core.c

/*
 * Performance events x86 architecture code
 *
 *  Copyright (C) 2008 Thomas Gleixner <tglx@linutronix.de>
 *  Copyright (C) 2008-2009 Red Hat, Inc., Ingo Molnar
 *  Copyright (C) 2009 Jaswinder Singh Rajput
 *  Copyright (C) 2009 Advanced Micro Devices, Inc., Robert Richter
 *  Copyright (C) 2008-2009 Red Hat, Inc., Peter Zijlstra
 *  Copyright (C) 2009 Intel Corporation, <markus.t.metzger@intel.com>
 *  Copyright (C) 2009 Google, Inc., Stephane Eranian
 *
 *  For licencing details see kernel-base/COPYING
 */

#include <linux/perf_event.h>
#include <linux/capability.h>
#include <linux/notifier.h>
#include <linux/hardirq.h>
#include <linux/kprobes.h>
#include <linux/export.h>
#include <linux/init.h>
#include <linux/kdebug.h>
#include <linux/sched/mm.h>
#include <linux/sched/clock.h>
#include <linux/uaccess.h>
#include <linux/slab.h>
#include <linux/cpu.h>
#include <linux/bitops.h>
#include <linux/device.h>
#include <linux/nospec.h>

#include <asm/apic.h>
#include <asm/stacktrace.h>
#include <asm/nmi.h>
#include <asm/smp.h>
#include <asm/alternative.h>
#include <asm/mmu_context.h>
#include <asm/tlbflush.h>
#include <asm/timer.h>
#include <asm/desc.h>
#include <asm/ldt.h>
#include <asm/unwind.h>

#include "perf_event.h"

struct x86_pmu x86_pmu __read_mostly;

DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events) = {
	.enabled = 1,
};

DEFINE_STATIC_KEY_FALSE(rdpmc_always_available_key);

u64 __read_mostly hw_cache_event_ids
				[PERF_COUNT_HW_CACHE_MAX]
				[PERF_COUNT_HW_CACHE_OP_MAX]
				[PERF_COUNT_HW_CACHE_RESULT_MAX];
u64 __read_mostly hw_cache_extra_regs
				[PERF_COUNT_HW_CACHE_MAX]
				[PERF_COUNT_HW_CACHE_OP_MAX]
				[PERF_COUNT_HW_CACHE_RESULT_MAX];

/*
 * Propagate event elapsed time into the generic event.
 * Can only be executed on the CPU where the event is active.
 * Returns the delta events processed.
 */
u64 x86_perf_event_update(struct perf_event *event)
{
	struct hw_perf_event *hwc = &event->hw;
	int shift = 64 - x86_pmu.cntval_bits;
	u64 prev_raw_count, new_raw_count;
	int idx = hwc->idx;
	u64 delta;

	if (idx == INTEL_PMC_IDX_FIXED_BTS)
		return 0;

	/*
	 * Careful: an NMI might modify the previous event value.
	 *
	 * Our tactic to handle this is to first atomically read and
	 * exchange a new raw count - then add that new-prev delta
	 * count to the generic event atomically:
	 */
again:
	prev_raw_count = local64_read(&hwc->prev_count);
	rdpmcl(hwc->event_base_rdpmc, new_raw_count);

	if (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
					new_raw_count) != prev_raw_count)
		goto again;

	/*
	 * Now we have the new raw value and have updated the prev
	 * timestamp already. We can now calculate the elapsed delta
	 * (event-)time and add that to the generic event.
	 *
	 * Careful, not all hw sign-extends above the physical width
	 * of the count.
	 */
	delta = (new_raw_count << shift) - (prev_raw_count << shift);
	delta >>= shift;

	local64_add(delta, &event->count);
	local64_sub(delta, &hwc->period_left);

	return new_raw_count;
}

/*
 * Find and validate any extra registers to set up.
 */
static int x86_pmu_extra_regs(u64 config, struct perf_event *event)
{
	struct hw_perf_event_extra *reg;
	struct extra_reg *er;

	reg = &event->hw.extra_reg;

	if (!x86_pmu.extra_regs)
		return 0;

	for (er = x86_pmu.extra_regs; er->msr; er++) {
		if (er->event != (config & er->config_mask))
			continue;
		if (event->attr.config1 & ~er->valid_mask)
			return -EINVAL;
		/* Check if the extra msrs can be safely accessed*/
		if (!er->extra_msr_access)
			return -ENXIO;

		reg->idx = er->idx;
		reg->config = event->attr.config1;
		reg->reg = er->msr;
		break;
	}
	return 0;
}

static atomic_t active_events;
static atomic_t pmc_refcount;
static DEFINE_MUTEX(pmc_reserve_mutex);

#ifdef CONFIG_X86_LOCAL_APIC

static bool reserve_pmc_hardware(void)
{
	int i;

	for (i = 0; i < x86_pmu.num_counters; i++) {
		if (!reserve_perfctr_nmi(x86_pmu_event_addr(i)))
			goto perfctr_fail;
	}

	for (i = 0; i < x86_pmu.num_counters; i++) {
		if (!reserve_evntsel_nmi(x86_pmu_config_addr(i)))
			goto eventsel_fail;
	}

	return true;

eventsel_fail:
	for (i--; i >= 0; i--)
		release_evntsel_nmi(x86_pmu_config_addr(i));

	i = x86_pmu.num_counters;

perfctr_fail:
	for (i--; i >= 0; i--)
		release_perfctr_nmi(x86_pmu_event_addr(i));

	return false;
}

static void release_pmc_hardware(void)
{
	int i;

	for (i = 0; i < x86_pmu.num_counters; i++) {
		release_perfctr_nmi(x86_pmu_event_addr(i));
		release_evntsel_nmi(x86_pmu_config_addr(i));
	}
}

#else

static bool reserve_pmc_hardware(void) { return true; }
static void release_pmc_hardware(void) {}

#endif

static bool check_hw_exists(void)
{
	u64 val, val_fail = -1, val_new= ~0;
	int i, reg, reg_fail = -1, ret = 0;
	int bios_fail = 0;
	int reg_safe = -1;

	/*
	 * Check to see if the BIOS enabled any of the counters, if so
	 * complain and bail.
	 */
	for (i = 0; i < x86_pmu.num_counters; i++) {
		reg = x86_pmu_config_addr(i);
		ret = rdmsrl_safe(reg, &val);
		if (ret)
			goto msr_fail;
		if (val & ARCH_PERFMON_EVENTSEL_ENABLE) {
			bios_fail = 1;
			val_fail = val;
			reg_fail = reg;
		} else {
			reg_safe = i;
		}
	}

	if (x86_pmu.num_counters_fixed) {
		reg = MSR_ARCH_PERFMON_FIXED_CTR_CTRL;
		ret = rdmsrl_safe(reg, &val);
		if (ret)
			goto msr_fail;
		for (i = 0; i < x86_pmu.num_counters_fixed; i++) {
			if (val & (0x03 << i*4)) {
				bios_fail = 1;
				val_fail = val;
				reg_fail = reg;
			}
		}
	}

	/*
	 * If all the counters are enabled, the below test will always
	 * fail.  The tools will also become useless in this scenario.
	 * Just fail and disable the hardware counters.
	 */

	if (reg_safe == -1) {
		reg = reg_safe;
		goto msr_fail;
	}

	/*
	 * Read the current value, change it and read it back to see if it
	 * matches, this is needed to detect certain hardware emulators
	 * (qemu/kvm) that don't trap on the MSR access and always return 0s.
	 */
	reg = x86_pmu_event_addr(reg_safe);
	if (rdmsrl_safe(reg, &val))
		goto msr_fail;
	val ^= 0xffffUL;
	ret = wrmsrl_safe(reg, val);
	ret |= rdmsrl_safe(reg, &val_new);
	if (ret || val != val_new)
		goto msr_fail;

	/*
	 * We still allow the PMU driver to operate:
	 */
	if (bios_fail) {
		pr_cont("Broken BIOS detected, complain to your hardware vendor.\n");
		pr_err(FW_BUG "the BIOS has corrupted hw-PMU resources (MSR %x is %Lx)\n",
			      reg_fail, val_fail);
	}

	return true;

msr_fail:
	if (boot_cpu_has(X86_FEATURE_HYPERVISOR)) {
		pr_cont("PMU not available due to virtualization, using software events only.\n");
	} else {
		pr_cont("Broken PMU hardware detected, using software events only.\n");
		pr_err("Failed to access perfctr msr (MSR %x is %Lx)\n",
		       reg, val_new);
	}

	return false;
}

static void hw_perf_event_destroy(struct perf_event *event)
{
	x86_release_hardware();
	atomic_dec(&active_events);
}

void hw_perf_lbr_event_destroy(struct perf_event *event)
{
	hw_perf_event_destroy(event);

	/* undo the lbr/bts event accounting */
	x86_del_exclusive(x86_lbr_exclusive_lbr);
}

static inline int x86_pmu_initialized(void)
{
	return x86_pmu.handle_irq != NULL;
}

static inline int
set_ext_hw_attr(struct hw_perf_event *hwc, struct perf_event *event)
{
	struct perf_event_attr *attr = &event->attr;
	unsigned int cache_type, cache_op, cache_result;
	u64 config, val;

	config = attr->config;

	cache_type = (config >> 0) & 0xff;
	if (cache_type >= PERF_COUNT_HW_CACHE_MAX)
		return -EINVAL;
	cache_type = array_index_nospec(cache_type, PERF_COUNT_HW_CACHE_MAX);

	cache_op = (config >>  8) & 0xff;
	if (cache_op >= PERF_COUNT_HW_CACHE_OP_MAX)
		return -EINVAL;
	cache_op = array_index_nospec(cache_op, PERF_COUNT_HW_CACHE_OP_MAX);

	cache_result = (config >> 16) & 0xff;
	if (cache_result >= PERF_COUNT_HW_CACHE_RESULT_MAX)
		return -EINVAL;
	cache_result = array_index_nospec(cache_result, PERF_COUNT_HW_CACHE_RESULT_MAX);

	val = hw_cache_event_ids[cache_type][cache_op][cache_result];

	if (val == 0)
		return -ENOENT;

	if (val == -1)
		return -EINVAL;

	hwc->config |= val;
	attr->config1 = hw_cache_extra_regs[cache_type][cache_op][cache_result];
	return x86_pmu_extra_regs(val, event);
}

int x86_reserve_hardware(void)
{
	int err = 0;

	if (!atomic_inc_not_zero(&pmc_refcount)) {
		mutex_lock(&pmc_reserve_mutex);
		if (atomic_read(&pmc_refcount) == 0) {
			if (!reserve_pmc_hardware())
				err = -EBUSY;
			else
				reserve_ds_buffers();
		}
		if (!err)
			atomic_inc(&pmc_refcount);
		mutex_unlock(&pmc_reserve_mutex);
	}

	return err;
}

void x86_release_hardware(void)
{
	if (atomic_dec_and_mutex_lock(&pmc_refcount, &pmc_reserve_mutex)) {
		release_pmc_hardware();
		release_ds_buffers();
		mutex_unlock(&pmc_reserve_mutex);
	}
}

/*
 * Check if we can create event of a certain type (that no conflicting events
 * are present).
 */
int x86_add_exclusive(unsigned int what)
{
	int i;

	/*
	 * When lbr_pt_coexist we allow PT to coexist with either LBR or BTS.
	 * LBR and BTS are still mutually exclusive.
	 */
	if (x86_pmu.lbr_pt_coexist && what == x86_lbr_exclusive_pt)
		return 0;

	if (!atomic_inc_not_zero(&x86_pmu.lbr_exclusive[what])) {
		mutex_lock(&pmc_reserve_mutex);
		for (i = 0; i < ARRAY_SIZE(x86_pmu.lbr_exclusive); i++) {
			if (i != what && atomic_read(&x86_pmu.lbr_exclusive[i]))
				goto fail_unlock;
		}
		atomic_inc(&x86_pmu.lbr_exclusive[what]);
		mutex_unlock(&pmc_reserve_mutex);
	}

	atomic_inc(&active_events);
	return 0;

fail_unlock:
	mutex_unlock(&pmc_reserve_mutex);
	return -EBUSY;
}

void x86_del_exclusive(unsigned int what)
{
	if (x86_pmu.lbr_pt_coexist && what == x86_lbr_exclusive_pt)
		return;

	atomic_dec(&x86_pmu.lbr_exclusive[what]);
	atomic_dec(&active_events);
}

int x86_setup_perfctr(struct perf_event *event)
{
	struct perf_event_attr *attr = &event->attr;
	struct hw_perf_event *hwc = &event->hw;
	u64 config;

	if (!is_sampling_event(event)) {
		hwc->sample_period = x86_pmu.max_period;
		hwc->last_period = hwc->sample_period;
		local64_set(&hwc->period_left, hwc->sample_period);
	}

	if (attr->type == PERF_TYPE_RAW)
		return x86_pmu_extra_regs(event->attr.config, event);

	if (attr->type == PERF_TYPE_HW_CACHE)
		return set_ext_hw_attr(hwc, event);

	if (attr->config >= x86_pmu.max_events)
		return -EINVAL;

	attr->config = array_index_nospec((unsigned long)attr->config, x86_pmu.max_events);

	/*
	 * The generic map:
	 */
	config = x86_pmu.event_map(attr->config);

	if (config == 0)
		return -ENOENT;

	if (config == -1LL)
		return -EINVAL;

	hwc->config |= config;

	return 0;
}

/*
 * check that branch_sample_type is compatible with
 * settings needed for precise_ip > 1 which implies
 * using the LBR to capture ALL taken branches at the
 * priv levels of the measurement
 */
static inline int precise_br_compat(struct perf_event *event)
{
	u64 m = event->attr.branch_sample_type;
	u64 b = 0;

	/* must capture all branches */
	if (!(m & PERF_SAMPLE_BRANCH_ANY))
		return 0;

	m &= PERF_SAMPLE_BRANCH_KERNEL | PERF_SAMPLE_BRANCH_USER;

	if (!event->attr.exclude_user)
		b |= PERF_SAMPLE_BRANCH_USER;

	if (!event->attr.exclude_kernel)
		b |= PERF_SAMPLE_BRANCH_KERNEL;

	/*
	 * ignore PERF_SAMPLE_BRANCH_HV, not supported on x86
	 */

	return m == b;
}

int x86_pmu_max_precise(void)
{
	int precise = 0;

	/* Support for constant skid */
	if (x86_pmu.pebs_active && !x86_pmu.pebs_broken) {
		precise++;

		/* Support for IP fixup */
		if (x86_pmu.lbr_nr || x86_pmu.intel_cap.pebs_format >= 2)
			precise++;

		if (x86_pmu.pebs_prec_dist)
			precise++;
	}
	return precise;
}

int x86_pmu_hw_config(struct perf_event *event)
{
	if (event->attr.precise_ip) {
		int precise = x86_pmu_max_precise();

		if (event->attr.precise_ip > precise)
			return -EOPNOTSUPP;

		/* There's no sense in having PEBS for non sampling events: */
		if (!is_sampling_event(event))
			return -EINVAL;
	}
	/*
	 * check that PEBS LBR correction does not conflict with
	 * whatever the user is asking with attr->branch_sample_type
	 */
	if (event->attr.precise_ip > 1 && x86_pmu.intel_cap.pebs_format < 2) {
		u64 *br_type = &event->attr.branch_sample_type;

		if (has_branch_stack(event)) {
			if (!precise_br_compat(event))
				return -EOPNOTSUPP;

			/* branch_sample_type is compatible */

		} else {
			/*
			 * user did not specify  branch_sample_type
			 *
			 * For PEBS fixups, we capture all
			 * the branches at the priv level of the
			 * event.
			 */
			*br_type = PERF_SAMPLE_BRANCH_ANY;

			if (!event->attr.exclude_user)
				*br_type |= PERF_SAMPLE_BRANCH_USER;

			if (!event->attr.exclude_kernel)
				*br_type |= PERF_SAMPLE_BRANCH_KERNEL;
		}
	}

	if (event->attr.branch_sample_type & PERF_SAMPLE_BRANCH_CALL_STACK)
		event->attach_state |= PERF_ATTACH_TASK_DATA;

	/*
	 * Generate PMC IRQs:
	 * (keep 'enabled' bit clear for now)
	 */
	event->hw.config = ARCH_PERFMON_EVENTSEL_INT;

	/*
	 * Count user and OS events unless requested not to
	 */
	if (!event->attr.exclude_user)
		event->hw.config |= ARCH_PERFMON_EVENTSEL_USR;
	if (!event->attr.exclude_kernel)
		event->hw.config |= ARCH_PERFMON_EVENTSEL_OS;

	if (event->attr.type == PERF_TYPE_RAW)
		event->hw.config |= event->attr.config & X86_RAW_EVENT_MASK;

	if (event->attr.sample_period && x86_pmu.limit_period) {
		if (x86_pmu.limit_period(event, event->attr.sample_period) >
				event->attr.sample_period)
			return -EINVAL;
	}

	return x86_setup_perfctr(event);
}

/*
 * Setup the hardware configuration for a given attr_type
 */
static int __x86_pmu_event_init(struct perf_event *event)
{
	int err;

	if (!x86_pmu_initialized())
		return -ENODEV;

	err = x86_reserve_hardware();
	if (err)
		return err;

	atomic_inc(&active_events);
	event->destroy = hw_perf_event_destroy;

	event->hw.idx = -1;
	event->hw.last_cpu = -1;
	event->hw.last_tag = ~0ULL;

	/* mark unused */
	event->hw.extra_reg.idx = EXTRA_REG_NONE;
	event->hw.branch_reg.idx = EXTRA_REG_NONE;

	return x86_pmu.hw_config(event);
}

void x86_pmu_disable_all(void)
{
	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
	int idx;

	for (idx = 0; idx < x86_pmu.num_counters; idx++) {
		u64 val;

		if (!test_bit(idx, cpuc->active_mask))
			continue;
		rdmsrl(x86_pmu_config_addr(idx), val);
		if (!(val & ARCH_PERFMON_EVENTSEL_ENABLE))
			continue;
		val &= ~ARCH_PERFMON_EVENTSEL_ENABLE;
		wrmsrl(x86_pmu_config_addr(idx), val);
	}
}

/*
 * There may be PMI landing after enabled=0. The PMI hitting could be before or
 * after disable_all.
 *
 * If PMI hits before disable_all, the PMU will be disabled in the NMI handler.
 * It will not be re-enabled in the NMI handler again, because enabled=0. After
 * handling the NMI, disable_all will be called, which will not change the
 * state either. If PMI hits after disable_all, the PMU is already disabled
 * before entering NMI handler. The NMI handler will not change the state
 * either.
 *
 * So either situation is harmless.
 */
static void x86_pmu_disable(struct pmu *pmu)
{
	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);

	if (!x86_pmu_initialized())
		return;

	if (!cpuc->enabled)
		return;

	cpuc->n_added = 0;
	cpuc->enabled = 0;
	barrier();

	x86_pmu.disable_all();
}

void x86_pmu_enable_all(int added)
{
	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
	int idx;

	for (idx = 0; idx < x86_pmu.num_counters; idx++) {
		struct hw_perf_event *hwc = &cpuc->events[idx]->hw;

		if (!test_bit(idx, cpuc->active_mask))
			continue;

		__x86_pmu_enable_event(hwc, ARCH_PERFMON_EVENTSEL_ENABLE);
	}
}

static struct pmu pmu;

static inline int is_x86_event(struct perf_event *event)
{
	return event->pmu == &pmu;
}

/*
 * Event scheduler state:
 *
 * Assign events iterating over all events and counters, beginning
 * with events with least weights first. Keep the current iterator
 * state in struct sched_state.
 */
struct sched_state {
	int	weight;
	int	event;		/* event index */
	int	counter;	/* counter index */
	int	unassigned;	/* number of events to be assigned left */
	int	nr_gp;		/* number of GP counters used */
	unsigned long used[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
};

/* Total max is X86_PMC_IDX_MAX, but we are O(n!) limited */
#define	SCHED_STATES_MAX	2

struct perf_sched {
	int			max_weight;
	int			max_events;
	int			max_gp;
	int			saved_states;
	struct event_constraint	**constraints;
	struct sched_state	state;
	struct sched_state	saved[SCHED_STATES_MAX];
};

/*
 * Initialize interator that runs through all events and counters.
 */
static void perf_sched_init(struct perf_sched *sched, struct event_constraint **constraints,
			    int num, int wmin, int wmax, int gpmax)
{
	int idx;

	memset(sched, 0, sizeof(*sched));
	sched->max_events	= num;
	sched->max_weight	= wmax;
	sched->max_gp		= gpmax;
	sched->constraints	= constraints;

	for (idx = 0; idx < num; idx++) {
		if (constraints[idx]->weight == wmin)
			break;
	}

	sched->state.event	= idx;		/* start with min weight */
	sched->state.weight	= wmin;
	sched->state.unassigned	= num;
}

static void perf_sched_save_state(struct perf_sched *sched)
{
	if (WARN_ON_ONCE(sched->saved_states >= SCHED_STATES_MAX))
		return;

	sched->saved[sched->saved_states] = sched->state;
	sched->saved_states++;
}

static bool perf_sched_restore_state(struct perf_sched *sched)
{
	if (!sched->saved_states)
		return false;

	sched->saved_states--;
	sched->state = sched->saved[sched->saved_states];

	/* continue with next counter: */
	clear_bit(sched->state.counter++, sched->state.used);

	return true;
}

/*
 * Select a counter for the current event to schedule. Return true on
 * success.
 */
static bool __perf_sched_find_counter(struct perf_sched *sched)
{
	struct event_constraint *c;
	int idx;

	if (!sched->state.unassigned)
		return false;

	if (sched->state.event >= sched->max_events)
		return false;

	c = sched->constraints[sched->state.event];
	/* Prefer fixed purpose counters */
	if (c->idxmsk64 & (~0ULL << INTEL_PMC_IDX_FIXED)) {
		idx = INTEL_PMC_IDX_FIXED;
		for_each_set_bit_from(idx, c->idxmsk, X86_PMC_IDX_MAX) {
			if (!__test_and_set_bit(idx, sched->state.used))
				goto done;
		}
	}

	/* Grab the first unused counter starting with idx */
	idx = sched->state.counter;
	for_each_set_bit_from(idx, c->idxmsk, INTEL_PMC_IDX_FIXED) {
		if (!__test_and_set_bit(idx, sched->state.used)) {
			if (sched->state.nr_gp++ >= sched->max_gp)
				return false;

			goto done;
		}
	}

	return false;

done:
	sched->state.counter = idx;

	if (c->overlap)
		perf_sched_save_state(sched);

	return true;
}

static bool perf_sched_find_counter(struct perf_sched *sched)
{
	while (!__perf_sched_find_counter(sched)) {
		if (!perf_sched_restore_state(sched))
			return false;
	}

	return true;
}

/*
 * Go through all unassigned events and find the next one to schedule.
 * Take events with the least weight first. Return true on success.
 */
static bool perf_sched_next_event(struct perf_sched *sched)
{
	struct event_constraint *c;

	if (!sched->state.unassigned || !--sched->state.unassigned)
		return false;

	do {
		/* next event */
		sched->state.event++;
		if (sched->state.event >= sched->max_events) {
			/* next weight */
			sched->state.event = 0;
			sched->state.weight++;
			if (sched->state.weight > sched->max_weight)
				return false;
		}
		c = sched->constraints[sched->state.event];
	} while (c->weight != sched->state.weight);

	sched->state.counter = 0;	/* start with first counter */

	return true;
}

/*
 * Assign a counter for each event.
 */
int perf_assign_events(struct event_constraint **constraints, int n,
			int wmin, int wmax, int gpmax, int *assign)
{
	struct perf_sched sched;

	perf_sched_init(&sched, constraints, n, wmin, wmax, gpmax);

	do {
		if (!perf_sched_find_counter(&sched))
			break;	/* failed */
		if (assign)
			assign[sched.state.event] = sched.state.counter;
	} while (perf_sched_next_event(&sched));

	return sched.state.unassigned;
}
EXPORT_SYMBOL_GPL(perf_assign_events);

int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
{
	struct event_constraint *c;
	unsigned long used_mask[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
	struct perf_event *e;
	int i, wmin, wmax, unsched = 0;
	struct hw_perf_event *hwc;

	bitmap_zero(used_mask, X86_PMC_IDX_MAX);

	if (x86_pmu.start_scheduling)
		x86_pmu.start_scheduling(cpuc);

	for (i = 0, wmin = X86_PMC_IDX_MAX, wmax = 0; i < n; i++) {
		cpuc->event_constraint[i] = NULL;
		c = x86_pmu.get_event_constraints(cpuc, i, cpuc->event_list[i]);
		cpuc->event_constraint[i] = c;

		wmin = min(wmin, c->weight);
		wmax = max(wmax, c->weight);
	}

	/*
	 * fastpath, try to reuse previous register
	 */
	for (i = 0; i < n; i++) {
		hwc = &cpuc->event_list[i]->hw;
		c = cpuc->event_constraint[i];

		/* never assigned */
		if (hwc->idx == -1)
			break;

		/* constraint still honored */
		if (!test_bit(hwc->idx, c->idxmsk))
			break;

		/* not already used */
		if (test_bit(hwc->idx, used_mask))
			break;

		__set_bit(hwc->idx, used_mask);
		if (assign)
			assign[i] = hwc->idx;
	}

	/* slow path */
	if (i != n) {
		int gpmax = x86_pmu.num_counters;

		/*
		 * Do not allow scheduling of more than half the available
		 * generic counters.
		 *
		 * This helps avoid counter starvation of sibling thread by
		 * ensuring at most half the counters cannot be in exclusive
		 * mode. There is no designated counters for the limits. Any
		 * N/2 counters can be used. This helps with events with
		 * specific counter constraints.
		 */
		if (is_ht_workaround_enabled() && !cpuc->is_fake &&
		    READ_ONCE(cpuc->excl_cntrs->exclusive_present))
			gpmax /= 2;

		unsched = perf_assign_events(cpuc->event_constraint, n, wmin,
					     wmax, gpmax, assign);
	}

	/*
	 * In case of success (unsched = 0), mark events as committed,
	 * so we do not put_constraint() in case new events are added
	 * and fail to be scheduled
	 *
	 * We invoke the lower level commit callback to lock the resource
	 *
	 * We do not need to do all of this in case we are called to
	 * validate an event group (assign == NULL)
	 */
	if (!unsched && assign) {
		for (i = 0; i < n; i++) {
			e = cpuc->event_list[i];
			e->hw.flags |= PERF_X86_EVENT_COMMITTED;
			if (x86_pmu.commit_scheduling)
				x86_pmu.commit_scheduling(cpuc, i, assign[i]);
		}
	} else {
		for (i = 0; i < n; i++) {
			e = cpuc->event_list[i];
			/*
			 * do not put_constraint() on comitted events,
			 * because they are good to go
			 */
			if ((e->hw.flags & PERF_X86_EVENT_COMMITTED))
				continue;

			/*
			 * release events that failed scheduling
			 */
			if (x86_pmu.put_event_constraints)
				x86_pmu.put_event_constraints(cpuc, e);
		}
	}

	if (x86_pmu.stop_scheduling)
		x86_pmu.stop_scheduling(cpuc);

	return unsched ? -EINVAL : 0;
}

/*
 * dogrp: true if must collect siblings events (group)
 * returns total number of events and error code
 */
static int collect_events(struct cpu_hw_events *cpuc, struct perf_event *leader, bool dogrp)
{
	struct perf_event *event;
	int n, max_count;

	max_count = x86_pmu.num_counters + x86_pmu.num_counters_fixed;

	/* current number of events already accepted */
	n = cpuc->n_events;

	if (is_x86_event(leader)) {
		if (n >= max_count)
			return -EINVAL;
		cpuc->event_list[n] = leader;
		n++;
	}
	if (!dogrp)
		return n;

	for_each_sibling_event(event, leader) {
		if (!is_x86_event(event) ||
		    event->state <= PERF_EVENT_STATE_OFF)
			continue;

		if (n >= max_count)
			return -EINVAL;

		cpuc->event_list[n] = event;
		n++;
	}
	return n;
}

static inline void x86_assign_hw_event(struct perf_event *event,
				struct cpu_hw_events *cpuc, int i)
{
	struct hw_perf_event *hwc = &event->hw;

	hwc->idx = cpuc->assign[i];
	hwc->last_cpu = smp_processor_id();
	hwc->last_tag = ++cpuc->tags[i];

	if (hwc->idx == INTEL_PMC_IDX_FIXED_BTS) {
		hwc->config_base = 0;
		hwc->event_base	= 0;
	} else if (hwc->idx >= INTEL_PMC_IDX_FIXED) {
		hwc->config_base = MSR_ARCH_PERFMON_FIXED_CTR_CTRL;
		hwc->event_base = MSR_ARCH_PERFMON_FIXED_CTR0 + (hwc->idx - INTEL_PMC_IDX_FIXED);
		hwc->event_base_rdpmc = (hwc->idx - INTEL_PMC_IDX_FIXED) | 1<<30;