xen/xen/arch/x86/oprofile/nmi_int.c

/**
 * @file nmi_int.c
 *
 * @remark Copyright 2002 OProfile authors
 * @remark Read the file COPYING
 *
 * @author John Levon <levon@movementarian.org>
 *
 * Modified for Xen: by Aravind Menon & Jose Renato Santos
 *   These modifications are:
 *   Copyright (C) 2005 Hewlett-Packard Co.
 */

#include <xen/event.h>
#include <xen/types.h>
#include <xen/errno.h>
#include <xen/init.h>
#include <xen/param.h>
#include <xen/string.h>
#include <xen/delay.h>
#include <xen/xenoprof.h>
#include <xen/xvmalloc.h>

#include <public/xenoprof.h>

#include <asm/msr.h>
#include <asm/apic.h>
#include <asm/regs.h>
#include <asm/current.h>
#include <asm/nmi.h>

#include "op_counter.h"
#include "op_x86_model.h"

struct op_counter_config counter_config[OP_MAX_COUNTER];
struct op_ibs_config ibs_config;

struct op_x86_model_spec const *__read_mostly model;
static struct op_msrs cpu_msrs[NR_CPUS];
static unsigned long saved_lvtpc[NR_CPUS];

static const char *cpu_type;

static DEFINE_PER_CPU(struct vcpu *, nmi_cont_vcpu);

static int passive_domain_msr_op_checks(unsigned int msr, int *typep, int *indexp)
{
	struct vpmu_struct *vpmu = vcpu_vpmu(current);
	if ( model == NULL )
		return 0;
	if ( model->is_arch_pmu_msr == NULL )
		return 0;
	if ( !model->is_arch_pmu_msr(msr, typep, indexp) )
		return 0;

	if ( !vpmu_is_set(vpmu, VPMU_PASSIVE_DOMAIN_ALLOCATED) )
		if ( ! model->allocated_msr(current) )
			return 0;
	return 1;
}

int passive_domain_do_rdmsr(unsigned int msr, uint64_t *msr_content)
{
	int type, index;

	if ( !passive_domain_msr_op_checks(msr, &type, &index))
		return 0;

	model->load_msr(current, type, index, msr_content);
	return 1;
}

int passive_domain_do_wrmsr(unsigned int msr, uint64_t msr_content)
{
	int type, index;

	if ( !passive_domain_msr_op_checks(msr, &type, &index))
		return 0;

	model->save_msr(current, type, index, msr_content);
	return 1;
}

void passive_domain_destroy(struct vcpu *v)
{
	struct vpmu_struct *vpmu = vcpu_vpmu(v);
	if ( vpmu_is_set(vpmu, VPMU_PASSIVE_DOMAIN_ALLOCATED) )
		model->free_msr(v);
}

bool nmi_oprofile_send_virq(void)
{
	struct vcpu *v = xchg(&this_cpu(nmi_cont_vcpu), NULL);

	if (v)
		send_guest_vcpu_virq(v, VIRQ_XENOPROF);

	return v;
}

static int cf_check nmi_callback(const struct cpu_user_regs *regs, int cpu)
{
	int xen_mode, ovf;

	ovf = model->check_ctrs(cpu, &cpu_msrs[cpu], regs);
	xen_mode = ring_0(regs);
	if (ovf && is_active(current->domain) && !xen_mode &&
	    !this_cpu(nmi_cont_vcpu)) {
		this_cpu(nmi_cont_vcpu) = current;
		trigger_nmi_continuation();
	}

	if ( ovf == 2 )
		current->arch.nmi_pending = true;
	return 1;
}


static void nmi_cpu_save_registers(struct op_msrs *msrs)
{
	unsigned int const nr_ctrs = model->num_counters;
	unsigned int const nr_ctrls = model->num_controls;
	struct op_msr *counters = msrs->counters;
	struct op_msr *controls = msrs->controls;
	unsigned int i;

	for (i = 0; i < nr_ctrs; ++i) {
		rdmsrl(counters[i].addr, counters[i].value);
	}

	for (i = 0; i < nr_ctrls; ++i) {
		rdmsrl(controls[i].addr, controls[i].value);
	}
}


static void cf_check nmi_save_registers(void *dummy)
{
	int cpu = smp_processor_id();
	struct op_msrs * msrs = &cpu_msrs[cpu];
	model->fill_in_addresses(msrs);
	nmi_cpu_save_registers(msrs);
}


static void free_msrs(void)
{
	unsigned int i;

	for (i = 0; i < nr_cpu_ids; ++i) {
		XVFREE(cpu_msrs[i].counters);
		XVFREE(cpu_msrs[i].controls);
	}
}


static int allocate_msrs(void)
{
	unsigned int i;
	int success = 1;

	for_each_online_cpu (i) {
		cpu_msrs[i].counters = xvmalloc_array(struct op_msr,
						      model->num_counters);
		if (!cpu_msrs[i].counters) {
			success = 0;
			break;
		}
		cpu_msrs[i].controls = xvmalloc_array(struct op_msr,
						      model->num_controls);
		if (!cpu_msrs[i].controls) {
			success = 0;
			break;
		}
	}

	if (!success)
		free_msrs();

	return success;
}


static void cf_check nmi_cpu_setup(void *dummy)
{
	int cpu = smp_processor_id();
	struct op_msrs * msrs = &cpu_msrs[cpu];
	model->setup_ctrs(msrs);
}


int nmi_setup_events(void)
{
	on_each_cpu(nmi_cpu_setup, NULL, 1);
	return 0;
}

int nmi_reserve_counters(void)
{
	if (!allocate_msrs())
		return -ENOMEM;

	/*
	 * We need to be careful to install our NMI handler
	 * without actually triggering any NMIs as this will
	 * break the core code horrifically.
	 */
	if (reserve_lapic_nmi() < 0) {
		free_msrs();
		return -EBUSY;
	}
	/* We need to serialize save and setup for HT because the subset
	 * of msrs are distinct for save and setup operations
	 */
	on_each_cpu(nmi_save_registers, NULL, 1);
	return 0;
}

int nmi_enable_virq(void)
{
	set_nmi_callback(nmi_callback);
	return 0;
}


void nmi_disable_virq(void)
{
	unset_nmi_callback();
}


static void nmi_restore_registers(struct op_msrs * msrs)
{
	unsigned int const nr_ctrs = model->num_counters;
	unsigned int const nr_ctrls = model->num_controls;
	struct op_msr * counters = msrs->counters;
	struct op_msr * controls = msrs->controls;
	unsigned int i;

	for (i = 0; i < nr_ctrls; ++i) {
		wrmsrl(controls[i].addr, controls[i].value);
	}

	for (i = 0; i < nr_ctrs; ++i) {
		wrmsrl(counters[i].addr, counters[i].value);
	}
}


static void cf_check nmi_cpu_shutdown(void *dummy)
{
	int cpu = smp_processor_id();
	struct op_msrs * msrs = &cpu_msrs[cpu];
	nmi_restore_registers(msrs);
}


void nmi_release_counters(void)
{
	on_each_cpu(nmi_cpu_shutdown, NULL, 1);
	release_lapic_nmi();
	free_msrs();
}


static void cf_check nmi_cpu_start(void *dummy)
{
	int cpu = smp_processor_id();
	struct op_msrs const * msrs = &cpu_msrs[cpu];
	saved_lvtpc[cpu] = apic_read(APIC_LVTPC);
	apic_write(APIC_LVTPC, APIC_DM_NMI);
	model->start(msrs);
}


int nmi_start(void)
{
	on_each_cpu(nmi_cpu_start, NULL, 1);
	return 0;
}


static void cf_check nmi_cpu_stop(void *dummy)
{
	unsigned int v;
	int cpu = smp_processor_id();
	struct op_msrs const * msrs = &cpu_msrs[cpu];
	model->stop(msrs);

	/* restoring APIC_LVTPC can trigger an apic error because the delivery
	 * mode and vector nr combination can be illegal. That's by design: on
	 * power on apic lvt contain a zero vector nr which are legal only for
	 * NMI delivery mode. So inhibit apic err before restoring lvtpc
	 */
	if ( (apic_read(APIC_LVTPC) & APIC_DM_MASK) != APIC_DM_NMI
	     || (apic_read(APIC_LVTPC) & APIC_LVT_MASKED) )
	{
		printk("nmi_stop: APIC not good %ul\n", apic_read(APIC_LVTPC));
		mdelay(5000);
	}
	v = apic_read(APIC_LVTERR);
	apic_write(APIC_LVTERR, v | APIC_LVT_MASKED);
	apic_write(APIC_LVTPC, saved_lvtpc[cpu]);
	apic_write(APIC_LVTERR, v);
}


void nmi_stop(void)
{
	on_each_cpu(nmi_cpu_stop, NULL, 1);
}


static int __init p4_init(const char **cpu_type)
{
	unsigned int cpu_model = current_cpu_data.x86_model;

	if ((cpu_model > 6) || (cpu_model == 5)) {
		printk("xenoprof: Initialization failed. "
		       "Intel processor model %u for pentium 4 family is not "
		       "supported\n", cpu_model);
		return 0;
	}

	switch (current_cpu_data.x86_num_siblings) {
		case 1:
			*cpu_type = "i386/p4";
			model = &op_p4_spec;
			return 1;

		case 2:
			*cpu_type = "i386/p4-ht";
			model = &op_p4_ht2_spec;
			return 1;
	}

	printk("Xenoprof ERROR: P4 HyperThreading detected with > 2 threads\n");

	return 0;
}


static int force_arch_perfmon;

static int __init cf_check force_cpu_type(const char *str)
{
	if (!strcmp(str, "arch_perfmon")) {
		force_arch_perfmon = 1;
		printk(KERN_INFO "oprofile: forcing architectural perfmon\n");
	}
	else
		return -EINVAL;

	return 0;
}
custom_param("cpu_type", force_cpu_type);

static int __init ppro_init(const char **cpu_type)
{
	if (force_arch_perfmon && cpu_has_arch_perfmon)
		return 0;

	switch (current_cpu_data.x86_model) {
	case 14:
		*cpu_type = "i386/core";
		break;
	case 15:
		*cpu_type = "i386/core_2";
		ppro_has_global_ctrl = 1;
		break;
	default:
		/* Unknown */
		return 0;
	}

	model = &op_ppro_spec;
	return 1;
}

static int __init arch_perfmon_init(const char **cpu_type)
{
	if (!cpu_has_arch_perfmon)
		return 0;
	*cpu_type = "i386/arch_perfmon";
	model = &op_arch_perfmon_spec;
	arch_perfmon_setup_counters();
	ppro_has_global_ctrl = 1;
	return 1;
}

static int __init cf_check nmi_init(void)
{
	unsigned int vendor = current_cpu_data.x86_vendor;
	unsigned int family = current_cpu_data.x86;

	if (!cpu_has_apic) {
		printk("xenoprof: Initialization failed. No APIC\n");
		return -ENODEV;
	}

	switch (vendor) {
		case X86_VENDOR_AMD:
			/* Needs to be at least an Athlon (or hammer in 32bit mode) */

			switch (family) {
			default:
				printk("xenoprof: Initialization failed. "
				       "AMD processor family %u is not "
				       "supported\n", family);
				return -ENODEV;
			case 0xf:
				model = &op_athlon_spec;
				cpu_type = "x86-64/hammer";
				break;
			case 0x10:
				model = &op_athlon_spec;
				cpu_type = "x86-64/family10";
				ibs_init();
				break;
			case 0x11:
				model = &op_athlon_spec;
				cpu_type = "x86-64/family11h";
				break;
                        case 0x12:
				model = &op_athlon_spec;
				cpu_type = "x86-64/family12h";
				break;
			case 0x14:
                                model = &op_athlon_spec;
                                cpu_type = "x86-64/family14h";
                                break;
                        case 0x15:
                                model = &op_amd_fam15h_spec;
                                cpu_type = "x86-64/family15h";
                                break;
			case 0x16:
				model = &op_athlon_spec;
				cpu_type = "x86-64/family16h";
				break;
			}
			break;

		case X86_VENDOR_INTEL:
			switch (family) {
				/* Pentium IV */
				case 0xf:
					p4_init(&cpu_type);
					break;

				/* A P6-class processor */
				case 6:
					ppro_init(&cpu_type);
					break;

				default:
				break;
			}
			if (!cpu_type && !arch_perfmon_init(&cpu_type)) {
				printk("xenoprof: Initialization failed. "
				       "Intel processor family %u model %d is not supported\n",
				       family, current_cpu_data.x86_model);
				return -ENODEV;
			}
			break;

		default:
			printk("xenoprof: Initialization failed. "
			       "Unsupported processor. Unknown vendor %u\n",
				vendor);
			return -ENODEV;
	}

	return 0;
}

__initcall(nmi_init);

int xenoprof_arch_init(int *num_events, char *_cpu_type)
{
	if (cpu_type == NULL)
		return -ENODEV;
	*num_events = model->num_counters;
	strlcpy(_cpu_type, cpu_type, XENOPROF_CPU_TYPE_SIZE);
	return 0;
}