[mirror_ubuntu-bionic-kernel.git] / arch / x86 / mm / tlb.c

#include <linux/init.h>

#include <linux/mm.h>
#include <linux/spinlock.h>
#include <linux/smp.h>
#include <linux/interrupt.h>
#include <linux/module.h>
#include <linux/cpu.h>

#include <asm/tlbflush.h>
#include <asm/mmu_context.h>
#include <asm/cache.h>
#include <asm/apic.h>
#include <asm/uv/uv.h>
#include <linux/debugfs.h>

DEFINE_PER_CPU_SHARED_ALIGNED(struct tlb_state, cpu_tlbstate)
			= { &init_mm, 0, };

/*
 *	Smarter SMP flushing macros.
 *		c/o Linus Torvalds.
 *
 *	These mean you can really definitely utterly forget about
 *	writing to user space from interrupts. (Its not allowed anyway).
 *
 *	Optimizations Manfred Spraul <manfred@colorfullife.com>
 *
 *	More scalable flush, from Andi Kleen
 *
 *	Implement flush IPI by CALL_FUNCTION_VECTOR, Alex Shi
 */

struct flush_tlb_info {
	struct mm_struct *flush_mm;
	unsigned long flush_start;
	unsigned long flush_end;
};

/*
 * We cannot call mmdrop() because we are in interrupt context,
 * instead update mm->cpu_vm_mask.
 */
void leave_mm(int cpu)
{
	struct mm_struct *active_mm = this_cpu_read(cpu_tlbstate.active_mm);
	if (this_cpu_read(cpu_tlbstate.state) == TLBSTATE_OK)
		BUG();
	if (cpumask_test_cpu(cpu, mm_cpumask(active_mm))) {
		cpumask_clear_cpu(cpu, mm_cpumask(active_mm));
		load_cr3(swapper_pg_dir);
	}
}
EXPORT_SYMBOL_GPL(leave_mm);

/*
 * The flush IPI assumes that a thread switch happens in this order:
 * [cpu0: the cpu that switches]
 * 1) switch_mm() either 1a) or 1b)
 * 1a) thread switch to a different mm
 * 1a1) set cpu_tlbstate to TLBSTATE_OK
 *	Now the tlb flush NMI handler flush_tlb_func won't call leave_mm
 *	if cpu0 was in lazy tlb mode.
 * 1a2) update cpu active_mm
 *	Now cpu0 accepts tlb flushes for the new mm.
 * 1a3) cpu_set(cpu, new_mm->cpu_vm_mask);
 *	Now the other cpus will send tlb flush ipis.
 * 1a4) change cr3.
 * 1a5) cpu_clear(cpu, old_mm->cpu_vm_mask);
 *	Stop ipi delivery for the old mm. This is not synchronized with
 *	the other cpus, but flush_tlb_func ignore flush ipis for the wrong
 *	mm, and in the worst case we perform a superfluous tlb flush.
 * 1b) thread switch without mm change
 *	cpu active_mm is correct, cpu0 already handles flush ipis.
 * 1b1) set cpu_tlbstate to TLBSTATE_OK
 * 1b2) test_and_set the cpu bit in cpu_vm_mask.
 *	Atomically set the bit [other cpus will start sending flush ipis],
 *	and test the bit.
 * 1b3) if the bit was 0: leave_mm was called, flush the tlb.
 * 2) switch %%esp, ie current
 *
 * The interrupt must handle 2 special cases:
 * - cr3 is changed before %%esp, ie. it cannot use current->{active_,}mm.
 * - the cpu performs speculative tlb reads, i.e. even if the cpu only
 *   runs in kernel space, the cpu could load tlb entries for user space
 *   pages.
 *
 * The good news is that cpu_tlbstate is local to each cpu, no
 * write/read ordering problems.
 */

/*
 * TLB flush funcation:
 * 1) Flush the tlb entries if the cpu uses the mm that's being flushed.
 * 2) Leave the mm if we are in the lazy tlb mode.
 */
static void flush_tlb_func(void *info)
{
	struct flush_tlb_info *f = info;

	inc_irq_stat(irq_tlb_count);

	if (f->flush_mm != this_cpu_read(cpu_tlbstate.active_mm))
		return;

	count_vm_event(NR_TLB_REMOTE_FLUSH_RECEIVED);
	if (this_cpu_read(cpu_tlbstate.state) == TLBSTATE_OK) {
		if (f->flush_end == TLB_FLUSH_ALL)
			local_flush_tlb();
		else if (!f->flush_end)
			__flush_tlb_single(f->flush_start);
		else {
			unsigned long addr;
			addr = f->flush_start;
			while (addr < f->flush_end) {
				__flush_tlb_single(addr);
				addr += PAGE_SIZE;
			}
		}
	} else
		leave_mm(smp_processor_id());

}

void native_flush_tlb_others(const struct cpumask *cpumask,
				 struct mm_struct *mm, unsigned long start,
				 unsigned long end)
{
	struct flush_tlb_info info;
	info.flush_mm = mm;
	info.flush_start = start;
	info.flush_end = end;

	count_vm_event(NR_TLB_REMOTE_FLUSH);
	if (is_uv_system()) {
		unsigned int cpu;

		cpu = smp_processor_id();
		cpumask = uv_flush_tlb_others(cpumask, mm, start, end, cpu);
		if (cpumask)
			smp_call_function_many(cpumask, flush_tlb_func,
								&info, 1);
		return;
	}
	smp_call_function_many(cpumask, flush_tlb_func, &info, 1);
}

void flush_tlb_current_task(void)
{
	struct mm_struct *mm = current->mm;

	preempt_disable();

	count_vm_event(NR_TLB_LOCAL_FLUSH_ALL);
	local_flush_tlb();
	if (cpumask_any_but(mm_cpumask(mm), smp_processor_id()) < nr_cpu_ids)
		flush_tlb_others(mm_cpumask(mm), mm, 0UL, TLB_FLUSH_ALL);
	preempt_enable();
}

/*
 * It can find out the THP large page, or
 * HUGETLB page in tlb_flush when THP disabled
 */
static inline unsigned long has_large_page(struct mm_struct *mm,
				 unsigned long start, unsigned long end)
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	unsigned long addr = ALIGN(start, HPAGE_SIZE);
	for (; addr < end; addr += HPAGE_SIZE) {
		pgd = pgd_offset(mm, addr);
		if (likely(!pgd_none(*pgd))) {
			pud = pud_offset(pgd, addr);
			if (likely(!pud_none(*pud))) {
				pmd = pmd_offset(pud, addr);
				if (likely(!pmd_none(*pmd)))
					if (pmd_large(*pmd))
						return addr;
			}
		}
	}
	return 0;
}

void flush_tlb_mm_range(struct mm_struct *mm, unsigned long start,
				unsigned long end, unsigned long vmflag)
{
	unsigned long addr;
	unsigned act_entries, tlb_entries = 0;

	preempt_disable();
	if (current->active_mm != mm)
		goto flush_all;

	if (!current->mm) {
		leave_mm(smp_processor_id());
		goto flush_all;
	}

	if (end == TLB_FLUSH_ALL || tlb_flushall_shift == -1
					|| vmflag & VM_HUGETLB) {
		local_flush_tlb();
		goto flush_all;
	}

	/* In modern CPU, last level tlb used for both data/ins */
	if (vmflag & VM_EXEC)
		tlb_entries = tlb_lli_4k[ENTRIES];
	else
		tlb_entries = tlb_lld_4k[ENTRIES];
	/* Assume all of TLB entries was occupied by this task */
	act_entries = mm->total_vm > tlb_entries ? tlb_entries : mm->total_vm;

	/* tlb_flushall_shift is on balance point, details in commit log */
	if ((end - start) >> PAGE_SHIFT > act_entries >> tlb_flushall_shift) {
		count_vm_event(NR_TLB_LOCAL_FLUSH_ALL);
		local_flush_tlb();
	} else {
		if (has_large_page(mm, start, end)) {
			local_flush_tlb();
			goto flush_all;
		}
		/* flush range by one by one 'invlpg' */
		for (addr = start; addr < end;	addr += PAGE_SIZE) {
			count_vm_event(NR_TLB_LOCAL_FLUSH_ONE);
			__flush_tlb_single(addr);
		}

		if (cpumask_any_but(mm_cpumask(mm),
				smp_processor_id()) < nr_cpu_ids)
			flush_tlb_others(mm_cpumask(mm), mm, start, end);
		preempt_enable();
		return;
	}

flush_all:
	if (cpumask_any_but(mm_cpumask(mm), smp_processor_id()) < nr_cpu_ids)
		flush_tlb_others(mm_cpumask(mm), mm, 0UL, TLB_FLUSH_ALL);
	preempt_enable();
}

void flush_tlb_page(struct vm_area_struct *vma, unsigned long start)
{
	struct mm_struct *mm = vma->vm_mm;

	preempt_disable();

	if (current->active_mm == mm) {
		if (current->mm)
			__flush_tlb_one(start);
		else
			leave_mm(smp_processor_id());
	}

	if (cpumask_any_but(mm_cpumask(mm), smp_processor_id()) < nr_cpu_ids)
		flush_tlb_others(mm_cpumask(mm), mm, start, 0UL);

	preempt_enable();
}

static void do_flush_tlb_all(void *info)
{
	count_vm_event(NR_TLB_REMOTE_FLUSH_RECEIVED);
	__flush_tlb_all();
	if (this_cpu_read(cpu_tlbstate.state) == TLBSTATE_LAZY)
		leave_mm(smp_processor_id());
}

void flush_tlb_all(void)
{
	count_vm_event(NR_TLB_REMOTE_FLUSH);
	on_each_cpu(do_flush_tlb_all, NULL, 1);
}

static void do_kernel_range_flush(void *info)
{
	struct flush_tlb_info *f = info;
	unsigned long addr;

	/* flush range by one by one 'invlpg' */
	for (addr = f->flush_start; addr < f->flush_end; addr += PAGE_SIZE) {
		count_vm_event(NR_TLB_LOCAL_FLUSH_ONE_KERNEL);
		__flush_tlb_single(addr);
	}
}

void flush_tlb_kernel_range(unsigned long start, unsigned long end)
{
	unsigned act_entries;
	struct flush_tlb_info info;

	/* In modern CPU, last level tlb used for both data/ins */
	act_entries = tlb_lld_4k[ENTRIES];

	/* Balance as user space task's flush, a bit conservative */
	if (end == TLB_FLUSH_ALL || tlb_flushall_shift == -1 ||
		(end - start) >> PAGE_SHIFT > act_entries >> tlb_flushall_shift)

		on_each_cpu(do_flush_tlb_all, NULL, 1);
	else {
		info.flush_start = start;
		info.flush_end = end;
		on_each_cpu(do_kernel_range_flush, &info, 1);
	}
}

#ifdef CONFIG_DEBUG_TLBFLUSH
static ssize_t tlbflush_read_file(struct file *file, char __user *user_buf,
			     size_t count, loff_t *ppos)
{
	char buf[32];
	unsigned int len;

	len = sprintf(buf, "%hd\n", tlb_flushall_shift);
	return simple_read_from_buffer(user_buf, count, ppos, buf, len);
}

static ssize_t tlbflush_write_file(struct file *file,
		 const char __user *user_buf, size_t count, loff_t *ppos)
{
	char buf[32];
	ssize_t len;
	s8 shift;

	len = min(count, sizeof(buf) - 1);
	if (copy_from_user(buf, user_buf, len))
		return -EFAULT;

	buf[len] = '\0';
	if (kstrtos8(buf, 0, &shift))
		return -EINVAL;

	if (shift < -1 || shift >= BITS_PER_LONG)
		return -EINVAL;

	tlb_flushall_shift = shift;
	return count;
}

static const struct file_operations fops_tlbflush = {
	.read = tlbflush_read_file,
	.write = tlbflush_write_file,
	.llseek = default_llseek,
};

static int __init create_tlb_flushall_shift(void)
{
	debugfs_create_file("tlb_flushall_shift", S_IRUSR | S_IWUSR,
			    arch_debugfs_dir, NULL, &fops_tlbflush);
	return 0;
}
late_initcall(create_tlb_flushall_shift);
#endif
Commit	Line	Data
c048fdfe GC	1	#include <linux/init.h>
	2
	3	#include <linux/mm.h>
c048fdfe GC	4	#include <linux/spinlock.h>
c048fdfe GC	5	#include <linux/smp.h>
c048fdfe	6	#include <linux/interrupt.h>
6dd01bed	7	#include <linux/module.h>
93296720	8	#include <linux/cpu.h>
c048fdfe	9
c048fdfe	10	#include <asm/tlbflush.h>
c048fdfe	11	#include <asm/mmu_context.h>
350f8f56	12	#include <asm/cache.h>
6dd01bed	13	#include <asm/apic.h>
bdbcdd48	14	#include <asm/uv/uv.h>
3df3212f	15	#include <linux/debugfs.h>
5af5573e	16
9eb912d1 BG	17	DEFINE_PER_CPU_SHARED_ALIGNED(struct tlb_state, cpu_tlbstate)
	18	= { &init_mm, 0, };
	19
c048fdfe GC	20	/*
	21	* Smarter SMP flushing macros.
	22	* c/o Linus Torvalds.
	23	*
	24	* These mean you can really definitely utterly forget about
	25	* writing to user space from interrupts. (Its not allowed anyway).
	26	*
	27	* Optimizations Manfred Spraul <manfred@colorfullife.com>
	28	*
	29	* More scalable flush, from Andi Kleen
	30	*
52aec330	31	* Implement flush IPI by CALL_FUNCTION_VECTOR, Alex Shi
c048fdfe GC	32	*/
c048fdfe GC	33
52aec330 AS	34	struct flush_tlb_info {
	35	struct mm_struct *flush_mm;
	36	unsigned long flush_start;
	37	unsigned long flush_end;
	38	};
93296720	39
c048fdfe GC	40	/*
	41	* We cannot call mmdrop() because we are in interrupt context,
	42	* instead update mm->cpu_vm_mask.
	43	*/
	44	void leave_mm(int cpu)
	45	{
02171b4a	46	struct mm_struct *active_mm = this_cpu_read(cpu_tlbstate.active_mm);
c6ae41e7	47	if (this_cpu_read(cpu_tlbstate.state) == TLBSTATE_OK)
c048fdfe	48	BUG();
a6fca40f SS	49	if (cpumask_test_cpu(cpu, mm_cpumask(active_mm))) {
	50	cpumask_clear_cpu(cpu, mm_cpumask(active_mm));
	51	load_cr3(swapper_pg_dir);
	52	}
c048fdfe GC	53	}
	54	EXPORT_SYMBOL_GPL(leave_mm);
	55
	56	/*
c048fdfe GC	57	* The flush IPI assumes that a thread switch happens in this order:
	58	* [cpu0: the cpu that switches]
	59	* 1) switch_mm() either 1a) or 1b)
	60	* 1a) thread switch to a different mm
52aec330 AS	61	* 1a1) set cpu_tlbstate to TLBSTATE_OK
	62	* Now the tlb flush NMI handler flush_tlb_func won't call leave_mm
	63	* if cpu0 was in lazy tlb mode.
	64	* 1a2) update cpu active_mm
c048fdfe	65	* Now cpu0 accepts tlb flushes for the new mm.
52aec330	66	* 1a3) cpu_set(cpu, new_mm->cpu_vm_mask);
c048fdfe GC	67	* Now the other cpus will send tlb flush ipis.
c048fdfe GC	68	* 1a4) change cr3.
52aec330 AS	69	* 1a5) cpu_clear(cpu, old_mm->cpu_vm_mask);
	70	* Stop ipi delivery for the old mm. This is not synchronized with
	71	* the other cpus, but flush_tlb_func ignore flush ipis for the wrong
	72	* mm, and in the worst case we perform a superfluous tlb flush.
c048fdfe	73	* 1b) thread switch without mm change
52aec330 AS	74	* cpu active_mm is correct, cpu0 already handles flush ipis.
52aec330 AS	75	* 1b1) set cpu_tlbstate to TLBSTATE_OK
c048fdfe GC	76	* 1b2) test_and_set the cpu bit in cpu_vm_mask.
	77	* Atomically set the bit [other cpus will start sending flush ipis],
	78	* and test the bit.
	79	* 1b3) if the bit was 0: leave_mm was called, flush the tlb.
	80	* 2) switch %%esp, ie current
	81	*
	82	* The interrupt must handle 2 special cases:
	83	* - cr3 is changed before %%esp, ie. it cannot use current->{active_,}mm.
	84	* - the cpu performs speculative tlb reads, i.e. even if the cpu only
	85	* runs in kernel space, the cpu could load tlb entries for user space
	86	* pages.
	87	*
52aec330	88	* The good news is that cpu_tlbstate is local to each cpu, no
c048fdfe GC	89	* write/read ordering problems.
	90	*/
	91
	92	/*
52aec330	93	* TLB flush funcation:
c048fdfe GC	94	* 1) Flush the tlb entries if the cpu uses the mm that's being flushed.
c048fdfe GC	95	* 2) Leave the mm if we are in the lazy tlb mode.
02cf94c3	96	*/
52aec330	97	static void flush_tlb_func(void *info)
c048fdfe	98	{
52aec330	99	struct flush_tlb_info *f = info;
c048fdfe	100
fd0f5869 TS	101	inc_irq_stat(irq_tlb_count);
fd0f5869 TS	102
52aec330 AS	103	if (f->flush_mm != this_cpu_read(cpu_tlbstate.active_mm))
52aec330 AS	104	return;
c048fdfe	105
9824cf97	106	count_vm_event(NR_TLB_REMOTE_FLUSH_RECEIVED);
52aec330	107	if (this_cpu_read(cpu_tlbstate.state) == TLBSTATE_OK) {
094ab1db	108	if (f->flush_end == TLB_FLUSH_ALL)
52aec330 AS	109	local_flush_tlb();
	110	else if (!f->flush_end)
	111	__flush_tlb_single(f->flush_start);
	112	else {
	113	unsigned long addr;
	114	addr = f->flush_start;
	115	while (addr < f->flush_end) {
	116	__flush_tlb_single(addr);
	117	addr += PAGE_SIZE;
e7b52ffd	118	}
52aec330 AS	119	}
	120	} else
	121	leave_mm(smp_processor_id());
c048fdfe	122
c048fdfe GC	123	}
c048fdfe GC	124
4595f962	125	void native_flush_tlb_others(const struct cpumask *cpumask,
e7b52ffd AS	126	struct mm_struct *mm, unsigned long start,
e7b52ffd AS	127	unsigned long end)
4595f962	128	{
52aec330 AS	129	struct flush_tlb_info info;
	130	info.flush_mm = mm;
	131	info.flush_start = start;
	132	info.flush_end = end;
	133
9824cf97	134	count_vm_event(NR_TLB_REMOTE_FLUSH);
4595f962	135	if (is_uv_system()) {
bdbcdd48	136	unsigned int cpu;
0e21990a	137
25542c64	138	cpu = smp_processor_id();
e7b52ffd	139	cpumask = uv_flush_tlb_others(cpumask, mm, start, end, cpu);
bdbcdd48	140	if (cpumask)
52aec330 AS	141	smp_call_function_many(cpumask, flush_tlb_func,
52aec330 AS	142	&info, 1);
0e21990a	143	return;
4595f962	144	}
52aec330	145	smp_call_function_many(cpumask, flush_tlb_func, &info, 1);
c048fdfe	146	}
c048fdfe GC	147
	148	void flush_tlb_current_task(void)
	149	{
	150	struct mm_struct *mm = current->mm;
c048fdfe GC	151
c048fdfe GC	152	preempt_disable();
c048fdfe	153
9824cf97	154	count_vm_event(NR_TLB_LOCAL_FLUSH_ALL);
c048fdfe	155	local_flush_tlb();
78f1c4d6	156	if (cpumask_any_but(mm_cpumask(mm), smp_processor_id()) < nr_cpu_ids)
e7b52ffd	157	flush_tlb_others(mm_cpumask(mm), mm, 0UL, TLB_FLUSH_ALL);
c048fdfe GC	158	preempt_enable();
	159	}
	160
611ae8e3 AS	161	/*
	162	* It can find out the THP large page, or
	163	* HUGETLB page in tlb_flush when THP disabled
	164	*/
d8dfe60d AS	165	static inline unsigned long has_large_page(struct mm_struct *mm,
	166	unsigned long start, unsigned long end)
	167	{
	168	pgd_t *pgd;
	169	pud_t *pud;
	170	pmd_t *pmd;
	171	unsigned long addr = ALIGN(start, HPAGE_SIZE);
	172	for (; addr < end; addr += HPAGE_SIZE) {
	173	pgd = pgd_offset(mm, addr);
	174	if (likely(!pgd_none(*pgd))) {
	175	pud = pud_offset(pgd, addr);
	176	if (likely(!pud_none(*pud))) {
	177	pmd = pmd_offset(pud, addr);
	178	if (likely(!pmd_none(*pmd)))
	179	if (pmd_large(*pmd))
	180	return addr;
	181	}
	182	}
	183	}
	184	return 0;
	185	}
e7b52ffd	186
611ae8e3 AS	187	void flush_tlb_mm_range(struct mm_struct *mm, unsigned long start,
	188	unsigned long end, unsigned long vmflag)
	189	{
	190	unsigned long addr;
	191	unsigned act_entries, tlb_entries = 0;
e7b52ffd AS	192
e7b52ffd AS	193	preempt_disable();
611ae8e3 AS	194	if (current->active_mm != mm)
611ae8e3 AS	195	goto flush_all;
e7b52ffd	196
611ae8e3 AS	197	if (!current->mm) {
	198	leave_mm(smp_processor_id());
	199	goto flush_all;
	200	}
c048fdfe	201
611ae8e3	202	if (end == TLB_FLUSH_ALL \|\| tlb_flushall_shift == -1
ddd32b42	203	\|\| vmflag & VM_HUGETLB) {
611ae8e3 AS	204	local_flush_tlb();
	205	goto flush_all;
	206	}
e7b52ffd	207
611ae8e3 AS	208	/* In modern CPU, last level tlb used for both data/ins */
	209	if (vmflag & VM_EXEC)
	210	tlb_entries = tlb_lli_4k[ENTRIES];
	211	else
	212	tlb_entries = tlb_lld_4k[ENTRIES];
	213	/* Assume all of TLB entries was occupied by this task */
	214	act_entries = mm->total_vm > tlb_entries ? tlb_entries : mm->total_vm;
	215
	216	/* tlb_flushall_shift is on balance point, details in commit log */
9824cf97 DH	217	if ((end - start) >> PAGE_SHIFT > act_entries >> tlb_flushall_shift) {
9824cf97 DH	218	count_vm_event(NR_TLB_LOCAL_FLUSH_ALL);
611ae8e3	219	local_flush_tlb();
9824cf97	220	} else {
611ae8e3 AS	221	if (has_large_page(mm, start, end)) {
	222	local_flush_tlb();
	223	goto flush_all;
e7b52ffd	224	}
611ae8e3	225	/* flush range by one by one 'invlpg' */
9824cf97 DH	226	for (addr = start; addr < end; addr += PAGE_SIZE) {
9824cf97 DH	227	count_vm_event(NR_TLB_LOCAL_FLUSH_ONE);
611ae8e3	228	__flush_tlb_single(addr);
9824cf97	229	}
611ae8e3 AS	230
	231	if (cpumask_any_but(mm_cpumask(mm),
	232	smp_processor_id()) < nr_cpu_ids)
	233	flush_tlb_others(mm_cpumask(mm), mm, start, end);
	234	preempt_enable();
	235	return;
e7b52ffd	236	}
611ae8e3 AS	237
611ae8e3 AS	238	flush_all:
e7b52ffd AS	239	if (cpumask_any_but(mm_cpumask(mm), smp_processor_id()) < nr_cpu_ids)
e7b52ffd AS	240	flush_tlb_others(mm_cpumask(mm), mm, 0UL, TLB_FLUSH_ALL);
c048fdfe GC	241	preempt_enable();
	242	}
	243
e7b52ffd	244	void flush_tlb_page(struct vm_area_struct *vma, unsigned long start)
c048fdfe GC	245	{
c048fdfe GC	246	struct mm_struct *mm = vma->vm_mm;
c048fdfe GC	247
c048fdfe GC	248	preempt_disable();
c048fdfe GC	249
	250	if (current->active_mm == mm) {
	251	if (current->mm)
e7b52ffd	252	__flush_tlb_one(start);
c048fdfe GC	253	else
	254	leave_mm(smp_processor_id());
	255	}
	256
78f1c4d6	257	if (cpumask_any_but(mm_cpumask(mm), smp_processor_id()) < nr_cpu_ids)
e7b52ffd	258	flush_tlb_others(mm_cpumask(mm), mm, start, 0UL);
c048fdfe GC	259
	260	preempt_enable();
	261	}
	262
	263	static void do_flush_tlb_all(void *info)
	264	{
9824cf97	265	count_vm_event(NR_TLB_REMOTE_FLUSH_RECEIVED);
c048fdfe	266	__flush_tlb_all();
c6ae41e7	267	if (this_cpu_read(cpu_tlbstate.state) == TLBSTATE_LAZY)
3f8afb77	268	leave_mm(smp_processor_id());
c048fdfe GC	269	}
	270
	271	void flush_tlb_all(void)
	272	{
9824cf97	273	count_vm_event(NR_TLB_REMOTE_FLUSH);
15c8b6c1	274	on_each_cpu(do_flush_tlb_all, NULL, 1);
c048fdfe	275	}
3df3212f	276
effee4b9 AS	277	static void do_kernel_range_flush(void *info)
	278	{
	279	struct flush_tlb_info *f = info;
	280	unsigned long addr;
	281
	282	/* flush range by one by one 'invlpg' */
9824cf97 DH	283	for (addr = f->flush_start; addr < f->flush_end; addr += PAGE_SIZE) {
9824cf97 DH	284	count_vm_event(NR_TLB_LOCAL_FLUSH_ONE_KERNEL);
effee4b9	285	__flush_tlb_single(addr);
9824cf97	286	}
effee4b9 AS	287	}
	288
	289	void flush_tlb_kernel_range(unsigned long start, unsigned long end)
	290	{
	291	unsigned act_entries;
	292	struct flush_tlb_info info;
	293
	294	/* In modern CPU, last level tlb used for both data/ins */
	295	act_entries = tlb_lld_4k[ENTRIES];
	296
	297	/* Balance as user space task's flush, a bit conservative */
	298	if (end == TLB_FLUSH_ALL \|\| tlb_flushall_shift == -1 \|\|
	299	(end - start) >> PAGE_SHIFT > act_entries >> tlb_flushall_shift)
	300
	301	on_each_cpu(do_flush_tlb_all, NULL, 1);
	302	else {
	303	info.flush_start = start;
	304	info.flush_end = end;
	305	on_each_cpu(do_kernel_range_flush, &info, 1);
	306	}
	307	}
	308
3df3212f AS	309	#ifdef CONFIG_DEBUG_TLBFLUSH
	310	static ssize_t tlbflush_read_file(struct file file, char __user user_buf,
	311	size_t count, loff_t *ppos)
	312	{
	313	char buf[32];
	314	unsigned int len;
	315
	316	len = sprintf(buf, "%hd\n", tlb_flushall_shift);
	317	return simple_read_from_buffer(user_buf, count, ppos, buf, len);
	318	}
	319
	320	static ssize_t tlbflush_write_file(struct file *file,
	321	const char __user user_buf, size_t count, loff_t ppos)
	322	{
	323	char buf[32];
	324	ssize_t len;
	325	s8 shift;
	326
	327	len = min(count, sizeof(buf) - 1);
	328	if (copy_from_user(buf, user_buf, len))
	329	return -EFAULT;
	330
	331	buf[len] = '\0';
	332	if (kstrtos8(buf, 0, &shift))
	333	return -EINVAL;
	334
d4c9dbc6	335	if (shift < -1 \|\| shift >= BITS_PER_LONG)
3df3212f AS	336	return -EINVAL;
	337
	338	tlb_flushall_shift = shift;
	339	return count;
	340	}
	341
	342	static const struct file_operations fops_tlbflush = {
	343	.read = tlbflush_read_file,
	344	.write = tlbflush_write_file,
	345	.llseek = default_llseek,
	346	};
	347
9611dc7a	348	static int __init create_tlb_flushall_shift(void)
3df3212f	349	{
094ab1db PA	350	debugfs_create_file("tlb_flushall_shift", S_IRUSR \| S_IWUSR,
094ab1db PA	351	arch_debugfs_dir, NULL, &fops_tlbflush);
3df3212f AS	352	return 0;
	353	}
	354	late_initcall(create_tlb_flushall_shift);
	355	#endif