULL);
	395	BUG();
	396	}
	397
a9e7062d	398	static void xen_smp_send_stop(void)
f87e4cac	399	{
8691e5a8	400	smp_call_function(stop_self, NULL, 0);
f87e4cac JF	401	}
f87e4cac JF	402
a9e7062d	403	static void xen_smp_send_reschedule(int cpu)
f87e4cac JF	404	{
	405	xen_send_IPI_one(cpu, XEN_RESCHEDULE_VECTOR);
	406	}
	407
bcda016e MT	408	static void xen_send_IPI_mask(const struct cpumask *mask,
bcda016e MT	409	enum ipi_vector vector)
f87e4cac JF	410	{
	411	unsigned cpu;
	412
bcda016e	413	for_each_cpu_and(cpu, mask, cpu_online_mask)
f87e4cac JF	414	xen_send_IPI_one(cpu, vector);
	415	}
	416
bcda016e	417	static void xen_smp_send_call_function_ipi(const struct cpumask *mask)
3b16cf87 JA	418	{
	419	int cpu;
	420
	421	xen_send_IPI_mask(mask, XEN_CALL_FUNCTION_VECTOR);
	422
	423	/* Make sure other vcpus get a chance to run if they need to. */
bcda016e	424	for_each_cpu(cpu, mask) {
3b16cf87	425	if (xen_vcpu_stolen(cpu)) {
1207cf8e	426	HYPERVISOR_sched_op(SCHEDOP_yield, NULL);
3b16cf87 JA	427	break;
	428	}
	429	}
	430	}
	431
a9e7062d	432	static void xen_smp_send_call_function_single_ipi(int cpu)
3b16cf87	433	{
bcda016e	434	xen_send_IPI_mask(cpumask_of(cpu),
e7986739	435	XEN_CALL_FUNCTION_SINGLE_VECTOR);
3b16cf87 JA	436	}
3b16cf87 JA	437
f87e4cac JF	438	static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id)
f87e4cac JF	439	{
f87e4cac	440	irq_enter();
3b16cf87	441	generic_smp_call_function_interrupt();
1b437c8c	442	inc_irq_stat(irq_call_count);
f87e4cac JF	443	irq_exit();
f87e4cac JF	444
f87e4cac JF	445	return IRQ_HANDLED;
	446	}
	447
3b16cf87	448	static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id)
f87e4cac	449	{
3b16cf87 JA	450	irq_enter();
3b16cf87 JA	451	generic_smp_call_function_single_interrupt();
1b437c8c	452	inc_irq_stat(irq_call_count);
3b16cf87	453	irq_exit();
f87e4cac	454
3b16cf87	455	return IRQ_HANDLED;
f87e4cac	456	}
a9e7062d JF	457
	458	static const struct smp_ops xen_smp_ops __initdata = {
	459	.smp_prepare_boot_cpu = xen_smp_prepare_boot_cpu,
	460	.smp_prepare_cpus = xen_smp_prepare_cpus,
a9e7062d JF	461	.smp_cpus_done = xen_smp_cpus_done,
a9e7062d JF	462
d68d82af AN	463	.cpu_up = xen_cpu_up,
	464	.cpu_die = xen_cpu_die,
	465	.cpu_disable = xen_cpu_disable,
	466	.play_dead = xen_play_dead,
	467
a9e7062d JF	468	.smp_send_stop = xen_smp_send_stop,
	469	.smp_send_reschedule = xen_smp_send_reschedule,
	470
	471	.send_call_func_ipi = xen_smp_send_call_function_ipi,
	472	.send_call_func_single_ipi = xen_smp_send_call_function_single_ipi,
	473	};
	474
	475	void __init xen_smp_init(void)
	476	{
	477	smp_ops = xen_smp_ops;
c7b75947	478	xen_fill_possible_map();
2d9e1e2f	479	xen_init_spinlocks();
a9e7062d	480	}