cpuidle-pseries.c 6.0 KB
Newer Older
1
/*
2
 *  cpuidle-pseries - idle state cpuidle driver.
3 4 5 6 7 8 9 10 11 12 13
 *  Adapted from drivers/idle/intel_idle.c and
 *  drivers/acpi/processor_idle.c
 *
 */

#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/moduleparam.h>
#include <linux/cpuidle.h>
#include <linux/cpu.h>
14
#include <linux/notifier.h>
15 16 17 18 19

#include <asm/paca.h>
#include <asm/reg.h>
#include <asm/machdep.h>
#include <asm/firmware.h>
20
#include <asm/runlatch.h>
21
#include <asm/plpar_wrappers.h>
22 23

struct cpuidle_driver pseries_idle_driver = {
24 25
	.name             = "pseries_idle",
	.owner            = THIS_MODULE,
26 27
};

28 29 30 31
static int max_idle_state __read_mostly;
static struct cpuidle_state *cpuidle_state_table __read_mostly;
static u64 snooze_timeout __read_mostly;
static bool snooze_timeout_en __read_mostly;
32

33
static inline void idle_loop_prolog(unsigned long *in_purr)
34
{
35
	ppc64_runlatch_off();
36 37 38 39 40 41 42 43
	*in_purr = mfspr(SPRN_PURR);
	/*
	 * Indicate to the HV that we are idle. Now would be
	 * a good time to find other work to dispatch.
	 */
	get_lppaca()->idle = 1;
}

44
static inline void idle_loop_epilog(unsigned long in_purr)
45
{
46 47 48 49 50
	u64 wait_cycles;

	wait_cycles = be64_to_cpu(get_lppaca()->wait_state_cycles);
	wait_cycles += mfspr(SPRN_PURR) - in_purr;
	get_lppaca()->wait_state_cycles = cpu_to_be64(wait_cycles);
51
	get_lppaca()->idle = 0;
52 53 54 55

	if (irqs_disabled())
		local_irq_enable();
	ppc64_runlatch_on();
56 57 58 59 60 61 62
}

static int snooze_loop(struct cpuidle_device *dev,
			struct cpuidle_driver *drv,
			int index)
{
	unsigned long in_purr;
63
	u64 snooze_exit_time;
64

65 66
	set_thread_flag(TIF_POLLING_NRFLAG);

67
	idle_loop_prolog(&in_purr);
68
	local_irq_enable();
69
	snooze_exit_time = get_tb() + snooze_timeout;
70

71
	while (!need_resched()) {
72 73
		HMT_low();
		HMT_very_low();
74 75
		if (snooze_timeout_en && get_tb() > snooze_exit_time)
			break;
76 77 78
	}

	HMT_medium();
79 80 81
	clear_thread_flag(TIF_POLLING_NRFLAG);
	smp_mb();

82 83
	idle_loop_epilog(in_purr);

84 85 86
	return index;
}

87 88 89
static void check_and_cede_processor(void)
{
	/*
90 91 92
	 * Ensure our interrupt state is properly tracked,
	 * also checks if no interrupt has occurred while we
	 * were soft-disabled
93
	 */
94
	if (prep_irq_for_idle()) {
95
		cede_processor();
96 97 98 99 100 101
#ifdef CONFIG_TRACE_IRQFLAGS
		/* Ensure that H_CEDE returns with IRQs on */
		if (WARN_ON(!(mfmsr() & MSR_EE)))
			__hard_irq_enable();
#endif
	}
102 103
}

104 105 106 107 108 109
static int dedicated_cede_loop(struct cpuidle_device *dev,
				struct cpuidle_driver *drv,
				int index)
{
	unsigned long in_purr;

110
	idle_loop_prolog(&in_purr);
111 112 113
	get_lppaca()->donate_dedicated_cpu = 1;

	HMT_medium();
114
	check_and_cede_processor();
115 116

	get_lppaca()->donate_dedicated_cpu = 0;
117 118 119

	idle_loop_epilog(in_purr);

120 121 122 123 124 125 126 127 128
	return index;
}

static int shared_cede_loop(struct cpuidle_device *dev,
			struct cpuidle_driver *drv,
			int index)
{
	unsigned long in_purr;

129
	idle_loop_prolog(&in_purr);
130 131 132 133 134 135 136 137

	/*
	 * Yield the processor to the hypervisor.  We return if
	 * an external interrupt occurs (which are driven prior
	 * to returning here) or if a prod occurs from another
	 * processor. When returning here, external interrupts
	 * are enabled.
	 */
138
	check_and_cede_processor();
139

140 141
	idle_loop_epilog(in_purr);

142 143 144 145 146 147
	return index;
}

/*
 * States for dedicated partition case.
 */
148
static struct cpuidle_state dedicated_states[] = {
149 150 151 152 153 154 155 156 157
	{ /* Snooze */
		.name = "snooze",
		.desc = "snooze",
		.exit_latency = 0,
		.target_residency = 0,
		.enter = &snooze_loop },
	{ /* CEDE */
		.name = "CEDE",
		.desc = "CEDE",
158 159
		.exit_latency = 10,
		.target_residency = 100,
160 161 162 163 164 165
		.enter = &dedicated_cede_loop },
};

/*
 * States for shared partition case.
 */
166
static struct cpuidle_state shared_states[] = {
167 168 169 170 171 172 173 174
	{ /* Shared Cede */
		.name = "Shared Cede",
		.desc = "Shared Cede",
		.exit_latency = 0,
		.target_residency = 0,
		.enter = &shared_cede_loop },
};

175
static int pseries_cpuidle_cpu_online(unsigned int cpu)
176
{
177
	struct cpuidle_device *dev = per_cpu(cpuidle_devices, cpu);
178

179
	if (dev && cpuidle_get_driver()) {
180 181 182 183 184 185
		cpuidle_pause_and_lock();
		cpuidle_enable_device(dev);
		cpuidle_resume_and_unlock();
	}
	return 0;
}
186

187 188 189
static int pseries_cpuidle_cpu_dead(unsigned int cpu)
{
	struct cpuidle_device *dev = per_cpu(cpuidle_devices, cpu);
190

191 192 193 194
	if (dev && cpuidle_get_driver()) {
		cpuidle_pause_and_lock();
		cpuidle_disable_device(dev);
		cpuidle_resume_and_unlock();
195
	}
196
	return 0;
197 198 199 200 201 202 203 204 205 206 207 208
}

/*
 * pseries_cpuidle_driver_init()
 */
static int pseries_cpuidle_driver_init(void)
{
	int idle_state;
	struct cpuidle_driver *drv = &pseries_idle_driver;

	drv->state_count = 0;

209 210
	for (idle_state = 0; idle_state < max_idle_state; ++idle_state) {
		/* Is the state not enabled? */
211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229
		if (cpuidle_state_table[idle_state].enter == NULL)
			continue;

		drv->states[drv->state_count] =	/* structure copy */
			cpuidle_state_table[idle_state];

		drv->state_count += 1;
	}

	return 0;
}

/*
 * pseries_idle_probe()
 * Choose state table for shared versus dedicated partition
 */
static int pseries_idle_probe(void)
{

230 231 232
	if (cpuidle_disable != IDLE_NO_OVERRIDE)
		return -ENODEV;

233
	if (firmware_has_feature(FW_FEATURE_SPLPAR)) {
234
		if (lppaca_shared_proc(get_lppaca())) {
235
			cpuidle_state_table = shared_states;
236 237
			max_idle_state = ARRAY_SIZE(shared_states);
		} else {
238
			cpuidle_state_table = dedicated_states;
239 240
			max_idle_state = ARRAY_SIZE(dedicated_states);
		}
241 242
	} else
		return -ENODEV;
243

244 245 246 247 248
	if (max_idle_state > 1) {
		snooze_timeout_en = true;
		snooze_timeout = cpuidle_state_table[1].target_residency *
				 tb_ticks_per_usec;
	}
249 250 251 252 253 254 255 256 257 258 259 260
	return 0;
}

static int __init pseries_processor_idle_init(void)
{
	int retval;

	retval = pseries_idle_probe();
	if (retval)
		return retval;

	pseries_cpuidle_driver_init();
261
	retval = cpuidle_register(&pseries_idle_driver, NULL);
262 263 264 265 266
	if (retval) {
		printk(KERN_DEBUG "Registration of pseries driver failed.\n");
		return retval;
	}

267 268 269 270 271 272 273 274
	retval = cpuhp_setup_state_nocalls(CPUHP_AP_ONLINE_DYN,
					   "cpuidle/pseries:online",
					   pseries_cpuidle_cpu_online, NULL);
	WARN_ON(retval < 0);
	retval = cpuhp_setup_state_nocalls(CPUHP_CPUIDLE_DEAD,
					   "cpuidle/pseries:DEAD", NULL,
					   pseries_cpuidle_cpu_dead);
	WARN_ON(retval < 0);
275 276 277 278
	printk(KERN_DEBUG "pseries_idle_driver registered\n");
	return 0;
}

279
device_initcall(pseries_processor_idle_init);