[deliverable/linux.git] / arch / x86 / kernel / process_32.c

/*
 *  Copyright (C) 1995  Linus Torvalds
 *
 *  Pentium III FXSR, SSE support
 *	Gareth Hughes <gareth@valinux.com>, May 2000
 */

/*
 * This file handles the architecture-dependent parts of process handling..
 */

#include <linux/stackprotector.h>
#include <linux/cpu.h>
#include <linux/errno.h>
#include <linux/sched.h>
#include <linux/fs.h>
#include <linux/kernel.h>
#include <linux/mm.h>
#include <linux/elfcore.h>
#include <linux/smp.h>
#include <linux/stddef.h>
#include <linux/slab.h>
#include <linux/vmalloc.h>
#include <linux/user.h>
#include <linux/interrupt.h>
#include <linux/delay.h>
#include <linux/reboot.h>
#include <linux/init.h>
#include <linux/mc146818rtc.h>
#include <linux/module.h>
#include <linux/kallsyms.h>
#include <linux/ptrace.h>
#include <linux/personality.h>
#include <linux/tick.h>
#include <linux/percpu.h>
#include <linux/prctl.h>
#include <linux/ftrace.h>
#include <linux/uaccess.h>
#include <linux/io.h>
#include <linux/kdebug.h>
#include <linux/cpuidle.h>

#include <asm/pgtable.h>
#include <asm/system.h>
#include <asm/ldt.h>
#include <asm/processor.h>
#include <asm/i387.h>
#include <asm/desc.h>
#ifdef CONFIG_MATH_EMULATION
#include <asm/math_emu.h>
#endif

#include <linux/err.h>

#include <asm/tlbflush.h>
#include <asm/cpu.h>
#include <asm/idle.h>
#include <asm/syscalls.h>
#include <asm/debugreg.h>
#include <asm/nmi.h>

asmlinkage void ret_from_fork(void) __asm__("ret_from_fork");

/*
 * Return saved PC of a blocked thread.
 */
unsigned long thread_saved_pc(struct task_struct *tsk)
{
	return ((unsigned long *)tsk->thread.sp)[3];
}

#ifndef CONFIG_SMP
static inline void play_dead(void)
{
	BUG();
}
#endif

/*
 * The idle thread. There's no useful work to be
 * done, so just try to conserve power and have a
 * low exit latency (ie sit in a loop waiting for
 * somebody to say that they'd like to reschedule)
 */
void cpu_idle(void)
{
	int cpu = smp_processor_id();

	/*
	 * If we're the non-boot CPU, nothing set the stack canary up
	 * for us.  CPU0 already has it initialized but no harm in
	 * doing it again.  This is a good place for updating it, as
	 * we wont ever return from this function (so the invalid
	 * canaries already on the stack wont ever trigger).
	 */
	boot_init_stack_canary();

	current_thread_info()->status |= TS_POLLING;

	/* endless idle loop with no priority at all */
	while (1) {
		tick_nohz_idle_enter_norcu();
		while (!need_resched()) {

			check_pgt_cache();
			rmb();

			if (cpu_is_offline(cpu))
				play_dead();

			local_touch_nmi();
			local_irq_disable();
			/* Don't trace irqs off for idle */
			stop_critical_timings();
			if (cpuidle_idle_call())
				pm_idle();
			start_critical_timings();
		}
		tick_nohz_idle_exit_norcu();
		preempt_enable_no_resched();
		schedule();
		preempt_disable();
	}
}

void __show_regs(struct pt_regs *regs, int all)
{
	unsigned long cr0 = 0L, cr2 = 0L, cr3 = 0L, cr4 = 0L;
	unsigned long d0, d1, d2, d3, d6, d7;
	unsigned long sp;
	unsigned short ss, gs;

	if (user_mode_vm(regs)) {
		sp = regs->sp;
		ss = regs->ss & 0xffff;
		gs = get_user_gs(regs);
	} else {
		sp = kernel_stack_pointer(regs);
		savesegment(ss, ss);
		savesegment(gs, gs);
	}

	show_regs_common();

	printk(KERN_DEFAULT "EIP: %04x:[<%08lx>] EFLAGS: %08lx CPU: %d\n",
			(u16)regs->cs, regs->ip, regs->flags,
			smp_processor_id());
	print_symbol("EIP is at %s\n", regs->ip);

	printk(KERN_DEFAULT "EAX: %08lx EBX: %08lx ECX: %08lx EDX: %08lx\n",
		regs->ax, regs->bx, regs->cx, regs->dx);
	printk(KERN_DEFAULT "ESI: %08lx EDI: %08lx EBP: %08lx ESP: %08lx\n",
		regs->si, regs->di, regs->bp, sp);
	printk(KERN_DEFAULT " DS: %04x ES: %04x FS: %04x GS: %04x SS: %04x\n",
	       (u16)regs->ds, (u16)regs->es, (u16)regs->fs, gs, ss);

	if (!all)
		return;

	cr0 = read_cr0();
	cr2 = read_cr2();
	cr3 = read_cr3();
	cr4 = read_cr4_safe();
	printk(KERN_DEFAULT "CR0: %08lx CR2: %08lx CR3: %08lx CR4: %08lx\n",
			cr0, cr2, cr3, cr4);

	get_debugreg(d0, 0);
	get_debugreg(d1, 1);
	get_debugreg(d2, 2);
	get_debugreg(d3, 3);
	printk(KERN_DEFAULT "DR0: %08lx DR1: %08lx DR2: %08lx DR3: %08lx\n",
			d0, d1, d2, d3);

	get_debugreg(d6, 6);
	get_debugreg(d7, 7);
	printk(KERN_DEFAULT "DR6: %08lx DR7: %08lx\n",
			d6, d7);
}

void release_thread(struct task_struct *dead_task)
{
	BUG_ON(dead_task->mm);
	release_vm86_irqs(dead_task);
}

/*
 * This gets called before we allocate a new thread and copy
 * the current task into it.
 */
void prepare_to_copy(struct task_struct *tsk)
{
	unlazy_fpu(tsk);
}

int copy_thread(unsigned long clone_flags, unsigned long sp,
	unsigned long unused,
	struct task_struct *p, struct pt_regs *regs)
{
	struct pt_regs *childregs;
	struct task_struct *tsk;
	int err;

	childregs = task_pt_regs(p);
	*childregs = *regs;
	childregs->ax = 0;
	childregs->sp = sp;

	p->thread.sp = (unsigned long) childregs;
	p->thread.sp0 = (unsigned long) (childregs+1);

	p->thread.ip = (unsigned long) ret_from_fork;

	task_user_gs(p) = get_user_gs(regs);

	p->thread.io_bitmap_ptr = NULL;
	tsk = current;
	err = -ENOMEM;

	memset(p->thread.ptrace_bps, 0, sizeof(p->thread.ptrace_bps));

	if (unlikely(test_tsk_thread_flag(tsk, TIF_IO_BITMAP))) {
		p->thread.io_bitmap_ptr = kmemdup(tsk->thread.io_bitmap_ptr,
						IO_BITMAP_BYTES, GFP_KERNEL);
		if (!p->thread.io_bitmap_ptr) {
			p->thread.io_bitmap_max = 0;
			return -ENOMEM;
		}
		set_tsk_thread_flag(p, TIF_IO_BITMAP);
	}

	err = 0;

	/*
	 * Set a new TLS for the child thread?
	 */
	if (clone_flags & CLONE_SETTLS)
		err = do_set_thread_area(p, -1,
			(struct user_desc __user *)childregs->si, 0);

	if (err && p->thread.io_bitmap_ptr) {
		kfree(p->thread.io_bitmap_ptr);
		p->thread.io_bitmap_max = 0;
	}
	return err;
}

void
start_thread(struct pt_regs *regs, unsigned long new_ip, unsigned long new_sp)
{
	set_user_gs(regs, 0);
	regs->fs		= 0;
	regs->ds		= __USER_DS;
	regs->es		= __USER_DS;
	regs->ss		= __USER_DS;
	regs->cs		= __USER_CS;
	regs->ip		= new_ip;
	regs->sp		= new_sp;
	/*
	 * Free the old FP and other extended state
	 */
	free_thread_xstate(current);
}
EXPORT_SYMBOL_GPL(start_thread);


/*
 *	switch_to(x,y) should switch tasks from x to y.
 *
 * We fsave/fwait so that an exception goes off at the right time
 * (as a call from the fsave or fwait in effect) rather than to
 * the wrong process. Lazy FP saving no longer makes any sense
 * with modern CPU's, and this simplifies a lot of things (SMP
 * and UP become the same).
 *
 * NOTE! We used to use the x86 hardware context switching. The
 * reason for not using it any more becomes apparent when you
 * try to recover gracefully from saved state that is no longer
 * valid (stale segment register values in particular). With the
 * hardware task-switch, there is no way to fix up bad state in
 * a reasonable manner.
 *
 * The fact that Intel documents the hardware task-switching to
 * be slow is a fairly red herring - this code is not noticeably
 * faster. However, there _is_ some room for improvement here,
 * so the performance issues may eventually be a valid point.
 * More important, however, is the fact that this allows us much
 * more flexibility.
 *
 * The return value (in %ax) will be the "prev" task after
 * the task-switch, and shows up in ret_from_fork in entry.S,
 * for example.
 */
__notrace_funcgraph struct task_struct *
__switch_to(struct task_struct *prev_p, struct task_struct *next_p)
{
	struct thread_struct *prev = &prev_p->thread,
				 *next = &next_p->thread;
	int cpu = smp_processor_id();
	struct tss_struct *tss = &per_cpu(init_tss, cpu);
	bool preload_fpu;

	/* never put a printk in __switch_to... printk() calls wake_up*() indirectly */

	/*
	 * If the task has used fpu the last 5 timeslices, just do a full
	 * restore of the math state immediately to avoid the trap; the
	 * chances of needing FPU soon are obviously high now
	 */
	preload_fpu = tsk_used_math(next_p) && next_p->fpu_counter > 5;

	__unlazy_fpu(prev_p);

	/* we're going to use this soon, after a few expensive things */
	if (preload_fpu)
		prefetch(next->fpu.state);

	/*
	 * Reload esp0.
	 */
	load_sp0(tss, next);

	/*
	 * Save away %gs. No need to save %fs, as it was saved on the
	 * stack on entry.  No need to save %es and %ds, as those are
	 * always kernel segments while inside the kernel.  Doing this
	 * before setting the new TLS descriptors avoids the situation
	 * where we temporarily have non-reloadable segments in %fs
	 * and %gs.  This could be an issue if the NMI handler ever
	 * used %fs or %gs (it does not today), or if the kernel is
	 * running inside of a hypervisor layer.
	 */
	lazy_save_gs(prev->gs);

	/*
	 * Load the per-thread Thread-Local Storage descriptor.
	 */
	load_TLS(next, cpu);

	/*
	 * Restore IOPL if needed.  In normal use, the flags restore
	 * in the switch assembly will handle this.  But if the kernel
	 * is running virtualized at a non-zero CPL, the popf will
	 * not restore flags, so it must be done in a separate step.
	 */
	if (get_kernel_rpl() && unlikely(prev->iopl != next->iopl))
		set_iopl_mask(next->iopl);

	/*
	 * Now maybe handle debug registers and/or IO bitmaps
	 */
	if (unlikely(task_thread_info(prev_p)->flags & _TIF_WORK_CTXSW_PREV ||
		     task_thread_info(next_p)->flags & _TIF_WORK_CTXSW_NEXT))
		__switch_to_xtra(prev_p, next_p, tss);

	/* If we're going to preload the fpu context, make sure clts
	   is run while we're batching the cpu state updates. */
	if (preload_fpu)
		clts();

	/*
	 * Leave lazy mode, flushing any hypercalls made here.
	 * This must be done before restoring TLS segments so
	 * the GDT and LDT are properly updated, and must be
	 * done before math_state_restore, so the TS bit is up
	 * to date.
	 */
	arch_end_context_switch(next_p);

	if (preload_fpu)
		__math_state_restore();

	/*
	 * Restore %gs if needed (which is common)
	 */
	if (prev->gs | next->gs)
		lazy_load_gs(next->gs);

	percpu_write(current_task, next_p);

	return prev_p;
}

#define top_esp                (THREAD_SIZE - sizeof(unsigned long))
#define top_ebp                (THREAD_SIZE - 2*sizeof(unsigned long))

unsigned long get_wchan(struct task_struct *p)
{
	unsigned long bp, sp, ip;
	unsigned long stack_page;
	int count = 0;
	if (!p || p == current || p->state == TASK_RUNNING)
		return 0;
	stack_page = (unsigned long)task_stack_page(p);
	sp = p->thread.sp;
	if (!stack_page || sp < stack_page || sp > top_esp+stack_page)
		return 0;
	/* include/asm-i386/system.h:switch_to() pushes bp last. */
	bp = *(unsigned long *) sp;
	do {
		if (bp < stack_page || bp > top_ebp+stack_page)
			return 0;
		ip = *(unsigned long *) (bp+4);
		if (!in_sched_functions(ip))
			return ip;
		bp = *(unsigned long *) bp;
	} while (count++ < 16);
	return 0;
}
Commit	Line	Data
1da177e4	1	/*
1da177e4 LT	2	* Copyright (C) 1995 Linus Torvalds
	3	*
	4	* Pentium III FXSR, SSE support
	5	* Gareth Hughes <gareth@valinux.com>, May 2000
	6	*/
	7
	8	/*
	9	* This file handles the architecture-dependent parts of process handling..
	10	*/
	11
5c79d2a5	12	#include <linux/stackprotector.h>
f3705136	13	#include <linux/cpu.h>
1da177e4 LT	14	#include <linux/errno.h>
	15	#include <linux/sched.h>
	16	#include <linux/fs.h>
	17	#include <linux/kernel.h>
	18	#include <linux/mm.h>
	19	#include <linux/elfcore.h>
	20	#include <linux/smp.h>
1da177e4 LT	21	#include <linux/stddef.h>
	22	#include <linux/slab.h>
	23	#include <linux/vmalloc.h>
	24	#include <linux/user.h>
1da177e4	25	#include <linux/interrupt.h>
1da177e4 LT	26	#include <linux/delay.h>
	27	#include <linux/reboot.h>
	28	#include <linux/init.h>
	29	#include <linux/mc146818rtc.h>
	30	#include <linux/module.h>
	31	#include <linux/kallsyms.h>
	32	#include <linux/ptrace.h>
c16b63e0	33	#include <linux/personality.h>
74167347	34	#include <linux/tick.h>
7c3576d2	35	#include <linux/percpu.h>
529e25f6	36	#include <linux/prctl.h>
8b96f011	37	#include <linux/ftrace.h>
befa9e78 JSR	38	#include <linux/uaccess.h>
	39	#include <linux/io.h>
	40	#include <linux/kdebug.h>
a0bfa137	41	#include <linux/cpuidle.h>
1da177e4	42
1da177e4 LT	43	#include <asm/pgtable.h>
1da177e4 LT	44	#include <asm/system.h>
1da177e4 LT	45	#include <asm/ldt.h>
	46	#include <asm/processor.h>
	47	#include <asm/i387.h>
1da177e4 LT	48	#include <asm/desc.h>
	49	#ifdef CONFIG_MATH_EMULATION
	50	#include <asm/math_emu.h>
	51	#endif
	52
1da177e4 LT	53	#include <linux/err.h>
1da177e4 LT	54
f3705136 ZM	55	#include <asm/tlbflush.h>
f3705136 ZM	56	#include <asm/cpu.h>
1eda8149	57	#include <asm/idle.h>
bbc1f698	58	#include <asm/syscalls.h>
66cb5917	59	#include <asm/debugreg.h>
b227e233	60	#include <asm/nmi.h>
f3705136	61
1da177e4 LT	62	asmlinkage void ret_from_fork(void) __asm__("ret_from_fork");
1da177e4 LT	63
1da177e4 LT	64	/*
	65	* Return saved PC of a blocked thread.
	66	*/
	67	unsigned long thread_saved_pc(struct task_struct *tsk)
	68	{
faca6227	69	return ((unsigned long *)tsk->thread.sp)[3];
1da177e4 LT	70	}
1da177e4 LT	71
913da64b AN	72	#ifndef CONFIG_SMP
	73	static inline void play_dead(void)
	74	{
	75	BUG();
	76	}
	77	#endif
	78
1da177e4 LT	79	/*
	80	* The idle thread. There's no useful work to be
	81	* done, so just try to conserve power and have a
	82	* low exit latency (ie sit in a loop waiting for
	83	* somebody to say that they'd like to reschedule)
	84	*/
f3705136	85	void cpu_idle(void)
1da177e4	86	{
5bfb5d69	87	int cpu = smp_processor_id();
f3705136	88
5c79d2a5 TH	89	/*
	90	* If we're the non-boot CPU, nothing set the stack canary up
	91	* for us. CPU0 already has it initialized but no harm in
	92	* doing it again. This is a good place for updating it, as
	93	* we wont ever return from this function (so the invalid
	94	* canaries already on the stack wont ever trigger).
	95	*/
	96	boot_init_stack_canary();
	97
495ab9c0	98	current_thread_info()->status \|= TS_POLLING;
64c7c8f8	99
1da177e4 LT	100	/* endless idle loop with no priority at all */
1da177e4 LT	101	while (1) {
2bbb6817	102	tick_nohz_idle_enter_norcu();
1da177e4	103	while (!need_resched()) {
1da177e4	104
f1d1a842	105	check_pgt_cache();
1da177e4	106	rmb();
1da177e4	107
f3705136 ZM	108	if (cpu_is_offline(cpu))
	109	play_dead();
	110
b227e233	111	local_touch_nmi();
7f424a8b	112	local_irq_disable();
6cd8a4bb SR	113	/* Don't trace irqs off for idle */
6cd8a4bb SR	114	stop_critical_timings();
a0bfa137 LB	115	if (cpuidle_idle_call())
a0bfa137 LB	116	pm_idle();
6cd8a4bb	117	start_critical_timings();
1da177e4	118	}
2bbb6817	119	tick_nohz_idle_exit_norcu();
5bfb5d69	120	preempt_enable_no_resched();
1da177e4	121	schedule();
5bfb5d69	122	preempt_disable();
1da177e4 LT	123	}
	124	}
	125
e2ce07c8	126	void __show_regs(struct pt_regs *regs, int all)
1da177e4 LT	127	{
1da177e4 LT	128	unsigned long cr0 = 0L, cr2 = 0L, cr3 = 0L, cr4 = 0L;
bb1995d5	129	unsigned long d0, d1, d2, d3, d6, d7;
65ea5b03	130	unsigned long sp;
9d975ebd PE	131	unsigned short ss, gs;
	132
	133	if (user_mode_vm(regs)) {
65ea5b03 PA	134	sp = regs->sp;
65ea5b03 PA	135	ss = regs->ss & 0xffff;
d9a89a26	136	gs = get_user_gs(regs);
9d975ebd	137	} else {
def3c5d0	138	sp = kernel_stack_pointer(regs);
9d975ebd PE	139	savesegment(ss, ss);
	140	savesegment(gs, gs);
	141	}
1da177e4	142
814e2c84	143	show_regs_common();
9d975ebd	144
d015a092	145	printk(KERN_DEFAULT "EIP: %04x:[<%08lx>] EFLAGS: %08lx CPU: %d\n",
92bc2056	146	(u16)regs->cs, regs->ip, regs->flags,
9d975ebd	147	smp_processor_id());
65ea5b03	148	print_symbol("EIP is at %s\n", regs->ip);
1da177e4	149
d015a092	150	printk(KERN_DEFAULT "EAX: %08lx EBX: %08lx ECX: %08lx EDX: %08lx\n",
65ea5b03	151	regs->ax, regs->bx, regs->cx, regs->dx);
d015a092	152	printk(KERN_DEFAULT "ESI: %08lx EDI: %08lx EBP: %08lx ESP: %08lx\n",
65ea5b03	153	regs->si, regs->di, regs->bp, sp);
d015a092	154	printk(KERN_DEFAULT " DS: %04x ES: %04x FS: %04x GS: %04x SS: %04x\n",
92bc2056	155	(u16)regs->ds, (u16)regs->es, (u16)regs->fs, gs, ss);
9d975ebd PE	156
	157	if (!all)
	158	return;
1da177e4	159
4bb0d3ec ZA	160	cr0 = read_cr0();
	161	cr2 = read_cr2();
	162	cr3 = read_cr3();
ff6e8c0d	163	cr4 = read_cr4_safe();
d015a092	164	printk(KERN_DEFAULT "CR0: %08lx CR2: %08lx CR3: %08lx CR4: %08lx\n",
9d975ebd	165	cr0, cr2, cr3, cr4);
bb1995d5 AS	166
	167	get_debugreg(d0, 0);
	168	get_debugreg(d1, 1);
	169	get_debugreg(d2, 2);
	170	get_debugreg(d3, 3);
d015a092	171	printk(KERN_DEFAULT "DR0: %08lx DR1: %08lx DR2: %08lx DR3: %08lx\n",
bb1995d5	172	d0, d1, d2, d3);
9d975ebd	173
bb1995d5 AS	174	get_debugreg(d6, 6);
bb1995d5 AS	175	get_debugreg(d7, 7);
d015a092	176	printk(KERN_DEFAULT "DR6: %08lx DR7: %08lx\n",
9d975ebd PE	177	d6, d7);
9d975ebd PE	178	}
bb1995d5	179
1da177e4 LT	180	void release_thread(struct task_struct *dead_task)
1da177e4 LT	181	{
2684927c	182	BUG_ON(dead_task->mm);
1da177e4 LT	183	release_vm86_irqs(dead_task);
	184	}
	185
	186	/*
	187	* This gets called before we allocate a new thread and copy
	188	* the current task into it.
	189	*/
	190	void prepare_to_copy(struct task_struct *tsk)
	191	{
	192	unlazy_fpu(tsk);
	193	}
	194
6f2c55b8	195	int copy_thread(unsigned long clone_flags, unsigned long sp,
1da177e4	196	unsigned long unused,
befa9e78	197	struct task_struct p, struct pt_regs regs)
1da177e4	198	{
befa9e78	199	struct pt_regs *childregs;
1da177e4 LT	200	struct task_struct *tsk;
	201	int err;
	202
07b047fc	203	childregs = task_pt_regs(p);
f48d9663	204	childregs = regs;
65ea5b03 PA	205	childregs->ax = 0;
65ea5b03 PA	206	childregs->sp = sp;
f48d9663	207
faca6227 PA	208	p->thread.sp = (unsigned long) childregs;
faca6227 PA	209	p->thread.sp0 = (unsigned long) (childregs+1);
1da177e4	210
faca6227	211	p->thread.ip = (unsigned long) ret_from_fork;
1da177e4	212
d9a89a26	213	task_user_gs(p) = get_user_gs(regs);
1da177e4	214
66cb5917	215	p->thread.io_bitmap_ptr = NULL;
1da177e4	216	tsk = current;
66cb5917	217	err = -ENOMEM;
24f1e32c FW	218
24f1e32c FW	219	memset(p->thread.ptrace_bps, 0, sizeof(p->thread.ptrace_bps));
66cb5917	220
b3cf2576	221	if (unlikely(test_tsk_thread_flag(tsk, TIF_IO_BITMAP))) {
52978be6 AD	222	p->thread.io_bitmap_ptr = kmemdup(tsk->thread.io_bitmap_ptr,
52978be6 AD	223	IO_BITMAP_BYTES, GFP_KERNEL);
1da177e4 LT	224	if (!p->thread.io_bitmap_ptr) {
	225	p->thread.io_bitmap_max = 0;
	226	return -ENOMEM;
	227	}
b3cf2576	228	set_tsk_thread_flag(p, TIF_IO_BITMAP);
1da177e4 LT	229	}
1da177e4 LT	230
efd1ca52 RM	231	err = 0;
efd1ca52 RM	232
1da177e4 LT	233	/*
	234	* Set a new TLS for the child thread?
	235	*/
efd1ca52 RM	236	if (clone_flags & CLONE_SETTLS)
efd1ca52 RM	237	err = do_set_thread_area(p, -1,
65ea5b03	238	(struct user_desc __user *)childregs->si, 0);
1da177e4	239
1da177e4 LT	240	if (err && p->thread.io_bitmap_ptr) {
	241	kfree(p->thread.io_bitmap_ptr);
	242	p->thread.io_bitmap_max = 0;
	243	}
	244	return err;
	245	}
	246
513ad84b IM	247	void
	248	start_thread(struct pt_regs *regs, unsigned long new_ip, unsigned long new_sp)
	249	{
d9a89a26	250	set_user_gs(regs, 0);
513ad84b	251	regs->fs = 0;
513ad84b IM	252	regs->ds = __USER_DS;
	253	regs->es = __USER_DS;
	254	regs->ss = __USER_DS;
	255	regs->cs = __USER_CS;
	256	regs->ip = new_ip;
	257	regs->sp = new_sp;
aa283f49 SS	258	/*
	259	* Free the old FP and other extended state
	260	*/
	261	free_thread_xstate(current);
513ad84b IM	262	}
	263	EXPORT_SYMBOL_GPL(start_thread);
	264
1da177e4 LT	265
1da177e4 LT	266	/*
ea70ef3d	267	* switch_to(x,y) should switch tasks from x to y.
1da177e4 LT	268	*
	269	* We fsave/fwait so that an exception goes off at the right time
	270	* (as a call from the fsave or fwait in effect) rather than to
	271	* the wrong process. Lazy FP saving no longer makes any sense
	272	* with modern CPU's, and this simplifies a lot of things (SMP
	273	* and UP become the same).
	274	*
	275	* NOTE! We used to use the x86 hardware context switching. The
	276	* reason for not using it any more becomes apparent when you
	277	* try to recover gracefully from saved state that is no longer
	278	* valid (stale segment register values in particular). With the
	279	* hardware task-switch, there is no way to fix up bad state in
	280	* a reasonable manner.
	281	*
	282	* The fact that Intel documents the hardware task-switching to
	283	* be slow is a fairly red herring - this code is not noticeably
	284	* faster. However, there _is_ some room for improvement here,
	285	* so the performance issues may eventually be a valid point.
	286	* More important, however, is the fact that this allows us much
	287	* more flexibility.
	288	*
65ea5b03	289	* The return value (in %ax) will be the "prev" task after
1da177e4 LT	290	* the task-switch, and shows up in ret_from_fork in entry.S,
	291	* for example.
	292	*/
8b96f011 FW	293	__notrace_funcgraph struct task_struct *
8b96f011 FW	294	__switch_to(struct task_struct prev_p, struct task_struct next_p)
1da177e4 LT	295	{
	296	struct thread_struct *prev = &prev_p->thread,
	297	*next = &next_p->thread;
	298	int cpu = smp_processor_id();
	299	struct tss_struct *tss = &per_cpu(init_tss, cpu);
2fcddce1	300	bool preload_fpu;
1da177e4 LT	301
	302	/* never put a printk in __switch_to... printk() calls wake_up() indirectly /
	303
2fcddce1 JF	304	/*
	305	* If the task has used fpu the last 5 timeslices, just do a full
	306	* restore of the math state immediately to avoid the trap; the
	307	* chances of needing FPU soon are obviously high now
	308	*/
	309	preload_fpu = tsk_used_math(next_p) && next_p->fpu_counter > 5;
1da177e4	310
2fcddce1	311	__unlazy_fpu(prev_p);
acc20761 CE	312
acc20761 CE	313	/* we're going to use this soon, after a few expensive things */
2fcddce1	314	if (preload_fpu)
86603283	315	prefetch(next->fpu.state);
acc20761	316
1da177e4	317	/*
e7a2ff59	318	* Reload esp0.
1da177e4	319	*/
faca6227	320	load_sp0(tss, next);
1da177e4 LT	321
1da177e4 LT	322	/*
464d1a78	323	* Save away %gs. No need to save %fs, as it was saved on the
f95d47ca JF	324	* stack on entry. No need to save %es and %ds, as those are
	325	* always kernel segments while inside the kernel. Doing this
	326	* before setting the new TLS descriptors avoids the situation
	327	* where we temporarily have non-reloadable segments in %fs
	328	* and %gs. This could be an issue if the NMI handler ever
	329	* used %fs or %gs (it does not today), or if the kernel is
	330	* running inside of a hypervisor layer.
1da177e4	331	*/
ccbeed3a	332	lazy_save_gs(prev->gs);
1da177e4 LT	333
1da177e4 LT	334	/*
e7a2ff59	335	* Load the per-thread Thread-Local Storage descriptor.
1da177e4	336	*/
e7a2ff59	337	load_TLS(next, cpu);
1da177e4	338
8b151144 ZA	339	/*
	340	* Restore IOPL if needed. In normal use, the flags restore
	341	* in the switch assembly will handle this. But if the kernel
	342	* is running virtualized at a non-zero CPL, the popf will
	343	* not restore flags, so it must be done in a separate step.
	344	*/
	345	if (get_kernel_rpl() && unlikely(prev->iopl != next->iopl))
	346	set_iopl_mask(next->iopl);
	347
1da177e4	348	/*
b3cf2576	349	* Now maybe handle debug registers and/or IO bitmaps
1da177e4	350	*/
cf99abac AA	351	if (unlikely(task_thread_info(prev_p)->flags & _TIF_WORK_CTXSW_PREV \|\|
	352	task_thread_info(next_p)->flags & _TIF_WORK_CTXSW_NEXT))
	353	__switch_to_xtra(prev_p, next_p, tss);
ffaa8bd6	354
2fcddce1 JF	355	/* If we're going to preload the fpu context, make sure clts
	356	is run while we're batching the cpu state updates. */
	357	if (preload_fpu)
	358	clts();
	359
9226d125 ZA	360	/*
	361	* Leave lazy mode, flushing any hypercalls made here.
	362	* This must be done before restoring TLS segments so
	363	* the GDT and LDT are properly updated, and must be
	364	* done before math_state_restore, so the TS bit is up
	365	* to date.
	366	*/
224101ed	367	arch_end_context_switch(next_p);
9226d125	368
2fcddce1 JF	369	if (preload_fpu)
2fcddce1 JF	370	__math_state_restore();
acc20761	371
9226d125 ZA	372	/*
	373	* Restore %gs if needed (which is common)
	374	*/
	375	if (prev->gs \| next->gs)
ccbeed3a	376	lazy_load_gs(next->gs);
9226d125	377
6dbde353	378	percpu_write(current_task, next_p);
9226d125	379
1da177e4 LT	380	return prev_p;
	381	}
	382
1da177e4 LT	383	#define top_esp (THREAD_SIZE - sizeof(unsigned long))
	384	#define top_ebp (THREAD_SIZE - 2*sizeof(unsigned long))
	385
	386	unsigned long get_wchan(struct task_struct *p)
	387	{
65ea5b03	388	unsigned long bp, sp, ip;
1da177e4 LT	389	unsigned long stack_page;
	390	int count = 0;
	391	if (!p \|\| p == current \|\| p->state == TASK_RUNNING)
	392	return 0;
65e0fdff	393	stack_page = (unsigned long)task_stack_page(p);
faca6227	394	sp = p->thread.sp;
65ea5b03	395	if (!stack_page \|\| sp < stack_page \|\| sp > top_esp+stack_page)
1da177e4	396	return 0;
65ea5b03 PA	397	/* include/asm-i386/system.h:switch_to() pushes bp last. */
65ea5b03 PA	398	bp = (unsigned long ) sp;
1da177e4	399	do {
65ea5b03	400	if (bp < stack_page \|\| bp > top_ebp+stack_page)
1da177e4	401	return 0;
65ea5b03 PA	402	ip = (unsigned long ) (bp+4);
	403	if (!in_sched_functions(ip))
	404	return ip;
	405	bp = (unsigned long ) bp;
1da177e4 LT	406	} while (count++ < 16);
	407	return 0;
	408	}
	409