[linux-2.6-block.git] / arch / x86 / kernel / process.c

#include <linux/errno.h>
#include <linux/kernel.h>
#include <linux/mm.h>
#include <linux/smp.h>
#include <linux/slab.h>
#include <linux/sched.h>
#include <linux/module.h>
#include <linux/pm.h>
#include <linux/clockchips.h>
#include <asm/system.h>

unsigned long idle_halt;
EXPORT_SYMBOL(idle_halt);
unsigned long idle_nomwait;
EXPORT_SYMBOL(idle_nomwait);

struct kmem_cache *task_xstate_cachep;

int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src)
{
	*dst = *src;
	if (src->thread.xstate) {
		dst->thread.xstate = kmem_cache_alloc(task_xstate_cachep,
						      GFP_KERNEL);
		if (!dst->thread.xstate)
			return -ENOMEM;
		WARN_ON((unsigned long)dst->thread.xstate & 15);
		memcpy(dst->thread.xstate, src->thread.xstate, xstate_size);
	}
	return 0;
}

void free_thread_xstate(struct task_struct *tsk)
{
	if (tsk->thread.xstate) {
		kmem_cache_free(task_xstate_cachep, tsk->thread.xstate);
		tsk->thread.xstate = NULL;
	}
}

void free_thread_info(struct thread_info *ti)
{
	free_thread_xstate(ti->task);
	free_pages((unsigned long)ti, get_order(THREAD_SIZE));
}

void arch_task_cache_init(void)
{
        task_xstate_cachep =
        	kmem_cache_create("task_xstate", xstate_size,
				  __alignof__(union thread_xstate),
				  SLAB_PANIC, NULL);
}

/*
 * Idle related variables and functions
 */
unsigned long boot_option_idle_override = 0;
EXPORT_SYMBOL(boot_option_idle_override);

/*
 * Powermanagement idle function, if any..
 */
void (*pm_idle)(void);
EXPORT_SYMBOL(pm_idle);

#ifdef CONFIG_X86_32
/*
 * This halt magic was a workaround for ancient floppy DMA
 * wreckage. It should be safe to remove.
 */
static int hlt_counter;
void disable_hlt(void)
{
	hlt_counter++;
}
EXPORT_SYMBOL(disable_hlt);

void enable_hlt(void)
{
	hlt_counter--;
}
EXPORT_SYMBOL(enable_hlt);

static inline int hlt_use_halt(void)
{
	return (!hlt_counter && boot_cpu_data.hlt_works_ok);
}
#else
static inline int hlt_use_halt(void)
{
	return 1;
}
#endif

/*
 * We use this if we don't have any better
 * idle routine..
 */
void default_idle(void)
{
	if (hlt_use_halt()) {
		current_thread_info()->status &= ~TS_POLLING;
		/*
		 * TS_POLLING-cleared state must be visible before we
		 * test NEED_RESCHED:
		 */
		smp_mb();

		if (!need_resched())
			safe_halt();	/* enables interrupts racelessly */
		else
			local_irq_enable();
		current_thread_info()->status |= TS_POLLING;
	} else {
		local_irq_enable();
		/* loop is done by the caller */
		cpu_relax();
	}
}
#ifdef CONFIG_APM_MODULE
EXPORT_SYMBOL(default_idle);
#endif

static void do_nothing(void *unused)
{
}

/*
 * cpu_idle_wait - Used to ensure that all the CPUs discard old value of
 * pm_idle and update to new pm_idle value. Required while changing pm_idle
 * handler on SMP systems.
 *
 * Caller must have changed pm_idle to the new value before the call. Old
 * pm_idle value will not be used by any CPU after the return of this function.
 */
void cpu_idle_wait(void)
{
	smp_mb();
	/* kick all the CPUs so that they exit out of pm_idle */
	smp_call_function(do_nothing, NULL, 1);
}
EXPORT_SYMBOL_GPL(cpu_idle_wait);

/*
 * This uses new MONITOR/MWAIT instructions on P4 processors with PNI,
 * which can obviate IPI to trigger checking of need_resched.
 * We execute MONITOR against need_resched and enter optimized wait state
 * through MWAIT. Whenever someone changes need_resched, we would be woken
 * up from MWAIT (without an IPI).
 *
 * New with Core Duo processors, MWAIT can take some hints based on CPU
 * capability.
 */
void mwait_idle_with_hints(unsigned long ax, unsigned long cx)
{
	if (!need_resched()) {
		__monitor((void *)&current_thread_info()->flags, 0, 0);
		smp_mb();
		if (!need_resched())
			__mwait(ax, cx);
	}
}

/* Default MONITOR/MWAIT with no hints, used for default C1 state */
static void mwait_idle(void)
{
	if (!need_resched()) {
		__monitor((void *)&current_thread_info()->flags, 0, 0);
		smp_mb();
		if (!need_resched())
			__sti_mwait(0, 0);
		else
			local_irq_enable();
	} else
		local_irq_enable();
}

/*
 * On SMP it's slightly faster (but much more power-consuming!)
 * to poll the ->work.need_resched flag instead of waiting for the
 * cross-CPU IPI to arrive. Use this option with caution.
 */
static void poll_idle(void)
{
	local_irq_enable();
	cpu_relax();
}

/*
 * mwait selection logic:
 *
 * It depends on the CPU. For AMD CPUs that support MWAIT this is
 * wrong. Family 0x10 and 0x11 CPUs will enter C1 on HLT. Powersavings
 * then depend on a clock divisor and current Pstate of the core. If
 * all cores of a processor are in halt state (C1) the processor can
 * enter the C1E (C1 enhanced) state. If mwait is used this will never
 * happen.
 *
 * idle=mwait overrides this decision and forces the usage of mwait.
 */

#define MWAIT_INFO			0x05
#define MWAIT_ECX_EXTENDED_INFO		0x01
#define MWAIT_EDX_C1			0xf0

static int __cpuinit mwait_usable(const struct cpuinfo_x86 *c)
{
	u32 eax, ebx, ecx, edx;

	if (force_mwait)
		return 1;

	if (c->cpuid_level < MWAIT_INFO)
		return 0;

	cpuid(MWAIT_INFO, &eax, &ebx, &ecx, &edx);
	/* Check, whether EDX has extended info about MWAIT */
	if (!(ecx & MWAIT_ECX_EXTENDED_INFO))
		return 1;

	/*
	 * edx enumeratios MONITOR/MWAIT extensions. Check, whether
	 * C1  supports MWAIT
	 */
	return (edx & MWAIT_EDX_C1);
}

/*
 * Check for AMD CPUs, which have potentially C1E support
 */
static int __cpuinit check_c1e_idle(const struct cpuinfo_x86 *c)
{
	if (c->x86_vendor != X86_VENDOR_AMD)
		return 0;

	if (c->x86 < 0x0F)
		return 0;

	/* Family 0x0f models < rev F do not have C1E */
	if (c->x86 == 0x0f && c->x86_model < 0x40)
		return 0;

	return 1;
}

/*
 * C1E aware idle routine. We check for C1E active in the interrupt
 * pending message MSR. If we detect C1E, then we handle it the same
 * way as C3 power states (local apic timer and TSC stop)
 */
static void c1e_idle(void)
{
	static cpumask_t c1e_mask = CPU_MASK_NONE;
	static int c1e_detected;

	if (need_resched())
		return;

	if (!c1e_detected) {
		u32 lo, hi;

		rdmsr(MSR_K8_INT_PENDING_MSG, lo, hi);
		if (lo & K8_INTP_C1E_ACTIVE_MASK) {
			c1e_detected = 1;
			mark_tsc_unstable("TSC halt in C1E");
			printk(KERN_INFO "System has C1E enabled\n");
		}
	}

	if (c1e_detected) {
		int cpu = smp_processor_id();

		if (!cpu_isset(cpu, c1e_mask)) {
			cpu_set(cpu, c1e_mask);
			/*
			 * Force broadcast so ACPI can not interfere. Needs
			 * to run with interrupts enabled as it uses
			 * smp_function_call.
			 */
			local_irq_enable();
			clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_FORCE,
					   &cpu);
			printk(KERN_INFO "Switch to broadcast mode on CPU%d\n",
			       cpu);
			local_irq_disable();
		}
		clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_ENTER, &cpu);

		default_idle();

		/*
		 * The switch back from broadcast mode needs to be
		 * called with interrupts disabled.
		 */
		 local_irq_disable();
		 clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_EXIT, &cpu);
		 local_irq_enable();
	} else
		default_idle();
}

void __cpuinit select_idle_routine(const struct cpuinfo_x86 *c)
{
#ifdef CONFIG_X86_SMP
	if (pm_idle == poll_idle && smp_num_siblings > 1) {
		printk(KERN_WARNING "WARNING: polling idle and HT enabled,"
			" performance may degrade.\n");
	}
#endif
	if (pm_idle)
		return;

	if (cpu_has(c, X86_FEATURE_MWAIT) && mwait_usable(c)) {
		/*
		 * One CPU supports mwait => All CPUs supports mwait
		 */
		printk(KERN_INFO "using mwait in idle threads.\n");
		pm_idle = mwait_idle;
	} else if (check_c1e_idle(c)) {
		printk(KERN_INFO "using C1E aware idle routine\n");
		pm_idle = c1e_idle;
	} else
		pm_idle = default_idle;
}

static int __init idle_setup(char *str)
{
	if (!str)
		return -EINVAL;

	if (!strcmp(str, "poll")) {
		printk("using polling idle threads.\n");
		pm_idle = poll_idle;
	} else if (!strcmp(str, "mwait"))
		force_mwait = 1;
	else if (!strcmp(str, "halt")) {
		/*
		 * When the boot option of idle=halt is added, halt is
		 * forced to be used for CPU idle. In such case CPU C2/C3
		 * won't be used again.
		 * To continue to load the CPU idle driver, don't touch
		 * the boot_option_idle_override.
		 */
		pm_idle = default_idle;
		idle_halt = 1;
		return 0;
	} else if (!strcmp(str, "nomwait")) {
		/*
		 * If the boot option of "idle=nomwait" is added,
		 * it means that mwait will be disabled for CPU C2/C3
		 * states. In such case it won't touch the variable
		 * of boot_option_idle_override.
		 */
		idle_nomwait = 1;
		return 0;
	} else
		return -1;

	boot_option_idle_override = 1;
	return 0;
}
early_param("idle", idle_setup);
Commit	Line	Data
61c4628b SS	1	#include <linux/errno.h>
	2	#include <linux/kernel.h>
	3	#include <linux/mm.h>
	4	#include <linux/smp.h>
	5	#include <linux/slab.h>
	6	#include <linux/sched.h>
7f424a8b PZ	7	#include <linux/module.h>
7f424a8b PZ	8	#include <linux/pm.h>
aa276e1c	9	#include <linux/clockchips.h>
c1e3b377 ZY	10	#include <asm/system.h>
	11
	12	unsigned long idle_halt;
	13	EXPORT_SYMBOL(idle_halt);
da5e09a1 ZY	14	unsigned long idle_nomwait;
da5e09a1 ZY	15	EXPORT_SYMBOL(idle_nomwait);
61c4628b	16
aa283f49	17	struct kmem_cache *task_xstate_cachep;
61c4628b SS	18
	19	int arch_dup_task_struct(struct task_struct dst, struct task_struct src)
	20	{
	21	dst = src;
aa283f49 SS	22	if (src->thread.xstate) {
	23	dst->thread.xstate = kmem_cache_alloc(task_xstate_cachep,
	24	GFP_KERNEL);
	25	if (!dst->thread.xstate)
	26	return -ENOMEM;
	27	WARN_ON((unsigned long)dst->thread.xstate & 15);
	28	memcpy(dst->thread.xstate, src->thread.xstate, xstate_size);
	29	}
61c4628b SS	30	return 0;
	31	}
	32
aa283f49	33	void free_thread_xstate(struct task_struct *tsk)
61c4628b	34	{
aa283f49 SS	35	if (tsk->thread.xstate) {
	36	kmem_cache_free(task_xstate_cachep, tsk->thread.xstate);
	37	tsk->thread.xstate = NULL;
	38	}
	39	}
	40
aa283f49 SS	41	void free_thread_info(struct thread_info *ti)
	42	{
	43	free_thread_xstate(ti->task);
1679f271	44	free_pages((unsigned long)ti, get_order(THREAD_SIZE));
61c4628b SS	45	}
	46
	47	void arch_task_cache_init(void)
	48	{
	49	task_xstate_cachep =
	50	kmem_cache_create("task_xstate", xstate_size,
	51	__alignof__(union thread_xstate),
	52	SLAB_PANIC, NULL);
	53	}
7f424a8b	54
00dba564 TG	55	/*
	56	* Idle related variables and functions
	57	*/
	58	unsigned long boot_option_idle_override = 0;
	59	EXPORT_SYMBOL(boot_option_idle_override);
	60
	61	/*
	62	* Powermanagement idle function, if any..
	63	*/
	64	void (*pm_idle)(void);
	65	EXPORT_SYMBOL(pm_idle);
	66
	67	#ifdef CONFIG_X86_32
	68	/*
	69	* This halt magic was a workaround for ancient floppy DMA
	70	* wreckage. It should be safe to remove.
	71	*/
	72	static int hlt_counter;
	73	void disable_hlt(void)
	74	{
	75	hlt_counter++;
	76	}
	77	EXPORT_SYMBOL(disable_hlt);
	78
	79	void enable_hlt(void)
	80	{
	81	hlt_counter--;
	82	}
	83	EXPORT_SYMBOL(enable_hlt);
	84
	85	static inline int hlt_use_halt(void)
	86	{
	87	return (!hlt_counter && boot_cpu_data.hlt_works_ok);
	88	}
	89	#else
	90	static inline int hlt_use_halt(void)
	91	{
	92	return 1;
	93	}
	94	#endif
	95
	96	/*
	97	* We use this if we don't have any better
	98	* idle routine..
	99	*/
	100	void default_idle(void)
	101	{
	102	if (hlt_use_halt()) {
	103	current_thread_info()->status &= ~TS_POLLING;
	104	/*
	105	* TS_POLLING-cleared state must be visible before we
	106	* test NEED_RESCHED:
	107	*/
	108	smp_mb();
	109
	110	if (!need_resched())
	111	safe_halt(); /* enables interrupts racelessly */
	112	else
	113	local_irq_enable();
	114	current_thread_info()->status \|= TS_POLLING;
	115	} else {
	116	local_irq_enable();
	117	/* loop is done by the caller */
	118	cpu_relax();
119	}
120	}
121	#ifdef CONFIG_APM_MODULE
122	EXPORT_SYMBOL(default_idle);
123	#endif
124
7f424a8b PZ	125	static void do_nothing(void *unused)
	126	{
	127	}
	128
	129	/*
	130	* cpu_idle_wait - Used to ensure that all the CPUs discard old value of
	131	* pm_idle and update to new pm_idle value. Required while changing pm_idle
	132	* handler on SMP systems.
	133	*
	134	* Caller must have changed pm_idle to the new value before the call. Old
	135	* pm_idle value will not be used by any CPU after the return of this function.
	136	*/
	137	void cpu_idle_wait(void)
	138	{
	139	smp_mb();
	140	/* kick all the CPUs so that they exit out of pm_idle */
127a237a	141	smp_call_function(do_nothing, NULL, 1);
7f424a8b PZ	142	}
	143	EXPORT_SYMBOL_GPL(cpu_idle_wait);
	144
	145	/*
	146	* This uses new MONITOR/MWAIT instructions on P4 processors with PNI,
	147	* which can obviate IPI to trigger checking of need_resched.
	148	* We execute MONITOR against need_resched and enter optimized wait state
	149	* through MWAIT. Whenever someone changes need_resched, we would be woken
	150	* up from MWAIT (without an IPI).
	151	*
	152	* New with Core Duo processors, MWAIT can take some hints based on CPU
	153	* capability.
	154	*/
	155	void mwait_idle_with_hints(unsigned long ax, unsigned long cx)
	156	{
	157	if (!need_resched()) {
	158	__monitor((void *)&current_thread_info()->flags, 0, 0);
	159	smp_mb();
	160	if (!need_resched())
	161	__mwait(ax, cx);
	162	}
	163	}
	164
	165	/* Default MONITOR/MWAIT with no hints, used for default C1 state */
	166	static void mwait_idle(void)
	167	{
	168	if (!need_resched()) {
	169	__monitor((void *)&current_thread_info()->flags, 0, 0);
	170	smp_mb();
	171	if (!need_resched())
	172	__sti_mwait(0, 0);
	173	else
	174	local_irq_enable();
	175	} else
	176	local_irq_enable();
	177	}
	178
7f424a8b PZ	179	/*
	180	* On SMP it's slightly faster (but much more power-consuming!)
	181	* to poll the ->work.need_resched flag instead of waiting for the
	182	* cross-CPU IPI to arrive. Use this option with caution.
	183	*/
	184	static void poll_idle(void)
	185	{
	186	local_irq_enable();
	187	cpu_relax();
	188	}
	189
e9623b35 TG	190	/*
	191	* mwait selection logic:
	192	*
	193	* It depends on the CPU. For AMD CPUs that support MWAIT this is
	194	* wrong. Family 0x10 and 0x11 CPUs will enter C1 on HLT. Powersavings
	195	* then depend on a clock divisor and current Pstate of the core. If
	196	* all cores of a processor are in halt state (C1) the processor can
	197	* enter the C1E (C1 enhanced) state. If mwait is used this will never
	198	* happen.
	199	*
	200	* idle=mwait overrides this decision and forces the usage of mwait.
	201	*/
09fd4b4e TG	202
	203	#define MWAIT_INFO 0x05
	204	#define MWAIT_ECX_EXTENDED_INFO 0x01
	205	#define MWAIT_EDX_C1 0xf0
	206
e9623b35 TG	207	static int __cpuinit mwait_usable(const struct cpuinfo_x86 *c)
e9623b35 TG	208	{
09fd4b4e TG	209	u32 eax, ebx, ecx, edx;
09fd4b4e TG	210
e9623b35 TG	211	if (force_mwait)
	212	return 1;
	213
09fd4b4e TG	214	if (c->cpuid_level < MWAIT_INFO)
	215	return 0;
	216
	217	cpuid(MWAIT_INFO, &eax, &ebx, &ecx, &edx);
	218	/* Check, whether EDX has extended info about MWAIT */
	219	if (!(ecx & MWAIT_ECX_EXTENDED_INFO))
	220	return 1;
	221
	222	/*
	223	* edx enumeratios MONITOR/MWAIT extensions. Check, whether
	224	* C1 supports MWAIT
	225	*/
	226	return (edx & MWAIT_EDX_C1);
e9623b35 TG	227	}
e9623b35 TG	228
aa276e1c TG	229	/*
	230	* Check for AMD CPUs, which have potentially C1E support
	231	*/
	232	static int __cpuinit check_c1e_idle(const struct cpuinfo_x86 *c)
	233	{
	234	if (c->x86_vendor != X86_VENDOR_AMD)
	235	return 0;
	236
	237	if (c->x86 < 0x0F)
	238	return 0;
	239
	240	/* Family 0x0f models < rev F do not have C1E */
	241	if (c->x86 == 0x0f && c->x86_model < 0x40)
	242	return 0;
	243
e9623b35 TG	244	return 1;
	245	}
	246
aa276e1c TG	247	/*
	248	* C1E aware idle routine. We check for C1E active in the interrupt
	249	* pending message MSR. If we detect C1E, then we handle it the same
	250	* way as C3 power states (local apic timer and TSC stop)
	251	*/
	252	static void c1e_idle(void)
7f424a8b	253	{
aa276e1c TG	254	static cpumask_t c1e_mask = CPU_MASK_NONE;
aa276e1c TG	255	static int c1e_detected;
7f424a8b	256
aa276e1c	257	if (need_resched())
7f424a8b	258	return;
aa276e1c TG	259
	260	if (!c1e_detected) {
	261	u32 lo, hi;
	262
	263	rdmsr(MSR_K8_INT_PENDING_MSG, lo, hi);
	264	if (lo & K8_INTP_C1E_ACTIVE_MASK) {
	265	c1e_detected = 1;
	266	mark_tsc_unstable("TSC halt in C1E");
	267	printk(KERN_INFO "System has C1E enabled\n");
	268	}
	269	}
	270
	271	if (c1e_detected) {
	272	int cpu = smp_processor_id();
	273
	274	if (!cpu_isset(cpu, c1e_mask)) {
	275	cpu_set(cpu, c1e_mask);
0beefa20 TG	276	/*
	277	* Force broadcast so ACPI can not interfere. Needs
	278	* to run with interrupts enabled as it uses
	279	* smp_function_call.
	280	*/
	281	local_irq_enable();
aa276e1c TG	282	clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_FORCE,
	283	&cpu);
	284	printk(KERN_INFO "Switch to broadcast mode on CPU%d\n",
	285	cpu);
0beefa20	286	local_irq_disable();
aa276e1c TG	287	}
aa276e1c TG	288	clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_ENTER, &cpu);
0beefa20	289
aa276e1c	290	default_idle();
0beefa20 TG	291
	292	/*
	293	* The switch back from broadcast mode needs to be
	294	* called with interrupts disabled.
	295	*/
	296	local_irq_disable();
	297	clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_EXIT, &cpu);
	298	local_irq_enable();
aa276e1c TG	299	} else
	300	default_idle();
	301	}
	302
7f424a8b PZ	303	void __cpuinit select_idle_routine(const struct cpuinfo_x86 *c)
7f424a8b PZ	304	{
7f424a8b PZ	305	#ifdef CONFIG_X86_SMP
	306	if (pm_idle == poll_idle && smp_num_siblings > 1) {
	307	printk(KERN_WARNING "WARNING: polling idle and HT enabled,"
	308	" performance may degrade.\n");
	309	}
	310	#endif
6ddd2a27 TG	311	if (pm_idle)
	312	return;
	313
e9623b35	314	if (cpu_has(c, X86_FEATURE_MWAIT) && mwait_usable(c)) {
7f424a8b	315	/*
7f424a8b PZ	316	* One CPU supports mwait => All CPUs supports mwait
7f424a8b PZ	317	*/
6ddd2a27 TG	318	printk(KERN_INFO "using mwait in idle threads.\n");
6ddd2a27 TG	319	pm_idle = mwait_idle;
aa276e1c TG	320	} else if (check_c1e_idle(c)) {
	321	printk(KERN_INFO "using C1E aware idle routine\n");
	322	pm_idle = c1e_idle;
6ddd2a27 TG	323	} else
6ddd2a27 TG	324	pm_idle = default_idle;
7f424a8b PZ	325	}
	326
	327	static int __init idle_setup(char *str)
	328	{
ab6bc3e3 CG	329	if (!str)
	330	return -EINVAL;
	331
7f424a8b PZ	332	if (!strcmp(str, "poll")) {
	333	printk("using polling idle threads.\n");
	334	pm_idle = poll_idle;
	335	} else if (!strcmp(str, "mwait"))
	336	force_mwait = 1;
c1e3b377 ZY	337	else if (!strcmp(str, "halt")) {
	338	/*
	339	* When the boot option of idle=halt is added, halt is
	340	* forced to be used for CPU idle. In such case CPU C2/C3
	341	* won't be used again.
	342	* To continue to load the CPU idle driver, don't touch
	343	* the boot_option_idle_override.
	344	*/
	345	pm_idle = default_idle;
	346	idle_halt = 1;
	347	return 0;
da5e09a1 ZY	348	} else if (!strcmp(str, "nomwait")) {
	349	/*
	350	* If the boot option of "idle=nomwait" is added,
	351	* it means that mwait will be disabled for CPU C2/C3
	352	* states. In such case it won't touch the variable
	353	* of boot_option_idle_override.
	354	*/
	355	idle_nomwait = 1;
	356	return 0;
c1e3b377	357	} else
7f424a8b PZ	358	return -1;
	359
	360	boot_option_idle_override = 1;
	361	return 0;
	362	}
	363	early_param("idle", idle_setup);
	364