[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;
static int force_mwait __cpuinitdata;

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.
 */
static int __cpuinitdata force_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;
9781f39f	18	static int force_mwait __cpuinitdata;
61c4628b SS	19
	20	int arch_dup_task_struct(struct task_struct dst, struct task_struct src)
	21	{
	22	dst = src;
aa283f49 SS	23	if (src->thread.xstate) {
	24	dst->thread.xstate = kmem_cache_alloc(task_xstate_cachep,
	25	GFP_KERNEL);
	26	if (!dst->thread.xstate)
	27	return -ENOMEM;
	28	WARN_ON((unsigned long)dst->thread.xstate & 15);
	29	memcpy(dst->thread.xstate, src->thread.xstate, xstate_size);
	30	}
61c4628b SS	31	return 0;
	32	}
	33
aa283f49	34	void free_thread_xstate(struct task_struct *tsk)
61c4628b	35	{
aa283f49 SS	36	if (tsk->thread.xstate) {
	37	kmem_cache_free(task_xstate_cachep, tsk->thread.xstate);
	38	tsk->thread.xstate = NULL;
	39	}
	40	}
	41
aa283f49 SS	42	void free_thread_info(struct thread_info *ti)
	43	{
	44	free_thread_xstate(ti->task);
1679f271	45	free_pages((unsigned long)ti, get_order(THREAD_SIZE));
61c4628b SS	46	}
	47
	48	void arch_task_cache_init(void)
	49	{
	50	task_xstate_cachep =
	51	kmem_cache_create("task_xstate", xstate_size,
	52	__alignof__(union thread_xstate),
	53	SLAB_PANIC, NULL);
	54	}
7f424a8b	55
00dba564 TG	56	/*
	57	* Idle related variables and functions
	58	*/
	59	unsigned long boot_option_idle_override = 0;
	60	EXPORT_SYMBOL(boot_option_idle_override);
	61
	62	/*
	63	* Powermanagement idle function, if any..
	64	*/
	65	void (*pm_idle)(void);
	66	EXPORT_SYMBOL(pm_idle);
	67
	68	#ifdef CONFIG_X86_32
	69	/*
	70	* This halt magic was a workaround for ancient floppy DMA
	71	* wreckage. It should be safe to remove.
	72	*/
	73	static int hlt_counter;
	74	void disable_hlt(void)
	75	{
	76	hlt_counter++;
	77	}
	78	EXPORT_SYMBOL(disable_hlt);
	79
	80	void enable_hlt(void)
	81	{
	82	hlt_counter--;
	83	}
	84	EXPORT_SYMBOL(enable_hlt);
	85
	86	static inline int hlt_use_halt(void)
	87	{
	88	return (!hlt_counter && boot_cpu_data.hlt_works_ok);
	89	}
	90	#else
	91	static inline int hlt_use_halt(void)
	92	{
	93	return 1;
	94	}
	95	#endif
	96
	97	/*
	98	* We use this if we don't have any better
	99	* idle routine..
	100	*/
	101	void default_idle(void)
	102	{
	103	if (hlt_use_halt()) {
	104	current_thread_info()->status &= ~TS_POLLING;
	105	/*
	106	* TS_POLLING-cleared state must be visible before we
	107	* test NEED_RESCHED:
	108	*/
	109	smp_mb();
	110
	111	if (!need_resched())
	112	safe_halt(); /* enables interrupts racelessly */
	113	else
	114	local_irq_enable();
	115	current_thread_info()->status \|= TS_POLLING;
	116	} else {
	117	local_irq_enable();
	118	/* loop is done by the caller */
	119	cpu_relax();
120	}
121	}
122	#ifdef CONFIG_APM_MODULE
123	EXPORT_SYMBOL(default_idle);
124	#endif
125
7f424a8b PZ	126	static void do_nothing(void *unused)
	127	{
	128	}
	129
	130	/*
	131	* cpu_idle_wait - Used to ensure that all the CPUs discard old value of
	132	* pm_idle and update to new pm_idle value. Required while changing pm_idle
	133	* handler on SMP systems.
	134	*
	135	* Caller must have changed pm_idle to the new value before the call. Old
	136	* pm_idle value will not be used by any CPU after the return of this function.
	137	*/
	138	void cpu_idle_wait(void)
	139	{
	140	smp_mb();
	141	/* kick all the CPUs so that they exit out of pm_idle */
127a237a	142	smp_call_function(do_nothing, NULL, 1);
7f424a8b PZ	143	}
	144	EXPORT_SYMBOL_GPL(cpu_idle_wait);
	145
	146	/*
	147	* This uses new MONITOR/MWAIT instructions on P4 processors with PNI,
	148	* which can obviate IPI to trigger checking of need_resched.
	149	* We execute MONITOR against need_resched and enter optimized wait state
	150	* through MWAIT. Whenever someone changes need_resched, we would be woken
	151	* up from MWAIT (without an IPI).
	152	*
	153	* New with Core Duo processors, MWAIT can take some hints based on CPU
	154	* capability.
	155	*/
	156	void mwait_idle_with_hints(unsigned long ax, unsigned long cx)
	157	{
	158	if (!need_resched()) {
	159	__monitor((void *)&current_thread_info()->flags, 0, 0);
	160	smp_mb();
	161	if (!need_resched())
	162	__mwait(ax, cx);
	163	}
	164	}
	165
	166	/* Default MONITOR/MWAIT with no hints, used for default C1 state */
	167	static void mwait_idle(void)
	168	{
	169	if (!need_resched()) {
	170	__monitor((void *)&current_thread_info()->flags, 0, 0);
	171	smp_mb();
	172	if (!need_resched())
	173	__sti_mwait(0, 0);
	174	else
	175	local_irq_enable();
	176	} else
	177	local_irq_enable();
	178	}
	179
7f424a8b PZ	180	/*
	181	* On SMP it's slightly faster (but much more power-consuming!)
	182	* to poll the ->work.need_resched flag instead of waiting for the
	183	* cross-CPU IPI to arrive. Use this option with caution.
	184	*/
	185	static void poll_idle(void)
	186	{
	187	local_irq_enable();
	188	cpu_relax();
	189	}
	190
e9623b35 TG	191	/*
	192	* mwait selection logic:
	193	*
	194	* It depends on the CPU. For AMD CPUs that support MWAIT this is
	195	* wrong. Family 0x10 and 0x11 CPUs will enter C1 on HLT. Powersavings
	196	* then depend on a clock divisor and current Pstate of the core. If
	197	* all cores of a processor are in halt state (C1) the processor can
	198	* enter the C1E (C1 enhanced) state. If mwait is used this will never
	199	* happen.
	200	*
	201	* idle=mwait overrides this decision and forces the usage of mwait.
	202	*/
08ad8afa	203	static int __cpuinitdata force_mwait;
09fd4b4e TG	204
	205	#define MWAIT_INFO 0x05
	206	#define MWAIT_ECX_EXTENDED_INFO 0x01
	207	#define MWAIT_EDX_C1 0xf0
	208
e9623b35 TG	209	static int __cpuinit mwait_usable(const struct cpuinfo_x86 *c)
e9623b35 TG	210	{
09fd4b4e TG	211	u32 eax, ebx, ecx, edx;
09fd4b4e TG	212
e9623b35 TG	213	if (force_mwait)
	214	return 1;
	215
09fd4b4e TG	216	if (c->cpuid_level < MWAIT_INFO)
	217	return 0;
	218
	219	cpuid(MWAIT_INFO, &eax, &ebx, &ecx, &edx);
	220	/* Check, whether EDX has extended info about MWAIT */
	221	if (!(ecx & MWAIT_ECX_EXTENDED_INFO))
	222	return 1;
	223
	224	/*
	225	* edx enumeratios MONITOR/MWAIT extensions. Check, whether
	226	* C1 supports MWAIT
	227	*/
	228	return (edx & MWAIT_EDX_C1);
e9623b35 TG	229	}
e9623b35 TG	230
aa276e1c TG	231	/*
	232	* Check for AMD CPUs, which have potentially C1E support
	233	*/
	234	static int __cpuinit check_c1e_idle(const struct cpuinfo_x86 *c)
	235	{
	236	if (c->x86_vendor != X86_VENDOR_AMD)
	237	return 0;
	238
	239	if (c->x86 < 0x0F)
	240	return 0;
	241
	242	/* Family 0x0f models < rev F do not have C1E */
	243	if (c->x86 == 0x0f && c->x86_model < 0x40)
	244	return 0;
	245
	246	return 1;
	247	}
	248
	249	/*
	250	* C1E aware idle routine. We check for C1E active in the interrupt
	251	* pending message MSR. If we detect C1E, then we handle it the same
	252	* way as C3 power states (local apic timer and TSC stop)
	253	*/
	254	static void c1e_idle(void)
	255	{
	256	static cpumask_t c1e_mask = CPU_MASK_NONE;
	257	static int c1e_detected;
	258
	259	if (need_resched())
	260	return;
	261
	262	if (!c1e_detected) {
	263	u32 lo, hi;
	264
	265	rdmsr(MSR_K8_INT_PENDING_MSG, lo, hi);
	266	if (lo & K8_INTP_C1E_ACTIVE_MASK) {
	267	c1e_detected = 1;
	268	mark_tsc_unstable("TSC halt in C1E");
	269	printk(KERN_INFO "System has C1E enabled\n");
	270	}
	271	}
	272
	273	if (c1e_detected) {
	274	int cpu = smp_processor_id();
	275
	276	if (!cpu_isset(cpu, c1e_mask)) {
	277	cpu_set(cpu, c1e_mask);
0beefa20 TG	278	/*
	279	* Force broadcast so ACPI can not interfere. Needs
	280	* to run with interrupts enabled as it uses
	281	* smp_function_call.
	282	*/
	283	local_irq_enable();
aa276e1c TG	284	clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_FORCE,
	285	&cpu);
	286	printk(KERN_INFO "Switch to broadcast mode on CPU%d\n",
	287	cpu);
0beefa20	288	local_irq_disable();
aa276e1c TG	289	}
aa276e1c TG	290	clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_ENTER, &cpu);
0beefa20	291
aa276e1c	292	default_idle();
0beefa20 TG	293
	294	/*
	295	* The switch back from broadcast mode needs to be
	296	* called with interrupts disabled.
	297	*/
	298	local_irq_disable();
	299	clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_EXIT, &cpu);
	300	local_irq_enable();
aa276e1c TG	301	} else
	302	default_idle();
	303	}
	304
7f424a8b PZ	305	void __cpuinit select_idle_routine(const struct cpuinfo_x86 *c)
7f424a8b PZ	306	{
7f424a8b PZ	307	#ifdef CONFIG_X86_SMP
	308	if (pm_idle == poll_idle && smp_num_siblings > 1) {
	309	printk(KERN_WARNING "WARNING: polling idle and HT enabled,"
	310	" performance may degrade.\n");
	311	}
	312	#endif
6ddd2a27 TG	313	if (pm_idle)
	314	return;
	315
e9623b35	316	if (cpu_has(c, X86_FEATURE_MWAIT) && mwait_usable(c)) {
7f424a8b	317	/*
7f424a8b PZ	318	* One CPU supports mwait => All CPUs supports mwait
7f424a8b PZ	319	*/
6ddd2a27 TG	320	printk(KERN_INFO "using mwait in idle threads.\n");
6ddd2a27 TG	321	pm_idle = mwait_idle;
aa276e1c TG	322	} else if (check_c1e_idle(c)) {
	323	printk(KERN_INFO "using C1E aware idle routine\n");
	324	pm_idle = c1e_idle;
6ddd2a27 TG	325	} else
6ddd2a27 TG	326	pm_idle = default_idle;
7f424a8b PZ	327	}
	328
	329	static int __init idle_setup(char *str)
	330	{
ab6bc3e3 CG	331	if (!str)
	332	return -EINVAL;
	333
7f424a8b PZ	334	if (!strcmp(str, "poll")) {
	335	printk("using polling idle threads.\n");
	336	pm_idle = poll_idle;
	337	} else if (!strcmp(str, "mwait"))
	338	force_mwait = 1;
c1e3b377 ZY	339	else if (!strcmp(str, "halt")) {
	340	/*
	341	* When the boot option of idle=halt is added, halt is
	342	* forced to be used for CPU idle. In such case CPU C2/C3
	343	* won't be used again.
	344	* To continue to load the CPU idle driver, don't touch
	345	* the boot_option_idle_override.
	346	*/
	347	pm_idle = default_idle;
	348	idle_halt = 1;
	349	return 0;
da5e09a1 ZY	350	} else if (!strcmp(str, "nomwait")) {
	351	/*
	352	* If the boot option of "idle=nomwait" is added,
	353	* it means that mwait will be disabled for CPU C2/C3
	354	* states. In such case it won't touch the variable
	355	* of boot_option_idle_override.
	356	*/
	357	idle_nomwait = 1;
	358	return 0;
c1e3b377	359	} else
7f424a8b PZ	360	return -1;
	361
	362	boot_option_idle_override = 1;
	363	return 0;
	364	}
	365	early_param("idle", idle_setup);
	366