[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 <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();
			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;
	set_fs(USER_DS);
	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>
1da177e4	41
1da177e4 LT	42	#include <asm/pgtable.h>
1da177e4 LT	43	#include <asm/system.h>
1da177e4 LT	44	#include <asm/ldt.h>
	45	#include <asm/processor.h>
	46	#include <asm/i387.h>
1da177e4 LT	47	#include <asm/desc.h>
	48	#ifdef CONFIG_MATH_EMULATION
	49	#include <asm/math_emu.h>
	50	#endif
	51
1da177e4 LT	52	#include <linux/err.h>
1da177e4 LT	53
f3705136 ZM	54	#include <asm/tlbflush.h>
f3705136 ZM	55	#include <asm/cpu.h>
1eda8149	56	#include <asm/idle.h>
bbc1f698	57	#include <asm/syscalls.h>
66cb5917	58	#include <asm/debugreg.h>
f3705136	59
1da177e4 LT	60	asmlinkage void ret_from_fork(void) __asm__("ret_from_fork");
1da177e4 LT	61
1da177e4 LT	62	/*
	63	* Return saved PC of a blocked thread.
	64	*/
	65	unsigned long thread_saved_pc(struct task_struct *tsk)
	66	{
faca6227	67	return ((unsigned long *)tsk->thread.sp)[3];
1da177e4 LT	68	}
1da177e4 LT	69
913da64b AN	70	#ifndef CONFIG_SMP
	71	static inline void play_dead(void)
	72	{
	73	BUG();
	74	}
	75	#endif
	76
1da177e4 LT	77	/*
	78	* The idle thread. There's no useful work to be
	79	* done, so just try to conserve power and have a
	80	* low exit latency (ie sit in a loop waiting for
	81	* somebody to say that they'd like to reschedule)
	82	*/
f3705136	83	void cpu_idle(void)
1da177e4	84	{
5bfb5d69	85	int cpu = smp_processor_id();
f3705136	86
5c79d2a5 TH	87	/*
	88	* If we're the non-boot CPU, nothing set the stack canary up
	89	* for us. CPU0 already has it initialized but no harm in
	90	* doing it again. This is a good place for updating it, as
	91	* we wont ever return from this function (so the invalid
	92	* canaries already on the stack wont ever trigger).
	93	*/
	94	boot_init_stack_canary();
	95
495ab9c0	96	current_thread_info()->status \|= TS_POLLING;
64c7c8f8	97
1da177e4 LT	98	/* endless idle loop with no priority at all */
1da177e4 LT	99	while (1) {
b8f8c3cf	100	tick_nohz_stop_sched_tick(1);
1da177e4	101	while (!need_resched()) {
1da177e4	102
f1d1a842	103	check_pgt_cache();
1da177e4	104	rmb();
1da177e4	105
f3705136 ZM	106	if (cpu_is_offline(cpu))
	107	play_dead();
	108
7f424a8b	109	local_irq_disable();
6cd8a4bb SR	110	/* Don't trace irqs off for idle */
6cd8a4bb SR	111	stop_critical_timings();
6ddd2a27	112	pm_idle();
6cd8a4bb	113	start_critical_timings();
1da177e4	114	}
74167347	115	tick_nohz_restart_sched_tick();
5bfb5d69	116	preempt_enable_no_resched();
1da177e4	117	schedule();
5bfb5d69	118	preempt_disable();
1da177e4 LT	119	}
	120	}
	121
e2ce07c8	122	void __show_regs(struct pt_regs *regs, int all)
1da177e4 LT	123	{
1da177e4 LT	124	unsigned long cr0 = 0L, cr2 = 0L, cr3 = 0L, cr4 = 0L;
bb1995d5	125	unsigned long d0, d1, d2, d3, d6, d7;
65ea5b03	126	unsigned long sp;
9d975ebd PE	127	unsigned short ss, gs;
	128
	129	if (user_mode_vm(regs)) {
65ea5b03 PA	130	sp = regs->sp;
65ea5b03 PA	131	ss = regs->ss & 0xffff;
d9a89a26	132	gs = get_user_gs(regs);
9d975ebd	133	} else {
def3c5d0	134	sp = kernel_stack_pointer(regs);
9d975ebd PE	135	savesegment(ss, ss);
	136	savesegment(gs, gs);
	137	}
1da177e4	138
814e2c84	139	show_regs_common();
9d975ebd	140
d015a092	141	printk(KERN_DEFAULT "EIP: %04x:[<%08lx>] EFLAGS: %08lx CPU: %d\n",
92bc2056	142	(u16)regs->cs, regs->ip, regs->flags,
9d975ebd	143	smp_processor_id());
65ea5b03	144	print_symbol("EIP is at %s\n", regs->ip);
1da177e4	145
d015a092	146	printk(KERN_DEFAULT "EAX: %08lx EBX: %08lx ECX: %08lx EDX: %08lx\n",
65ea5b03	147	regs->ax, regs->bx, regs->cx, regs->dx);
d015a092	148	printk(KERN_DEFAULT "ESI: %08lx EDI: %08lx EBP: %08lx ESP: %08lx\n",
65ea5b03	149	regs->si, regs->di, regs->bp, sp);
d015a092	150	printk(KERN_DEFAULT " DS: %04x ES: %04x FS: %04x GS: %04x SS: %04x\n",
92bc2056	151	(u16)regs->ds, (u16)regs->es, (u16)regs->fs, gs, ss);
9d975ebd PE	152
	153	if (!all)
	154	return;
1da177e4	155
4bb0d3ec ZA	156	cr0 = read_cr0();
	157	cr2 = read_cr2();
	158	cr3 = read_cr3();
ff6e8c0d	159	cr4 = read_cr4_safe();
d015a092	160	printk(KERN_DEFAULT "CR0: %08lx CR2: %08lx CR3: %08lx CR4: %08lx\n",
9d975ebd	161	cr0, cr2, cr3, cr4);
bb1995d5 AS	162
	163	get_debugreg(d0, 0);
	164	get_debugreg(d1, 1);
	165	get_debugreg(d2, 2);
	166	get_debugreg(d3, 3);
d015a092	167	printk(KERN_DEFAULT "DR0: %08lx DR1: %08lx DR2: %08lx DR3: %08lx\n",
bb1995d5	168	d0, d1, d2, d3);
9d975ebd	169
bb1995d5 AS	170	get_debugreg(d6, 6);
bb1995d5 AS	171	get_debugreg(d7, 7);
d015a092	172	printk(KERN_DEFAULT "DR6: %08lx DR7: %08lx\n",
9d975ebd PE	173	d6, d7);
9d975ebd PE	174	}
bb1995d5	175
1da177e4 LT	176	void release_thread(struct task_struct *dead_task)
1da177e4 LT	177	{
2684927c	178	BUG_ON(dead_task->mm);
1da177e4 LT	179	release_vm86_irqs(dead_task);
	180	}
	181
	182	/*
	183	* This gets called before we allocate a new thread and copy
	184	* the current task into it.
	185	*/
	186	void prepare_to_copy(struct task_struct *tsk)
	187	{
	188	unlazy_fpu(tsk);
	189	}
	190
6f2c55b8	191	int copy_thread(unsigned long clone_flags, unsigned long sp,
1da177e4	192	unsigned long unused,
befa9e78	193	struct task_struct p, struct pt_regs regs)
1da177e4	194	{
befa9e78	195	struct pt_regs *childregs;
1da177e4 LT	196	struct task_struct *tsk;
	197	int err;
	198
07b047fc	199	childregs = task_pt_regs(p);
f48d9663	200	childregs = regs;
65ea5b03 PA	201	childregs->ax = 0;
65ea5b03 PA	202	childregs->sp = sp;
f48d9663	203
faca6227 PA	204	p->thread.sp = (unsigned long) childregs;
faca6227 PA	205	p->thread.sp0 = (unsigned long) (childregs+1);
1da177e4	206
faca6227	207	p->thread.ip = (unsigned long) ret_from_fork;
1da177e4	208
d9a89a26	209	task_user_gs(p) = get_user_gs(regs);
1da177e4	210
66cb5917	211	p->thread.io_bitmap_ptr = NULL;
1da177e4	212	tsk = current;
66cb5917	213	err = -ENOMEM;
24f1e32c FW	214
24f1e32c FW	215	memset(p->thread.ptrace_bps, 0, sizeof(p->thread.ptrace_bps));
66cb5917	216
b3cf2576	217	if (unlikely(test_tsk_thread_flag(tsk, TIF_IO_BITMAP))) {
52978be6 AD	218	p->thread.io_bitmap_ptr = kmemdup(tsk->thread.io_bitmap_ptr,
52978be6 AD	219	IO_BITMAP_BYTES, GFP_KERNEL);
1da177e4 LT	220	if (!p->thread.io_bitmap_ptr) {
	221	p->thread.io_bitmap_max = 0;
	222	return -ENOMEM;
	223	}
b3cf2576	224	set_tsk_thread_flag(p, TIF_IO_BITMAP);
1da177e4 LT	225	}
1da177e4 LT	226
efd1ca52 RM	227	err = 0;
efd1ca52 RM	228
1da177e4 LT	229	/*
	230	* Set a new TLS for the child thread?
	231	*/
efd1ca52 RM	232	if (clone_flags & CLONE_SETTLS)
efd1ca52 RM	233	err = do_set_thread_area(p, -1,
65ea5b03	234	(struct user_desc __user *)childregs->si, 0);
1da177e4	235
1da177e4 LT	236	if (err && p->thread.io_bitmap_ptr) {
	237	kfree(p->thread.io_bitmap_ptr);
	238	p->thread.io_bitmap_max = 0;
	239	}
	240	return err;
	241	}
	242
513ad84b IM	243	void
	244	start_thread(struct pt_regs *regs, unsigned long new_ip, unsigned long new_sp)
	245	{
d9a89a26	246	set_user_gs(regs, 0);
513ad84b IM	247	regs->fs = 0;
	248	set_fs(USER_DS);
	249	regs->ds = __USER_DS;
	250	regs->es = __USER_DS;
	251	regs->ss = __USER_DS;
	252	regs->cs = __USER_CS;
	253	regs->ip = new_ip;
	254	regs->sp = new_sp;
aa283f49 SS	255	/*
	256	* Free the old FP and other extended state
	257	*/
	258	free_thread_xstate(current);
513ad84b IM	259	}
	260	EXPORT_SYMBOL_GPL(start_thread);
	261
1da177e4 LT	262
	263	/*
	264	* switch_to(x,yn) should switch tasks from x to y.
	265	*
	266	* We fsave/fwait so that an exception goes off at the right time
	267	* (as a call from the fsave or fwait in effect) rather than to
	268	* the wrong process. Lazy FP saving no longer makes any sense
	269	* with modern CPU's, and this simplifies a lot of things (SMP
	270	* and UP become the same).
	271	*
	272	* NOTE! We used to use the x86 hardware context switching. The
	273	* reason for not using it any more becomes apparent when you
	274	* try to recover gracefully from saved state that is no longer
	275	* valid (stale segment register values in particular). With the
	276	* hardware task-switch, there is no way to fix up bad state in
	277	* a reasonable manner.
	278	*
	279	* The fact that Intel documents the hardware task-switching to
	280	* be slow is a fairly red herring - this code is not noticeably
	281	* faster. However, there _is_ some room for improvement here,
	282	* so the performance issues may eventually be a valid point.
	283	* More important, however, is the fact that this allows us much
	284	* more flexibility.
	285	*
65ea5b03	286	* The return value (in %ax) will be the "prev" task after
1da177e4 LT	287	* the task-switch, and shows up in ret_from_fork in entry.S,
	288	* for example.
	289	*/
8b96f011 FW	290	__notrace_funcgraph struct task_struct *
8b96f011 FW	291	__switch_to(struct task_struct prev_p, struct task_struct next_p)
1da177e4 LT	292	{
	293	struct thread_struct *prev = &prev_p->thread,
	294	*next = &next_p->thread;
	295	int cpu = smp_processor_id();
	296	struct tss_struct *tss = &per_cpu(init_tss, cpu);
2fcddce1	297	bool preload_fpu;
1da177e4 LT	298
	299	/* never put a printk in __switch_to... printk() calls wake_up() indirectly /
	300
2fcddce1 JF	301	/*
	302	* If the task has used fpu the last 5 timeslices, just do a full
	303	* restore of the math state immediately to avoid the trap; the
	304	* chances of needing FPU soon are obviously high now
	305	*/
	306	preload_fpu = tsk_used_math(next_p) && next_p->fpu_counter > 5;
1da177e4	307
2fcddce1	308	__unlazy_fpu(prev_p);
acc20761 CE	309
acc20761 CE	310	/* we're going to use this soon, after a few expensive things */
2fcddce1	311	if (preload_fpu)
86603283	312	prefetch(next->fpu.state);
acc20761	313
1da177e4	314	/*
e7a2ff59	315	* Reload esp0.
1da177e4	316	*/
faca6227	317	load_sp0(tss, next);
1da177e4 LT	318
1da177e4 LT	319	/*
464d1a78	320	* Save away %gs. No need to save %fs, as it was saved on the
f95d47ca JF	321	* stack on entry. No need to save %es and %ds, as those are
	322	* always kernel segments while inside the kernel. Doing this
	323	* before setting the new TLS descriptors avoids the situation
	324	* where we temporarily have non-reloadable segments in %fs
	325	* and %gs. This could be an issue if the NMI handler ever
	326	* used %fs or %gs (it does not today), or if the kernel is
	327	* running inside of a hypervisor layer.
1da177e4	328	*/
ccbeed3a	329	lazy_save_gs(prev->gs);
1da177e4 LT	330
1da177e4 LT	331	/*
e7a2ff59	332	* Load the per-thread Thread-Local Storage descriptor.
1da177e4	333	*/
e7a2ff59	334	load_TLS(next, cpu);
1da177e4	335
8b151144 ZA	336	/*
	337	* Restore IOPL if needed. In normal use, the flags restore
	338	* in the switch assembly will handle this. But if the kernel
	339	* is running virtualized at a non-zero CPL, the popf will
	340	* not restore flags, so it must be done in a separate step.
	341	*/
	342	if (get_kernel_rpl() && unlikely(prev->iopl != next->iopl))
	343	set_iopl_mask(next->iopl);
	344
1da177e4	345	/*
b3cf2576	346	* Now maybe handle debug registers and/or IO bitmaps
1da177e4	347	*/
cf99abac AA	348	if (unlikely(task_thread_info(prev_p)->flags & _TIF_WORK_CTXSW_PREV \|\|
	349	task_thread_info(next_p)->flags & _TIF_WORK_CTXSW_NEXT))
	350	__switch_to_xtra(prev_p, next_p, tss);
ffaa8bd6	351
2fcddce1 JF	352	/* If we're going to preload the fpu context, make sure clts
	353	is run while we're batching the cpu state updates. */
	354	if (preload_fpu)
	355	clts();
	356
9226d125 ZA	357	/*
	358	* Leave lazy mode, flushing any hypercalls made here.
	359	* This must be done before restoring TLS segments so
	360	* the GDT and LDT are properly updated, and must be
	361	* done before math_state_restore, so the TS bit is up
	362	* to date.
	363	*/
224101ed	364	arch_end_context_switch(next_p);
9226d125	365
2fcddce1 JF	366	if (preload_fpu)
2fcddce1 JF	367	__math_state_restore();
acc20761	368
9226d125 ZA	369	/*
	370	* Restore %gs if needed (which is common)
	371	*/
	372	if (prev->gs \| next->gs)
ccbeed3a	373	lazy_load_gs(next->gs);
9226d125	374
6dbde353	375	percpu_write(current_task, next_p);
9226d125	376
1da177e4 LT	377	return prev_p;
	378	}
	379
1da177e4 LT	380	#define top_esp (THREAD_SIZE - sizeof(unsigned long))
	381	#define top_ebp (THREAD_SIZE - 2*sizeof(unsigned long))
	382
	383	unsigned long get_wchan(struct task_struct *p)
	384	{
65ea5b03	385	unsigned long bp, sp, ip;
1da177e4 LT	386	unsigned long stack_page;
	387	int count = 0;
	388	if (!p \|\| p == current \|\| p->state == TASK_RUNNING)
	389	return 0;
65e0fdff	390	stack_page = (unsigned long)task_stack_page(p);
faca6227	391	sp = p->thread.sp;
65ea5b03	392	if (!stack_page \|\| sp < stack_page \|\| sp > top_esp+stack_page)
1da177e4	393	return 0;
65ea5b03 PA	394	/* include/asm-i386/system.h:switch_to() pushes bp last. */
65ea5b03 PA	395	bp = (unsigned long ) sp;
1da177e4	396	do {
65ea5b03	397	if (bp < stack_page \|\| bp > top_ebp+stack_page)
1da177e4	398	return 0;
65ea5b03 PA	399	ip = (unsigned long ) (bp+4);
	400	if (!in_sched_functions(ip))
	401	return ip;
	402	bp = (unsigned long ) bp;
1da177e4 LT	403	} while (count++ < 16);
	404	return 0;
	405	}
	406