[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>

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_stop_sched_tick(1);
		while (!need_resched()) {

			check_pgt_cache();
			rmb();

			if (cpu_is_offline(cpu))
				play_dead();

			local_irq_disable();
			/* Don't trace irqs off for idle */
			stop_critical_timings();
			if (cpuidle_idle_call())
				pm_idle();
			start_critical_timings();
		}
		tick_nohz_restart_sched_tick();
		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,yn) 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>
f3705136	60
1da177e4 LT	61	asmlinkage void ret_from_fork(void) __asm__("ret_from_fork");
1da177e4 LT	62
1da177e4 LT	63	/*
	64	* Return saved PC of a blocked thread.
	65	*/
	66	unsigned long thread_saved_pc(struct task_struct *tsk)
	67	{
faca6227	68	return ((unsigned long *)tsk->thread.sp)[3];
1da177e4 LT	69	}
1da177e4 LT	70
913da64b AN	71	#ifndef CONFIG_SMP
	72	static inline void play_dead(void)
	73	{
	74	BUG();
	75	}
	76	#endif
	77
1da177e4 LT	78	/*
	79	* The idle thread. There's no useful work to be
	80	* done, so just try to conserve power and have a
	81	* low exit latency (ie sit in a loop waiting for
	82	* somebody to say that they'd like to reschedule)
	83	*/
f3705136	84	void cpu_idle(void)
1da177e4	85	{
5bfb5d69	86	int cpu = smp_processor_id();
f3705136	87
5c79d2a5 TH	88	/*
	89	* If we're the non-boot CPU, nothing set the stack canary up
	90	* for us. CPU0 already has it initialized but no harm in
	91	* doing it again. This is a good place for updating it, as
	92	* we wont ever return from this function (so the invalid
	93	* canaries already on the stack wont ever trigger).
	94	*/
	95	boot_init_stack_canary();
	96
495ab9c0	97	current_thread_info()->status \|= TS_POLLING;
64c7c8f8	98
1da177e4 LT	99	/* endless idle loop with no priority at all */
1da177e4 LT	100	while (1) {
b8f8c3cf	101	tick_nohz_stop_sched_tick(1);
1da177e4	102	while (!need_resched()) {
1da177e4	103
f1d1a842	104	check_pgt_cache();
1da177e4	105	rmb();
1da177e4	106
f3705136 ZM	107	if (cpu_is_offline(cpu))
	108	play_dead();
	109
7f424a8b	110	local_irq_disable();
6cd8a4bb SR	111	/* Don't trace irqs off for idle */
6cd8a4bb SR	112	stop_critical_timings();
a0bfa137 LB	113	if (cpuidle_idle_call())
a0bfa137 LB	114	pm_idle();
6cd8a4bb	115	start_critical_timings();
1da177e4	116	}
74167347	117	tick_nohz_restart_sched_tick();
5bfb5d69	118	preempt_enable_no_resched();
1da177e4	119	schedule();
5bfb5d69	120	preempt_disable();
1da177e4 LT	121	}
	122	}
	123
e2ce07c8	124	void __show_regs(struct pt_regs *regs, int all)
1da177e4 LT	125	{
1da177e4 LT	126	unsigned long cr0 = 0L, cr2 = 0L, cr3 = 0L, cr4 = 0L;
bb1995d5	127	unsigned long d0, d1, d2, d3, d6, d7;
65ea5b03	128	unsigned long sp;
9d975ebd PE	129	unsigned short ss, gs;
	130
	131	if (user_mode_vm(regs)) {
65ea5b03 PA	132	sp = regs->sp;
65ea5b03 PA	133	ss = regs->ss & 0xffff;
d9a89a26	134	gs = get_user_gs(regs);
9d975ebd	135	} else {
def3c5d0	136	sp = kernel_stack_pointer(regs);
9d975ebd PE	137	savesegment(ss, ss);
	138	savesegment(gs, gs);
	139	}
1da177e4	140
814e2c84	141	show_regs_common();
9d975ebd	142
d015a092	143	printk(KERN_DEFAULT "EIP: %04x:[<%08lx>] EFLAGS: %08lx CPU: %d\n",
92bc2056	144	(u16)regs->cs, regs->ip, regs->flags,
9d975ebd	145	smp_processor_id());
65ea5b03	146	print_symbol("EIP is at %s\n", regs->ip);
1da177e4	147
d015a092	148	printk(KERN_DEFAULT "EAX: %08lx EBX: %08lx ECX: %08lx EDX: %08lx\n",
65ea5b03	149	regs->ax, regs->bx, regs->cx, regs->dx);
d015a092	150	printk(KERN_DEFAULT "ESI: %08lx EDI: %08lx EBP: %08lx ESP: %08lx\n",
65ea5b03	151	regs->si, regs->di, regs->bp, sp);
d015a092	152	printk(KERN_DEFAULT " DS: %04x ES: %04x FS: %04x GS: %04x SS: %04x\n",
92bc2056	153	(u16)regs->ds, (u16)regs->es, (u16)regs->fs, gs, ss);
9d975ebd PE	154
	155	if (!all)
	156	return;
1da177e4	157
4bb0d3ec ZA	158	cr0 = read_cr0();
	159	cr2 = read_cr2();
	160	cr3 = read_cr3();
ff6e8c0d	161	cr4 = read_cr4_safe();
d015a092	162	printk(KERN_DEFAULT "CR0: %08lx CR2: %08lx CR3: %08lx CR4: %08lx\n",
9d975ebd	163	cr0, cr2, cr3, cr4);
bb1995d5 AS	164
	165	get_debugreg(d0, 0);
	166	get_debugreg(d1, 1);
	167	get_debugreg(d2, 2);
	168	get_debugreg(d3, 3);
d015a092	169	printk(KERN_DEFAULT "DR0: %08lx DR1: %08lx DR2: %08lx DR3: %08lx\n",
bb1995d5	170	d0, d1, d2, d3);
9d975ebd	171
bb1995d5 AS	172	get_debugreg(d6, 6);
bb1995d5 AS	173	get_debugreg(d7, 7);
d015a092	174	printk(KERN_DEFAULT "DR6: %08lx DR7: %08lx\n",
9d975ebd PE	175	d6, d7);
9d975ebd PE	176	}
bb1995d5	177
1da177e4 LT	178	void release_thread(struct task_struct *dead_task)
1da177e4 LT	179	{
2684927c	180	BUG_ON(dead_task->mm);
1da177e4 LT	181	release_vm86_irqs(dead_task);
	182	}
	183
	184	/*
	185	* This gets called before we allocate a new thread and copy
	186	* the current task into it.
	187	*/
	188	void prepare_to_copy(struct task_struct *tsk)
	189	{
	190	unlazy_fpu(tsk);
	191	}
	192
6f2c55b8	193	int copy_thread(unsigned long clone_flags, unsigned long sp,
1da177e4	194	unsigned long unused,
befa9e78	195	struct task_struct p, struct pt_regs regs)
1da177e4	196	{
befa9e78	197	struct pt_regs *childregs;
1da177e4 LT	198	struct task_struct *tsk;
	199	int err;
	200
07b047fc	201	childregs = task_pt_regs(p);
f48d9663	202	childregs = regs;
65ea5b03 PA	203	childregs->ax = 0;
65ea5b03 PA	204	childregs->sp = sp;
f48d9663	205
faca6227 PA	206	p->thread.sp = (unsigned long) childregs;
faca6227 PA	207	p->thread.sp0 = (unsigned long) (childregs+1);
1da177e4	208
faca6227	209	p->thread.ip = (unsigned long) ret_from_fork;
1da177e4	210
d9a89a26	211	task_user_gs(p) = get_user_gs(regs);
1da177e4	212
66cb5917	213	p->thread.io_bitmap_ptr = NULL;
1da177e4	214	tsk = current;
66cb5917	215	err = -ENOMEM;
24f1e32c FW	216
24f1e32c FW	217	memset(p->thread.ptrace_bps, 0, sizeof(p->thread.ptrace_bps));
66cb5917	218
b3cf2576	219	if (unlikely(test_tsk_thread_flag(tsk, TIF_IO_BITMAP))) {
52978be6 AD	220	p->thread.io_bitmap_ptr = kmemdup(tsk->thread.io_bitmap_ptr,
52978be6 AD	221	IO_BITMAP_BYTES, GFP_KERNEL);
1da177e4 LT	222	if (!p->thread.io_bitmap_ptr) {
	223	p->thread.io_bitmap_max = 0;
	224	return -ENOMEM;
	225	}
b3cf2576	226	set_tsk_thread_flag(p, TIF_IO_BITMAP);
1da177e4 LT	227	}
1da177e4 LT	228
efd1ca52 RM	229	err = 0;
efd1ca52 RM	230
1da177e4 LT	231	/*
	232	* Set a new TLS for the child thread?
	233	*/
efd1ca52 RM	234	if (clone_flags & CLONE_SETTLS)
efd1ca52 RM	235	err = do_set_thread_area(p, -1,
65ea5b03	236	(struct user_desc __user *)childregs->si, 0);
1da177e4	237
1da177e4 LT	238	if (err && p->thread.io_bitmap_ptr) {
	239	kfree(p->thread.io_bitmap_ptr);
	240	p->thread.io_bitmap_max = 0;
	241	}
	242	return err;
	243	}
	244
513ad84b IM	245	void
	246	start_thread(struct pt_regs *regs, unsigned long new_ip, unsigned long new_sp)
	247	{
d9a89a26	248	set_user_gs(regs, 0);
513ad84b	249	regs->fs = 0;
513ad84b IM	250	regs->ds = __USER_DS;
	251	regs->es = __USER_DS;
	252	regs->ss = __USER_DS;
	253	regs->cs = __USER_CS;
	254	regs->ip = new_ip;
	255	regs->sp = new_sp;
aa283f49 SS	256	/*
	257	* Free the old FP and other extended state
	258	*/
	259	free_thread_xstate(current);
513ad84b IM	260	}
	261	EXPORT_SYMBOL_GPL(start_thread);
	262
1da177e4 LT	263
	264	/*
	265	* switch_to(x,yn) should switch tasks from x to y.
	266	*
	267	* We fsave/fwait so that an exception goes off at the right time
	268	* (as a call from the fsave or fwait in effect) rather than to
	269	* the wrong process. Lazy FP saving no longer makes any sense
	270	* with modern CPU's, and this simplifies a lot of things (SMP
	271	* and UP become the same).
	272	*
	273	* NOTE! We used to use the x86 hardware context switching. The
	274	* reason for not using it any more becomes apparent when you
	275	* try to recover gracefully from saved state that is no longer
	276	* valid (stale segment register values in particular). With the
	277	* hardware task-switch, there is no way to fix up bad state in
	278	* a reasonable manner.
	279	*
	280	* The fact that Intel documents the hardware task-switching to
	281	* be slow is a fairly red herring - this code is not noticeably
	282	* faster. However, there _is_ some room for improvement here,
	283	* so the performance issues may eventually be a valid point.
	284	* More important, however, is the fact that this allows us much
	285	* more flexibility.
	286	*
65ea5b03	287	* The return value (in %ax) will be the "prev" task after
1da177e4 LT	288	* the task-switch, and shows up in ret_from_fork in entry.S,
	289	* for example.
	290	*/
8b96f011 FW	291	__notrace_funcgraph struct task_struct *
8b96f011 FW	292	__switch_to(struct task_struct prev_p, struct task_struct next_p)
1da177e4 LT	293	{
	294	struct thread_struct *prev = &prev_p->thread,
	295	*next = &next_p->thread;
	296	int cpu = smp_processor_id();
	297	struct tss_struct *tss = &per_cpu(init_tss, cpu);
2fcddce1	298	bool preload_fpu;
1da177e4 LT	299
	300	/* never put a printk in __switch_to... printk() calls wake_up() indirectly /
	301
2fcddce1 JF	302	/*
	303	* If the task has used fpu the last 5 timeslices, just do a full
	304	* restore of the math state immediately to avoid the trap; the
	305	* chances of needing FPU soon are obviously high now
	306	*/
	307	preload_fpu = tsk_used_math(next_p) && next_p->fpu_counter > 5;
1da177e4	308
2fcddce1	309	__unlazy_fpu(prev_p);
acc20761 CE	310
acc20761 CE	311	/* we're going to use this soon, after a few expensive things */
2fcddce1	312	if (preload_fpu)
86603283	313	prefetch(next->fpu.state);
acc20761	314
1da177e4	315	/*
e7a2ff59	316	* Reload esp0.
1da177e4	317	*/
faca6227	318	load_sp0(tss, next);
1da177e4 LT	319
1da177e4 LT	320	/*
464d1a78	321	* Save away %gs. No need to save %fs, as it was saved on the
f95d47ca JF	322	* stack on entry. No need to save %es and %ds, as those are
	323	* always kernel segments while inside the kernel. Doing this
	324	* before setting the new TLS descriptors avoids the situation
	325	* where we temporarily have non-reloadable segments in %fs
	326	* and %gs. This could be an issue if the NMI handler ever
	327	* used %fs or %gs (it does not today), or if the kernel is
	328	* running inside of a hypervisor layer.
1da177e4	329	*/
ccbeed3a	330	lazy_save_gs(prev->gs);
1da177e4 LT	331
1da177e4 LT	332	/*
e7a2ff59	333	* Load the per-thread Thread-Local Storage descriptor.
1da177e4	334	*/
e7a2ff59	335	load_TLS(next, cpu);
1da177e4	336
8b151144 ZA	337	/*
	338	* Restore IOPL if needed. In normal use, the flags restore
	339	* in the switch assembly will handle this. But if the kernel
	340	* is running virtualized at a non-zero CPL, the popf will
	341	* not restore flags, so it must be done in a separate step.
	342	*/
	343	if (get_kernel_rpl() && unlikely(prev->iopl != next->iopl))
	344	set_iopl_mask(next->iopl);
	345
1da177e4	346	/*
b3cf2576	347	* Now maybe handle debug registers and/or IO bitmaps
1da177e4	348	*/
cf99abac AA	349	if (unlikely(task_thread_info(prev_p)->flags & _TIF_WORK_CTXSW_PREV \|\|
	350	task_thread_info(next_p)->flags & _TIF_WORK_CTXSW_NEXT))
	351	__switch_to_xtra(prev_p, next_p, tss);
ffaa8bd6	352
2fcddce1 JF	353	/* If we're going to preload the fpu context, make sure clts
	354	is run while we're batching the cpu state updates. */
	355	if (preload_fpu)
	356	clts();
	357
9226d125 ZA	358	/*
	359	* Leave lazy mode, flushing any hypercalls made here.
	360	* This must be done before restoring TLS segments so
	361	* the GDT and LDT are properly updated, and must be
	362	* done before math_state_restore, so the TS bit is up
	363	* to date.
	364	*/
224101ed	365	arch_end_context_switch(next_p);
9226d125	366
2fcddce1 JF	367	if (preload_fpu)
2fcddce1 JF	368	__math_state_restore();
acc20761	369
9226d125 ZA	370	/*
	371	* Restore %gs if needed (which is common)
	372	*/
	373	if (prev->gs \| next->gs)
ccbeed3a	374	lazy_load_gs(next->gs);
9226d125	375
6dbde353	376	percpu_write(current_task, next_p);
9226d125	377
1da177e4 LT	378	return prev_p;
	379	}
	380
1da177e4 LT	381	#define top_esp (THREAD_SIZE - sizeof(unsigned long))
	382	#define top_ebp (THREAD_SIZE - 2*sizeof(unsigned long))
	383
	384	unsigned long get_wchan(struct task_struct *p)
	385	{
65ea5b03	386	unsigned long bp, sp, ip;
1da177e4 LT	387	unsigned long stack_page;
	388	int count = 0;
	389	if (!p \|\| p == current \|\| p->state == TASK_RUNNING)
	390	return 0;
65e0fdff	391	stack_page = (unsigned long)task_stack_page(p);
faca6227	392	sp = p->thread.sp;
65ea5b03	393	if (!stack_page \|\| sp < stack_page \|\| sp > top_esp+stack_page)
1da177e4	394	return 0;
65ea5b03 PA	395	/* include/asm-i386/system.h:switch_to() pushes bp last. */
65ea5b03 PA	396	bp = (unsigned long ) sp;
1da177e4	397	do {
65ea5b03	398	if (bp < stack_page \|\| bp > top_ebp+stack_page)
1da177e4	399	return 0;
65ea5b03 PA	400	ip = (unsigned long ) (bp+4);
	401	if (!in_sched_functions(ip))
	402	return ip;
	403	bp = (unsigned long ) bp;
1da177e4 LT	404	} while (count++ < 16);
	405	return 0;
	406	}
	407