time.c 4.1 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0
2 3 4 5 6 7 8
/*
 * Xen stolen ticks accounting.
 */
#include <linux/kernel.h>
#include <linux/kernel_stat.h>
#include <linux/math64.h>
#include <linux/gfp.h>
9
#include <linux/slab.h>
10

11
#include <asm/paravirt.h>
12 13 14 15 16 17 18 19 20 21 22 23
#include <asm/xen/hypervisor.h>
#include <asm/xen/hypercall.h>

#include <xen/events.h>
#include <xen/features.h>
#include <xen/interface/xen.h>
#include <xen/interface/vcpu.h>
#include <xen/xen-ops.h>

/* runstate info updated by Xen */
static DEFINE_PER_CPU(struct vcpu_runstate_info, xen_runstate);

24 25
static DEFINE_PER_CPU(u64[4], old_runstate_time);

26 27 28 29 30 31 32
/* return an consistent snapshot of 64-bit time/counter value */
static u64 get64(const u64 *p)
{
	u64 ret;

	if (BITS_PER_LONG < 64) {
		u32 *p32 = (u32 *)p;
S
Stefano Stabellini 已提交
33
		u32 h, l, h2;
34 35 36 37 38 39 40 41

		/*
		 * Read high then low, and then make sure high is
		 * still the same; this will only loop if low wraps
		 * and carries into high.
		 * XXX some clean way to make this endian-proof?
		 */
		do {
S
Stefano Stabellini 已提交
42 43 44 45
			h = READ_ONCE(p32[1]);
			l = READ_ONCE(p32[0]);
			h2 = READ_ONCE(p32[1]);
		} while(h2 != h);
46 47 48

		ret = (((u64)h) << 32) | l;
	} else
S
Stefano Stabellini 已提交
49
		ret = READ_ONCE(*p);
50 51 52 53

	return ret;
}

54 55
static void xen_get_runstate_snapshot_cpu_delta(
			      struct vcpu_runstate_info *res, unsigned int cpu)
56 57 58 59 60 61
{
	u64 state_time;
	struct vcpu_runstate_info *state;

	BUG_ON(preemptible());

62
	state = per_cpu_ptr(&xen_runstate, cpu);
63 64 65

	do {
		state_time = get64(&state->state_entry_time);
66
		rmb();	/* Hypervisor might update data. */
67
		*res = __READ_ONCE(*state);
68 69 70 71 72
		rmb();	/* Hypervisor might update data. */
	} while (get64(&state->state_entry_time) != state_time ||
		 (state_time & XEN_RUNSTATE_UPDATE));
}

73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137
static void xen_get_runstate_snapshot_cpu(struct vcpu_runstate_info *res,
					  unsigned int cpu)
{
	int i;

	xen_get_runstate_snapshot_cpu_delta(res, cpu);

	for (i = 0; i < 4; i++)
		res->time[i] += per_cpu(old_runstate_time, cpu)[i];
}

void xen_manage_runstate_time(int action)
{
	static struct vcpu_runstate_info *runstate_delta;
	struct vcpu_runstate_info state;
	int cpu, i;

	switch (action) {
	case -1: /* backup runstate time before suspend */
		if (unlikely(runstate_delta))
			pr_warn_once("%s: memory leak as runstate_delta is not NULL\n",
					__func__);

		runstate_delta = kmalloc_array(num_possible_cpus(),
					sizeof(*runstate_delta),
					GFP_ATOMIC);
		if (unlikely(!runstate_delta)) {
			pr_warn("%s: failed to allocate runstate_delta\n",
					__func__);
			return;
		}

		for_each_possible_cpu(cpu) {
			xen_get_runstate_snapshot_cpu_delta(&state, cpu);
			memcpy(runstate_delta[cpu].time, state.time,
					sizeof(runstate_delta[cpu].time));
		}

		break;

	case 0: /* backup runstate time after resume */
		if (unlikely(!runstate_delta)) {
			pr_warn("%s: cannot accumulate runstate time as runstate_delta is NULL\n",
					__func__);
			return;
		}

		for_each_possible_cpu(cpu) {
			for (i = 0; i < 4; i++)
				per_cpu(old_runstate_time, cpu)[i] +=
					runstate_delta[cpu].time[i];
		}

		break;

	default: /* do not accumulate runstate time for checkpointing */
		break;
	}

	if (action != -1 && runstate_delta) {
		kfree(runstate_delta);
		runstate_delta = NULL;
	}
}

138 139 140 141 142 143
/*
 * Runstate accounting
 */
void xen_get_runstate_snapshot(struct vcpu_runstate_info *res)
{
	xen_get_runstate_snapshot_cpu(res, smp_processor_id());
144 145 146 147 148 149 150 151
}

/* return true when a vcpu could run but has no real cpu to run on */
bool xen_vcpu_stolen(int vcpu)
{
	return per_cpu(xen_runstate, vcpu).state == RUNSTATE_runnable;
}

152
u64 xen_steal_clock(int cpu)
153 154 155
{
	struct vcpu_runstate_info state;

156
	xen_get_runstate_snapshot_cpu(&state, cpu);
157 158 159
	return state.time[RUNSTATE_runnable] + state.time[RUNSTATE_offline];
}

160 161 162 163 164 165 166
void xen_setup_runstate_info(int cpu)
{
	struct vcpu_register_runstate_memory_area area;

	area.addr.v = &per_cpu(xen_runstate, cpu);

	if (HYPERVISOR_vcpu_op(VCPUOP_register_runstate_memory_area,
167
			       xen_vcpu_nr(cpu), &area))
168 169 170
		BUG();
}

171 172
void __init xen_time_setup_guest(void)
{
173 174 175 176 177
	bool xen_runstate_remote;

	xen_runstate_remote = !HYPERVISOR_vm_assist(VMASST_CMD_enable,
					VMASST_TYPE_runstate_update_flag);

178
	pv_ops.time.steal_clock = xen_steal_clock;
179 180

	static_key_slow_inc(&paravirt_steal_enabled);
181 182
	if (xen_runstate_remote)
		static_key_slow_inc(&paravirt_steal_rq_enabled);
183
}